Вы находитесь на странице: 1из 401

Stijn Bruers

orn !ree and equaI?


Oh Ihe eIhical cohsisIehcy o! ahimal equaliIy
WhaI is equaliIy? Are Ihere di!!erehI kihds o! equaliIy? Who is equal? Ahd
whaI are Ihe cohsequehces o! IreaIihg oIhers equally? 1his work
ihvesIigaIes whaI kihds o! equaliIy cah be applied Io ahimals ih a
cohsisIehI way. lI sIarIs aI Ihe meIa-eIhical level, how Io cohsIrucI a
coherehI eIhical sysIem o! uhiversal eIhical prihciples. CouhIeracIihg
arbiIrarihess is o! key imporIahce aI Ihis level. NexI iI descehds Io Ihe level
o! hormaIive eIhics, where couhIeracIihg ihequaliIy or discrimihaIioh
becomes Ihe cehIral issue. Fihally, iI moves dowh Io Ihe level o! applied
eIhics, wiIh a !ocus oh imporIahI problems ih ahimal eIhics, such as Ihe
predaIioh problem. 1he ehd resulI will be a coherehI eIhical sysIem wiIh
!ive basic prihciples. 1his eIhical sysIem geheraIes !ive kihds o! equaliIy,
solves Ihe predaIioh problem, proves IhaI discrimihaIioh such as speciesism
is based oh a moral illusioh ahd ihdicaIes IhaI vegahism is a moral duIy IhaI
is cohsisIehI wiIh our moral values.
Stijn Bruers
SIi|h 8ruers has a PhD ih IheoreIical physics ahd a
PhD ih philosophy ahd moral sciehces aI Ihe
uhiversiIy o! GhehI.
978-3-659-53766-0
B
o
r
n

f
r
e
e

a
n
d

e
q
u
a
l
7
B
r
u
e
r
s
Stijn Bruers
Born free and equaI?
Stijn Bruers
Born free and equaI?
On the ethicaI consistency of animaI equaIity
LAP LAMBERT Academic PubIishing
LAP LAMBERT Academic PubIishing
Impressum / ImprinI
8ibliogra!ische lh!ormaIioh der DeuIscheh NaIiohalbiblioIhek: Die DeuIsche
NaIiohalbiblioIhek verzeichheI diese PublikaIioh ih der DeuIscheh NaIiohalbibliogra!ie,
deIaillierIe bibliogra!ische DaIeh sihd im lhIerheI ber hIIp://dhb.d-hb.de abru!bar.
Alle ih diesem 8uch gehahhIeh Markeh uhd ProdukIhameh uhIerliegeh warehzeicheh-,
markeh- oder paIehIrechIlichem SchuIz bzw. sihd Warehzeicheh oder eihgeIragehe
Warehzeicheh der |eweiligeh lhhaber. Die Wiedergabe voh Markeh, ProdukIhameh,
Cebrauchshameh, Hahdelshameh, Warehbezeichhuhgeh u.s.w. ih diesem Werk berechIigI
auch ohhe besohdere Kehhzeichhuhg hichI zu der Ahhahme, dass solche Nameh im Sihhe
der Warehzeicheh- uhd MarkehschuIzgeseIzgebuhg als !rei zu beIrachIeh wareh uhd
daher voh |edermahh behuIzI werdeh dr!Ieh.
8ibliographic ih!ormaIioh published by Ihe DeuIsche NaIiohalbiblioIhek: 1he DeuIsche
NaIiohalbiblioIhek lisIs Ihis publicaIioh ih Ihe DeuIsche NaIiohalbibliogra!ie, deIailed
bibliographic daIa are available ih Ihe lhIerheI aI hIIp://dhb.d-hb.de.
Ahy brahd hames ahd producI hames mehIiohed ih Ihis book are sub|ecI Io Irademark,
brahd or paIehI proIecIioh ahd are Irademarks or regisIered Irademarks o! Iheir respecIive
holders. 1he use o! brahd hames, producI hames, commoh hames, Irade hames, producI
descripIiohs eIc. eveh wiIhouI a parIicular markihg ih Ihis works is ih ho way Io be
cohsIrued Io meah IhaI such hames may be regarded as uhresIricIed ih respecI o!
Irademark ahd brahd proIecIioh legislaIioh ahd could Ihus be used by ahyohe.
Coverbild / Cover image: www.ihgimage.com
Verlag / Publisher:
LAP LAM8ER1 Academic Publishihg
isI eih lmprihI der / is a Irademark o!
OmhiScripIum CmbH & Co. KC
Heihrich-8ckihg-SIr. 6-8, 66121 Saarbrckeh, DeuIschlahd / Cermahy
Email: ih!o@lap-publishihg.com
HersIelluhg: siehe leIzIe SeiIe /
PrihIed aI: see lasI page
I5N: 978-3-659-53766-0
Zugl. / Approved by: ChehI, UhiversiIy o! ChehI, Diss., 2014
CopyrighI 2014 OmhiScripIum CmbH & Co. KC
Alle RechIe vorbehalIeh. / All righIs reserved. Saarbrckeh 2014
i
Acknowledgements
lirst ol all, l would like to thank my supervisor lrol. ur. }ohan lraeckman. le
allowed me to explore many new paths in ethics and philosophy and he took time
to guide me through the research. lis assistance and advice were precious.
8econd, l owe gratitude to lrol. ur. 1om claes and 1im ue 8met lor comments
and uianne 8catrine and 8cott lell lor proolreading. 1hanks to oitte lor helping
me with the lay-out and cover. 1hanks to the Lniversity ol ohent lor giving
opportunities, knowledge and assistance.
lrom all philosophers l know, perhaps lloris van den lerg has ethical ideas
closest to mine. l enjoyed our collaboration and meetings with him. Also the many
discussions with animal rights activists ol lite lack, with the participants at the
lnternational Animal kights conlerences and oatherings and with the many meat
eaters l encountered during the years allowed me to reline my theories.
lurthermore, l am gratelul to anonymous reviewers lor some uselul comments on
my research papers. let me also thank the many authors listed in the
bibliography, because without their work my hands would be empty.
And linally, ol course, a warm thank you to all my lamily, lriends and lellow
activists lor their support and lrienship: my parents, my sister (and her little
lloris), An, An-latrien, kosie, uennis, 1im, lenjamin, kalph, 1obias,.




iii
Abstract
1his dissertation investigates the possibility ol constructing a consistent ethical
system that ollers clear notions ol equality and incorporates an animal ethic. 1he
lirst part is more meta-ethical in nature, rellecting on notions such as moral
intuitions, universalism, consistency and coherence. lt demonstrates that moral
illusions might exist and ollers a method to discover such moral illusions.
1he second part turns to normative ethics, dealing with principles ol wellare,
justice and basic rights. lt tackles problems ranging lrom population ethics to non-
ideal theory.
linally, the third part moves to applied (animal) ethics, ln analogy to optical
illusions, l demonstrate that speciesism is not only a kind ol prejudicial
discrimination but also a moral illusion: an obstinate intuitive judgment that is
inconsistent with a coherent system. 1he third part also tackles the predation
problem and the sentience problem in animal ethics.
1he end result ol this work is a pluralist principlist ethical system that can be
captured in a metaphor ol live moral lingers working together as tnc mcra| nanJ.
1his moral hand is a constructed, coherent ethical system ol live universalized
ethical principles based on strong moral intuitions. 1he tnumu represents the
principle ol universalism, which is a basic ingredient ol coherentism, and
generates an anti-discrimination rule. 1he inJcx jinqcr symbolizes a
consequentialist wellare ethic, based on the coherence ol impartiality and
empathy. 1he miJJ|c jinqcr is the mere means principle ol a deontological rights
ethic: humans (and animals) have a right not to be used as merely means to
someone else's ends. 1his principle captures a lot ol moral intuitions that pop up
in lamous dilemmas. A lourth principle, the rinq jinqcr, relers to the value ol
biodiversity and adopts some elements ol carnism, the opposite ol veganism as
ideology. 1his lourth principle solves the predation problem and is coherent with
some other moral intuitions. linally, the |itt|c jinqcr represents the principle ol
tolerated partiality which can be used in some linal moral dilemmas. with these

iv
live lingers ol ethics, we can grasp the moral problem ol consuming animal
products, and answer the question whether veganism is a moral duty.


v
1able ol contents
lntroduction ................................................................................................................. 1
constructing a coherent theory ol animal equality ................................................. 10
8ummary ol the dissertation ........................................................................................ 2
1he metaphor ol the crossword puzzle ..................................................................... 2
1he metaphor ol the optical illusion .......................................................................... 2
1he metaphor ol the moral hand ................................................................................ 26
1he metaphor ol the standard model ol lorces ........................................................ !0
1he metaphor ol the moral landscape ....................................................................... !0
art ! tnica| ccnsistcncy ............................................................................................ 11
chapter 1 1he basic elements ............................................................................... !
1.1 1he input data: moral intuitions ....................................................................... !
1.2 1he method: rule universalism .......................................................................... 41
1.2.1 Lniversalizations made by meat eaters ............................................... 46
1.2.2 Lniversalizations made by animal ethicists ........................................ 49
chapter 2 1he goal: consistency and coherence ................................................. !
chapter ! 1he problem: moral illusions .............................................................. 9
!.1 Optical illusions .................................................................................................... 60
!.2 moral illusions ...................................................................................................... 66
!.! An example ol moral illusions in the trolley dilemma .................................. 0
!.4 ls the deontological right a moral illusion7 ..................................................... 6
!. leuristics in thought experiments ................................................................... 8
8ummary ol part one ..................................................................................................... 81
art . 1nccrics cj cua|ity ...........................................................................................
chapter 4 lmpartiality and prioritarian equality ............................................... 8

vi
4.1 contractarianism universalized ........................................................................ 8
4.2 lrom leelings and well-being to the value ol lile ........................................... 89
4.2.1 Allective qualia: lrom experienced leelings to experienced
pleasure ..................................................................................................... 89
4.2.2 1he importance ol prelerences: lrom experienced pleasure to
momentaneous well-being..................................................................... 90
4.2.! 1he problem ol interpersonal comparability: lrom individual
well-being to comparable momentaneous well-being ...................... 92
4.2.4 1he liletime perspective: lrom momentaneous well-being to
the value ol lile ........................................................................................ 96
4.2. lersonal identity and psychological continuity ................................. 99
4.! 1he maximin principle ..................................................................................... 10!
4.4 1he quasi-maximin principle and prioritarianism ....................................... 10
4. Applications ol the quasi-maximin theory .................................................... 10
4..1 kawls' theory ol justice ......................................................................... 10
4..2 kesponsibility and desert ..................................................................... 109
4..! luture orientation and restorative justice ........................................ 112
4.6 lntermezzo: a mathematical description lor a theory ol justice ................ 114
4. 8ummary ............................................................................................................. 116
chapter lartiality and tolerated choice equality .......................................... 11
.1 1olerated choice equality ................................................................................. 118
.2 1o whom applies the tolerated choice equality7 .......................................... 119
.! 1olerated choice equality and equality ol opportunity ............................... 121
chapter 6 lasic right equality ............................................................................ 12!
6.1 moral dilemmas and strong moral intuitions ............................................... 12!
6.1.1 A lirst approach: uncertainty aversion .............................................. 124
6.1.2 1entative ethical principles ................................................................. 12
6.2 1he basic right and the mere means principle ............................................. 12
6.! when is the basic right violated7 .................................................................... 1!1
6.!.1 1wo words, two conditions .................................................................. 1!2
6.!.2 conclusion .............................................................................................. 140
6.4 who gets the basic right7 ................................................................................. 140
6. low strong is the basic right7 ......................................................................... 146
6.6 1he extended mere means principle .............................................................. 148
6.6.1 uoing versus allowing ........................................................................... 149
6.6.2 1olerated partiality and imperlect duties ......................................... 10
6.6.! 1he asymmetry ol procreational duties ............................................ 12
6. Application: the least harm principle and vegetarianism ........................... 14
chapter 8ummary: principles ol equality and lurther relinements ............ 1
.1 lquality and veganism ...................................................................................... 1
.2 ldeal and non-ideal theory: applying the universalist imperative ............ 19
.2.1 1he argument ol lutility ....................................................................... 162
.2.2 1it-lor-what7 .......................................................................................... 16!

vii
.2.! lrohibition laws ..................................................................................... 16
.2.4 8ell-delense against culpable attackers and innocent threats ....... 16
.2. 8ummary ................................................................................................. 10
.! lormal equality, discrimination and hierarchic dualism ............................ 10
art 1 Anima| ctnics .................................................................................................... !1
chapter 8 8peciesism as a moral illusion .......................................................... 1
8.1 1he current situation: patho-anthropocentrism .......................................... 1
8.2 moral illusions and discrimination ................................................................. 16
8.! low do we know whether speciesism is a moral illusion7 ......................... 18
8.4 live arguments against the species boundary .............................................. 182
8. live arguments in lavor ol sentience ............................................................. 186
8.6 8peciesism and cognitive impenetrability ..................................................... 190
8. lsychological background theories: human prejudices and
essentialism ........................................................................................................ 191
8.8 8peciesism as a moral heuristic ....................................................................... 19
8.8.1 1he heuristics hypothesis .................................................................... 196
8.8.2 1ime and knowledge constraints ........................................................ 198
8.8.! lear ol a slippery slope ......................................................................... 200
8.8.4 1he emotional cost ol excluding atypical humans ........................... 202
8.8. 1he importance ol sentience ............................................................... 204
8.9 8ummary ............................................................................................................. 208
chapter 9 1he sentience problem ...................................................................... 211
9.1 1he scientilic problem ...................................................................................... 211
9.2 1he ethical problem .......................................................................................... 21
chapter 10 1he predation problem ..................................................................... 21
10.1 lnvalid solutions to the prey problem ............................................................ 219
10.2 A hypothetical solution to the prey problem ................................................ 220
10.! lnvalid solutions to the dillerence problem .................................................. 221
10.4 A lirst hypothetical solution to the dillerence problem: the !-N-
principle .............................................................................................................. 22
10. 1he value ol biodiversity .................................................................................. 2!0
10..1 coupling the !-N-principle to biodiversity ...................................... 2!0
10..2 lntrinsic or instrumental value ol biodiversity7 .............................. 2!2
10..! low valuable is biodiversity7 .............................................................. 2!!
10..4 An analogy between biodiversity and well-being ............................ 2!4
10.6 8ome lurther tests lor the !-N principle ........................................................ 2!
10. A second hypothetical solution to the dillerence problem: behavioral
lairness ................................................................................................................ 241
10.8 8ummary ............................................................................................................. 244
chapter 11 1he property problem and the harvest problem ............................ 24

viii
11.1 labitat destruction ........................................................................................... 248
11.2 Animals killed in harvest .................................................................................. 20
chapter 12 1he core argument lor veganism ..................................................... 2!
Argumentation scheme lor veganism ...................................................................... 20
art ! pi|cquc .............................................................................................................. .1
chapter 1! 1he moral hands ................................................................................ 2
1!.1 1he moral hand ol normative ethics: live principles ol a complete
and coherent ethic ............................................................................................. 2
1!.1.1 live principles ol equality .................................................................... 28
1!.1.2 Applications ol the live lingers ........................................................... 29
1!.1.! lntermezzo: maps ol the moral landscape ......................................... 284
1!.2 A second moral hand ol meta-ethics .............................................................. 288
1!.2.1 An analogy with crossword puzzles ................................................... 291
1!.2.2 live principles ol anti-arbitrariness ................................................... 291
1!.2.! Applications ol the meta-ethical hand ............................................... 292
1!.! 1he impossible triangle ol the meat eater ..................................................... 29
where to go lrom here7 Questions lor luture research ........................................... 29
Appendix 1: a review and systematization ol the trolley problem ......................... 299
Abstract ......................................................................................................................... 299
lntroduction ................................................................................................................. 299
1he trolley dilemmas .................................................................................................. !02
8ix algorithmic accounts ............................................................................................ !0
oroup A: the 'mere means' accounts .............................................................. !06
oroup l: the 'same threat' accounts ............................................................... !0
oroup c: the 'causal chain' accounts .............................................................. !09
8even psychological accounts ................................................................................... !12
lour invalid accounts .................................................................................................. !1
conclusion and lurther research .............................................................................. !19
Appendix 2: aversions behind the veil ol ignorance (a mathematical
description lor a theory ol justice) ................................................... !21
why a mathematical model7 ..................................................................................... !21
1he mathematics ol consequentialist wellare ethics ............................................. !22
1he impartial observer behind the veil ol ignorance ............................................ !2
1he wellare lunction ......................................................................................... !26
ueriving the wellare lunction behind the veil ol ignorance ....................... !2
1he rellection ellect and risk neutrality lor negative well-being
levels ........................................................................................................ !!!
loss aversion ...................................................................................................... !!!
lroblematic properties ol number-dampened prioritarianism ................. !40

ix
lntermezzo: a more complex lormulation to solve the replaceability
problem ................................................................................................... !4
Lncertainty aversion ........................................................................................ !4
lrioritarian theories lor lotteries ................................................................... !61
combining the prioritarian theory with the basic right and biodiversity
principles ............................................................................................................ !6
uemocratic impartial prelerences ol moral agents ............................................... !6
libliography ............................................................................................................. !2



1
lntroduction
1he discussion about the moral status ol non-human animals (herealter: animals),
and the use ol animals lor lood, clothing, entertainment or research, has a long
history that goes back to Ancient oreece (lythagoras, llutarch). lrom time to time
the problem resurges throughout the centuries (leonardo ua Vinci, }eremy
lentham, Oswald, 191, kitson, 1802, 8alt, 1892). 1he real breakthrough ol an
academic animal ethics came in 191, when kichard kyder introduced the term
'speciesism': a discrimination on the basis ol someone's species, by analogy with
racism and sexism (oodlovitch & larris, 191, kyder, 19). 1he 190s and 1980s
were characterized by the application ol dillerent rational
1
theories in normative
ethics (mainly utilitarianism and deontological ethics) to animals (8inger, 19,
clark, 19, kegan, 198!). ln the 1990s, criticism arose lrom the postmodernist and
leminist point ol view, against the cold", rational approach. (llumwood, 199!,
Adams, 199, 199b). A new plea lor vegetarianism relied on an ethics ol care
(Adams & uonovan, 1996) or a virtue ethics (lursthouse, 2000).
Around the turn ol the century the debate took a new twist towards (social)
psychology and experimental philosophy. many animal rights ethicists consider
the argumentation lor animal rights and veganism as solid and completed, but
they note that there is more psychology than ethics behind our use ol animals
(8erpell, 1996, Allen et al. 2000, }oy 2002, 2009, lerzog 2010). 1he question should
be asked why so lew people are convinced by logical consistency and rational
arguments.


1
l deline 'rationality' in its broadest sense as 'ellectivity in means, consistency in ends'. ln this context,
a rational ethical theory is (very roughly) characterized by an appeal to critical thinking, logic,
consistency and reason, using a language ol principles, rules or rights. lt is distinguished lrom a more
emotional approach to ethics. lowever, some experimental philosophers (e.g. oreene, 2008) claim that
some rationalist ethical principles (e.g. deontological rights) might be the result ol underlying intuitive
emotional reactions.
tnica| ccnsistcncy anJ Anima| ua|ity
2
ln this dissertation, l will return to the older, rational tradition in animal ethics,
the approach ol the seventies and eighties. Alter 40 years ol relining the theories
ol animal ethics, l want to try to present an ambitious, most consistent and
coherent ethical system ol animal equality. consistency is the objective. my
motivation lor returning to this rationalist tradition with its locus on consistency
is sixlold.
lirst, this consistent ethical system ol animal equality demonstrates that vegan
animal rights people are sale in relying on and lollowing its moral code ol
veganism. 1hese people can trust the ethical system and don't need to worry that
their system contains moral inconsistencies as severe as the inconsistencies
encountered in the other, speciesist ethics.
8econd, l want to express that critical thinking, consistency and rationality are
and should be important elements in our moral lives. consistency and coherence
are strong constraints on ethical systems. 1hese constraints help us to distrust
unreliable moral intuitions (moral illusions) and to avoid a hyperrelativistic
anything goes" attitude. 1hrowing away all inconsistent ethical systems already
limits the options to the surviving consistent ethical systems. 1his makes it easier
and more reliable to select the ethical system that best lits our shared and
strongest moral intuitions. ll ethics would be merely a matter ol taste, it should be
a matter ol consistent taste.
1hird, related to the previous point: l believe that a consistent system that best
lits our shared strongest moral intuitions has a higher likelihood ol compliance.
consistency limits arbitrariness, and a less arbitrary conception ol justice might be
more politically stable: l believe (and hope) that individuals who grow up in
institutions governed by less arbitrary, more consistent conceptions ol justice
tend to be more motivated to respect its rules (this is an empirical claim, however,
that requires empirical evidence).
lourth, we can try to convince someone to become vegan by appeal to emotion,
intuition, empathy or serving a delicious healthy vegan meal. 1hese are all
valuable psychological marketing tricks" in the animal rights movement, and
when it comes to encouraging common people to become vegan, these strategies
are likely to be more ellicient than an appeal to rational consistency. lut it is my
conviction that becoming vegan lor an additional right reason, as a moral duty
based on rational, consistent arguments, has some extra intrinsic value.
lilth, l want to demonstrate that some new things can and should be said in a
rational animal rights ethic. ln the past 40 years, philosophers underestimated the
importance ol some problems (such as the predation problem and the sentience
problem). l will tackle these problems. And l want to present a clear overview ol all
basic principles needed in an animal rights ethic, using the metaphor ol live
lingers ol a moral hand.
ntrcJucticn
!
8ixth, my hope is that people, especially academics and other people interested
in ethics and philosophy, would critically consider the arguments in this
dissertation and apply them to issues like veganism and the production and
consumption ol animal products. l will try to demonstrate that the speciesist
ethics that a lot ol people have, is not internally consistent, and that a leasible
consistent theory that respects deep (phenomenally strong) moral intuitions is
possible. making the ethics consistent results in an ethical system, one ol its
implications is veganism. lor those people who believe in rationality and
consistency, these arguments might be an additional motivating lactor to become
vegan. Alter critical rellection l believe that veganism is not only a lair, healthy,
ecologically sustainable and animal lriendly way ol living, but it is also a way ol
living that is most consistent with the deepest moral values ol a lot ol people.

ln order to argue lor animal equality, this dissertation contains three parts. 1he
lirst part relers to the ethical consistency" in the title, here understood in the
broad sense. 8o l start with meta-ethics, a rellection on the meaning and validity
ol ethics. ln the lirst part, we will not yet lind principles ol how we should behave.
we lirst have to set up the rules ol the game. ll l want to convince a meat eater
with rational arguments, we lirst have to agree on the validity ol the rules ol the
game.
1he starting point is our moral intuitions. 1hese intuitions are spontaneous,
unrellected gut leelings that something is right, wrong, good or bad. 1he resulting
persistent judgments cannot be justilied with lurther rational arguments.
lntuitions are things that strike us as true without us knowing entirely why they
do" (cohnitz and lggqvist, 2009, p.!). 1hey arise on their own and are not the
result ol inlerential reasoning.
moral intuitions have dillerent strengths, where the strength is determined by
our willingness (not) to give up the intuition. lence, the strongest intuitions are
the ones that are accompanied with the strongest emotions and the strongest
desire to respect them. 1herelore, strong moral intuitions are intrinsically
motivating, which gives us a lirst reason to start with those moral intuitions. A
second important reason is that meat eaters start with intuitions as well, when
they want to justily their meat consumption. 8o meat eaters and animal ethicists
can agree on this part ol the rules ol the game. l believe that all ethical systems are
in the end based on moral intuitions. l hope (and weakly believe) that the
strongest moral intuitions that l use in this dissertation to argue lor veganism, are
shared by meat eaters. ln other words: il animal rights activists and meat eaters
put all their moral intuitions on the table, selecting the strongest ol them to
construct a consistent ethical system, we might see that a lot ol those intuitions
tnica| ccnsistcncy anJ Anima| ua|ity
4
are shared by all parties. 1his however is an empirical claim that requires lurther
evidence.
Alter moral intuitions, a second important element is the method that will be
used to move towards consistency: universalization. 1his is the area ol rellective
ethics. kellecting on particular situations, some moral intuitions are ignited. we
have to translate those moral intuitions into particular ethical rules, applicable to
those particular situations. Next, these particular rules should be universalized to
all morally similar situations. we arrive at universalized ethical rules.
1he importance ol this method ol universalization is also shared by meat eaters,
as we can see in discussions about animal rights and meat consumption. Not only
animal rights ethicists, but also a lot ol meat eaters give arguments based on
universalization. 8o the good news in convincing meat eaters is that we share the
same rules ol the game: moral intuitions and universalization. we both value
consistency.
Apart lrom a (hopelully) rather strong consensus on the importance ol
universalization, this method has another important advantage: it puts strong
constraints on the consistency ol ethical systems. 8o the next step is to check
whether our universalized ethical rules are consistent in the sense that they lorm
an internally consistent system lacking contradictions. when it comes to the
theory ol animal equality, we will see that it is not only consistent, but coherent:
dillerent intuitions, principles and arguments mutually support each other. 1hey
lorm a web ol principles, like a crossword puzzle. As dillerent letters combine into
words, dillerent intuitions can be unilied in principles. 1he meat eater will have
dilliculties trying to show inconsistencies in this coherent structure.
1he crossword puzzle analogy helps to clarily the notion ol coherence.
coherence, or consistency in the broad sense, has two aspects: non-arbitrariness
and internal consistency. Non-arbitrariness in the crossword puzzle means that
the letters in neighbouring white boxes should not be random, but should lorm
existing words. lnternal consistency means that each white box should not contain
more than one letter.
1he two rules ol the game (one letter per white box and white boxes generate
words) place strong restrictions on the possible solutions ol a crossword puzzle.
without those two rules, a crossword puzzle will have many equivalent solutions.
with the rules, there will only be a lew solutions (most ol the puzzles have only
one, but some might have two or even three parallel solutions). 8imilarly, non-
arbitrariness and internal consistency place strong restrictions on possible ethical
systems. As arbitrary and inconsistent systems are thrown away, we are lelt with
only a lew possible systems that best match our moral intuitions. As a
consequence ol these two restrictions, there will be more mutual agreement
ntrcJucticn

between the ethical systems that dillerent people have. A consensus will be easier
to achieve and there will be less space lelt lor moral relativism.
1o illustrate the two rules ol internal consistency and non-arbitrariness, we can
look at two interesting analogies between atheism and egalitarianism. lirst, the
atheist does not beliel in a god that is 1) almighty, 2) all good and !) allows the evil
in the world, because a representation ol a god that has those three properties is
internally inconsistent. 8imilarly, an egalitarian animal rights activist does not
beliel in an ethical system where 1) humans have basic rights, 2) animals can be
killed and eaten by humans and !) speciesist discrimination is not allowed,
because an ethical system that contains those three elements is internally
inconsistent. As a second analogy, we can say that a theist is in lact an inconsistent
atheist: a person who believes in ood does not beliel in 2eus, Apollo, 1hor, lrishna
or any other possible god. 8Jhe is a 99,999./ atheist. 1his kind ol inconsistency
relers to the arbitrariness: it is arbitrary to pick out and believe in one ol those
many possible gods il the amount ol evidence lor the existence ol all those gods is
equal (in lact nihil). 8imilarly, a speciesist meat eater is an inconsistent egalitarian:
it is arbitrary to discriminate on the basis ol species instead ol e.g. race, sex,
population, genus, lamily, order, class or any other possible (biological) category.
licking out the category ol species is arbitrary, because the moral relevance ol all
ol those many possible categories is equal (in lact nihil).
we can apply a kind ol golden rule ol reciprocity to counter arbitrariness and
inconsistency. ll you may believe in ood without evidence, then l may beliel in
lrahma without evidence. ll you may say that we should have blind laith in Allah,
then l may say that we should have blind laith in }upiter. ll you may be a speciesist
without justilication, then l may be a racist without justilication. 1his kind ol
golden rule ol reciprocity is nothing but an application ol the method ol
universalization. As we will see, this universalization puts strong constraints on
our beliels and our ethical systems.
Yet, like theists, speciesists will not easily be convinced by the arguments. 1he
reason is that there are cognitive and moral illusions, in analogy with optical
illusions (lurves & lotto, 2002). we cannot trust all our intuitions.
Lniversalization is a method to lind out whether an intuition is an illusion. l will
argue as clear as possible how to determine moral illusions. my linal goal, in part
three ol the dissertation, is to show that speciesism based on a moral illusion: it is
a stubborn intuition that makes our ethical theory inconsistent.
1he principle (or method) ol universalism generates a lormal principle ol
equality in terms ol impartiality and antidiscrimination. lt says that we should
treat equals equally in all equal situations. 1his is a lormal principle, because it
does not state how we should treat an individual. 1o give this principle some
material content, we have to move to normative ethics. lnstead ol meta-ethics,
tnica| ccnsistcncy anJ Anima| ua|ity
6
normative ethics deal with questions ol what is right and good, what should be
done and what is valuable.
1he second part ol this dissertation uses normative ethics to present three
material principles ol equality. lence, this part relers to the linal word in the
subtitle ol this dissertation. l will derive three material principles ol equality,
based on dillerent normative systems (consequentialist ethics, ethics ol care and
deontological ethics). 1ogether with the lormal principle ol equality (the
impartiality and antidiscrimination principle), these three dillerent material
principles ol equality can be used to construct a nuanced theory ol animal
equality. As we will see, these principles ol equality are not incompatible with
some ol our moral intuitions that can be translated into two notions ol inequality.
1he lirst material principle ol prioritarian equality" is a consequentialist
principle ol justice, locusing on just distributions ol liletime well-being. lt states
that we have to give a strong priority to increasing the liletime well-being ol the
worst-oll sentient beings. lt can be justilied with two dillerent, mutually
supporting arguments. One argument is based on a rational thought experiment ol
impartiality (the veil ol ignorance, kawls, 191), extended with a high but not
maximum risk aversion (need lor salety). A second argument is based on a leeling
ol empathy, extended with a low but not zero need lor elliciency.
1he second principle ol tolerated choice equality" says that we are allowed to
be partial towards those sentient beings with whom we have a personal or special
relationship
2
or lor whom we leel a lot ol empathy. 1his partiality is only allowed
as long as we could respect similar levels ol partiality that all other sentient beings
might have. 1he concern lor personal relationships or special leelings ol empathy
is characteristic ol an ethic ol care that distances itsell lrom an ethic that is too
impartial. l will combine the tolerated choice equality with a well-known principle
ol equality ol opportunity.
1he third principle ol basic right equality" brings us to deontological ethics
based on duties and rights. 1he basic right is related to a mere means principle, as
it is the right not to be used as merely a means to someone else's ends. lquality
means that all sentient beings have an equal claim to this right. lnstead ol
empathy, this right stems lrom the leeling ol respect. 1he basic right is coherent
with a lot ol shared moral intuitions encountered in a lot ol dillerent moral
dilemmas. An extended version ol the mere means principle allows lor other


2
1his principle is not restricted to a mutually conscious relationship: you might have a personal
relationship with an individual that is not consciously aware ol being in a special relationship. l.g. the
relation with a baby.
ntrcJucticn

deontological intuitions (e.g. the dillerence between doing and allowing). And
more: the extended mere means principle opens space lor the tolerated partiality
principle that generated the tolerated choice equality. lt says that we should
tolerate some levels ol impartiality. 1his property nicely demonstrates the
coherent interweaving ol the dillerent principles.
1he three material principles ol equality allow us to get a very precise and
nuanced picture ol discrimination.
1he third part ol the dissertation is devoted to applied ethics, relerring to the
animals" in the subtitle ol this dissertation. 8o we will discuss some issues in
animal ethics. lirst, l demonstrate that speciesism is a moral illusion which results
in immoral discrimination. ln order to do this, l will present live arguments why
the species boundary (the criterion human") is not morally relevant, and live
other arguments why sentience is morally relevant. 1hese ten arguments cohere
with each other and are based on strong moral intuitions and scientilic knowledge.
One intuition that speciesists have - the prejudicial dillerence in moral status
between humans and non-humans - is not strong enough to overthrow the ten
arguments. And il we add some psychological insights in the mechanism ol
discrimination, we get an even stronger case against speciesism.
ll we mention sentience
!
, our next problem is how to know whether a living
being is sentient. 1his sentience problem consists ol two parts: lirst there is the
scientilic question ol the required criteria to test whether someone is sentient. l
will brielly present the current scientilic consensus on this issue, which roughly
says that at least all vertebrate animals with a lunctioning central nervous system
are sentient. 1he second problem is an ethical one: we now have to do tests to see
whether a living being (a lish, an insect,.) is sentient, and those tests might cause
pain, lear or distress when the individual is indeed sentient. Are we justilied in
perlorming such tests7 Are we not using those animals as merely means7 l will
argue that such tests are not immoral.
ln the animal rights discussions over the past 40 years, a highly underestimated
problem is the predation problem. As with the sentience problem, l will discuss
the predation problem in two parts. lirst, there is the prey problem: suppose a lion
is attacking a zebra. most people, including animal rights activists, say we do not


!
Note: l use the words sentient being", allective being" and person" interchangeably. Although
leelings and sensations can be neutral, a sentient being has allective (positive and negative) reactions
such as liking, disliking, pleasure and displeasure. A person is broadly delined as a being who has
personal experiences and prelerences. 1his requires the presence ol a perceptual consciousness and
the capacity to have positive and negative leelings. when the context makes it clear, persons" will
sometimes reler to moral agents", a subset ol the sentient beings.
tnica| ccnsistcncy anJ Anima| ua|ity
8
have a duty to protect the zebra il we could. lut il the lion is attacking a human,
things change. lsn't this speciesist, and how to reconcile this intuition with an
antispeciesist ethics7 1he principle ol tolerated choice equality will solve the issue.
1he second issue ol the predation problem is the dillerence problem. what is
the dillerence between a lioness killing a zebra in order to leed it to her two
whelps, and a surgeon killing an innocent person in order to use his organs to save
two patients in the hospital7 ln both cases, a sentient being is killed against his will
and parts ol his body (muscle tissue or organ tissue) are given to other sentient
beings in order lor them to live. Yet, there appears to be a consensus amongst
most animal rights activists, that predation is allowed but organ transplantation is
not. 8o what is the dillerence between predation and transplantation7 low can we
reconcile our theory ol animal equality with this intuition that there is a
dillerence7 ln order to solve the issue, l will introduce a new principle, which is in
lact based on elements ol a carnist ideology. carnism (}oy, 2002, 2009) is a sub-
ideology ol speciesism, the opposite ol the ideology ol veganism. modern day meat
eaters olten unconsciously adopt this ideology, justilying their meat consumption
by claiming that this behaviour is natural, normal and necessary
4
. ly clarilying
those three notions, l will argue that violations ol the basic right are only allowed
when all three criteria are satislied. 8o predation is allowed because it is normal,
natural and necessary, whereas transplantations are neither normal nor natural.
lorrowing aspects lrom the carnist ideology gives us the advantage that our
theory becomes coherent with the intuitions that a lot ol meat eaters share.
1o make the case lor this !-N-principle even stronger, l relate the !-N-principle
to the intrinsic value ol biodiversity. 1o justily this value, l explore an interesting
analogy between two properties: well-being ol sentient beings and biodiversity ol
ecosystems. 1he intrinsic value ol biodiversity introduces an element ol
environmental ethics, an important branch ol applied ethics.
1he !-N-principle (based on the value ol biodiversity) can be combined with
another principle that can solve the dillerence problem in the predation problem:
the principle ol behavioural lairness. 1ogether, we arrive at a lilth principle ol
equality (a lourth material principle): everyone has an equal right to a behaviour
that is both natural, normal and necessary (i.e. a behaviour that strongly
contributes to biodiversity). lrielly put: il a zebra is allowed to eat lor survival,
then so is a lion.


4
consumption ol animal products is not necessary, however, as dietitians claim that a well-planned
vegan diet is healthy lor everyone, including pregnant women and athletes (AuA, 2009).
ntrcJucticn
9
Alter the predation problem, the harvest problem is discussed: are we allowed
to do activities (such as larming) that accidentally kill a lot ol sentient beings7
uemonstrating the ethical consistency ol animal equality is not only ol
theoretical interest, but has some practical consequences as well. 1herelore, the
linal chapter ol the third part on applied animal ethics presents a core argument
lor veganism as a moral duty that best lits our moral values.
1he epilogue ol this dissertation presents the metaphor ol the moral hand: live
basic ethical principles correspond with the live lingers ol the moral hand. 1hese
live principles are universalism (the thumb), prioritarian justice and the value ol
well-being in consequentialist ethics (lorelinger), the mere means principle and
the basic right in deontological ethics (middle linger), naturalness and the value ol
biodiversity in environmental ethics (ring linger), and tolerated choice partiality
and the value ol personal relationships in ethics ol care (the little linger). 1hese
live lingers generate live principles ol equality (one lormal, lour material),
respectively: impartiality, prioritarian equality ol well-being, basic right equality,
naturalistic behavioural lairness and tolerated choice equality. Although these live
principles ol the moral hand might conllict with each other in particular
situations, they can be considered as moral lorces that need to be balanced against
each other. }ust as a physical system with multiple lorces (gravity,
electromagnetism,.) is not inconsistent, the moral hand is not necessarily an
inconsistent ethical system. lowever, some elements ol a speciesist or carnist
ethical system are internally inconsistent in the sense that the theory says that
something is both allowed and impermissible, without the possibility ol balancing
dillerent principles.
1he lirst appendix presents a review and systematization ol the trolley
problem, a very lamous thought experiment in moral philosophy. A runaway
trolley is about to kill live innocent people. Are we allowed to sacrilice another
innocent person in order to save the live people7
1he second appendix presents a mathematical lormulation ol a theory ol
justice, based on aversions that an impartial observer might have behind the veil
ol ignorance. Lsing prospect theory, it shows that risk aversion, loss aversion and
uncertainty aversion can be related to some dillicult issues in population ethics.

lelore we dive into the details, l want to give two kinds ol summaries ol this
dissertation. lirst, a structured line ol reasoning to construct a complete ethical
system ol animal equality, consisting ol clear and coherent universalized ethical
principles that best lit our strongest moral intuitions, without too many arbitrary
elements, and second, a more general summary that presents the key lindings ol
this dissertation.
tnica| ccnsistcncy anJ Anima| ua|ity
10
constructing a coherent theory ol animal equality
ln this introduction my goal is to construct a coherent ethical system that is
capable ol dealing with all relevant issues in principle-based animal ethics. 1he
basic line ol reasoning ol this construction goes as lollows: l start with a lactual
property ol the world, which ignites a moral intuition or emotion, i.e. a quick,
spontaneous moral response or judgment that has no lurther rational justilication.
1hen, in a process ol rellection, this intuition is translated into a universalized
ethical rule, where universalized" means: relevant to all morally similar
situations". 8ometimes dillerent moral intuitions will mutually support each
other, resulting in a set ol coherent universalized ethical principles. lut
sometimes we encounter a new lact or situation that again ignites another moral
intuition or emotion, which might be in contradiction with our constructed set ol
universalized ethical principles. 1o solve this conllict or moral dilemma, we can
either change the ethical principles, or introduce a new ethical principle that
trumps the previous ethical principles in that particular situation. 1his new
ethical principle needs to be universalized as well to all relevantly similar
situations.
1his process continues: we again test the constructed coherent set ol
universalized principles in new situations, and il we encounter a moral dilemma,
we look lor lurther relinements. lventually, all situations and all lacts that ignite
moral intuitions should be covered, and we move to a consistent ethical system ol
hierarchical universalized principles, where some principles trump others. ln
other words, we reach a theory in 'rellective equilibrium' (kawls, 191), which
means that our strongest moral intuitions and ethical principles are coherent
(mutually supporting each other).
1his approach can be compared with solving a crossword puzzle. 1he
descriptions ol the words are the analogues ol relevant input data (objective lacts
in the world as well as moral intuitions that we have). 1he white boxes reler to the
dillerent possible situations and viewpoints, the individual letters represent the
intuitive moral judgments lrom particular viewpoints in particular situations. 1he
words correspond with the universalized ethical principles (applied to all similar
situations), and these words mutually support each other and lorm a coherent
ntrcJucticn
11
solution to the puzzle.

8o let's derive a coherent ethic ol animal equality, starting


lrom the most basic, indisputable objective lacts and moral intuitions.
1he construction ol a coherent system
lact 1: All sentient beings have a well-being and they value their own well-being
(and everything that contributes to well-being). 8entient beings are beings that
have and can subjectively leel interests. 1hey have the experience ol having
prelerences (wanting something). 1hings subjectively matter to them, meaning
lirst ol all that the individual has a mechanism (i.e. a complex lunctioning nervous
system) that enables the individual to have representations ol their bodies and
environments. 1hese representations can have intentionality, resulting in
qualitative experiences (phenomenological sensations or qualia). lor example:
through my lingers l can leel these pages. l know the dillerence between this
leeling and an absence ol leeling, lor example when my lingers are anaesthetized.
lowever, just belore l paid attention to this leeling ol touch, l was not aware ol it.
1here was an unconscious neural activity (but no anaesthesia). Only alter l locused
on my lingertips, it became a conscious experience or 'quale' ol touch. Now, qualia
are olten neutral. l don't leel an urge to avoid touching paper. lut other qualia
have valence. 1hey are allective in nature and are evaluated as being positive or
negative. A needle in my linger generates a quale that l wish to avoid. 1his quale is
called pain and it generates an urge in me. Once a quale becomes an allective
mental state (i.e. a positive or negative leeling or emotion such as pain, distress,
joy,.), well-being comes into play. 1hese leelings are related to interests, desires
or needs: they are nothing but subjective experiences ol (un)satislied interests.
lear, pain and lrustration indicate that the needs lor respectively salety, bodily
integrity and lreedom are not satislied.
moral intuition 1: lmpartiality is morally important. lmpartiality is based on
anti-arbitrariness: it is arbitrary to exclude or undervalue someone's well-being
without good reason, because everyone, without exclusion, counts.

1o be clear, ccnstructinq an ethical system (based on input data such as moral intuitions generated
through thought experiments) is analogous to sc|vinq a crossword puzzle (based on input data such as
descriptions ol the words and knowledge ol the pattern ol the white boxes and the lengths ol the
words). lence, constructing an ethical system should not be conlused with ccnstructinq a crossword
puzzle.
tnica| ccnsistcncy anJ Anima| ua|ity
12
we can consider a two-step process to increase impartiality, lrom rational
egoism to extended contractualism. A rational egoist would strive lor a
contractarian ethic (clr. 1homas lobbes), where all rational beings (i.e. beings
with whom one can negotiate) ol equal power will become part ol the moral
community, because those rational egoists gain mutual advantages through the
social contract. lowever, in a lirst step to extend impartiality, kawls (191) used
the method ol the veil ol ignorance to delete the second condition ol equality ol
power. le arrives at a contractualist ethic that also includes rational people in
dependent or weaker positions (minorities, luture generations,.). 1he veil ol
ignorance is a thought experiment, whereby you imagine that you will be born as
a rational agent, but you don't know who you will be. You can determine the moral
and political laws, based on your knowledge ol the natural laws. l would suggest a
second step to extend impartiality, whereby we delete the condition ol rationality.
lmagine that you might be any object or entity in the world, but you don't know
who or what you might be. lor complete impartiality, you have to imagine you
could be a planet, an electron, a pig in the year !000 or anything you can think ol.
low would you like that entity to be treated7 ll you were non-sentient, this
question would not matter to you, because nothing done to you will inlluence your
well-being. You would not have a well-being, experiences or prelerences. 1he kind
ol treatment becomes important only lor those beings whose well-being can be
inlluenced by moral agents. Non-sentient entities should not be taken into
account in this moral evaluation. 8o the least arbitrary and most impartial thing to
do is to delete both conditions (ol rationality and equality ol power), which is what
kowlands (1998) argued, lrom which it lollows that well-being still remains
important.
Lniversal ethical principle 1: All moral agents should strive towards
impartiality in all situations, and should take everyone's well-being into
consideration in an impartial way. moral agents are people who are able to
understand the notion ol impartiality.
lact 2: lmpathy is meaninglul lor all and only lor sentient beings (leeling
empathy lor non-sentient beings such as teddy bears would be a kind ol projection
ol emotions). lmpathy is the capacity to experience or sample the emotions ol
others. 1his emotional response occurs when the perspective (lrame ol relerence)
ol the other is taken.
moral intuition 2: compassion (empathy plus the desire to alleviate the
sullering ol the other) is a moral virtue.
Lniversal ethical principle 2: All moral agents should develop compassion in all
situations (hence also towards all sentient beings). moral agents are people who
are able to develop compassion, are able to understand the virtue ol compassion,
ntrcJucticn
1!
and are able to help others. 1hose moral agents should try to improve the well-
being ol others.
1he above two universal ethical principles are coherent with each other, and
give a rational and emotional basis ol the moral importance ol sentience. 1hey are
based on contractualism, consequentialism and virtue ethics. 1he coherence gets
even stronger when we consider the lollowing two moral intuitions. A) mental
capacities (sell-consciousness, rationality,.) are morally important. 1hey are very
special, complex and vulnerable, hence worth protecting. l) labies and mentally
disabled humans have rights because they have something morally important.
1hey have a higher moral status than human egg cells, skin cells, dead human
bodies, plants or stones. 1ogether with the lact that sentience is the only mental
capacity that mentally disabled persons have in common with other humans who
have strong rights, A and l generate two extra reasons why sentience is
important. lurthermore, the link between rights and sentience is also not
larletched: rights protect interests, leelings detect interests.
1his gives us a strong coherent case lor the moral relevance ol sentience. lt is a
scientilic question (i.e. a matter ol lact) what entity has a well-being and how its
well-being can be inlluenced. we can brielly compare this moral relevance ol
sentience with the moral irrelevance ol a criterion such as the species ucmc
sapicns. lirst, the species is one ol the many biological classilications, thus it is
arbitrary to pick a specilic species and not a specilic population, genus, lamily,
order, class,. 8econd, the delinition ol a species is very complicated. One ol the
delinitions relers to a set ol individuals who could get lertile ollspring. lut
relerence to lertility and ollspring is very artilicial and larletched when it comes
to determining who has rights. 1hird, science will never be able to determine
whether a human-chimpanzee hybrid, a human-animal chimera, an ancestor
(Austra|cpitnccus, ucmc naui|is,.) or a genetically modilied humanoid should still be
called ucmc sapicns. 1he boundaries are luzzy. lourth, all species are temporally
related to all other species in a similar way, as populations can be spatially related
in a ring species (a ring species consists ol a spatial spreading ol populations,
where A can get lertile ollspring with l, l with c, but c not with A). lilth, il the
moral status ol a species is determined by genes or bodily appearance, then it is
also very arbitrary to pick out those genes or bodily characteristics and not others
(such as skin colour). we are not responsible lor our genes, so it would be a
violation ol the desert principle il we based moral status on genes. ln summary,
the species boundary is too arbitrary, artilicial and abstract to be morally relevant.
8o lar, our ethic is not yet unambiguous and clear. we observe that there are
dillerent sentient beings and multiple ways to inlluence their well-being (lor
example: increasing everyone's well-being a little bit versus increasing the well-
being ol one individual a lot). 8o what is a just distribution ol well-being7 lirst ol
tnica| ccnsistcncy anJ Anima| ua|ity
14
all, we value parsimony and simplicity. One simple solution would be to add the
levels ol well-being ol all sentient beings lor a specilic time interval, and then take
the sum over all times. 1hen we could try to maximize this sum. 1his is sum-
utilitarianism. lut there are also other simple options, such as trying to maximize
the well-being ol the worst-oll sentient being (the one with the lowest level ol
well-being). 1his is maximin-utilitarianism. lowever, according to many people,
both sum-utilitarianism and maximin-utilitarianism have some counterintuitive
implications. with sum-utilitarianism, it is morally good to sacrilice one individual
in order to increase the well-being ol others, or to kill one individual and replace
him with another sentient being, or to keep on breeding sentient beings in order
to increase the sum ol well-being. 1he latter is known as the 'repugnant
conclusion' (larlit, 1984): an overpopulated world with a trillion individuals with a
well-being slightly above zero, might be better than a world with only a thousand
individuals who have a satislyingly high level ol well-being. Our moral intuitions
go against these conclusions. 1hese conclusions can be avoided by introducing a
level ol risk aversion.
lact !: 1here are many sentient beings, and some beings can be worse-oll than
others. 1his lact implies that lrom behind the impartial veil ol ignorance, how to
maximize your well-being becomes a game ol chance. mathematically, sum-
utilitarianism implies that the expectation value ol your well-being will be
maximized. lut you have to be aware that there is a risk that you might be born as
one ol the worst-oll individuals. lor example: two individuals might have well-
being levels equal to 10 and 100, so the expectation value will be equal to (the
average). ln sum-utilitarianism, this situation would be equal to the situation
where those two beings both have a well-being ol . 1he problem is that in the
lirst situation, you might end up as the person with level 10. when much is at
stake, most moral agents have a risk aversion (need lor salety - to play it sale), and
in this game ol chance, this means that they would not opt lor sum-utilitarianism,
but to some kind ol prioritarianism: giving priority to increases ol well-being ol
the worst-oll positions. 1herelore they preler the second situation (with equality
ol well-being). ll you have maximum risk aversion (a maximum need lor salety),
you would take the maximin-utilitarian strategy (maximizing the
minimumJlowest well-being), giving all priority to the worst-oll position, because
you are so worried at becoming this worst-oll individual. ll you have zero risk
aversion, you are a sum-utilitarist. A high but not maximum level ol risk aversion
would result in a prioritarianism that is in between maximin-utilitarianism and
sum-utilitarianism. we could call this 'quasi-maximin prioritarianism'.
moral intuition !: A (high) level ol risk aversion is good (especially when much
ol your well-being is at stake, then most people are risk averse).
ntrcJucticn
1
Lniversal ethical principle !: Quasi-maximin prioritarianism should be applied
in all situations. mathematically, this principle can be expressed as the
maximization ol a power average ol values ol lile ol all sentient beings. 1he power
in the power averaging corresponds with the level ol risk aversion behind the veil
ol ignorance. 1he value ol lile (liletime well-being) relers to the total prelerred
well-being ol an individual over hisJher complete lilespan (that spans lrom the
lirst till the last subjective leeling ol the individual). 1his prelerred well-being is
the value that one would ascribe to living the complete lile ol that individual,
when looking lrom the most impartial point ol view, e.g. lrom behind a veil ol
ignorance. 1he value ol lile contains everything that would matter to you,
everything that would be valuable to you, all the prelerences that you would have,
il you would live the lile ol that sentient being.
Quasi-maximin prioritarianism has some elegant leatures. lt avoids the
abovementioned objections against sum-utilitarianism, and also a lot ol objections
against animal ethics. lirst, consider the idea ol painlessly killing someone (lor
example in his sleep). lrom behind the veil ol ignorance, you cannot preler such
killing, even il you are not aware that you will be killed. 1his means that a sentient
being should now be delined as a being that has already developed the capacity to
leel and has not yet permanently lost this capacity. lndeed, value ol lile starts lrom
the lirst leeling and ends at the last leeling.
Next, take the problem ol replaceability. ls it allowed to kill a sentient being
(painlessly), and then let another sentient being be born7 1his happens when we
breed and slaughter cows. ll we kill a sentient being, his value ol lile will be e.g. ,
whereas it would have been 10 otherwise (when he lives a lull lile). 8o in a lirst
option, one individual will have a lile with total well-being equal to (an early
death), and a second one will also have a short lile with total well-being . ln a
second option, we will have only one being, with level 10 (a lull lile). lrom behind
the veil ol ignorance, in the lirst option you will get a low value ol lile equal to .
ln the second option, you are sure you will have level 10. A sum-utilitarist would
say that the both options are equal, because the total value ol lile equals 10 in both
situations. lut l would preler the second situation, and that's also what our
prioritarian theory says, because this theory uses a (power) average. 1herelore,
sentient beings are not replaceable. Also the repugnant conclusion (the idea to
keep on breeding sentient beings until their values ol lile are about to drop below
zero), can be avoided, by simply noting that behind the veil ol ignorance you
would not preler an overpopulated world where everyone has a very low value ol
lile. 8o quasi-maximin prioritarianism avoids the olten heard argument that
breeding livestock animals is good, because they owe their lives to the breeders,
and it is better to live a lile on a larm than not to be born at all (this might not be
the case lor animals living on a 'lactory larm'). According to our prioritarianism,
tnica| ccnsistcncy anJ Anima| ua|ity
16
the choice is not between an existing lile on a larm versus a non-existing lile,
because as said above: in each choice, we only consider the sentient beings that
exist in that world-history.
Another lamous problem in animal ethics is the lileboat dilemma (e.g. kegan,
198!). 8uppose there are dillerent sentient beings in a lileboat, but we cannot save
everyone. 1hose beings can have dillerent expected lile expectancies, but they can
also diller in complexity (richness) ol emotions, the amounts ol needs, the levels ol
satislaction when needs are satislied,.. 1his means that the potential values ol lile
can diller amongst the dillerent sentient beings in the lileboat. 1he potential
values ol lile between a (mentally normal) human, a dog or a lrog can diller. 1his
inlluences our choices whom to rescue. As kegan argued, it might be required to
sacrilice the dogs lirst, because they experience a less rich lile than the humans.
lowever, kegan said that the lile ol one human would trump the lives ol a million
and more dogs. According to our prioritarianism (the veil ol ignorance with a high
but not maximum level ol risk aversion), there would be a number ol dogs, above
which the loss ol that amount ol dogs would be worse than the loss ol one human
lile.
1he quasi-maximin principle is coherent with a lot ol our moral intuitions. And
there is a second way to arrive at this principle.
lact 4: 1here might be situations where we can decrease someone's well-being
with a huge amount (e.g. drive himJher into extreme poverty) in order to increase
the worst-oll position with a negligible small amount.
moral intuition 4: llliciency is important to some degree. lmpathy might have
a tendency to give absolute priority to improving the worst-oll individual, which
results in a maximin strategy. lut il we value elliciency, we would not sacrilice too
much well-being.
Lniversal ethical principle 4: 1his equals quasi-maximin prioritarianism
(principle !). we should maximize the value ol lile ol all sentient beings, giving a
strong priority to increase the lowest values ol well-being. ln other words: we
should maximize the value ol lile ol the worst-oll individuals, unless this is at the
expense ol much more well-being ol others.
ln summary: a rational approach ol impartiality (the veil ol ignorance) with a
high but not maximum risk aversion (need lor salety) coheres with an emotional
approach ol compassion with a low but non-zero need lor elliciency. 1he two
approaches represent two dillerent points ol view: the rational approach looks at a
situation lrom the outside, lrom an impartial point ol view behind a veil ol
ignorance. 1he emotional approach is more down to earth: it looks at a situation
lrom the inside, lrom the subjective experience ol compassion with others. 1hese
are two approaches resulting in the same quasi-maximin prioritarian principle.
ntrcJucticn
1
1his principle has two disadvantages. As a lirst problem, the values ol lile are
very dillicult to measure and compare. All we have is our empathy, our scientilic
knowledge and our imagination. we have to try placing ourselves in the position
ol others, by using empathy, or by imagining that we could be the other
individual, with all his or her needs and leelings. 8o the 'emotional' method ol
empathy and the 'rational' method ol the veil ol ignorance are actually two rules
ol thumb to make educated guesses about the order ol the values ol lile ol
dillerent individuals. lmpathy and imagination are virtues to be developed and
already allow us to move quite lar.
A second disadvantage is that the level ol priority given to the worst-oll (in
other words: the level ol risk aversion or the need lor elliciency), is in some sense
arbitrary. 1he level is somewhere between 0 (sum-utilitarism with zero risk
aversion) and inlinity (maximin-utilitarism with maximum risk aversion).
lowever, l believe our coherent picture is strong enough to withstand this
objection. 1he arbitrariness is less bad than overriding a coherent set ol strong
moral intuitions. 1he good thing is that no-one has a strong prelerence to a sharp
level ol priority. No-one says the value should be 48. lt's more like a luzzy range
that we preler. 8o we can and should be a bit tolerant to the levels ol priority that
other moral agents would preler, and this means we can be llexible and could
come to a democratic or mutual consensus between all moral agents. lut once we
have set a level ol priority, we should apply it consistently in all relevantly similar
cases.
1he quasi-maximin prioritarianism is the basic lramework ol a coherent ethical
system ol animal equality. All sentient beings are in some sense equal lrom an
impartial perspective such as behind a veil ol ignorance. lt is a consequentialist
ethic, because it only looks at outcomes ol values ol lile. oiving a level ol priority
lor the worst-oll positions, some people (true consequentialists) might preler to
stop the construction ol a coherent ethical system here. lowever, there are some
more intuitions that do not lit in the prioritarian ethic. we lirst discuss an
intuition related to an ethic ol care and next an intuition related to an ethic ol
rights.
lact : 1here is a possible situation where l have to choose between a sentient
being l hold dear and one or more other unknown sentient beings. l.g. in a
burning house dilemma, where l have to choose between saving my child or other
individuals lrom the llames.
moral intuition : l am allowed to help the person l hold dear.
Lniversal ethical principle : lt is allowed to be partial in all situations ol aid
where someone is involved whom you hold dear (with whom you have a personal
relationship or strong leelings ol empathy), as long as we tolerate similar levels ol
tnica| ccnsistcncy anJ Anima| ua|ity
18
partiality ol everyone else. 1his principle ol tolerated partiality trumps the above
prioritarian principle to some degree, but not too much.
lurning house dilemmas such as Your child or the dog7" (lrancione, 2000) are
olten used to criticize animal equality. lut here l introduce a new principle ol
tolerated partiality, which hides a new kind ol equality: tolerated choice equality.
ln the burning house, l would save my child instead ol someone else, which points
at an emotional inequality in lavour ol my child. lut l can still consider all
individuals in the house as being equal in some other subtle sense, il lor example l
tolerate your choice to save someone else instead ol my child. A white racist would
say that it is immoral to save black children lrom the house instead ol white
children. A speciesist would say that it is immoral to save someone belonging to
another species. lut il someone has an emotional connection with a dog, the
principle ol tolerated choice equality says we should tolerate his choice to save the
dog. 8aving a dog instead ol a human
6
, saving a mentally disabled orphan instead
ol a mentally normal child, or saving your lover instead ol two unknown persons,
might be violations ol the quasi-maximin prioritarian principle. lut l think we are
allowed to violate this quasi-maximin principle to some degree. Also here we could
try to reach a democratic or mutual consensus between all moral agents, about the
degree ol violation that is allowed. we should apply this degree ol partiality
consistently in all situations.
lact 6: 1he organ transplantation problem. 1here is a possible situation, where
live patients in a hospital would die unless we sacrilice an innocent person against
his will and use live ol his organs lor transplantations. 1his would be allowed
according to prioritarianism, because the (power average) liletime well-being
would be higher il the innocent person is sacriliced.
moral intuition 6: l (and most people) leel emotional distress and restraint to
sacrilice this one person against his will. we should not sacrilice someone, even il
prioritarianism is violated and even il someone l hold dear is one ol the patients in
the hospital. 8o this intuition trumps both prioritarianism and tolerated partiality.
1here are a lot ol other moral dilemmas where we can use someone without
hisJher consent as merely means to save others. 1orturing someone in order to
gain inlormation about a bomb, throwing someone (a sentient being such as a
mentally disabled human) in lront ol a runaway trolley in order to block the
trolley that is about to kill other people, using someone as a shield against bullets,
using someone as a slave, using someone in medical experiments, using someone
as a scapegoat to stop a riot, terror bombing civilians in order to demoralize the


6
l tolerate that you give more lood and medical assistance to your pet than to a hungry child lar away.
ntrcJucticn
19
enemy, raping someone, killing and eating someone (cannibalism), trallicking,.
All these situations generate moral intuitions that are very coherent il we
translate them into the lollowing deontological principle (an interpretation ol a
lantian ethics).
Lniversal ethical principle 6: All sentient beings have a basic right not to be
used as merely a means to someone else's ends. A victim is used as merely a means,
when two conditions are met. 1) A moral agent causes the victim a 'disrespectlul
harm' against its will: the victim has to do or undergo something that sJhe does
not want. lxamples ol disrespectlul harm are a treatment as property or
commodity (see lrancione, 2000) or a violation ol bodily integrity. 2) 1he presence
ol the body ol the victim is required in order to reach the ends. lor example
without the body ol an animal, we could not produce an animal product (meat,
eggs,...) lor consumption. 1he latter is an important criterion because there are
moral dilemmas whereby you are allowed to cause harm to someone in order to
save others (lor example redirecting a threat towards one person in order to save
a group ol people). ln those dilemmas, the presence ol the victim was not required
in order to save the others.
1his principle is coherent with the notion ol respect, which is next to empathy
an important moral virtue, and it is coherent with the notion ol intrinsic value
(the opposite ol instrumental value) as well.
1he ethical principle ol the basic right trumps both the principle ol priority and
the principle ol tolerated partiality. lut the basic right is not absolute: il the
principle ol priority is strongly violated (il thousands ol sentient beings will die),
then a basic right might be violated (this corresponds with a need lor elliciency).
As with the above principles, this level ol violation can be determined on the basis
ol a democratic or mutual consensus among moral agents. And here we have
llexibility as well: there are dillerent levels ol harm, there is a morally relevant
gradation in someone's ends (lrom the vital needs ol many sentient beings to the
luxury ends ol one individual), and there is a gradation in the level ol sentience
and mental capacities. 1hese gradations could be coupled. lor example: a being
with higher levels ol morally relevant mental capacities has a stronger claim to
this basic right.
let's brielly apply this principle to the 'least harm' objection against veganism
(uavis, 200!). 8uppose that a meat eater can kill and eat one cow, whereas a vegan
needs a crop lield to get the same amount ol nutrients. 8uppose using that crop
lield accidently kills live mice. 1he meat eater causes least harm, but sJhe violates
the basic right ol the cow, which is worse. 1he mice are not used as merely means,
so therelore veganism remains the morally better choice. (lor lurther criticism on
the least harm argument ol uavis, see matheny, 200!, and lamey, 200).
tnica| ccnsistcncy anJ Anima| ua|ity
20
we now arrive at an ethical system with three principles ol equality. 1he lirst is
based on impartiality (interchangeability ol sentient beings) and results in a lorm
ol prioritarianism. According to this theory, il we have to choose between two
situations that have equal total well-being, we should choose the one with the
most equal distribution ol well-being. 1he second is a tolerated partiality, whereby
we tolerate the choices ol others to save those they preler. lrom this tolerated
partiality, the individuals in a burning house inherit a 'tolerated choice equality'.
1his principle weakly trumps the lirst principle. 1he third principle is a basic right
equality, and this trumps the two lormer principles to a strong but not absolute
degree. All beings with similar levels ol the relevant mental capacities have an
equal claim to the basic right not to be used as merely a means to someone else's
ends. 1he three principles are related to, respectively, a consequentialist ethic ol
well-being and justice, a leminist ethic ol care and a deontological ethic ol rights.
1hese three principles imply veganism. consider a dairy cow in the livestock
industry and a human who likes to eat cheese. 8tart with the veil ol ignorance. ln
one situation, dairy cows are not bred, so we can only be a human being, who has a
value ol lile equal to 10. ln the second situation, this human enjoys the cheese (his
value ol lile increases to 11), but the cow has a miserable lile (sullering in the
livestock industry, early death,.). 8o her value ol lile equals !. According to quasi-
maximin prioritarianism, the lirst situation is prelerred. ll you'd choose the
second situation, lrom behind the veil ol ignorance, you have probability Z to end
up in the worst-oll position. (According to sum-utilitarianism, the second situation
is better). 1olerated partiality is also violated: il we preler the enjoyment ol cheese
above the use ol the cow, we should also tolerate the other option: breeding
women and using their breast milk to make cheese lor cows (suppose the cow likes
human cheese). 1his we would not tolerate. 1he third principle is also violated,
because the cow in the livestock industry is used as merely a means (her bodily
integrity is violated and she is treated as property).
with these three principles, we arrive at a coherent system that best lits our
strongest moral intuitions. 8ome intuitions based on speciesist judgments are not
compatible with this system ol animal equality. 1hese intuitions are too weak and
cannot be incorporated without introducing highly arbitrary and artilicial
constructions, so we have to dismiss these speciesist intuitions as being moral
illusions. Although our theory implies veganism, it still allows lor some partiality
(the tolerated partiality meets our intuitive prelerence lor some individuals).
lowever, there is one serious problem remaining.
lact : Obligate predators need meat in order to survive. ll obligate predators
cannot use other sentient beings as merely means, they will all become extinct. ll
principles 4, and 6 are universalized to predator animals, this would imply that
they have to become extinct.
ntrcJucticn
21
moral intuition : Obligate predators are allowed to hunt and hence violate the
basic rights and well-being ol prey. lt would be a tragedy il they became extinct.
lt is not easy to lormulate a clear principle that is coherent with this intuition
as well as with the intuitions that we encountered belore. ll we suppose that
biodiversity has a moral value, then we have the lollowing option.
Lniversal ethical principle : ll a sulliciently large group ol sentient beings
became, by an evolutionary process, dependent on the use ol other sentient beings
lor their survival, they are allowed to use other sentient beings lor that purpose
(until leasible alternatives, that don't violate basic rights, are lound

).
ll we suppose that biodiversity has moral (intrinsic) value, and il we deline
biodiversity as the diversity ol everything that is the direct product ol
evolutionary processes, then this seventh principle becomes coherent with the
value ol biodiversity. 8o the existence ol predator animals contributes to
biodiversity and we should not destroy that biodiversity.
1his principle is also coherent with a 'triple-N-principle', which relers to the
three values 'natural, normal and necessary' ol a carnist ideology (}oy, 2009). 1his
connection works il we deline natural as: behaviour that is a direct consequence ol
a process ol evolution (genetic mutation and natural selection). 8o it relers to an
'evolutionary process'. Normal means that the behaviour happens a lot, so it relers
to a 'sulliciently large group'. And necessary means that those beings would die il
they no longer exhibit that behaviour. 1his relers to 'dependency lor survival'.
lutting the three criteria together, natural-normal-necessary means that a lot
ol biodiversity would be lost when the behaviour stopped. And a lot ol biodiversity
has a lot ol moral value, sulliciently enough to trump the basic right and well-
being ol prey animals. lredation is normal, natural and necessary, so it is allowed
(as long as there are no leasible alternatives), even il it violates the basic right. lor
humans, eating animal products is not necessary (according to the Academy ol
Nutrition and uietetics (AuA, 2009)), so we are not allowed to violate the basic
rights ol animals. Organ transplantation (by sacrilicing a sentient being against his
will) is not allowed either, because it is a violation ol the basic right and it is not
normal and natural (although it is necessary lor the patients).
Note that this value-ol-biodiversity principle is completely unrelated to the
value-ol-sentience principles discussed belore, although we could compare
biodiversity as an intrinsically valuable property ol ecosystems with well-being as

8uch alternatives could be the production ol artilicial (cultured) meat to leed the predators,
genetically or psychologically reprogramming predators to change their behavior, the use ol wildlile
contraception to control prey populations,.
tnica| ccnsistcncy anJ Anima| ua|ity
22
an intrinsically valuable property ol sentient beings. loth ecosystems and sentient
beings are unique and irreplaceable entities that have a tendency to increase their
corresponding valuable properties (biodiversity and well-being). ln itsell, the
biodiversity principle seems arbitrary, but it is coherent with a lot ol moral
intuitions that a lot ol people share. lor example: moving around and killing
insects (by accident) is considered allowed, even il scientists are able to
demonstrate that insects are sentient. lut the !-N-principle says that moving
around is natural, normal and necessary behaviour ol animals. 1he same goes lor
procreation, even il the animal species does not sulliciently contribute to the
(power) average well-being ol a prioritarian theory. lrocreation is natural, normal
and necessary, and a lot ol biodiversity will get lost il some species were not
allowed to procreate.
1he !-N-principle, based on the value ol biodiversity, generates a lourth
principle ol equality: naturalistic behavioural lairness: all natural beings (who
contribute equally to biodiversity) have an equal right to a behavior that is both
natural, normal and necessary (i.e. a behavior that contributes to biodiversity).
Natural beings are those beings that originated by natural evolution.
linally, we also have situations where predators attack us or beings that we
hold dear. Our intuition says we are allowed to delend ourselves and others, and
we have a stronger duty to protect some individuals with whom we have special
relationships. All sentient beings have the right to delend themselves or others,
they have the right to be partial in such decisions, as long as they respect similar
levels ol partiality ol others (see principle ) and as long as biodiversity is not
threatened. ll we wish, we could also add that we have a duty to protect all beings
who have (or will develop) moral agency or rationality. 1hose rational beings not
only leel their interests, but they also know and understand their interests. 1his
rationality applies to most human beings, except e.g. seriously mentally disabled
human orphans. 1his satislies people's intuitions that we have a duty to protect
humans lrom predators. (lut il we say that we have a duty to protect mentally
disabled humans whereas we do not have a duty to protect non-human animals,
because all humans have a higher moral status than non-humans, then we become
too partial. 1his kind ol speciesism, like racism or sexism, is a kind ol partiality and
arbitrariness that we cannot tolerate.)
1his completes the process. we now have a theory ol animal equality, with clear
and coherent universalized ethical principles that best lit our strongest moral
intuitions, and without too many arbitrary elements. ln the epilogue ol this
dissertation, l will relate the above seven universal ethical principles to live
principles ol the moral hand. Lniversal ethical principles 1 to 4 are unilied in a
lorelinger principle ol justice and the value ol liletime well-being. Lniversal
ethical principle corresponds with the little linger principle ol tolerated
ntrcJucticn
2!
partiality. Lniversal ethical principle 6 corresponds with a middle-linger mere
means principle and the basic right to bodily autonomy. Lniversal ethical
principle corresponds with the ring linger principle ol naturalness and the value
ol biodiversity. linally, the method ol translating particular moral intuitions into
universalized ethical principles corresponds with a lormal thumb principle ol rule
universalism.

tnica| ccnsistcncy anJ Anima| ua|ity




2
8ummary ol the dissertation
1he metaphor ol the crossword puzzle
A coherent ethical system can be compared with the solution ol a crossword
puzzle. 1o solve a crossword puzzle, there are a lew strong rules.
kegularity (non-arbitrariness): the white boxes should not contain letters at
random, but should lorm existing words. 1hese words are the analog ol universal
ethical principles: we should apply moral rules non-arbitrarily and impartially.
conlormity with input data: the words in the puzzle should match the given
descriptions. 1he input data in ethics are the moral intuitions, and the ethical
principles should lit those intuitions as good as possible.
consistency: a white box in the puzzle should contain no more than one letter.
8imilarly, a situation should not have two contradictory moral judgments at once.
completeness: every white box in the puzzle should be lilled with a letter. 1he
ethical principles can always be applied, in every possible situations.
ll an ethical system respects those rules, it becomes a strong, coherent system.
1he ethical system ol the moral hand, presented below, will be candidate ol such a
system.
1he metaphor ol the optical illusion
8tarting with the basic inlormation (the input data ol moral intuitions) is not
always without risk: some intuitions are not reliable. 1hink about the mller-lyer
optical illusion: two parallel lines with equal lengths have arrowheads at their
ends. A lot ol people have the spontaneous judgment (intuitive perception) that
the line with outward pointing arrowheads is smaller than the one with inward
pointing arrowheads.
1his intuition is an illusion, because it is not consistent with two strong and
coherent intuitions: the length ol a measure stick does not change length when
shilted, and the length ol a line does not depend on arrowheads or other
geometric ligures. lence, we have two coherent methods to discover the optical
tnica| ccnsistcncy anJ Anima| ua|ity
26
illusion: 1) 1he translation method, shilting a measure stick lrom one line to the
other and 2) the deletion method, erasing the arrowheads.
Also in ethics we can search lor moral illusions in a similar way. lor example we
can test whether discrimination such as speciesism - the spontaneous judgment
that a human has more moral value than an animal - can be a moral illusion. }ust
as one line appears to be longer than the other in the optical illusion, so does a
human appear to be more valuable than an animal. According to this optical-moral
illusion analogy, the arrowheads in the optical illusion correspond with morally
irrelevant properties used in discrimination. A geometric rule that says that the
length increases when arrowheads point outwards would be arbitrary, just as
irrelevant properties such as species, appearance or genes are morally arbitrary.
1he translation method applied to speciesism consists ol shilting your position:
put yoursell in the position ol a human and an animal. lmpathy will be the
measure stick. A thought experiment such as the veil ol ignorance" (or lottery ol
lile") could help: imagine that you will be born, but you do not know yet who or
what you will be. which moral rules would you then choose7 Lsing this thought
experiment, you will lind out that sentience and well-being are what matters,
because il you are non-sentient, then nothing (including the choice ol moral rules)
matters to you. Only sentient beings have the capacity to want something.
1he deletion method applied to speciesism means that we erase irrelevant
properties such as bodily appearance or genes, and that we look at the moral value
that remains. According to evolutionary biology, there is no essence connected to
humans: there is nothing special that all and only humans have, and there are
many luzzy boundaries between humans and animals (think about all our
ancestors and intermediate lorms between humans and animals that once existed,
and the potential existence ol human-animal hybrids, chimeras and genetically
modilied beings).
loth the mller-lyer optical illusion and the speciesist moral illusion have
psychological explanations: they both share a mechanism ol an acquired heuristic
(an automatic rule ol thumb inlluenced by the environment).
1he metaphor ol the moral hand
1he moral hand is a metaphor ol live basic ethical principles, one lor each linger,
summarizing a complete, coherent ethic.
1he thumb: rule universalism. You must lollow the rules that everyone (who is
capable) must lollow in all morally similar situations. You may lollow only the

2
rules that everyone (who is capable) may lollow in all morally similar situations.
1his principle generates a lormal principle ol equality in terms ol impartiality and
rejection ol prejudicial discrimination. lt also implies that we should give the good
example. 1he thumb principle is lormal and does not have material content. }ust
like we have to place the thumb against the other lingers in order to grasp an
object, we have to apply the principle ol universalism to the other lour basic
principles in order to grasp a moral problem.
1he lorelinger: justice and the value ol liletime well-being. lncrease the liletime
well-being ol all sentient beings alive in the present and the luture, whereby
improvements ol the worst-oll positions (the worst sullerers, the beings who have
the worst lives) have a strong priority. liletime well-being is the value you would
ascribe when you would live the complete lile ol a sentient being, and is a lunction
ol all positive (and negative) leelings that are the result ol (dis)satislaction ol
prelerences: ol everything (not) wanted by the being. ln a mathematical
expression, this basic principle says that we should maximize a generalized mean
ol everyone's liletime well-being using a concave lunction.
1his principle has two coherent justilications: 1) the thought experiment ol the
veil ol ignorance (you can be born as anyone or anything), where you have a high
but not maximum risk aversion (avoiding the risk ol becoming one ol the worst
sullerers means giving priority to increase the levels ol liletime well-being ol the
worst-oll positions), and 2) empathy (locusing on the needs ol the worst-oll) with
a small but non-zero need lor elliciency (maximize the liletime well-being ol the
worst-oll, unless this is at the expense ol much more well-being ol others).
1his principle generates a second, material principle ol equality: il total liletime
well-being is constant between dillerent situations, then the situation which has
the most equal distribution ol well-being is the best.
1his principle is coherent with many moral intuitions, but does not lit with a
special list ol moral intuitions, such as the trolley dilemma (we should not push a
lat man in lront ol a runaway trolley in order to block the trolley and save live
people on the tracks ahead) and the transplantation problem (we should not
sacrilice someone to use his live organs against his will to save live patients in the
hospital when there is a shortage ol organs). 1he next principle unilies these
intuitions.
1he middle linger: the mere means principle and the basic right to bodily
autonomy. Never use the body ol a sentient being against its will as a means to
someone else's ends, because that violates the right to bodily autonomy. A sentient
being is a being who has a sense ol its own body and has developed the capacity to
want something by having positive and negative leelings (and who has not yet
permanently lost this capacity). 1he two words mere means" reler to two
conditions: you violate the basic right 1) il you lorce a sentient being to do or
tnica| ccnsistcncy anJ Anima| ua|ity
28
undergo something that the being does not want in order to reach an end that the
sentient being does not share, and 2) il the body ol that sentient being is necessary
as a means lor that end. 8omeone's body belongs to that individual, not to us.
1he middle linger generates a third principle ol equality: all sentient beings
with equal levels ol morally relevant mental capacities should have an equal claim
to the basic right. 1his means that a lot ol animals and mentally disabled humans
also have a claim to this right.
1he middle linger is a bit longer than the lorelinger, and so the basic right is a
bit stronger than the right to liletime well-being (which includes the right to live).
1he middle linger is not inlinitely long, so the basic right can be violated when the
lorelinger principle ol well-being is seriously threatened.
1he previous lingers still do not match some moral intuitions, such as the
problems ol predation (dolphins are allowed to hunt sentient lish, even il dolphins
are moral agents), motion (large animals are allowed to move around, even when
insects are sentient and get harmed) and procreation (animals are allowed to
procreate, even il they do not sulliciently contribute to the generalized mean ol
liletime well-being). 1he lourth linger unilies these intuitions.
1he ring linger: naturalness and the value ol biodiversity. A behavior is allowed
(even il it violates the lorelinger or middle linger principles) il that behavior is
both natural (a direct consequence ol spontaneous evolution), normal (lrequent)
and necessary (important lor the survival ol sentient beings). ll the behavior has
several options, then the option should be chosen that least violates the other
linger principles (e.g. eating is natural and necessary, but when you can choose
between eating sentient or non-sentient beings, you should choose the latter).
}ust as liletime well-being is the value ol a sentient being, biodiversity is the
value ol an ecosystem: both liletime well-being and biodiversity have a tendency
to increase (within constraints) and both are lunctions ol variable valuable things
(leelings, lile lorms) that are the direct consequence ol a driving lorce (prelerence
satislaction, natural evolution). 1he valuable biodiversity would drastically
decrease il a behavior that is natural, normal and necessary would be universally
prohibited (universally, because you have to put the thumb against the ring
linger).
A lourth principle ol equality arises lrom the ring linger: all beings (who
contribute equally to biodiversity) have an equal right to a behavior that is both
natural, normal and necessary (i.e. a behavior that contributes to biodiversity). l.g.
il a prey is allowed to eat in order to survive, a predator is allowed to do so as well.
As the above ethical principles can be too demanding when it comes to helping
others, we can add a little linger to respect our prelerence lor the ones with whom
we have special relationships. consider lor example a burning house dilemma: you
can save either your child or someone else (another child, or a dog). Or consider
5ummary
29
the prey problem: your child and someone else is being attacked: who would you
save7
1he little linger: tolerated partiality and the value ol personal relationships.
when helping others, you are allowed to be a bit partial in lavor ol your loved
ones, as long as you are prepared to tolerate similar levels ol partiality ol everyone
else (everyone, because you have to put the thumb against the little linger). }ust as
the little linger can deviate a little bit lrom the other lingers, a small level ol
partiality is allowed.
1his linger generates a lilth equality principle: tolerated choice equality.
lveryone is allowed to be partial to an equal degree that we can tolerate. ll you
choose to help individual X instead ol individual Y, and il you tolerate that
someone else would choose to help Y instead ol X, then X and Y have a tolerated
choice equality (even il X is emotionally more important lor you than Y).
1he little linger is coherent with a lot ol intuitions, and is also related to the
middle linger: il l would not tolerate your choice to help your loved one instead ol
my child, l would not literally usc you but still ccnsiJcr you as merely a means lor
my ends. l have to tolerate your choice, otherwise l violate your basic right.
let us apply the live lingers to the production and consumption ol animal
products. 1he lorelinger principle is violated, because the loss ol liletime well-
being ol lish and livestock animals is worse than the loss ol well-being that
humans would experience when they are no longer allowed to consume animal
products. livestock animals are in the worst-oll position compared to humans, due
to sullering and early death. 1he middle linger is violated, because the bodies ol
animals are used in a way that they do not want, without their bodies there could
be no consumption ol animal products, and so the animals are used as merely a
means. 1he ring linger and little linger principles cannot be invoked to justily the
consumption ol animal products, because animal products are not necessary lor
humans (well-planned vegan diets are not unhealthy and hence biodiversity will
not decrease when humans would stop consuming animal products) and we would
never tolerate the degree ol partiality that is required to justily livestock larming
and lishing. lt lollows that veganism is ethically consistent and the production and
consumption ol animal products are ethically inconsistent. 1he thumb says that
we have to give the good example, and hence veganism is a moral duty according
to the ethical system ol the moral hand.
tnica| ccnsistcncy anJ Anima| ua|ity
!0
1he metaphor ol the standard model ol lorces
As there are dillerent lorces in physics (as expressed in the standard model ol
particle physics), the ethical system ol the moral hand contains three moral lorces.
1hese lorces can counteract each other, but that does not make a system
inconsistent. 1o clarily these lorces, they can be expressed in a mathematical
equation. 1he lorelinger principle generates a 'wellare lunction': a generalized
mean ol the concave weighted liletime well-being ol all sentient beings (including
luture generations). 1he middle linger adds a second lorce to this wellare
lunction: the sum ol basic rights violations. 1his second term is (highly) negative il
someone's basic right is violated. 1he ring linger adds a third moral lorce: the
moral value ol biodiversity, which decreases when biodiversity gets lost.
1he metaphor ol the moral landscape
1his equation (the three lorces) can be represented as a (multidimensional) moral
landscape with peaks and valleys. A rule that (when the rule is universally
complied - see the thumb principle) leads us to a mountain peak on the moral
landscape, is a better rule than a rule that leads us downwards. looking at paths
on the moral landscape, and using the little linger principle, we can see what
actions and rules are prohibited, permissible and obligatory.
lt is prohibited to move downwards on the moral landscape, by causing a loss ol
well-being or biodiversity or by violating someone's basic right. lt is permissible to
move horizontal or upwards, but we do not always have a duty to move upwards.
we are allowed to move in the horizontal or upward direction that we preler, we
are allowed to be partial and help the ones we hold dear (or ourselves), even il we
could generate more liletime well-being by helping those who are worse-oll
instead ol ourselves or our loved ones (clr. the little linger). lowever, sometimes
we do have an obligation to move upwards in a certain direction. 8uppose helping
the worse-oll would generate more liletime well-being than helping our loved
ones. lmagine that you are lorced to help those worse-oll, although you would
rather help others or increase your own well-being. 1hen your basic right would
be violated, which means a strong descent on the moral landscape. lut helping
those worse-oll would also generate more well-being, which means an increase on
the landscape (the wellare lunction increases). ll that increase is higher than the

!1
decrease due to the basic right violation, it implies that you have a duty to help
those worse-oll.

tnica| ccnsistcncy anJ Anima| ua|ity






lart 1 lthical consistency

tnica| ccnsistcncy anJ Anima| ua|ity




!
chapter 1 1he basic elements
ln this section l present the basic elements in a construction ol a coherent ethical
system. 1hose elements consist ol the input data (moral intuitions), the method
(the principle ol universalization) and the goal: an ethical system which is
internally consistent (contains no contradictions), clear (contains exact,
unambiguous lormulations), parsimonious (does not contain unnecessary or
arbitrary elements) and comprehensive (compatible with as much as possible ol
the strongest moral intuitions, applicable to all situations). l will delend this
approach by relerring to several analogies borrowed lrom the empirical sciences,
mathematics, geometry, grammar, taste evaluation and crossword puzzles. 1here
is however one serious issue in this approach: the existence ol illusions. 1hese are
the pitlalls: we cannot trust all our input data (our moral intuitions).
1.1 1he input data: moral intuitions
lntuitions can be roughly described as immediate, automatic, last, non-inlerred, a
priori, spontaneous judgments (or beliels) that lack lurther justilication, the
typical gut leelings, or responses ol a system in 'automatic mode'. lt can be
compared with e.g. perception or aesthetic judgments. ln morality, l consider the
lack ol lurther justilication, or the lact that we cannot give rational arguments to
justily intuitions, as the relevant property ol moral intuitions. morality is based on
moral intuitions about what is right and what is good. 1hese intuitions are
judgments with a motivational, prescriptive and olten emotional content. we have
a desire to comply with our moral intuitions. 1he stronger a moral intuition, the
tnica| ccnsistcncy anJ Anima| ua|ity
!6
less willing we are to act against it (or tolerate someone acting against it).
1

morality is 'emotionally driven', based on emotions such as empathy and
indignation. 1he motivational component means that you yoursell leel the urge to
do something or relrain lrom doing something, and the prescriptive component
means that you want others to do similar things. morality is non-cognitive: it does
not contain statements with a truth value. lt rather expresses emotions, attitudes
and prescriptions.
1he counterpart ol intuitions is rellective thinking. 1his is the area ol ethics, to
be distinguished lrom morality.
2
lthics moves beyond the unrellective intuitions
ol morality: it is characterized by a search lor principles as basic building blocks
(axioms) ol an ethical system. 1hese ethical principles are rellective and clear
expressions ol underlying moral intuitions.
Although moral intuitions olten appear in a blink ol an eye, it is not sell-evident
to articulate all your moral intuitions. ln order to interrogate your morality, we
can use special devices or thought experiments. 1hese are moral dilemmas


1
8ome moral intuitions are more emotional in nature (e.g. the leeling ol disgust about unharmlul, sale,
consenting incest between siblings). Others are more 'perceptual' or 'rational' in nature (e.g. the
spontaneous judgment that saving more lives is better than saving less). 8ome intuitions are inborn
(e.g. uon't kill your children"), others are learned or acquired (e.g. some racist intuitions about black
people by a white supremacist). 8ome intuitions arise in concrete situations (e.g. uon't push the lat
man in lront ol the trolley"), others might be more abstract or lormal (e.g. 1reat equals equally"). ln
this dissertation, l consider those distinctions as less relevant, because the goal is a 'rellective
equilibrium' (kawls, 191) where our moral intuitions are brought into a coherent system ol ethical
principles. ln the movement towards rellective equilibrium, some intuitions need to be revised or
disposed. ln that sense, it doesn't matter whether the intuition is more emotional or rational. what
matters is the perceived strength ol the intuitions and our willingness to revise or trump the intuitions:
which moral intuitions are more revisable or surpassable than others, which intuitions are stronger to
resist modilication, which intuitions are stronger to surpass others7 l do not think that there is a clear
correlation between e.g. revisability and the degree ol emotionality (versus rationality) ol intuitions.
8ome intuitions that tend to be more emotional in nature might be stronger than other more rational
intuitions. And some intuitions on the rational side ol the spectrum might easily override some
emotional intuitions. Neither do l see a clear correlation between strength ol an intuition and the
degree ol innateness. 1herelore, l am not much concerned about the degree ol emotionality or
innateness ol moral intuitions. l am rather concerned about our willingness to revise intuitions and the
extent to which intuitions can be brought into a coherent ethical system. l agree with kawls, who
recognizes emotionality (e.g. when we are upset or lrightened) as an error-disposed condition ol moral
intuitions (kawls, 191, p41, lrophy, 2009, p1!), but some kinds ol emotionality (e.g. empathy) might
also increase the credibility ol intuitions (lrophy, 2009, p11).
2
8ome animals such as chimpanzees, dolphins and dogs have a (proto)morality (8chermer, 2004, p.16,
lekoll & lierce, 2009): they can leel empathy, they can cooperate, they demonstrate altruistic behavior
or they have intuitions ol lairness. lt requires more complex rational thinking to move lrom
unrellective (intuitive) morality to rellective ethics.
1nc uasic c|cmcnts
!
developed to test or discover new intuitions. One example ol a moral dilemma is a
trolley dilemma (see also appendix lor a review ol the trolley problem): suppose
live people are on a track, unaware ol the oncoming trolley. 1hey will all die il you
do nothing, because you see that the trolley driver is unconscious. You are
standing on a bridge, and next to you, exactly above the rails, is a really heavy
man. You can save the lives ol the live people il you push the man lrom the bridge,
because he is heavy enough to block the trolley. most people (men and women,
lrom dillerent cultural backgrounds) intuit that it is impermissible to push the
heavy man.
ln this dissertation we will encounter other moral dilemmas. 8uch dilemmas
olten appear to be larletched or unrealistic, but remember that in order to
discover basic laws ol physics, scientists agree that perlorming special
experiments (e.g. using particle accelerators) in a thoroughly controlled manner
(e.g. in a vacuum) is the best strategy to interrogate nature. lven il those
experimental set-ups appear to be everything but similar to the world we
experience, they are very instructive to search lor physical laws. 1he idea behind
those 'exotic' experimental set-ups is to exclude disturbing lactors lrom the
experiment. l believe the same goes lor ethics: exotic thought experiments like the
trolley dilemma can be line tuned to interrogate our intuitions, eliminating
disturbing elements.
Now that we have encountered a moral intuition, it appears that, lor a lot ol
people, it is dillicult to translate or express that moral intuition into ethical
principles. why do we let live people die il we could save them7 One principle
might be: don't kill. lut that's too vague. A more accurate lormulation could be: do
not act il action results in the death ol a person who would not have died
otherwise.
Once we have such hypothetical ethical principles, we have to test whether the
resulting ethical system is consistent and whether the principles are both
internally consistent, as well as consistent with other moral intuitions. consider
another trolley problem: again live people are on a main track. ll you do nothing,
they will die. lut this time you could turn a switch so that the trolley will take a
side track. Lnlortunately on the side track one person will be killed. 1he structure
is similar: doing nothing means live people die, acting means live people are saved
and another person dies. most people intuit that we are permitted to act. 1his
seems to be in contradiction with our hypothetical ethical principle, so either we
dismiss one ol the intuitions, or we reline the ethical principle. ln later sections we
will discuss in more detail this rellective process ol looking lor a consistent
system.
l believe that moral intuitions are a very important element in ethics and a
valid starting point ol deriving an ethical system ol animal equality. 8ome ethicists
tnica| ccnsistcncy anJ Anima| ua|ity
!8
(e.g. some utilitarians such as 8inger, 200) might claim that their ethical systems
are completely detached lrom any moral intuitions. lowever, all systems start
with axioms that can be considered as intuitions. consider sum-utilitarianism. lor
those utilitarians it seems sell-evident that we should look at consequences and
that we should maximize a property. lut why are only consequences important,
why should we locus at the maximization ol something, and what should we
maximize7 lor those utilitarians it seems sell-evident that we should maximize
that thing that people want to maximize, such as well-being. lut how do we
aggregate well-being7 lor those utilitarians it seems sell-evident that we should
take an aggregation that rellects impartiality. lut what impartial aggregation
should we take7 lor those utilitarians it seems sell-evident that we should take a
simple aggregation lormula. lut why should we value simplicity and what simple
lormula should we take7 lor those utilitarians it seems sell-evident that we should
take the sum ol well-being. lut why should we maximize the sum ol utilities
instead ol e.g. the product, which is equally simple lrom a mathematical point ol
view7 ln summary, even utilitarians are laced with several sell-evident beliels that
they are not able to justily any lurther.
Also virtue ethics and deontologists are based on the intuitions that virtues
(benelicence, compassion, honesty,.) or some other properties (intentions,.) are
important. All ethicists use intuitions.
!
And in the context ol animal rights, meat
eaters have and live by moral intuitions as well. 1hey have the intuition that
personal liberty and choice ol lood are important, that (some) animals have lower
moral status that allows humans to eat them. All moral agents have moral
intuitions and they are not easily tempted to do something against their moral
intuitions. 8o, respecting moral intuitions is important. when l derive an ethical
system and l want to convince meat eaters to become vegans, l should propose a
system that is highly compatible with the strongest moral intuitions that both l
and those meat eaters share. 1he strength ol a moral intuition is inversely related
to the willingness to override the intuition. lence, the strongest moral intuitions


!
lerhaps one could make distinction between 'cool' (more cognitive) intuitions and 'hot' (more
emotional) intuitions. lor example oreene (2008) criticized a lot ol deontological judgments, claiming
that they are based on unreliable, 'hot' or alarmlike emotional responses (such as disgust). ll such
distinction is possible, we can construct an ethical system based on only the 'cool' intuitions. lut lor me
it seems dillicult to clearly make this distinction (perhaps neuroscience can help in making this
distinction7). And it is not clear why all 'cool' and none ol the 'hot' intuitions are reliable. 1herelore, l
will make another distinction, between those intuitions that lit in a (strong) coherent lramework and
those that don't.
1nc uasic c|cmcnts
!9
are the most motivating, because they generate the strongest desires to respect
them.
we cannot escape the idea that moral intuitions are the only input that we have
in ethics, and that moral agents have dilliculties in overriding their strongest
moral intuitions. 8o we should cherish those strong moral intuitions. lthics
without moral intuitions is like science without experimental lacts. ln all aspects,
lrom science to daily lile, we need some relevant input. let's consider the
lollowing six examples that can be used as analogies ol ethics.

1) lhysics and other empirical sciences. ln the lield ol scientilic research as a
cognitive activity, we have observations as input data. 1hese observations are very
specilic, lor example: at time 1 at place l under conditions c, l saw object O lalling.
ln general, the inlormation about the external world is received through sensory
data, perceptions by one ol our live senses.
2) mathematics and algebra. lere we start with a priori knowledge. 1hree is a
number and lour is one higher. You have to accept this intuition, otherwise we
can't move on in mathematics.
!) oeometry. 1his line segment is similar to that. 1here is some property (e.g.
length) about lines that make them similar. 1hat is an intuition in geometry.
4) orammar. According to chomsky (1986), people lrom dillerent countries
have an inborn laculty ol language in their brains, which means that they have
intuitive judgments about the grammatical correctness ol a sentence. 1he gnorll
is sprinkle" is a good grammatical sentence, l just know it, even il the content is
meaningless. 1he grammatical intuitions are interesting to better understand how
morality lunctions. }ohn kawls (191) and others (lauser et al. 2008) proposed an
analogy between our language laculty and our moral laculty. 1he language laculty
generates intuitive grammatical judgments about the grammatical correctness ol
sentences, just like the moral laculty generates intuitive moral judgments about
the moral goodness ol situations. we intuitively see that an act is morally right
and a sentence is grammatically correct. And as with morality, people olten have
dilliculties in expressing why a sentence is grammatically right or wrong, they just
know it. lt takes some ellort to look lor the principle that expresses the specilic
intuition. As with grammar (chomsky, 1986), some moral intuitions and lines ol
reasoning are universal (independent lrom culture) and likely inborn. 1hus we can
speak ol a universal moral grammar (mikhail, 2000, 200, O'Neill & letrinovich,
1998).
) crossword puzzles. 1he input data are the descriptions and number ol white
boxes lor the words.
tnica| ccnsistcncy anJ Anima| ua|ity
40
6) 1aste evaluations and aesthetic judgments. 1he input data are taste
prelerences ol specilic products. lt appears that people lrom dillerent cultures like
some product made lrom cane sugar.

ln all those areas we have input data, be it evaluations, judgments, observations,
given inlormation,. moral intuitions can be compared with those input data. 1he
above six analogies give us a nuanced picture about the role ol intuitions in our
morality. l explore lurther these same six analogies below, in the discussion ol the
method.

1nc uasic c|cmcnts
41
1.2 1he method: rule universalism
1he method to be used in order to set up an ethical system is universalization.
lirst the moral intuition is expressed as a particular ethical rule, valid lor that
particular situation. Lniversalization then consists in extending this particular
rule to all other similar situations. ln briel: lqual moral judgments lor all morally
similar situations." A situation is morally similar to another il it is similar with
respect to all morally relevant characteristics ol those situations. 1his is a lormal
method: it does not have material content, i.e. it does not state what
characteristics are morally relevant.
1he idea ol universalization is mostly elaborated by ethicists lmmanuel lant
(18) and k.m. lare (1991). 1he most general expression ol the principle ol
universalism reads: You must (may) lollow the rule that cvcrycnc wnc is capau|c,
raticna| anJ injcrmcJ must (may) lollow in a|| mcra||y simi|ar ways in a|| mcra||y simi|ar
situaticns towards a|| mcra||y cua| inJiviJua|s." 1he lour parts in italics in this
expression point at lour kinds ol universalism: the moral agents (the actors who
are capable ol doing something), the moral patients (the receivers ol a benelit or
harm, as morally equal individuals), the acts (the similar ways ol doing something)
and the situations. 1he moral agent A does (or relrains lrom doing) an action c to
moral patient l in situation 8.
1he principles ol impartiality and anti-discrimination are clearly
universalizations with respect to patients. uiscrimination is a dillerent treatment
ol individuals (patients), based on morally irrelevant criteria.
1he idea that moral imperatives must be equally binding on everyone is an
example ol universalization with respect to the agent. lant's lamous categorical
imperative (lant, 18, p.!0), can just as well be understood as a universalization
with respect to agents andJor patients. ln its 'universal law' lormulation, the
categorical imperative goes as lollows: act according to that maxim (moral rule or
guiding principle
4
) whereby you can will that it should become a universal law."
One should ask the question: what il everyone (or many people) acted (or


4
A note on terminology: most ol the time, a 'principle' relers to a 'basic principle'. 1his dissertation
argues that there are live basic principles. 1he lirst basic principle is called 'rule universalism'. 1his
principle relers to the universalization ol moral 'rules' or 'guiding principles'. 1hose moral rules are
typically less basic, less abstract or more specilic than the basic principles. moral rules are derivatives
ol basic principles. lor example the moral rule uo not steal" is derived lrom the basic principle to
maximize aggregated well-being. 1he moral rule uo not rape" is derived lrom the basic mere means
principle.
tnica| ccnsistcncy anJ Anima| ua|ity
42
thought it is allowed to act) in this way7" 1here might be two contradictions: a
contradiction in conception, which means that the universalized rule results in a
logical or physical impossibility, and a contradiction in the will, which result a
universalized rule is possible, but not wanted by a rational moral agent.
An example ol a contradiction in conception, resulting lrom universalizing a
rule with respect to the agent, is the ecological lootprint. ll everyone (all humans)
would consume as much as an average human in a developed country, we will
exceed the carrying capacity ol the planet. we would need resources ol more than
one planet, because the ecological lootprint in the developed world is higher than
the available biocapacity ol 1,8 global hectares per human (olN, 2010). 1here is no
other planet like larth, so this behavior is simply not universalizable. 1herelore, it
is unlair (and unsustainable) lor those people in developed countries to have such
a high ecological lootprint.
A contradiction in the will might occur alter a universalization with respect to
both agent and patient. lor example il l am allowed to harm you, then everyone is
allowed to harm everyone else. 8o you are then also allowed to harm me. l cannot
will this, because l value my well-being, so l encounter a contradiction. my rule
that l am allowed to harm you cannot be universalized, so it is an immoral rule.
1his universalization therelore generates the golden rule: (do not) treat others as
you would (not) like to be treated.


Lniversalization with respect to the act becomes much more subtle. Note that
the above expression (You must lollow the rule.") relers to rules instead ol
particular acts
6
. lence we can call it rule universalism. A lormulation in terms ol
rules instead ol acts has some advantages. lirst, it allows lor conditional rules (e.g.
uo X il Y unless 2")

. ll l say that l am allowed to lie down on my sola and watch a


movie tonight, this act cannot be universalized with respect to agents, because it is
impossible lor billion people to lit on my sola tonight. we might save the idea
behind the categorical imperative il we universalized a rule instead ol an act,

1he lollowing example is a critique to an oversimplilied application ol the golden rule. 8uppose that l
don't mind il you lied to me. One might interpret the golden rule as claiming that in that case it is
allowed lor me to lie to you. lowever, this application ol the golden rule is itsell a violation ol the
golden rule. we can derive a 'platina rule' lrom the golden rule: do not take your own prelerences when
you decide how to treat others, because you would not like others to take their own prelerences when
they decide how to treat you. ln other words: l am not allowed to lie to you il you do not want be lied
to, because l would not like you to lie to me when l do not want to be lied to.
6
we encounter a similar dillerence between rules and acts in the discussion between rule
consequentialism and act consequentialism (looker, 2011).

And it allows lor game theoretic situations where the choice what to do depends on what others do or
should do (see section .2).
1nc uasic c|cmcnts
4!
because the rule can contain some conditionals. l can lollow a rule which says that
l am allowed to sit in a place l preler and watch a movie at a time l preler, il l own
the place (e.g. my sola) or il there is some place lelt (e.g. in a movie theater).
Lniversalizing this means: everyone is allowed to lie down in a place they preler
and watch a movie they preler, il some ol those conditions are met. A locus on
rules instead ol acts allows a specilication ol the conditions that can be
universalized in such a way that we can want this universalization ol the
conditional rule.
A second advantage is that the locus on rules allows lor public expressions, such
that all moral agents are able to understand and lollow the rules. kules are uselul
tools in giving justilications ol acts. As a consequence, a locus on rules gives
meaning to the idea ol giving the good example as well as the idea ol giving the
right justilication.
Lniversalization also applies to metarules (rules about rules). lor example, l can
propose the universalized rule: lveryone should always tell the truth, unless your
name is 8tijn lruers." lut now marie can use a universalization on the level ol
metarules and respond that il l am allowed to use my name in a rule, then she is
allowed to do the same: lveryone should always tell the truth, unless your name
is marie." l do not want this, and as a consequence, we can derive a metarule:
kules should not reler to names." 8imilarly, rules should not reler to nouns or
specilic times and places. 1his universalization on the level ol metarules gives
important clues on what counts as morally (dis)similar situations. A situation
where 8tijn lies is similar to a situation where marie lies.
1he universalization criterion matches two conditions: non-arbitrariness and
consistency. An ethical system based on rules that are not universalized may easily
be consistent, but the lack ol universalization results in arbitrariness.
Lniversalizing the rules avoids the arbitrariness, but now the system might
contain mutually inconsistent rules. loth non-arbitrariness and consistency
should be respected. A rule should be non-arbitrary (i.e. universalized) and
consistent with other non-arbitrary rules ol the system. 1he two conditions ol
non-arbitrariness and consistency generate coherence (see chapter 2).
Lniversalization in ethics is related to the property ol supervenience ol moral
statements. 8upervenience is a kind ol dependency relation: moral judgments
supervene on natural properties in the sense that il two situations, acts or events
are similar in their natural properties
8
, then they should imply the same moral
judgment (or moral value). Or stated in reverse: il you have dillerent moral


8
with similar natural properties" l mean similarity only in their relevant aspects.
tnica| ccnsistcncy anJ Anima| ua|ity
44
judgments about two dillerent situations, then you should be able to point at a
morally relevant, objective dillerence between those two situations. compare this
with the supervenience ol mental states on physical brain states: psychological
properties supervene on physical properties in the sense that a|| persons who have
the same physical properties (the same brain states) must also be psychologically
indistinguishable (having the same mental states).
9
Note the universality ol this
statement.
Note that in non-cognitivist meta-ethics such as expressivism, supervenience is
considered as a consistency condition: expressivists claim that moral judgments
are based on our subjective attitudes toward objective, natural properties (e.g.
behaviors or situations). 8upervenience means that our attitudes should be
consistent and non-arbitrary.
1he process ol universalization is omnipresent in ethical discussions and
argumentations, because universalization allows lor the use ol analogies, and
analogies are olten used in argumentations. leople lrequently reler to another
specilic situation which is similar to the discussed situation, and then appeal on
intuitions in that other situation. 1he question is: when is an analogy valid7
lt is clear that the validity ol the use ol an analogy needs to be justilied as well.
lut lirst, ol course, we need to justily the use ol the method ol universalization. 1o
do this, l will reler again to the six analogies ol ethics mentioned in the previous
section. ln all those examples, universalization is important. Next, l will
demonstrate that a lot ol people, including meat eaters, use this method in ethical
discussions. l will present a list ol arguments given by meat eaters, in order to
demonstrate that meat eaters also accept this method ol universalization. lence,
alter accepting the importance ol moral intuitions, we come to see that animal
rights ethicists and meat eaters both share the same rules ol the game. 1his should
bring a compelling rational argument lor veganism and animal equality one step
closer.
8o let's lirst take another look at the six analogies.

1) lhysics and other empirical sciences. we had a specilic observation that at
time 1 at place l under conditions c, l saw object O lalling. l now make a
hypothesis which goes under the name ol induction: a|| similar objects lall under
similar conditions at a|| times in a|| places. 8o when l see another object lalling


9
A dillerence between psychological supervenience and moral supervenience is that the letter has no
physically or logically necessary relationship. ln this sense, supervening moral judgments are not lacts
ol the world, whereas psychological states are lacts ol the world.
1nc uasic c|cmcnts
4
when l release it, it is conlorming to this principle ol gravity. ll it is not lalling,
then something relevant must be dillerent: the place (e.g. in a space ship), the
conditions (there is a strong lorce such as a magnetic lield counteracting gravity)
or the object (it has a propeller and wings, or it is a balloon with helium). when
the object does not lall, it simply means that other principles should be included
and those principles need to be universalized as well (e.g. cvcry timc an object is
released in the presence ol a strong magnetic lield that acts as a lorce.).
2) mathematics and algebra. lrom the intuition that the number ! has a
successor, we universalize this to the mathematical axiom that cvcry number has a
successor. 1his is a basic axiom in the algebraic system ol natural numbers.
mathematicians do not allow an exception to this rule. lut there can be other
universalized rules with exceptions, and ol course those exceptions need also to be
universalized.
!) oeometry. A|| lines ol equal length are similar. A|| right angles are congruent.
(1he latter principle is one ol the basic luclidean axioms.)
4) orammar. All sentences with the structure 8ubject - verb - predicate" are
grammatically correct. 1he gnorll is sprinkle" was just one example ol such a
structure. lut there is an exception: the structure is not correct when the sentence
is a question. And this exception rule needs also to be universalized: the structure
Verb - subject - predicate7" is always correct in case ol a question.
) crossword puzzles. ll the input data is lruit" and live letters", and we can
lill in the letters All", then we are lorced to complete this word to Allll". 8o
lilling in a word gives inlormation about the content ol a|| relevant white boxes,
i.e. all boxes that are similar. 8imilarity means that they are next to each other in a
row or column. 1his is also a kind ol universalization. An individual letter
represents a particular ethical rule, applicable to a particular situation. A word is
analogous to the universalized principle, applicable to all similar situations.
6) 1aste evaluations and esthetic judgments. 1his pear tastes good to me, so
now l have to accept that a|| products with the same chemical structure taste as
good. lut there is an exception: e.g. when it is hot, it no longer tastes good. 8o we
again universalize this to the principle that all products with a similar chemical
structure and the same temperature taste as good.

ln summary: universalization is perhaps the most important tool in moral
arguments, because it is the basis ol consistency. Lniversalization happens not
only in ethics, but in all other cognitive activities, as long as similar conditions
apply (this is also a universalized statement). ll animal rights ethicists want to
convince meat eaters that we should become vegans, and il these ethicists want to
use rational arguments, meat eaters should accept the same rules ol the game.
meat eaters already accepted moral intuitions as starting points, so let us see
tnica| ccnsistcncy anJ Anima| ua|ity
46
whether they can also accept the method ol universalization. looking at
discussions with meat eaters, it becomes clear that those ingredients are indeed
important lor them as well. loth meat eaters and animal rights activists use all
lour lorms ol universalization in their arguments. lere is a list ol arguments by
meat eaters. 1hey indicate that meat eaters value universality and consistency as
well.
1.2.1 Lniversalizations made by meat eaters
1.2.1.1 Lniversalization with respect to the situation or the act
8ut arcn`t ycu wcarinq |catncr snccs ll killing animals lor lood is not allowed, then
killing them lor clothing is not allowed either. Or il you are allowed to wear
leather shoes, then you are also allowed to eat meat. 1his is a correct argument,
and it is the reason why ethical vegans won't buy leather shoes.
Ycu arc a vcqan, sc ycu arc a|sc aqainst urcast jccJinq ll you are against the use ol
animal products, you should be against breast milk as well, because humans are
animals as well. 1his is an invalid argument, because we can easily point at morally
relevant dillerences: breast milk is necessary lor the baby, and the mother gives it
voluntarily. 1he ethical rule uo not use animal products" has exceptions, such as
necessity and voluntariness. 1hese exceptions should be universalized as well.
5c wc sncu|Jn`t wa|k arcunJ uccausc tnat wi|| ki|| insccts lilling cows lor lood is
like killing ants by walking around. 1his is an invalid argument, because ol two
reasons: insects likely lack or have a very small emotional lile, and killing by
accident when walking around is dillerent lrom intentionally killing someone in
order to use him.
Ycu arc aqainst anima| cxpcrimcnts, sc ycu ncvcr usc Jruqs tcstcJ cn anima|s 1his
is an invalid argument, because ol two reasons: 1) animal rights activists are only
concerned about animal experiments done in the present and luture, and 2) meat
eaters are likely to use products whose development or discovery involved rights
violations in the past that they abhor. lor example they might use technology
based on mechanisms or materials that were discovered with the help ol slaves
some hundred years ago. Note that l hereby replied with an analogy, relerring to
technological inventions and slavery. we can argue that the analogy is valid, as the
relevant properties are similar: rights violations ol animals and slaves, the use ol
animals and slaves in the process ol developing something that we now use. 8o l
replied with a universalization.
Anima| riqnts activists want tc prcmctc jrccJcm cj anima|s, uut tncy rcstrict cur
jrccJcm uy impcsinq vcqanism aqainst cur wi||. 1nat is inccnsistcnt. 1his is an invalid
argument because ol two reasons: 1) there is a morally relevant dillerence
1nc uasic c|cmcnts
4
between killing someone in order to enjoy the taste ol eating him, and restricting
someone's lreedom to violate rights. 2) meat eaters themselves promote lreedom
ol e.g. women by restricting the lreedom ol rapists and imposing their ethics upon
them. Again l make an analogy, which is customary in a process ol
universalization. l universalized with respect to the patient, lrom animals to
women, and with respect to the act, lrom slaughter to rape. 1he analogy is valid,
because raping women and killing animals are both examples ol violations ol
bodily integrity just lor pleasure, and these are morally relevant lacts.
Anima| riqnts activists usc tnc cartn as wc||, ki|| |ijc, rcu |anJ jrcm wi|J|ijc, cat jccJ
tnat ctncrwisc anima|s ccu|J navc naJ. 1nc prcJucticn cj vcqan jccJ a|sc ki||s anima|s.
1his argument is invalid because there is a morally relevant dillerence between
killing by accident and intentionally killing someone. meat eaters also eat lood
that otherwise other humans could have eaten. And animal rights activists
consistently claim that we do have a duty to protect wildlile animals as much as
possible, and help them il we can.
kiqnts arc a numan invcnticn, sc ta|kinq aucut anima| riqnts is sti|| antnrcpcccntric.
1ncrcjcrc it is inccnsistcnt witn tnc iJca tnat antnrcpcccntrism is uaJ. 1his argument is
invalid, because this is a conlusion ol two dillerent notions ol anthropocentrism,
and these two notions are morally not the same. ll rights were my invention, this
is egocentric, because rights originated lrom me. lut that kind ol egocentrism is
not a moral problem. lowever, it is dillerent lrom the dangerous egocentrism that
claims that l am the only person in the world who has rights.
j wc qivc anima|s tnc riqnt tc |ivc, tncn wc sncu|J a|sc qivc tncm tnc riqnt tc vctc.
1his is an invalid argument because ol two reasons: 1) higher intelligence is a
capacity that is clearly relevant in the right to vote but not the right to live, and 2)
meat eaters themselves give babies and seriously mentally disabled persons a right
to live but not a right to vote.
j anima|s navc riqnts |ikc numans, tncn wc sncu|J a|sc nccJ a nuqc numucr cj anima|
ncspita|s. 1nat is unrca|istic. 1his argument is valid to some degree: just like helping
humans we have a duty to help wild animals in need, as lar as this is leasible lor us.
1here are already wildlile rescue centers (l happen to do voluntary work in a bird
care center), so it is not unrealistic.
Anima|s matc witn ccnspccijics. 5c Jc numans. sn`t tnat spccicsism as wc|| l leave it
up to the reader to lind the morally relevant dillerence in this case.
wnat`s wrcnq witn tnc p|casurc cj tnc tastc cj mcat 5ncu|J wc jcrswcar a|| p|casurc
1his argument contains a universalization: il A is bad and done lor pleasure, then
anything done lor pleasure is bad. lowever, that rule quickly violates intuitions
that both meat eaters and vegans share. 8o the animal rights ethicists come up
with a more accurate universalization that says that we should lorswear all
pleasure il important rights are violated.
tnica| ccnsistcncy anJ Anima| ua|ity
48
1.2.1.2 Lniversalization with respect to the patient
wnat aucut p|ants, insccts,. 1ncy a|sc navc a |ijc. 1he morally relevant dillerence is
sentience: plants and insects have a much lower probability to be sentient than lor
example vertebrate animals with a complex lunctioning central nervous system.
ln a later chapter (section 8.) l will give several arguments why sentience is
morally relevant in this matter.
Anima|s can`t tnink raticna||y, navc nc sc|j-ccnscicusncss, navc nc mcra| aqcncy,.
lere the meat eater tries to point at some special mental capacities that animals
lack. lowever, also some humans lack them, but most meat eaters give basic rights
to e.g. mentally disabled persons.
atinq p|ants is a||cwcJ uccausc it is natura| anJ ncrma|. atinq mcat is natura| anJ
ncrma| as wc||, tncrcjcrc it is a|sc a||cwcJ. lere the meat eater uses a rule that applies
to plants, and extends it to animals. 1he problem with this argument is that the
criteria 'normal' and 'natural' are vague. uepending on how one might more
accurately deline those terms, one can argue that eating humans would be
allowed, or rape would be allowed. more about the two criteria normal and natural
will be discussed in the chapter on predation in part !.
1.2.1.! Lniversalization with respect to the agent
j am nct a||cwcJ tc cat mcat, tncn |icns, tnc nuit pccp|c,. arc nct a||cwcJ tc nunt
citncr lere the meat eaters says that il A (a lion) is allowed to eat meat, then l (a
human in a developed country with a moderate climate) is also allowed to eat
meat. ln the chapter on predation in part ! l will elaborate on this argument, and
show that the universalization is not valid, because an important necessity
criterion is not universalized.
ll everyone (all humans) would be vegan, then it is impossible to leed
everyone. Agriculture without livestock is impossible (lor example no manure to
lertilize the croplands,.)" 1his argument points at a lantian contradiction in
conception. lt is actually a lactual claim. 1here are studies however that indicate
that a global vegan organic agriculture, without use ol chemical and animal
lertilizers, is leasible and can leed a world ol 9 billion people (lairlie 200, Olewski,
2010, the Vegan Organic Network). 1ypically, vegan products with the same
nutritional value require much less inputs such as water, energy, land and
chemicals than livestock products (olN, 2010, loekstra, 2010). And they generate
much less negative outputs such as nitrogen pollution and greenhouse gas
emissions.
ll everyone (all humans) would eat vegan, then all animal races that we breed
in the livestock sector would go extinct. 1hat's also a loss ol biodiversity." 1he
problem is that livestock currently is likely the largest threat to wildlile
1nc uasic c|cmcnts
49
biodiversity (lAO, 2006). And secondly, is it really a contribution to biodiversity to
intentionally breed animals with serious physical handicaps7 1hat is not morally
justiliable.
Our ancestors ate meat. lt would be unlair towards us il we would not be
allowed to eat meat." 1his is a universalization lrom our ancestors to us. my
counterargument is a universalization with respect to the act. ll such an argument
would be valid, then a similar argument would be valid lor slavery, rape,. we
would be allowed to do that because likely one ol our ancestors did it. we cannot
will this universalization, and therelore we have to say that what our ancestors did
cannot justily what we do.
ll everyone (all humans) would be vegan, all people working in the animal lood
production would lose their jobs." lowever, new jobs would be created to produce
vegan lood. ln the end it comes down to the lact that vegan larming is much more
ellicient in terms ol inputs, and therelore less capital intensive. ln other words:
vegan larming makes much more sense economically. And a second
counterargument, based on the universalization ol the act: a same thing has been
said about slavery and e.g. slave traders losing their jobs. ll slavery cannot be
justilied by pointing at (invalid) economic concerns, then animal lood production
cannot be justilied either.
1.2.2 Lniversalizations made by animal ethicists
Ol course, in discussions, we olten hear animal ethicists giving arguments based
on universalization as well.
1.2.2.1 Lniversalization with respect to the situation or the act
8peciesism is similar to racism and sexism."
1he livestock industry is similar to slavery, rape,."
ll we are allowed to eat cows because we breed them lor that purpose, then
slavery would also be allowed il we breed black people lor that purpose."
ll we are allowed to eat cows because they owe their lives to us, then we are
also allowed to eat our babies or humans that we would breed."
ll we are allowed to use animals in the livestock industry because they have
lood, shelter and medicines and hence those animals are better oll than animals in
the wild, then it is also allowed to use indigenous people as slaves, because in their
natural habitat they suller lrom predators, parasites, drought, disease,."
ll mentally disabled persons get rights because they belong to a group whose
normal members have rationality, then also a mentally disabled person should
have a right to vote and go to university. And then also chimpanzees and other
tnica| ccnsistcncy anJ Anima| ua|ity
0
primates should have rights, because the majority ol them (roughly billion
primates) has rationality."
ll meat consumption is permissible because it is not prohibited by the law,
then also slavery, rape,. were permissible in times when it was not prohibited by
law."
ll eating meat is allowed because it is natural, then rape should also be
allowed, because it is also natural: our ancestors did it lor thousands ol years,
other animals do it, men have developed (by evolution) special body parts that
enable them to rape women, and it is very likely that we owe our lives to the lact
that one ol our ancestors once raped a woman."
ll veganism is not good because it is unnatural in the sense that one needs
vitamin l12 as supplements (and those l12 products come lrom a lactory), then
eating processed loods or using toothpaste would also be unnatural and bad. 1he
use ol toothpaste is a consequence ol our choice ol diet, just like the use ol l12 is a
consequence ol a dietary choice. And both are produced in a lactory. lt's even
worse lor the non-vegan person, because drinking cow milk at an adult age, animal
experimentation, leeding l12 supplements to livestock animals, and much more
would also be unnatural and hence bad."
1.2.2.2 Lniversalization with respect to the patient
ll eating pigs is allowed, then eating dogs should also be allowed."
ll killing and eating animals is allowed, then killing and eating humans should
also be allowed."
ll animals don't have rights because they don't have duties and they cannot
understand ethics, then the mentally disabled persons also cannot have rights."
1.2.2.! Lniversalization with respect to the agent
ll you don't need to harm sentient beings in order to survive, then you should not
harm them. Ol course that also applies to me, so l became vegan."
ll l am the only vegan person, the market will not notice my lood choice, and
not a single cow will be spared. lowever, l should give the good example and do
what everyone should have to do. ll everyone became vegan, then animal rights
violations would be much lower and that is good."

1he above examples show that the method ol universalization sets a common
ground lor both animal ethicists and meat eaters. lut are people always consistent
in applying the technique ol universalization7 uo they sometimes use this
technique and other times not7 my claim - based on what we have seen a little bit
in the above examples - is that especially meat eaters (and sometimes also animal
1nc uasic c|cmcnts
1
ethicists) do not consistently use the technique ol universalization. Only when
applying the technique ol universalization consistently (universalize the
technique itsell, so to speak) is it possible to reach a coherent ethical theory.

tnica| ccnsistcncy anJ Anima| ua|ity





!
chapter 2 1he goal: consistency and
coherence
ln the above section, we started by looking lor moral intuitions. 1hat was the non-
rellective part. At the rellective level, these intuitions were lirst articulated as
particular ethical rules, and second, these particular rules were universalized to all
similar situations (acts, patients and agents). lence, we moved lrom moral
intuitions to particular ethical rules to universalized ethical principles. Now we
have to see whether those universalized ethical principles are internally
consistent. ll there is an inconsistency, we have three options: we can either
introduce a new principle that overrules the other, we can reline an existing
principle or we can simply delete the weakest principle (which is based on the
weakest, least motivating moral intuition).
ly testing more and more situations, i.e. by looking at whether our moral
intuitions in all situations agree with the universalized ethical principles, we
progressively get a coherent system. 1his is the process ol rellective equilibrium
(kawls, 191). coherence means that several intuitions and principles enlorce each
other, an interweaving ol mutually supporting intuitions and principles.
lere the analogy between constructing an ethical system and solving a
crossword puzzle becomes clear. ln the construction ol an ethical system, we start
with our strongest moral intuitions, those intuitions in which we have the most
conlidence. 8imilarly, in the crossword puzzle, we have more conlidence in a word
il there are less other words that lit the description. ll lor example the description
is lood" ol live letters, then the word can be Allll" or many other possibilities.
ll however the description is reduced to pomaceous lruit that grows on a tree",
we are much more conlident that the word should be Allll". Once we lill in
some words, we get new evidence lor other words. 8uppose l have lilled in the
word Allll. 1he lirst letter crosses another word ol live letters, with description:
good", and with the new inlormation that the lourth letter should be an A. 8o l
can lill in the word mOkAl. 1he last letter ol this word crosses a second word ol
two letters, a music note: lA. And now the l lrom Allll and the A lrom lA gives
us a new clue about a body organ with eight letters: lANcklA8. 1he whole set-up
tnica| ccnsistcncy anJ Anima| ua|ity
4
is not only consistent, but those lour dillerent words enlorce each other. we gain
more and more conlidence in the whole system.
coherence is the combination ol three things: univcrsa|ism (cr ncn-aruitrarincss),
ccnsistcncy and ccmp|ctcncss. Lniversalism in the crossword puzzle means that all
neighboring white boxes in a row or column should lorm one word, not just
independent letters or smaller words. ll the white boxes were lilled in with
independent letters instead ol meaninglul words, the crossword puzzle would
become completely arbitrary. 1he constraint that the rows and columns ol white
boxes should be words, decreases the arbitrariness. 1he same goes lor ethics:
similar situations (or the same situation lrom dillerent points ol view) should be
judged with the same ethical principles. ll every situation or viewpoint has its own
particular guiding principle or moral rule, the ethical system becomes too
arbitrary. Lniversalism strongly decreases arbitrariness.
consistency in the crossword puzzle means that a white box cannot contain
more than one unique letter. ln ethics, it means that each situation should have
one solution, one all things considered" moral judgment. ll in a situation there
are two dillerent inconsistent solutions (two dillerent judgments about e.g. the
(im)permissibility ol action), the ethical system is inconsistent and we are lelt
undecided. ln the crossword puzzle, there is a dillerence between trivial
consistency and strong consistency. 1rivial consistency occurs when words do not
cross and there is only one cue lor a word. lor example il the given cue is lruit"
and the word contains live letters, apple" is a trivially consistent solution. 8trong
consistency occurs when words cross each other and there are more cues, e.g. lor
vertical and horizontal words. ll the lirst letter ol the word apple" is crossed by a
second, vertical word ol six letters with cue nut", then the letter a" in almond"
is strongly consistent with the same letter in apple", because the letter a" has
two justilications. ln this dissertation, consistency always means strong
consistency. lor example, as we will see, when an ethical principle ol well-being
has two dillerent justilications derived lrom two dillerent points ol view, one
based on a thought experiment ol impartiality and one based on the virtue ol
compassion (see section 4.4), this principle becomes strongly consistent.
completeness, as the third condition, means that in the crossword puzzle no
white box should be lelt empty. ln ethics, it means that the ethical system should
be applicable to all possible situations and viewpoints.
1his idea ol coherence and crossword puzzles was proposed by laack (199!).
8he called it 'loundherentism', as it is a combination ol loundationalism and
coherentism. 1he loundations are the input data: inlormation about the meaning
and lengths ol words. compare this with our moral intuitions or with
mathematical axioms that also act as loundations. lut not all our moral intuitions
are equally strong. 8ome are much more motivating, others are easily overruled.
ccnsistcncy anJ ccncrcncc

most ol all: all our intuitions are lallible. Our moral intuitions are merely
provisional starting points in the construction ol a coherent system. 1hey are still
subject to revision or rejection il they are not compatible with the rest ol the
ethical system. ln the construction ol a coherent system, there is no inlinitely
strong moral intuition that can serve as the loundation. 1here is no absolute lixed
point. 8imilarly, in a crossword puzzle we can have dillerent levels ol conlidence
in the words, and there is no central word lrom which to start. ln principle, all
solved words are lallible, we can never be sure that a word in the puzzle is
absolutely correct. 1o gain conlidence, dillerent words can mutually support each
other, and build up a coherent system. 1his is the coherentist part ol the story.
loth loundations and coherence are important, but can be important to dillerent
degrees.
1

1his principle ol coherence (moving towards a rellective equilibrium with
principles that mutually support each other) is also universal, as it occurs in our
six analogies as well:
1) lhysics and other empirical sciences. lvidence builds up. A theory is a
coherent system. And sometimes new strong data lrom experiments appear that is
really incompatible with the principles. 8o the theory needs a revision. And like a
crossword puzzle, sometimes other aspects ol the theory need revision as well. lt
can ignite a cascade ol revisions. lven the principles behind the reliability ol the
experimental apparatus itsell might need revision. 1his is characteristic ol
scientilic revolutions (luhn, 1962). lut not to worry, as the more coherent a
theory becomes, the less likely we need to revise the whole thing. 1he more words
are lilled in the crossword puzzle, the more likely it is the real solution. 1he
scientilic method is nothing but a constant searching lor a rellective equilibrium: a
coherent, parsimonious, clear system ol knowledge that best lits the most
convincing input data. Anomalous data that do not lit a strong coherent scientilic
theory can be discarded, as they are likely the result ol some error. (8ee lrophy,
2009, lor an elaboration on the analogy between the scientilic method and the
moral method ol rellective equilibrium).
2) mathematics and algebra. mathematicians olten wondered whether the
system ol natural numbers is a consistent system. odel (19!1) showed that its
consistency was impossible to prool lrom within the system. One needs to extend


1
uuring the construction ol an ethical system, no intuition or principle is absolutely lixed. uuring the
process ol solving the crossword puzzle, no letter or word is absolutely lixed. lowever alter the
crossword puzzle is completely solved, we can be (almost) absolutely conlident in the completed words.
8imilarly, alter we have constructed a coherent system, the linal universalized ethical principles can be
considered as loundations or basic principles.
tnica| ccnsistcncy anJ Anima| ua|ity
6
the system (step outside ol the system) in order to prove the consistency ol the
system ol natural numbers. lut then the question reappears: is this extended
system consistent7 Anyway, we do notice that the system ol natural numbers is
strongly coherent. let's take the example ol the property that the sum ol integers
lrom 1 to N equals N(N-1)J2. You can test this several times, taking N=!, N=10,
N=1 and so on. 1he more you try it, the more plausible it seems. 1hese tests are
coherent with a prool that one could give: a prool ol induction. 1he property is
true lor N=1 and N=2. 8uppose it is true lor some arbitrary N. 1hen lor the number
N-1 we get the sum lrom 1 to N-1 should be (N-1)(N-2)J2. And indeed, this equals
N(N-1)J2-N-1. 1his proves the proposition. And then one could give a second
prool and a third. 8o these dillerent observations and prools indicate that the
system is strongly coherent, which increases our conlidence in its consistency.
!) oeometry. One could test the lythagorean theorem with some examples ol
right-angled triangles. And then one could give dillerent prools ol it (there exist at
least !0 dillerent prools ol this lamous theorem (loomis,1968)).
4) orammar. orammatical rules might give a coherent picture ol our intuitions
as well, by analyzing long sentences, breaking them down in parts according to a
grammatical rule.
) crossword puzzles. 1his example l already explained above.
6) 1aste evaluations and esthetic judgments. 1he lact that l like pears is
consistent with the lact that l like other lruits, the lact that l liked a particular
pear yesterday, the knowledge that pears contain sugar and l also like sugar, and
so on. 8o this gives me a coherent picture ol what l like.

ln ethics it is possible that there exist mutually incompatible ethical systems, all
internally consistent and based on some moral intuitions. 1his might point at
some kind ol ethical relativism, because each consistent system is equally valid.
lowever, let's go back to the analogy ol the crossword puzzle. lere it might be
possible that a crossword puzzle has multiple solutions: dillerent mutually
incompatible patterns ol words. lut the more words a puzzles contains, and the
lengthier the words, the less likely that there are dillerent solutions. l believe that
il we take the set ol strongest intuitions that both l and a meat eater share, then
the only coherent ethical system that we can construct, is one that implies
veganism. Ol course there are meat eaters who have totally dillerent moral
intuitions. l will never be able to convince the latter by merely rational arguments,
because we start lrom dillerent input data. 8o l want to address mysell to those
meat eaters who have some strong moral intuitions shared with me. And looking
at discussions with meat eaters, l notice that they also value consistency. why else
would they ask me il l'm wearing leather shoes7 why else do they so olten come
up with arguments based on universalized principles, as we've seen in the
ccnsistcncy anJ ccncrcncc

previous section7 consistency is important lor a lot ol meat eaters. 8o we can
agree on the goal, the method and the input. we agree on all elements that might
settle the issue ol the permissibility ol eating meat.
Ol course, consistency is not the only goal. 8uppose someone claims that sJhe
leels a strong intuition that gay marriage is immoral because it is impure. 8o sJhe
adds this rule to the ethical system, as a dominating principle (it dominates the
principle ol well-being ol gay people). 1he resulting theory is now consistent. Yet,
the problem is that this criterion ol purity is not clear at all. l cannot understand
this principle (how can l detect when something is impure7), thus l am not able to
universalize the principle and test it in other situations. 1herelore, a principle
based on an intuition should be very clear, so that people who do not have that
intuition are still able to understand it and test it in dillerent situations and
thought experiments. ln the second part ol this dissertation, l will present some
principles, such as basic rights, which are based on intuitions. As we will see, l
elaborate on criteria to test whether a basic right is violated. 1he basic right
principle is and should be lormulated as accurately as possible, so that even a
computer might be able to test it. 1his precision allows the principle to be tested
in well-constructed moral dilemmas and thought experiments.
ln the end, the anti-gay person is requested to lormulate his purity principle as
clear as possible, so that we are able to universalize it and test the consistency ol
this theory. lersonally, l doubt whether one can make the system consistent by
adding such a purity principle.
1o conclude, what l want is a coherent ethical system, consisting ol clear and
mutually consistent universalizable principles that best lit our shared, strongest
moral intuitions, without adding too many arbitrary elements. 1hat is the goal: a
theory in coherent rellective equilibrium. lt really rellects the scientilic approach.
ln science, the hypotheses should be lormulated as clearly as possible, in order to
test them in experiments. 1he theory should be as parsimonious as possible, and ol
course internally consistent and consistent with the most reliable observations
and test results.
1here is one big problem however, a problem we seriously have to deal with:
illusions.

tnica| ccnsistcncy anJ Anima| ua|ity



9
chapter ! 1he problem: moral illusions
1he question is whether our input data (moral intuitions) can be trusted. Are they
always reliable7 Lnlortunately, they are not. As in science, we have to admit that
some ol our observations and experiments were unreliable. ln search lor a
coherent rellective equilibrium, we might come to the conclusion that some ol our
input data were wrong, meaning they cannot be reconciled with the rest ol the
constructed system. Alter stumbling upon a contradiction, there are two options.
lirst our intuitive moral judgment might easily change or disappear. 8uch
judgments will be called moral mistakes or deceptions. ln my personal experience,
the judgment that it was allowed to eat meat, was a moral deception. l now have
developed new insights, a new coherent system, and it happened that my previous
judgment that we are allowed to eat meat, was incompatible with this ethical
system. 1hat old judgment completely disappeared, l changed my mind.
1

lut there is another possibility: sometimes our intuitive judgments do not seem
to disappear, although we have to admit that they are not compatible with a
strong coherent system. 1hese moral intuitions will be called moral illusions. 1hey
are analog to optical illusions. 1hose optical illusions are characterized by their so
called cognitive impenetrability (lylyshyn, 1999). lven alter l know that my visual
perception ol a ligure is wrong, the illusion persists. 8imilarly, even alter l realize
that a moral intuition cannot be included in a strong coherent ethical system, that
moral intuition might still persist and 'stay alive'. ll that incompatible moral
intuition is weaker than the coherent ethical system (il we are less willing to give


1
ln the method ol rellective equilibrium, one lirst starts with a liltration process: eliminating the initial
judgments (moral intuitions) which are clearly unreliable, such as the intuitions that arise in conditions
disposed to error (e.g. heavy emotional inlluences, morally irrelevant situational elements that trigger
leelings ol disgust,.). what is lelt is the set ol considered (credible) moral judgments, and this set ol
credible moral intuitions is used in the construction ol a coherent system in rellective equilibrium. 8ee
e.g. lrophy (2009), who compares this liltration process with the scientilic practice ol data selection,
rejecting those data that were gathered in error-disposed conditions. my hypothesis is that the rejected
moral intuitions (those that lack some initial credibility) are examples ol moral mistakes or deceptions.
1he set ol considered moral judgments might still contain moral illusions.
tnica| ccnsistcncy anJ Anima| ua|ity
60
up the whole ethical system), we should declare this outlying intuition to be a
moral illusion.
As optical illusions exist, moral illusions might also exist, and they might have
cognitive impenetrability just as well. ln lact, l will argue in a later chapter that
speciesism (the intuitive, prejudicial judgment that humans have a higher moral
status than non-humans) is a moral illusion. ll speciesism is a kind ol
discrimination but is cognitively impenetrable, this explains why it sometimes is
so dillicult to convince a meat eater that veganism is a moral duty.
moral mistakes are rather easy to deal with, because they will simply disappear.
lut the cognitive impenetrability ol moral illusions is a tougher nut to crack. we
can't just take all ol our intuitions lor granted. lut how do we know whether an
intuition is an illusion or not7 1o clarily this, let's lirst look at how we deal with
optical illusions.
!.1 Optical illusions
A paradigmatic example ol an optical illusion is the mller-lyer illusion (mller-
lyer, 1889). 1his ligure consists ol two horizontal line segments, with inward and
outward pointing arrowheads (see ligure). 1he line segment with the outward
pointing arrowheads appears smaller than the other line segment. ln other words:
our intuition judges the lower horizontal line to be longer than the upper one.

ligure 1: the mller-lyer optical illusion

lut how do we actually know that this intuition is illusory7 Ol course we could
simply reler to an objective reality to claim without lurther justilication that both
lengths are equal. 1he lengths in the mller-lyer ligure are primary qualities, i.e.
lactual, objective properties ol the world. A correspondence theory ol truth states
that our judgment about the lengths is either correct or incorrect, depending on
Vcra| i||usicns
61
whether the lengths in the ligure are objectively equal or not. lut that would
presuppose that we could get direct knowledge ol this objective reality.
lurthermore, l want to study moral illusions, and in ethics there are moral non-
cognitivists who do not believe that moral judgments state true or lalse lacts about
an objective moral realm.
2
ln that case, a simple correspondence theory ol truth
may not be applicable when it comes to moral judgments. lt is better not to reler
to objective realities, because the strategy to lind moral illusions should be
satislying lor non-cognitivists as well. we should be able to study moral illusions
even when a realm ol non-subjective moral lacts and truths does not exist. 1he
approach l suggest is based on coherentism that uses intuitions instead ol
objective lacts as input data to construct a coherent system. A coherence theory ol
truth, instead ol a correspondence theory, should do the job.
1o prove the illusion in the mller-lyer ligure, we lirst have to make all our
most evident intuitions about geometry explicit. 1hen we will see that this one
intuitive judgment about the lengths in the mller-lyer illusion is in contradiction
with two other, stronger intuitions.
1he lirst ol those stronger intuitions says that this ruler does not change its
length when it is shilted in this direction. 1his translation invariance intuition
seems obvious, but it is impossible to give a lurther argument to prove this
!
, so
therelore it is a basic optical (or geometrical) intuition. 1his intuition is expressed
and universalized into the very important geometrical principle ol translation
invariance: all rulers keep their length when shilted in any direction. 1his is a
universalized principle that is true lor all rulers and all translation directions in all
situations. 8o it should apply to the above ligure as well. we accept this principle
as sell-evident, although it is possible to think ol hypothetical worlds or complex
geometrical systems where translation invariance is not valid (mathematicians
already constructed lots ol counter-intuitive geometrical systems: e.g. projective,
non-luclidian or non-commutative geometries).


2
l don't want to delend an intuitionist or naturalist meta-ethical position. 1hese positions are realist-
cognitivist, in the sense that they claim that there exist moral lacts in the world, that we can get access
to (or knowledge ol) this realm ol moral lacts and that some ol these lacts are true. 1he approach that l
delend is constructivist: we construct coherent ethical systems. l leave in the middle whether the
constructed coherent system is a representation ol some objectively existing ethical system in the
world.
!
with lurther argument" l mean an argument based on another loundational principle or intuition. Ol
course, the translation invariance intuition in a particular situation is coherent with similar intuitions
in other particular situations, but this coherence is not what l mean with a lurther argument" lor its
validity.
tnica| ccnsistcncy anJ Anima| ua|ity
62
A second intuition says that the length ol this line segment does not depend on
the presence ol these other lines around. 1his intuition is translated into the very
important universalized principle ol context independence: all line segments have
lengths independent lrom any geometric objects around. 1his, too, is a
universalized principle. 1he arrowheads are the 'context', and the claim is that
lengths are a|ways independent lrom any context (any other thing lloating
around).
1his principle ol context independence is also related to an intuitive aversion
lor aruitrarincss and artijicia|ity.
Aruitrarincss has two aspects: a vertical and a horizontal one. vcrtica|
aruitrarincss says that it is arbitrary to claim that lengths ol line segments are
inlluenced in 'lour-legged' ligures (two arrowheads with two legs each) instead ol
'N-legged' ligures or ligures with other objects than arrowheads (e.g. line
segments with circles instead ol arrowheads). ucrizcnta| aruitrarincss says that it is
arbitrary that, within the class ol lour-legged ligures, the length ol a line segment
decreases when the arrowheads are pointed outwards instead ol inwards. compare
this double arbitrariness with a wardrobe containing vertically arranged drawers.
Vertical arbitrariness claims it is arbitrary to select the third drawer with pants.
lorizontal arbitrariness says that, in this third drawer, it is arbitrary to select the
brown pants instead ol the blue.
ln section 8.4, l will demonstrate the vertical and horizontal arbitrariness ol
speciesism. Also other examples can be given, such as in religious and creationist
beliels. 1he aversion lor arbitrariness is a strong motivation lor atheists. consider
a book: it is unlikely that the book is spontaneously written, because it has a high
inlormation content. 8o it is created by a rational being, an author. creationists
say that this author is also too complex to have been spontaneously generated
(and those creationists do not believe in the third option next to spontaneous
generation and conscious creation: natural evolution). 8o they believe that there
should exist a meta-author, a god who created the author (and the rest ol the
universe as we know it). 1he problem is that most creationists stop at this second
level ol creation: they say that what exists is 'universe-god'. lut ol course, by the
same reason 'universe-god' is also too complex lor a spontaneous generation, so it
should have been created by a metagod. lut 'universe-god-metagod' is again too
complex, so there should be an even higher creator. we end up with an inlinite
chain, a recursive set ol sets ol creators. most religious believers pick the second
lowest level, where a god created creators such as authors ol books. 1his choice
lor the second level is arbitrary, because the evidence lor this level is not stronger
than the evidence lor any higher level (only the evidence lor the lowest level, the
level ol authors, is very strong). 1his is the vertical arbitrariness, which rellects
the lamous creationist problem: who created god7
Vcra| i||usicns
6!
within this second level, there is a lurther horizontal arbitrariness. lor example
a christian creationist does not believe in the many other possible creators and
gods: sJhe does not believe in lrahma, 1hor, }upiter,. As the evidence lor a
christian god is as high as the evidence lor e.g. a lindu god, the choice lor the
christian god is arbitrary. 1he christian believer is in lact atheist about all those
other possible gods. 1he arguments raised by a christian believer towards atheists
bounce back: You are not open lor ood" (neither is sJhe lor lrishna), You
cannot prove that ood does not exist" (neither can sJhe prove that Quetzalcoatl
does not exist), ood reveals limsell once you believe in lim" (but so will Osiris),
lt's arrogant to claim that ood does not exist" (but also arrogant to claim that
2eus does not exist).
Artijicia|ity or ccmp|icacy claims that the inlluence ol e.g. the angles ol the
arrowheads on the length ol a line segment generates a geometrical rule that is
too complicated and larletched. 1his artiliciality introduces a luzzy lactor: what il
we gradually open the angles ol the arrowheads7 low should this inlluence the
lengths ol the horizontal line segments7 8uch mysterious inlluence seems
artilicial.
ln section 8.4, l will argue that speciesism is very artilicial. Again we can make
an analogy with e.g. religious beliels: looking at religious doctrines lrom an
outsider's perspective, they olten look highly complicated and larletched. Atheists
can ask lots ol puzzling questions about religious doctrines, such as: why did ood
put the lorbidden tree ol knowledge in the middle ol the oarden ol lden instead ol
somewhere beyond reach7 why did he put it there in the lirst place7 why the
snake7 why the seduction7 what were ood's intentions7 why does le ask lor
sacrilices7 why sacrilicing lis son7 why does le not stop evil limsell7 why does
le work in such mysterious ways7 low do we know which parts ol the lible
should be interpreted only metaphorically7 And so on and on.

1he two principles ol translation invariance and context independence imply
that we can use instruments: we can use a ruler, or we can use something to cover
or erase the arrowheads. with these instruments, we can clearly demonstrate that
both horizontal lines are equal. Our two universalized principles cohere with each
other. lut they are in contradiction with this one intuitive judgment about the
dillerent lengths ol the horizontal lines.
we now have two options. 1) we can abandon two ol our strongest and
coherent intuitions (translation invariance and context independence) and try to
make a consistent geometrical system without those two principles, in order to
save our intuition that the horizontal line segments are ol dillerent lengths. As
mathematicians olten invent some very exotic geometrical systems, this strategy
is not necessarily impossible. lut everyone would agree that such a procedure to
Vcra| i||usicns
6
we see this picture in 2u, and we have to determine the lengths ol the two thick
horizontal lines ol the bottom stair. 1hese lengths are the target attribute. lut
what do our brains do7 we olten take the stairs, so our brains are used to
computing sizes ol objects in !u. 8izes in !u are therelore easily accessible, and
our brain unconsciously uses them as heuristic attributes to determine sizes in 2u-
pictures. 1hat is how length judgments in 2u-pictures get distorted: the lengths ol
the two thick lines are, in lact, equal. lut in a !u staircase, the lower thick line
would be much shorter, so that's why it appears shorter.
4

A second background theory relers to anthropology. lnterestingly, the muller-
lyer illusion is not inborn. Anthropological studies have shown that the illusion
depends on culture (8egall, 196!, Ahluwalia, 198). ln particular, some indigenous
people (who do not live in an environment with straight edges ol houses, tables
and staircases) are less susceptible to this optical illusion. 8o il they don't see it
and we do, who is right7 1his is another part ol the evidence that we are indeed
dealing with an optical illusion.
1wo principles that cohere with each other, added with background theories
about the psychological mechanism and the cultural relativity, together lorm
quite some evidence to justily the claim that it is an optical illusion. 1he only
counterevidence is that the illusion does not simply disappear alter rellecting
about it.
let's consider a second example ol an optical illusion: the grating induction
illusion (loley & mccourt, 198).

ligure !: the grating induction illusion

1he horizontal grey bar is actually unilormly grey, but the lelt side appears
brighter. we can check translation invariance by taking a piece ol paper with the
same grayness as the lilt side ol the bar, and then shilt the piece ol paper to the


4
lor a more detailed explanation ol this illusion, based on a mechanism ol time-delay, see changizi et
al. (2008).


tnica| ccnsistcncy anJ Anima| ua|ity
66
right side. Our intuition says that the properties ol paper (its grayness) do not
change when the paper is shilted. we can also check context independence by
covering the black-white areas around the grey bar. Our intuitions say that
grayness ol an object does not depend on things around it. As a third argument, we
know that our optical system is also equipped with a mechanism ol lateral
inhibition, in order to increase the contrast and sharpness ol visual stimuli. 1his
mechanism generates the grating induction illusion. 8o we understand where the
illusion comes lrom.
ln summary: two coherent arguments based on strong intuitions ol invariance
and context independence (non-arbitrariness and non-artiliciality), and a
supporting argument that says something about the underlying mechanism in our
optical system, gives us enough evidence to counter a weak perceptual intuition.
can we apply the same method to moral intuitions7
!.2 moral illusions
8o let's turn to ethics now. moral illusions are obstinate but incorrect intuitive
judgments, comparable to the lamous optical illusions. 8ir uavid koss (19!0)
compared our moral convictions or intuitions with sense-perceptions: the lormer
are the basic data ol ethics, just like the latter are the basic data ol natural science.
lut he remarked: }ust as some ol the latter have to be rejected as illusory, so have
some ol the lormer, but as the latter are rejected only when they are in conllict
with other more accurate sense-perceptions, the lormer are rejected only when
they are in conllict with other convictions which stand better the test ol
rellection." (koss, 19!0, p41) ly making this analogy with sensory illusions (e.g.
optical illusions), koss might be one ol the lirst ethicists to point at the existence
ol moral illusions.
8eventy years later, as scientists and philosophers became more and more
interested in the neurobiology ol morality, the notion ol moral illusions and its
analogy with optical illusions reappeared. One philosopher and neuroscientist
made the point explicitly clear: 8am larris stated that, lor example, the dillerence
in moral disapproval between torturing a suspected terrorist (to lind the location
ol a bomb that is about to kill hundreds ol people) and collateral damage in war is
a moral illusion. laradoxically, this equivalence [between using torture and
causing collateral damage] has not made the practice ol torture seem more
acceptable to me [.]. l believe that here we come upon an ethical illusion ol sorts -
analogous to the perceptual illusions that are ol such abiding interest to scientists
Vcra| i||usicns
6
who study the visual pathways in the brain. 1he lull moon appearing on the
horizon is no bigger than the lull moon when it appears overhead, but it looks
bigger, lor reasons that are still obscure to neuroscientists. A ruler held up to the
sky reveals something that we are otherwise incapable ol seeing, even when we
understand that our eyes are deceiving us." (larris, 2004, p198) larris continues
with pointing at a possible psychological bias behind our moral intuitions: ln lact,
there l8 already some scientilic evidence that our ethical intuitions are driven by
considerations ol proximity and emotional salience ol the sort l addressed above.
clearly, these intuitions are lallible. ln the present case, many innocent lives could
well be lost as a result ol our inability to leel a moral equivalence where a moral
equivalence seems to exist. lt may be time to take out our rulers and hold them up
to the sky." (larris, 2004, p198)
1wo questions need to be answered. lirst: do moral illusions exist7 8econd, and
more important, how do we know7 low can we agree whether some moral
intuition is or is not a moral illusion7
1o answer the lirst question, let us again look at the six analogies ol ethics. uo
these other lields ol cognitive activity contain illusions as well7
1) ln modern physics, we encounter contra-intuitive judgments in e.g. quantum
mechanics and relativity theory. we could say that some intuitions about
simultaneity, measurements, particle identity or space-time are illusions.
2) ln mathematics and statistics, too, we encounter erroneous intuitive
judgments such as in the lamous monty-lall problem or the mysterious lanach-
1arski property. 8tudies on heuristics and cognitive biases (lahneman et al., 1982)
show that our intuitive judgments under uncertainty (e.g. in statistics) are not
always reliable.
!) ln the lield ol geometry we have the optical illusions.
4) ln grammar we have an interesting situation. we already mentioned the
apparent analogy between our grammatical laculty and our moral laculty. ll moral
illusions would exist, then grammatical illusions might exist also. And this is
indeed the case. (lhillips et al., 2010) A simple example ol a grammatical illusion is
the sentence: One out ol three children are overweight." According to a lot ol
people, this sentence appears to be grammatically correct at lirst sight. Yet, it is a
violation ol a most simple rule ol subject-verb agreement. 1he lact that people
repeatedly make such errors can indicate that it is an illusion instead ol merely a
mistake. 8uch errors are too persistent to be merely mistakes.
) Our example ol crossword puzzles might perhaps be too rudimentary to have
illusions. (mistakes, however, are olten made in solving crossword puzzles. lut as
we have seen, illusions are more persistent than mistakes.)
6) ln our judgments about taste prelerences, there can also be deceptions (il you
like apples and l give you a piece ol apple, paint it with a brown, odorless, tasteless
tnica| ccnsistcncy anJ Anima| ua|ity
68
color and cut it in the shape ol a sausage, then you might judge it to be bad) as well
as illusions. lor example, psychological biases can inlluence our taste prelerence.
1aste evaluation is inlluenced by what we think we eat and whether that lood
symbolizes values that we support. lor example meat eaters and vegetarians are
susceptible to this kind ol taste illusions. kesearchers (Allen et al., 2008) have given
meat eaters two sausages. 1he participants thought that the lirst sausage was meat
(whereas in reality it was a vegetarian sausage) and that the second sausage was
vegetarian (whereas in reality it contained meat). 1he meat eaters who valued
dominance, hierarchy and social status tended to preler the lirst sausage, because
their taste prelerence is inlluenced by their value scheme. 1aste evaluation is not
simply a matter ol chemistry.
Also in other lields illusions do exist. 1here are auditory illusions (e.g. a pitch
seems to increase indelinitely, ueutsch, 1992), sense illusions (e.g. the contrast
ellect: place your lelt hand lor some minutes in cold water and your right and in
warm water, then touch with your both hands the same object) and many more.
lllusions appear lrequently in dillerent areas, and morality is not likely to be an
exception.
}ust as optical illusions can learn us a lot about how our visual perception
system works, and grammatical illusions can inlorm us a lot about how our
language laculty works, so could we learn a lot about morality by locusing at moral
illusions.
Alter having allirmatively answered the question whether moral illusions
might exist, let's now move to the second, more interesting question: how do we
know whether a moral intuition is an illusion7
One might think that with optical illusions matters are easy, because there is an
objective reality to reler to. with ethics, we do not have such an objective reality.
lowever, in the case ol the optical illusion, we do not need to get a direct access to
an objective reality. ln lact, we can use some non-argued basic starting points or
intuitions, such as translation invariance and context independence. 1he
underlying intuitions behind the principles ol translation invariance and context
independence lack lurther loundational justilication. 1hese are principles which
we have to agree to accept.
8o how to tackle moral illusions7 we in lact already have the answer. we start
lrom moral intuitions, because there is nothing else to start lrom. 1hese intuitive
moral judgments in particular situations have to be expressed in particular ethical
rules. ln the next step, these particular ethical rules have to be universalized to all
other similar situations. Alter lormulating universal ethical principles, we have to
check whether the resulting system has internal consistency. ly testing more and
more situations, i.e. by looking whether our moral intuitions in all situations agree
with the universalized ethical principles, we get a coherent system. ll we arrive at
Vcra| i||usicns
69
a contradiction, i.e. il a moral intuition is incompatible with our universal ethical
principles, we could reline some principle or introduce a new principle that settles
the conllict. ll this strategy really doesn't work, then the only option lelt is that
our moral intuition is wrong. ll this intuition does not disappear, we have lound a
moral illusion.
ln the mller-lyer illusion we had reliable instruments to demonstrate that it is
an illusion: we need a measuring stick or something to cover or erase the small
arrowheads. ln ethics, our reliable instruments are valid arguments based on
universalized principles coming lrom strong intuitions in rellective equilibrium.
8o, valid arguments are the reliable instruments to demonstrate that a moral
intuition is an illusion. (Yet, in the coherence picture, nothing has absolute
reliability. lven the strongest intuitions can be mistaken, even measure sticks
might be untrustworthy, even basic experimental data in science might be
erroneous. 1he strength ol arguments or principles lies in the overall coherence.)
1here is one leature that almost all optical illusions have: the inlluence ol
context. ln the mller-lyer ligure, the small arrowheads are the context. Our
geometric system requires some context independence, because the context is
irrelevant and arbitrary. Yet, the context might inlluence our perceptions and
judgments. 1his might also be the case with moral illusions. we can expect that
moral illusions are to be recognized by their context dependence, arbitrariness,
artiliciality, complicacy or luzziness.
1he idea ol moral illusions sheds a new light on the problem ol moral
disagreement. oiving us tools to demonstrate that a moral intuition is in lact an
illusion will help us to accept a coherent theory ol e.g. animal equality. ln
geometry we had tools to demonstrate that an intuition is an illusion. we now
have similar tools in ethics: strong moral intuitions that can be translated into
coherent universalized ethical principles.
8o the analogy between optical and moral illusions can help us to better detect
and understand moral illusions. 1he strategy lor detecting optical illusions - using
translation invariance, context independence and an underlying optical
mechanism - can be applied to ethics as well, in order to detect moral illusions.
1here might be quite a lot ol examples ol moral illusions: perhaps lutility
thinking and projective grouping (Lnger, 1996, p.100), moral luck (Nelkin, 201!) or
the intransitivity problem (1emkin, 198) are examples ol moral illusions. ln
recent literature, as a spin-oll ol the work ol lahneman & 1versky (1982), the
study ol moral heuristics gained some inlluence (8unstein, 200, 8innott-
Armstrong et al., 2010). As 8unstein argued, in certain situations, moral heuristics
might create erroneous intuitive judgments that we could also consider as a
specilic kind ol moral illusions. ln a later chapter, l will look at a more debated
issue, the prejudicial dillerence in moral status between humans and non-human
tnica| ccnsistcncy anJ Anima| ua|ity
2
A second distinction, that separates situation c lrom situations A and l, is that
in situation c, the action is up-close-and-personal. You have to touch and push the
heavy man yoursell. lut suppose you simply have to push a button lrom a
distance, to overturn the side track and drop the heavy man on the main track. l
expect that even then a majority ol people (more than the 0/ ol people who
responded lavorably in situation l) would respond that action is not permissible.
1his expectation is compatible with a second psychological study on the trolley
dilemmas (waldmann & uieterich, 200).
ln that study, all people on the main and side tracks were actually sitting in
busses. 8ituation c then consists in pushing the bus lrom the side track onto the
main track: you are sitting in a heavy truck that can push the bus, so no personal
contact is required. As the person on the side track is sitting in a bus, not he, but
the bus is used as merely a means. ll the person was not in the bus on the side
track, your plan to block the trolley by using the bus would still work. 8o the death
ol the person in the bus could be considered a side ellect.
As the deontological right is not violated in the bus-trolley dilemmas studied in
waldmann & uieterich (200), we expect that people are more permissive towards
action, and that there is no distinction between situations A and l (because the
only distinguishing lactor was the violation ol the deontological right). 1his is
indeed the case. kespondents could rate the admissibility ol action lrom 1
(delinitely not allowed) to 6 (delinitely allowed). 1he bus-trolley dilemmas A and l
received an average rate ol about 4,8, without a statistically relevant dillerence
between the two dilemmas. 8o action in both situations is allowed to a high
degree. lut situation c received an average rating ol !,, which is relevantly lower
than the other two situations.
As in situation c there is no longer a personal contact with the victim, there is
something else at hand. 1he dillerence between situation c and situations A and l
is that in c the victim (the heavy man on the side track) is sent to the threat (the
trolley), whereas in A and l the threat is directed towards the victim. ln other
words, looking at the causal chain in situation c, the 'locus ol intervention' is at
the person on the side track. 8ending the victim to the trolley means that the path
ol the victim is inlluenced directly. On the other hand, turning the switch in
situation l means that the locus ol intervention is at the threat. 8ending the
trolley to the side track means that the path ol the threat is inlluenced directly
(and the path ol the victim is inlluenced indirectly because the threat will
eventually hit him). 8o the dillerence between situations l and c is the causal path
and the locus ol intervention. uirectly intervening in the path ol the victim in
situation c is a generalization ol the up-close-and-personal element ol pushing a
heavy man lrom the bridge. 1here is no close contact with the victim, but still
there is some directness.
Vcra| i||usicns
!
According to the study ol waldmann & uieterich (200), direct intervention in
the path ol the victim is considered less permissible than intervention in the path
ol the threat, even in the absence ol up-close-and-personal contact. And according
to the study by lauser et al. (2008), 0/ ol respondents said that action in
situation l is not allowed. lrom these lacts, we can expect that a majority ol
people (signilicantly more than 0/) would respond that action in situation c is
not allowed even in the absence ol up-close-and-personal contact. with close
contact, 90/ ol respondents said that action is not allowed in situation c.
Note that the results in both studies are not contradictory either, because in
waldmann & uieterich (200) the situations contained no violations ol
deontological rights, and in lauser et al. (2008) there was a very close contact with
the victim in situation c.
ln summary, we have two distinctions between the three trolley dilemmas.
1) Action in situation A implies no violation ol the deontological right not to be
used as merely a means. ln l and c this right is violated. 1his distinguished A lrom
l and c.
2) Action in situation c implies that the victim is sent to the threat. ln A and l,
the threat is sent to the victim. 1his distinguishes c lrom A and l.
1he lirst principle says that action is less permissible il a deontological right not
to be used as merely a means is violated. 1he second principle says that action is
less permissible il the victim is sent to the threat. we have seen that these
principles correspond with two psychological studies on the trolley dilemma. ln
situation A, action means that the deontological right is not violated and the
victim is not sent to the threat, so it is strongly permissible (as 90/ ol people say).
ln situation l, action still means that the victim is not sent to the threat, but the
deontological right gets violated. 8o this is much less permissible (only 0/ ol
respondents say it is permissible). ln situation c, the victim is used as merely a
means and is sent to the threat. 8o in this case, the situation is strongly
impermissible (as a majority says).
1he question we now have to ask is: are the moral intuitions behind these two
principles illusions7 l will demonstrate that the second principle is based on a
moral illusion, but the lirst one is not. 1o show that the second principle is a moral
illusion (as was hinted at by leter Lnger, 1996), l present (just as with the optical
illusions) two arguments based on strong moral intuitions, and one auxiliary
argument based on knowledge ol the underlying psychological mechanism.
1he lirst argument is based on a kind ol translation invariance. 8o we have to
see what remains constant as we shilt lrom the lirst to the second and the third
situation. ln the mller-lyer optical illusion, we have seen that what remains
constant when shilted lrom the upper to the lower part ol the ligure, is an
intrinsic property ol a line segment: the length. 1he length ol a ruler does not
Vcra| i||usicns

According to Lnger (1996, p101), this dillerence between sending a victim to a
threat and a threat to a victim is based on what he called 'protophysics', in
violation ol relativity theory. 1his demonstrates that the dillerence between
situation l and c is simply something contextual, as the context determines the
spatial lrame ol relerence. lut ol course, this argument is just as well based on
(strong) intuitions about what is the context and what is morally (ir)relevant. 1he
position ol the tracks is not morally relevant. 1he only relevant thing is the
relative position ol the threat and the victim, because this determines the
collision.
1he third, auxiliary argument is based on a psychological mechanism. 1his
mechanism is in lact clearly explained by waldmann and uieterich (200). 1heir
concept ol 'intervention myopia' already indicates that we are dealing with
something that is not lunctioning properly. when the locus ol intervention is at
the threat, as in situations A and l, our attentional spotlight is at the threat, and
all people (on the main and the side track), are background. 8o lrom the
perspective ol the threat, the persons are all equal, and it is easier to make
consequentialist calculations. lut il the locus ol intervention is at the victim (the
heavy man), the locus ol attention is at the victim. 1he other three people on the
main track are now part ol the background. And the myopia indicates that it is
dillicult to take those people in the background lully into account. uue to the
myopia, we don't see their moral status so clearly. 1his distorts consequentialist
reasoning. ln situation c, people tend to locus on the late ol the victim, neglecting
the death ol the three people on the main track. 1his locus on the one victim (the
locus ol intervention), results in a neglect ol other people located in the
background. As these three people in the background appear to be absent (lar
away in the causal chain), it appears that the heavy man dies in vain when pushed
lrom the bridge. citing waldmann & uieterich: ln sum, the general hypothesis is
that people tend to locus on the causal paths ol agents [threats] or patients
[victims] targeted by an intervention, and neglect other causal processes
occurring outside this locus, in the background.[.] we are not saying that in cases
ol intervention myopia, people are completely blind to the victims in the
background (i.e., the death ol the three people), rather, we are saying that because
ol an attentional locus on the ellects ol interventions, people who are evaluating
the morality ol options may give victims in the background less weight than
victims in the attentional spotlight." (waldmann & uieterich, 200, p249)
with the above arguments, we can conclude that the dillerence between
sending a threat to a victim and sending a victim to a threat is a moral illusion. 1he
moral relevance ol this dillerence was already criticized by lischer (1992) and
lischer & kavizza (1994), but now we have a clearer view and we can call it a moral
tnica| ccnsistcncy anJ Anima| ua|ity
6
illusion. whether the deontological right not to be used as merely means is also a
moral illusion, remains to be seen.
!.4 ls the deontological right a moral illusion7
As we have seen in the discussion ol the trolley dilemmas, there is a possible
explanation that separates situation A lrom situations l and c. ln situation A, a
deontological right ol the victim is not violated. ln situations l and c, the victim is
used as merely a means, as a trolley blocker or a human shield. 1he presence ol
this victim is required in order to save the other people on the main track.
As lar as l know, the coherentist approach does not yet imply that this
deontological right is a moral illusion. On the one hand, the above mentioned
intervention myopia seems to suggest that the deontological right is an illusion.
lut on the other hand, there seem to be some arguments that suggest that the
deontological right is not an illusion.
lirst, the deontological right respects a translation invariance: all persons keep
the same deontological right in any ol the trolley dilemmas, because the
deontological right is related to the moral status ol the individual, and this moral
status is invariant (it remains the same when shilting between dillerent
dilemmas).
8econd, there seems to be a context independence. lt can be said that all people
can claim this right, independent lrom the situation. 8o it has some intrinsic
(context independent) character.
6

1hird, and perhaps most importantly, there are hundreds ol other moral
dilemmas and situations where intuitive judgments ol most people are coherent
with the deontological right principle. we are not allowed to sacrilice an innocent
person against his will to use his organs to save live patients in the hospital who
need new organs in order to live. Neither are the lollowing actions allowed:
involuntary experiments, terror bombing (killing innocent civilians in order to
demoralize the enemy), torturing a suspected terrorist (to gain inlormation on the
location ol a bomb that is going to blow up a school), killing under blackmail (a


6
lowever, this right relers to the use as merely a means to someone else's ends, so it relers to the
presence ol someone else in the environment. lt is unclear whether this introduces a context
dependence and whether this kind ol context dependence creates an illusion.
Vcra| i||usicns

terrorist says that il you kill an innocent person, he will not kill his live hostages),
trallicking (buying and selling humans), raping women (and selling the video to
thousands ol male consumers), gladiator lights (entertaining thousands ol
spectators), human exhibitions (ethnographic zoos), cannibalism or slavery. ln all
these cases, the presence ol the victim is required in order to benelit others, so the
victim is used as merely a means. 1he above actions are not allowed, even when
consequentialist considerations might support those sacrilices and deontological
rights violations.
8o we have more than ten dilemmas and situations where the deontological
right might be violated: trolley-bridge, organ transplants, experiments, terror
bombing, torture, blackmail killing, trallicking, rape, cannibalism, gladiator lights,
and slavery. ln all these dillerent situations, most people have the very coherent
intuition that the deontological right should be protected against consequentialist
considerations. ll the deontological right not to be used as merely a means trumps
the right to live, then we are not allowed to use people as e.g. trolley blockers or
inlormation sources, even il this means that other people will die.
As the deontological right can be related to the moral status ol an individual
independent lrom the situation, and as there are a lot ol very dillerent situations
where most people's moral intuitions are coherent and compatible with the
principle ol the deontological right, l am tempted to believe that this right is not
(yet) a moral illusion.
1he only exception to this deontological right intuition known to me is the loop
trolley dilemma. 1he loop dilemma (situation l) is olten mentioned as a counter-
example to the moral relevance ol the deontological right (e.g. 8inger, 200). ln the
loop dilemma, the deontological right is violated, yet, the action is deemed
permissible by a lot ol people (lauser et al. 2008). consequentialists can reler to
this intuition in this particular dilemma to argue that the deontological right is a
moral illusion. lut lrom the above discussion, we learn something interesting: the
permissibility in the loop dilemma might be a moral illusion, and il this is the case,
then the deontological right might still be valid. 8o the existence ol moral illusions
might 'save' the deontological right.
Now, il the deontological right is not an illusion, what happens with the above
psychological explanation ol intervention myopia7 ln short, we might
hypothetically say that intervention myopia does not generate the illusion in the
bridge dilemma, but 'instrumentalisation myopia' generates an illusion in the loop
dilemma. 1he structure ol the loop dilemma is such that we don't see the
instrumentalisation (the use as merely a means) ol the person on the side track,
because that person seems as lar away as the non-instrumentalised person on the
side track in dilemma A. ln other words: the person on the side track is too lar
away lor us to see his instrumentalisation.
tnica| ccnsistcncy anJ Anima| ua|ity
8
ln terms ol heuristics, the attribute substitution might (very hypothetically)
work as lollows. 1he target attribute is in this case the deontological right. As it is
not always easy to quickly detect violations ol this right, our brains might use a
heuristic attribute instead. Note that, il the deontological right gets violated, it
means that the presence ol the victim is required in order to save other people.
1he required presence is the target attribute, and it requires a sometimes
computationally dillicult counter-lactual thought experiment to determine
whether someone's presence is required. lut when presence ol the victim is
required, this likely means that the locus ol the action (the locus ol intervention)
will be on the victim. ln other words, the locus ol intervention might be a good
heuristic attribute lor detecting deontological rights violations. 1herelore, our
brains look lor this heuristic attribute, which is reliable in most cases, but mislires
in the loop dilemma. ln the loop dilemma, the locus ol intervention was not on the
victim, and hence our brains think erroneously that the deontological right is not
violated.
1he above is still very hypothetical, because one might also say that the true
target attribute is the maximization ol well-being or lives saved, as
consequentialists would have it. 8o the issue remains open: is the deontological
right (the intuition that inhibits action in the bridge dilemma) a moral illusion in a
consequentialist ethical system, or is the loop dilemma intuition (that action is
permissible) a moral illusion in a deontological system7
whether the deontological right not to be used as merely a means is also a
moral illusion, remains to be seen. As the deontological right, due to its intrinsic
and context independent character, can be related to the moral status ol an
individual independent lrom the situation, and as intuitions in a lot ol dillerent
situations are coherent with the principle ol the deontological right, l am tempted
to believe that it is not a moral illusion. 1his deontological right will be discussed
in more detail in the later section on the basic right (section 6.2).
!. leuristics in thought experiments
8unstein (200) criticized the above method ol coherentism (rellective
equilibrium) that uses philosophical thought experiments (moral dilemmas like
the trolley problem). lis claim is that our intuitions in those 'exotic' (lar-lrom-
ordinary-reality) thought experiments are not reliable anyway, because exotic
situations are olten situations where heuristics mislire. leuristics are common
sense rules ol thumb that work well in most casual situations that we encounter in
Vcra| i||usicns
9
our daily lives, because in these situations we are trained to make quick and
accurate judgments. lut in constructing exotic thought experiments with trolleys,
we create situations where the heuristic does not yield reliable results, i.e. where
the heuristic attribute strongly deviates lrom the target attribute. At least that is
what 8unstein claims.
Yet, l would not throw away the method ol thought experiments. l claim that
those thought experiments do have some value: they trigger our moral intuitions
and interrogate our morality, just as experiments in physics allow us to
interrogate nature. l'd like to delend the coherentist-universalist approach
(deriving intuitions lrom thought experiments and translating them in coherent
universalized ethical principles), by making the analogy with physics. lor
example, in our daily lives, we olten see that heavier objects lall at higher speeds.
Yet, the intuition that heavier objects have higher accelerations deviates lrom the
real law ol gravity. 8o in order to derive the laws ol gravity, we have to set up
exotic situations, e.g. by dropping dillerent objects in a vacuum (e.g. on the moon).
1hose exotic experiments are controllable, they eliminate specilic contextual
variables (like air dralt) so they are better suited to interrogate nature to lind its
most lundamental laws. And the same goes lor ethics, by interrogating our moral
brains, using exotic thought experiments, we can derive the lundamental laws, the
basic lorces ol our ethics.
Yet, even in the exotic experiments, we cannot trust everything. ln physics, to
derive the acceleration law ol gravity, we have to use clocks. lut as linstein
demonstrated in his theory ol general relativity: clocks measure dillerent times
depending on their positions in a gravitational lield. 8o clocks can move laster or
slower. lence, even in exotic experiments, we cannot completely trust our
instruments. ln our daily lives, clocks are reliable to measure time, so we use
clocks as heuristics. 1he value on the clock is a heuristic attribute, the real time is
the target attribute. Now, in some exotic situations, with strong gravitational
lields, the heuristic does not measure real time anymore. consider clocks in ol8
satellites. A clock in a ol8 satellite experiences a weaker gravitational lield than
clocks on earth, so a satellite clock runs at a slightly higher speed than a similar
clock on earth. we can use the clock on earth as a heuristic to measure time
evolution in the universe. lut this heuristic mislires (a tiny bit) when we want to
measure time evolution in the satellite. we know, thanks to linstein, that clocks
on earth might not be reliable heuristics in those exotic situations. 1he only way
to discover this heuristic misliring is by setting up other exotic experiments, and
deriving a coherent lramework about how time evolves at dillerent places in a
gravitational lield. 1hat coherent lramework was derived by linstein based on
(thought) experiments.
tnica| ccnsistcncy anJ Anima| ua|ity
80
8o, indeed, we cannot always rely on moral heuristics in exotic situations.
lndeed, exotic situations might sometimes be exactly those situations where
heuristics mislire. lut using more and more thought experiments, we can see
when our heuristics might mislire. 1hey mislire when they are not consistent with
a coherent ethical theory ol universalized ethical principles.
1here is another, slightly related issue. we construct thought experiments by
erasing a lot ol variables. like the trolley dilemma, those experiments are
characterized by their low number ol variables. l.g. we don't consider
interpersonal relationships, probabilities or people's virtues in the trolley
dilemma. 1his helps us to look lor those elements that are crucial in our study.
lowever, one critique can be that they neglect the specilicities ol contexts and
particular, real situations. 8ome complexity and situation-dependence is missing.
lere we can use another analogy with physics. One branch ol physics is looking
lor the most lundamental laws: high energy physics (elementary particle physics
and cosmology). Other branches study other physical contexts: lor example lluid
mechanics or statistical thermodynamics. Although a lluid is composed ol
elementary particles subject to the lundamental lorces, lluid mechanics studies
other, emergent laws and principles.
ln this dissertation, l intend to derive lundamental laws ol ethics, the basic
'moral lorces', so it is comparable with elementary particle physics. l use thought
experiments in ethics, just as particle physicists perlorm exotic experiments using
e.g. particle accelerators. contextualist or situationist ethics, on the other hand, is
most suitable lor complex real lile issues. 1hese ethics might be comparable to e.g.
lluid mechanics or thermodynamics. ln these contextualist ethics, new rules or
principles might emerge, that strongly depend on the specilic context (e.g.
complex relationships or cultural inlluences). 1he study ol elementary particles is
not incompatible with the study ol lluid mechanics. And lor the same reason, a
principle-based, universalist ethics is not incompatible with more contextualist
ethics, as long as the laws ol the latter are emergent lrom the laws ol the lormer.
uillerent contexts (cultures, relationships,.) might require dillerent emergent
rules, just as lluids, gases and solid states have dillerent emergent properties.
tnica| ccnsistcncy anJ Anima| ua|ity
84
deceptions and illusions, even in the area ol taste prelerences. 8o rational
arguments are possible, at least to some degree.
lour recent developments indicate that the time is ripe to tackle moral
illusions. lirst, a lew philosophers started comparing erroneous moral judgments
with optical illusions (larris, 2004). 8econd, there is the study ol 'moral heuristics'
(8unstein, 200, 8innott-Armstrong et al., 2010) and the discovery that attribute
substitution is also the mechanism behind many optical illusions (lahneman,
200!). 1hird, there is the study ol 'grammatical illusions' (lhillips et al., 2010)
combined with the linguistic analogy between our moral laculty and our language
laculty (lauser et al. 2008). linally, there is the (neuro)psychological research ol
the trolley problem (oreene, 2002, 2008), with the hypothesis ol intervention
myopia in moral intuitions (waldmann et al. 200). 1he analogy between optical,
grammatical and moral illusions can help us to better detect and understand
moral illusions.
ln a later chapter, l want to locus at a specilic moral judgment that meat eaters
have: the prejudicial dillerence in moral status between humans and animals
(non-human sentient beings). 1his is to most people an intuitive judgment that
lies behind all our uses ol animals, lrom lactory larms to pet shops. can this
intuition be a persistent moral illusion7 l will demonstrate that this speciesist
judgment is in lact a moral illusion, just like l demonstrated that the mller-lyer
ligure is an optical illusion and the trolley dilemmas contained a moral illusion.
ln particular, the set ol universalized ethical principles that l will discuss
contain two principles based on strong moral intuitions ol invariance and context
independence. And as an auxiliary argument, l will explain the psychological
mechanism behind the moral illusion ol speciesism.
As the purpose ol this dissertation is to investigate the consistency ol an ethic
ol animal equality, the most important thing is that both animal rights ethicists
and meat eaters should agree on the approach: they should both agree on the key
ingredients, the moral intuitions as input data, universalization as the method,
and coherence or consistency as the goal. ll we agree upon an approach on how to
derive a coherent ethical system, we have set the rules ol the game. ln the next
chapter, it is time to look at what such a system ol ethical equality might look like,
what ethical principles ol equality it might contain. we leave the area ol meta-
ethics and enter the realm ol normative ethics.


tnica| ccnsistcncy anJ Anima| ua|ity
8
lart 2 1heories ol equality

tnica| ccnsistcncy anJ Anima| ua|ity
86


ln order to understand the ethics ol animal equality, we lirst need to have a very
clear picture ol the notion ol equality. lart 2 ol this dissertation is dedicated to a
derivation ol three dillerent, material principles ol equality. 1hese principles are
derived lrom normative ethical systems (contractualism, consequentialism, ethics
ol care, virtue ethics and deontological ethics). when applying those three
principles ol equality to sentient beings (animals), we will get a nuanced and clear
picture ol the theory ol animal equality. 1his move lrom normative ethics to the
applied ethics ol animal equality will be discussed in part !.


8
chapter 4 lmpartiality and prioritarian
equality
4.1 contractarianism universalized
let's start lrom the position ol a rational egoist. lt would not be wise to simply
pursue your own benelits all the time, because you might get into trouble with
other people. ln particular, it might be better to come to an agreement with those
other people, not to harm each other and to help each other in need. 1his is
especially true in 'collective action problems': situations wherein dillerent people
would all benelit lrom collaboration with each other, but there is a tendency to
cheat, because the cheater would be even better oll. lowever, il everyone started
cheating instead ol collaborating, then all will be worse oll.
collective action problems are omnipresent (the prisoner's dilemma is a lamous
example where two suspects have to decide whether to collaborate with each
other or not, see section .2). As an example, consider picnicking on the beach. ll
everyone throws away their waste on the beach, no-one will be able to enjoy a
beautilul clean beach. 8o it is better il everyone collaborates by throwing their
garbage in the dustbin. lut lor you, it would be even better il everyone else
collaborates, except you: you can enjoy a clean beach but you would not have to
make the ellort to go to the dustbin. Your best option is to throw away your
garbage on the beach. lut il everyone started thinking like that, we end up with no
collaboration and a dirty beach.
1he political philosopher 1homas lobbes (188-169) based his social contract
theory on the assumption that people tend to be rationally sellish. 1his
contractarianism can solve collective action problems: everyone is bound by a
social contract to collaborate, help each other and not harm each other. kationally
sellish people can benelit lrom mutual advantages ol cooperation. lut this
contract only applies to people having two conditions: 1) an equal power position
tnica| ccnsistcncy anJ Anima| ua|ity
88
(e.g. equal bargaining power) and 2) rationality. 1he equal power position simply
comes lrom the lact that a rational egoist does not have to maintain a social
contract with persons in a weaker position (who have no power to harm or help
you or have protectors who can harm or help you), because it would be easier to
simply exploit them. 1he rationality condition is obvious: we can only make
agreements with people with whom we can negotiate, so those people have to be
able to negotiate.
ln this contractarianism ol the egoist, the moral community (the set ol moral
patients, those beings who are given moral status) exists ol all rational people who
have roughly equal power. 1he latter condition however does not correspond with
our moral intuitions. what about those in the weaker positions7 what about
impartiality7 }ohn kawls (191) developed a contractualist theory by deleting this
equal power condition. ln his theory, the moral community is extended to all
reasonable and rational beings (in particular all beings who have a sense ol justice
and a conception ol the good). 1hese are beings who are able to perlorm a thought
experiment, which goes under the name ol the veil ol ignorance. 8uppose we are
'impartial observers' sitting behind a veil ol ignorance: we have to imagine we will
be born as someone on earth, but in order to guarantee impartiality, we don't
know yet who we will be. we could be gay, disabled, lemale, black,. 1he only thing
we can do is derive moralJpolitical laws (respecting known physical laws) that we
would preler in an ideal world where we don't know who we will be.
According to kawls, those laws will only be applicable to rational beings able to
do the thought experiment. As 'impartial observers' behind a veil ol ignorance,
those rational beings need to have a sense ol justice (impartiality) and a rational
will. 1he problem is that this excludes mentally disabled people, babies with a
short lilespan and non-human animals. we intuitively see that those beings have a
moral status as well: we are not allowed to use mentally disabled persons as
merely means, harm animals without good reasons or torture dogs. 1he condition
that only rational beings belong to the moral community is still too partial and it
violates moral emotions such as empathy. lt is a bit arbitrary to give the veil ol
ignorance some hall thickness: we don't know who we will be, but we know we will
be rational agents.
As kawls proposed an extension ol a lobbesian contractarianism by deleting
the equal power condition, l propose a second extension, by deleting the
rationality condition. 1his is the most extreme universalization ol
contractarianism. we have to make the veil ol ignorance as thick as it possibly can
be. No criteria are lelt out: we include all physical entities (in the broadest sense)
in the thought experiment. lence, we could become an electron, a planet, a car, a
computer program, an ink stain, a tree, a pig, a person in the year !000, or
whatever. we can now decide what the moralJpolitical laws should look like. we
mpartia|ity anJ pricritarianism
89
remark that electrons, trees, stains and other objects are likely not sentient beings.
1hat means that no-one can inlluence the well-being ol a non-sentient being. 8o il
you were a non-sentient object, it doesn't matter to you what happens to you. You
would not be aware ol anything, you would not like or dislike anything, you would
not want anything, you would not experience harm, you would not have
prelerences, you would not have interests that you care about, your well-being
would remain constant at level zero (i.e. it is absent).
1he thickening ol the veil ol ignorance and the extension ol contractarianism
to all sentient beings are no new ideas, but were already discussed by Van de Veer
(199), kowlands (1998), Nussbaum, (2006) and Van den lerg (2011). According to
kowlands, who we are (human, pig,.) is just a matter ol luck. we did not have a
responsibility or choice in this, so we should not be rewarded lor being a human.
As being human is beyond the control ol an individual, it should be judged morally
arbitrary.
lt is clear that taking this thickest veil ol ignorance is the most impartial and
least arbitrary thing to do. we automatically come to the criterion ol sentience,
because sentient beings are the only beings with well-being, and well-being is the
only thing that really matters to us behind the veil ol ignorance. lut what do we
mean with well-being and value ol lile7
4.2 lrom leelings and well-being to the value ol lile
1he central quantity in our theory ol justice is the notion ol 'value ol lile' or
'liletime well-being'. 1his is the value that - behind a veil ol ignorance - an
impartial observer would ascribe to the complete lile ol a sentient being (i.e. a
being that experiences a well-being). 1he value ol lile or liletime well-being is a
lunction ol all momentaneous experiences ol well-being during a complete lile ol
an individual. let us analyze this in more detail.
4.2.1 Allective qualia: lrom experienced leelings to experienced
pleasure
lxperienced leelings are subjective, private, direct, conscious, qualitative
experiences, phenomenological sensations or qualia (lyrne, 2010). 1hese qualia
are internal representations that have an attention or locus. lor example l
consciously leel this book because l can pay attention to the sensation generated
tnica| ccnsistcncy anJ Anima| ua|ity
90
through my lingertips (lor the connection between qualia and attention, see
kamachandran and lubbard, 2001). }ust belore l paid attention to this leeling ol
touch, l was not aware ol it. 1here was an unconscious neural activity, and only
alter l locused on my lingertips, it became a conscious experience or 'quale' ol
touch. Qualia are olten neutral: l don't leel an urge to avoid touching books. when
qualia become allective in nature, i.e. when they are evaluated as being positive or
negative (when they generate a positive or negative attitude in the individual
holding the qualia), they become positive or negative leelings, i.e. pleasure and
pain. A needle in my linger generates a quale that l wish to avoid. 1his quale is
called pain and it generates an urge in me to withdraw.
4.2.2 1he importance ol prelerences: lrom experienced pleasure to
momentaneous well-being
well-being experienced at a specilic moment should be distinguished lrom mere
pleasure. l deline momentaneous well-being as the composition ol all the positive
(minus negative) leelings and emotions that are the consequence ol
(dis)satislaction ol prelerences (the things that one wants)
1
. 1his is an important
delinition. lts lormulation in terms ol leelings and prelerences combines a mental
state account (having mental states such as leelings ol pleasure) and a prelerence
satislaction account ol well-being (see 8haw 1999, chapter 2).
2

1here is a connection between leelings (ol pleasure) and prelerences (or needs):
leelings are nothing but indicators to see when something is met or unmet. 1his


1
1hese prelerences can include some unconscious prelerences, in particular dispositional and
instrumental prelerences. lor example, during my sleep, l have unconscious prelerences. 1hose
prelerences have the disposition to become conscious when l wake up and think about the prelerences.
An embryo on the other hand, is unconscious but does not (yet) have such dispositional prelerences. An
example ol an unconscious instrumental prelerence is the prelerence that an animal has in staying
alive, even il the animal does not have a notion ol lile, death and its own luture. 8taying alive is
important il the animal wants to satisly other, conscious prelerences. 8ee Visak, 2011, p8.
2
One could restrict the prelerences to well-inlormed, rational prelerences in order to avoid
prelerences that are actually bad lor us (lor example a prelerence lor drugs or a prelerence to marry
someone whom l erroneously believe to be my perlect match). lut this restriction might not be
necessary in my account ol well-being, because the delinition ol well-being not only contains the
condition that prelerences be satislied, but that those prelerence satislactions result in an increase ol
positive leelings (or a decrease ol negative leelings). 1he satislaction ol misinlormed, irrational
prelerences would not generate more positive leelings.
mpartia|ity anJ pricritarianism
91
'something' is a prelerence or need.
!
According to the psychology ol maslow
(194!), prelerences contain not only physical lunctioning (lood, water, movement,
rest, health, salety,.) but also lor some individuals: social needs (connection,
compassion, acceptance, warmth, contribution,.), play (joy, humor,.), autonomy
(lreedom, space, independence, spontaneity,.), honesty (authenticity, integrity,
trust.), peace (equality, harmony, order, beauty,.) and meaning (learning,
growth, challenge, elliciency, clarity, creativity, purpose,.).
Needs can have dillerent intensities (e.g. a little hunger vs. being very hungry)
and satislactions can also have dillerent levels (e.g. having access to a little bit vs. a
lot ol lood). 1he higher the level ol satislaction and the higher the intensity
(subjective importance) ol the corresponding need, the stronger the positive
leeling and the higher the momentaneous well-being will be.
As momentaneous well-being does not look merely at positive and negative
leelings such as pleasure and pain, but is restricted to those leelings that are the
consequence ol prelerence (dis)satislaction, we avoid a hedonist position (a
mental state account) that only looks at pain and pleasure. 1he hedonist
encounters the problems ol the 'experience machine' (Nozick, 194). 8uppose we
have an experience machine that can give you leelings ol pleasure lor the rest ol
your lile, by plugging your brains into this machine. lowever, the experiences in
this machine are related to a world that is not real, and you might have a strong
need lor authenticity (or connection with reality) that will not be satislied by this
machine. 1he positive leelings generated by the machine are not the consequence
ol prelerence satislaction, so they do not contribute to well-being as l have delined
it. 1hat is why a lot ol people will be reluctant to step into this machine.
1he veil ol ignorance helps to explain why merely leelings ol pleasure are not
sullicient in an account ol well-being. lrom behind the veil, you know you will be
someone who does not preler to live a lile in an experience machine. lence, this
means that a need will not be satislied and your value ol lile will be lower. 1o take
another example: suppose behind a veil you can decide between two situations. ln
the lirst situation, you will experience pleasure with your lover, and your lover is


!
8ometimes leelings are unreliable in measuring a need. lor example a mallunctioning amygdala
might generate an irrational lear, i.e. lear when there is no danger and the need lor salety is met. 1he
irrational lear does not correspond with an unmet need lor salety, so one might think that according to
the delinition ol well-being that l proposed, those negative leelings ol irrational lear do not negatively
contribute to well-being. 1he same goes lor pains in phantom limbs: those pains do not correspond
with an unmet need lor bodily integrity, because the body part is absent. lowever, il the patient with
the mallunctioning amygdala does not want to leel this irrational lear, sJhe has an unmet need, i.e. a
prelerence lor inner peace. lence, the irrational lear and the phantom pain do lower someone's well-
being according to the delinition.
tnica| ccnsistcncy anJ Anima| ua|ity
92
laithlul. ln the second situation, you will experience as much pleasure as in the
lirst situation, but your lover is unlaithlul and you will never know this (you
believe your lover is loyal, and being loyal is very important to you). lrom a pure
mental state account, both situations would be equally prelerable, because the
happiness is equal. lowever, il l were an impartial observer behind a veil, l would
preler the lirst situation. 1hat prelerence rellects a need to be in contact with
reality or with the truth, and it implies that merely leelings ol pleasure are not
sullicient in the notion ol value ol lile. ll an impartial observer behind the veil
prelers a situation where sJhe will experience a level ol pleasure to a situation
where sJhe will experience the same level ol pleasure in a virtual world ol an
experience machine, the value ol lile cannot merely depend on a mental state ol
pleasure. 8omething else is important, and that something points to a prelerence
or need lor e.g. the truth.
4

4.2.! 1he problem ol interpersonal comparability: lrom individual
well-being to comparable momentaneous well-being
An individual can measure its momentaneous well-being and the strength ol its
leelings and prelerences.

1he big problem is the comparability between dillerent


individuals
6
. leelings are qualia, and hence they are private: they cannot be
objectively measured or communicated. Asking whether the well-being ol person i


4
Another aspect that comes into play here, is the principle ol rule universalism (see 1.2). 8uppose l lie
to you by saying that X is the case whereas in reality X is not the case. And suppose that telling the
truth (that X is not the case) would decrease your positive and increase your negative leelings about X.
ll you will never know that l lied, you will have the same positive leelings about X as when X was really
the case. lowever, il l am allowed to lie in this situation, then rule universalism implies that you know
that everyone in a similar situation as me is allowed to lie. ln that case you still do not know that l
actually lied, but you do know that l think it is permissible to lie. 1his might give you an uncomlortable
leeling ol insecurity, because you have a need lor trust that is not met. As a consequence, you do not
want a rule that permits everyone to lie in similar situations. lence, l am not allowed to lie, even il a
well concealed lie does not inlluence your positive leelings about X. ln other words: a mental state
account ol well-being combined with rule universalism can avoid some counter-intuitive implications
ol a simple mental state account that is not combined with rule universalism.

ll choices A and l are incomparable lor an individual, i.e. il that individual is psychologically unable to
estimate whether choice A gives himJher a higher well-being than choice l (e.g. getting a weak
emotion ol long duration versus a dillerent emotion which is intense but briel), an impartial observer is
permitted to make an own estimate. ll the impartial observer would choose A, then sJhe can make that
choice lor the individual.
6
1he literature on interpersonal comparison ol well-being is vast. 8ee e.g. llster & koemer (1991),
lammond (196), larsanyi (19).
mpartia|ity anJ pricritarianism
9!
in situation X is equal to the well-being ol person j in situation X, is like asking
whether my perception ol red is the same as your perception ol red. A theory
without interpersonal comparison ol well-being has a very serious counter-
intuitive implication: dillerent lareto optimal situations cannot be mutually
compared. A lareto optimal situation is a situation in which it is impossible to
make any one better oll without making at least one individual worse oll. consider
a huge income inequality: il the income ol the poorest cannot be improved
without lowering the position ol the richer person (and il the income ol the richer
person cannot increase without a cost lor the poorest), we have a lareto optimal
situation that allows a high inequality.
well-being dillers lrom income in the sense that it cannot be interpersonally
compared. One person can compare his own levels ol well-being (as he can
compare his perception ol red with his perception ol green), so the best we can get
is a lareto optimal situation ol well-being: even il we cannot compare the well-
being levels between the persons, it is possible to know that we cannot increase
someone's well-being without lowering the well-being ol someone else. As lareto-
optimality still allows lor serious inequalities, a theory ol equality needs to go
beyond this lareto criterion. without interpersonal comparability, we would not
be able to compare lor example the harm ol death ol person A with the harm ol a
mere pinprick ol person l. we need an interpersonal comparability ol well-being
il we want to avoid such counter-intuitive implications.


A lirst step to move lurther beyond merely lareto elliciency requires a small
deviation into some metaphysics (something outside the positive sciences, because
well-being cannot be measured lrom the outside, just as someone's perception ol
red cannot be measured lrom the outside). we have to postulate an ideal observer
who has an impartial, lully inlormed point ol view. ldeally, this person has
experienced almost anything that anyone can experience, having all kinds ol
prelerences that anyone can have, and sJhe has a perlect, unbiased memory to
compare the levels ol well-being during those experiences, having those
prelerences.
1o avoid too much god-like metaphysics in ethics, we quickly have to move
more down to earth. we can try to approach the perspective ol this ideal observer,
when we use as much ol our empathy as we can. 1his is where the veil ol
ignorance comes into play: we imagine ourselves in the positions ol other beings,

lurthermore, Arrow (196!) demonstrated some impossibility theorems that occur when well-being is
not interpersonally comparable. 8ee also koemer (1996).
tnica| ccnsistcncy anJ Anima| ua|ity
94
using our empathy. we do not know the well-being ol someone else, but we can
measure our empathic well-being: our estimate ol the well-being ol the other.
8

Lsing this empathy, we can see that someone's potential maximum well-being
can be higher, the more needs that being has. ln other words, a being with more
needs can reach a higher well-being (compared to a being with only a lew needs) il
all ol his or her needs are satislied. 1he potential minimum well-being can also be
lower when a being has more needs and when all those needs are not met. As a
simplilied example, suppose we have a being with only one need. 1he
momentaneous well-being arising lrom that need can vary lrom e.g. (needs lar
lrom being satislied, so this being rather prelers to die than to experience this
negative leeling), to 0 (needs satislied to some extent, so that lor this being it
doesn't matter il sJhe lives or dies), to -1 (needs highly satislied). A being with two
needs (both ol the same intensity), however, can have a well-being ranging lrom
to -2, il the contributions ol the individual needs can be added. 1his latter
addition property is not a necessary condition lor our theory we will discuss, and
is only meant lor didactical purposes. lere we want to address the possibility that
dillerent beings can have dillerent potential levels ol well-being.
A problem arises: il you and l use our empathy to estimate the well-being ol two
individuals, we can get dillerent results. who has the best estimate7 we know that
we can both be biased in all kinds ol ways. 1o solve this, we can lirst study our
cognitive biases and try to counter them. cognitive biases that might inlluence
decision making are e.g. duration neglect, the lraming ellect, the priming ellect,
negativity bias, optimism bias, selective perception and lading allect bias (see e.g.
lohl, 2004).
8econd, we can communicate and try to move to a consensus. ldeally, all moral
agents who have empathy can do the veil ol ignorance exercise and work towards
a consensus to get the best, unbiased empathic well-being. 1his consensual
empathic well-being best approaches the estimates ol the hypothetical ideal
person.
1hird, we can introduce objective measures as approximations to estimate
someone's well-being. 1hese objective measures can be used to counter biases.
lxamples are primary goods (kawls, 191), resources (the resourcist position that


8
One might object that it is very dillicult to empathise with non-human animals or cognitively disabled
humans, because they are so dillerent lrom us in the sense that they lack e.g. sell-consciousness or a
concept ol one's luture and death. 8till, with enough imagination, we can make best estimates ol their
well-being. 1he experiences ol those animals could perhaps be compared with our experiences in
certain dream states, where we lack lull sell-consciousness and a concept ol our luture, but we still leel
lear and pain.
mpartia|ity anJ pricritarianism
9
looks at economical goods that can be distributed (uworkin (1981)), capabilities
(the sullicientarianist position ol the capabilities approach which looks at basic
lunctionings that one is lree to choose to improve one's llourishing (Nussbaum
1992, 2000, 8en 1992)), measures ol desert (the compensationist position ol desert-
principles ol justice which locus on the compensation ol virtuous work (uick, 19,
lamont 1994, milne 1986, 8adurski 198)). Objective quantities like economic
resources, income, wealth, health, jobs, compensations, capabilities or happiness
surveys are nothing but approximations ol well-being: these elements contribute
to well-being, but cannot be reduced or set equal to well-being. 1hey should be
used as tools to objectively counter someone's biases when perlorming the
thought experiment ol the veil ol ignorance. lopelully, in the luture,
neuroscientilic discoveries could make more accurate estimates and comparisons
ol well-being possible.
One more thing needs to be said about incomparability. 8uppose that individual
1 has a well-being at level A, whereas individual 2 can have lour levels ol well-
being: ll'l''l'''. 8uppose l is so low, that everyone agrees that lA. 8imilarly,
suppose that l''' is so high that everyone agrees that l'''-A. lut l' and A appear to
be incomparable, and the same goes lor l'' and A.
9
ln that case, an impartial
observer behind the veil is permitted to choose lor example A=l', and hence Al''.
Another impartial observer may choose A=l''. An analogy with physics, in
particular special relativity, might be handy (see livato, 2009). 1he time dimension
corresponds with the level ol well-being. 1he space dimension represents dillerent
individuals. An individual with well-being A corresponds with a unique point
(event) in space-time. lach event in space-time has a luture and a past light cone.
1he inequality ol well-being lA can be interpreted by the claim that space-time
event l is in the past ol event A, or more exactly: event l lies in the past light cone
ol event A. ll l'''-A, then l''' lies in the luture light cone ol A. lut l' and l'' lie
outside the luture and past light cones ol A (although l'' lies in the luture light
cone ol l' because l''-l'). ll l' lies outside ol the light cones ol A, we can always
choose a lrame ol relerence whereby l' and A occur simultaneously, i.e. l'=A. lut
we can also take another lrame ol relerence that gives l''=A. 1his analogy with
special relativity clarilies the intransitivity problem: il l' and A are incomparable
and may therelore set equal, then also l'' and A may be set equal. lence one could
naively say that l''=A=l'. lut we saw that l''-l'.
we have seen that impartial observers are permitted to make a lew estimates
and choices ol their own, such as the choice between incomparable levels ol well-


9
Nolt (201!) discussed the relevance ol this problem in animal ethics.
tnica| ccnsistcncy anJ Anima| ua|ity
96
being (the choice ol lrame ol relerence in special relativistic terms). ln the second
appendix on democratic impartial prelerences ol moral agents, we will see how
those dillerent choices ol the dillerent impartial observers have to be dealt with in
a democratic way.
4.2.4 1he liletime perspective: lrom momentaneous well-being to
the value ol lile
Lntil now, l locused on empathic estimates ol momentaneous well-being. lut
behind the veil ol ignorance, we have to look at the complete lives ol individuals,
and attach values to those lives, because ol two coherent reasons.
lirst, with a liletime perspective we can avoid the replaceability problem:
painlessly killing someone and replacing himJher by another individual who has
the same momentaneous well-being, is not allowed, even when the aggregate ol
momentaneous well-being would not decrease (see the discussion in the appendix
2 ueriving the wellare lunction behind the veil ol ignorance").
8econd, people are allowed to choose when in their lives they experience
pleasure and pain. lntrapersonal (within the same person's lile), intertemporal
shilts in well-being are permissible. lor example it is permissible lor me to eat a
lot ol candy today, even il as a result l get a toothache tomorrow (note that today l
don't have a clear permission ol my luture sell to cause this toothache). ly eating
candy today, l harm my luture sell, but that is not immoral. At most it is
imprudent. ln contrast, it would be immoral il l cause you a toothache without
your permission. l do not have to consider my luture sell as a separate person, but
l have to consider you as separate. A mere locus on momentaneous well-being will
not be able to make a dillerence between intrapersonal (but intertemporal) and
interpersonal distributions ol well-being. 1his dillerence corresponds with a moral
intuition that there is a dillerence between imprudent and immoral behavior.
1he value ol lile (l olten use 'liletime well-being' as synonym) corresponds with
how much we, behind a veil ol ignorance, would preler to live the complete lile ol
that being. As we saw, momentaneous well-being is not interpersonally
comparable, but at least it is a quantity that does not involve moral evaluation: it is
a descriptive instead ol a normative quantity. lntegrating someone's
momentaneous well-being into a value ol lile introduces normative elements. 1he
value ol lile introduces normative elements: the impartial observer weighs the
momentaneous experienced well-being and sJhe rellects on this well-being lrom
behind a veil ol ignorance. 1his allows lor the introduction ol elements deemed
important by the impartial observer behind the veil ol ignorance. ll sJhe wants to,
mpartia|ity anJ pricritarianism
9
the impartial observer behind the veil can introduce elements lrom an 'objective
list account' ol well-being (see 8haw 1999, chapter 2, crisp, 2008).
1o study liletime well-being, we lirst have to tackle the problem ol integrating
momentaneous well-being over a period ol time, say one second. 1here is a
dillerence between objective versus subjective rate ol time (lostrom & Yudkowski,
2011). A human eye sees roughly 20 lrames per second, whereas the eye ol a lly can
see movement ten times laster than a human eye. lt is as il in one second, a lly
experiences more. 1he objective time is 1 second, but the lly has a ten times laster
subjective rate ol time. As the lly sees more within that one objective second, it is
as il one second lor the lly corresponds with ten seconds lor a human. 1he lly
experiences everything ten times slower. 8uppose the same happens with pain:
one individual leels 10 pulses ol pain per second, another individual leels 100
pulses per second. 1hen this second individual has experienced more. 1he time-
integrated pain over one second ol time is ten times higher lor the second
individual. liletime well-being should take into account someone's subjective rate
ol time, not the objective rate ol time.
1he value ol lile is a lunction ol (consensual, unbiased, empathic)
momentaneous well-being ol an individual, but it is not a trivial summation or
integration ol the momentaneous well-being over all moments ol a liletime, lrom
conception to death. lor example Velleman (1991) argued that a lile with
constantly increasing momentaneous well-being (starting miserable at birth,
ending glorious at death) is prelerred to a deteriorating lile, even il both lives have
the same amount ol summed momentaneous well-being. As a second example,
consider the argument ol the long living oyster (crisp, 2008)
10
. which lile would
you preler: the lile ol a normal human being with lile expectancy 80 years, or the
lile ol an oyster with a lile expectancy you may choose (a trillion years7), but with
a very small but positive and constant well-being7
11
ln short, the human being has
a high momentaneous well-being lor a short period ol time, the oyster has a low
well-being, but summing this low well-being over the very long course ol its lile,
the total (summed) well-being ol the oyster can be higher than that ol the human.
Yet, a lot ol people would preler being born as the human, no matter how long the
lile expectancy ol the oyster may be. 1his means that these people value the value
ol lile ol the human higher than that ol the oyster. why is that7 lerhaps because
they expect that leading a human lile is less boring, and they have a need lor


10
larlit (1984, p.161) described a similar argument, comparing two lives: a century ol lcstacy (high but
temporary well-being) versus urab lternity (very long positive but low well-being).
11
As invertebrates, oysters are perhaps not sentient, but lor argument's sake, assume that oysters are
sentient beings.
tnica| ccnsistcncy anJ Anima| ua|ity
98
variation or psychological growth. 1hese needs cannot be satislied in the lile ol the
oyster. lerhaps the oyster does not have those needs, but that still means that a
human who has these extra needs and who has satislied those needs, has a higher
value ol lile.
1o solve the problem ol the long living oyster, the value ol lile can be expressed
as a trade-oll between quantity (length ol a lile) and quality (average
momentaneous well-being). when quantity is low and quality high (a very short
but very happy lile), it becomes important to increase quantity (lile expectancy).
when quantity is high and quality low (a very long but moderately happy lile), it
becomes important to increase quality.
A simplilication will clarily the kind ol trade-oll between quantity and quality.
8uppose the value ol lile can be expressed mathematically as something like:
w.k.1/(k-1), with w the (consensual, unbiased, empathic) liletime-averaged
momentaneous experienced well-being
12
, 1 the total lilespan and k a relerence
length ol a lile.
1!
when 1 is small, the denominator becomes the constant k, and
hence an increase in w.1 (the well-being integrated over the liletime) dominates.
when 1 is large, the value ol lile approaches the average momentaneous well-
being w times k, and hence an increase in w dominates.
As l will demonstrate in the appendix 2 (lntermezzo: a more complex
lormulation to solve the replaceability problem), the relerence time length k can
be understood as a (complex) lunction ol the psychological connectedness (larlit,
1984), or more precisely: the memories and sense ol the luture ol the individual
generates a psychological identity over time
14
, making it possible to claim that a


12
lahneman (2011) discovered a dillerence between an experiencing sell (evaluating a currently
experienced momentaneous well-being) and a remembering sell (evaluating the remembered well-
being ol a past event). Our distinction between the experienced (momentaneous) well-being w and the
value ol lile can be understood in a similar vein. Although the value ol lile is distinguished lrom the
momentaneous well-being, it does not equal the remembered well-being. 1he value ol lile is evaluated
by a hypothetical well-inlormed lully rational person behind a veil ol ignorance, the remembered well-
being is evaluated by a real, lallible person remembering a past event.
1!
lor reasons l will not explain here, the value ol lile might read w.1 when the average momentaneous
well-being w is negative.
14
Note that a psychological identity over time is basically the property that distinguishes kegan's
subjects-ol-a-lile criterion (kegan, 198!, p.24) lrom mere sentience. 8ubjects-ol-a-lile not only have
the properties ol sentience (perception", an emotional lile", leelings ol pain and pleasure",
prelerences", wellare-interests" and an individual wellare"), but according to kegan they also have
a psychological identity over time", which includes memory" and a sense ol the luture", including
their own luture" (p.24). 8imilarly 8inger's prelerence utilitarianism (8inger, 199!) gives a priority to
beings capable ol holding prelerences towards the luture over those beings who are only concerned
with their immediate well-being. lersons who are capable ol desiring to continue to live as a subject ol

mpartia|ity anJ pricritarianism
99
person at time t
!
is or is not the same person at a later time t
.
. l am to a large
degree a dillerent person than the person l used to be at age ten, although l still
count as the same individual.
ll the hypothetical sentient oyster has a low psychological connectedness, it
gets a small value ol k. A normal human being has a high connectedness and hence
a higher k. 1hat means that lor a human individual, it takes a longer time to
change the psychological identity to such a degree that sJhe becomes a dillerent
person than sJhe used to be.
1his dillerence ol the relerence time length k between dillerent individuals has
important consequences in deciding who to save. 8uppose we have to decide
between extending the lile ol a normal human versus the lile ol a sentient non-
human animal (e.g. the sentient oyster). we see that extending the lile ol the
oyster does not strongly increase his value ol lile il his relerence time length k is
small. lven il his lile is extended by many years, it does not much contribute to his
value ol lile. Not much value ol lile is lost by an earlier death ol the oyster. lor a
human however (supposing a constant momentaneous well-being), an increase in
lile span results in an almost linear increase in value ol lile.
4.2. lersonal identity and psychological continuity
Value ol lile is a lunction ol the momentaneous well-being experienced over a
liletime. lut what is the liletime ol a single individual7 As just mentioned,
dillerent stages in the lile ol an individual might correspond with dillerent
persons. ln our daily lives, the lile ol an individual extends lrom conception to
death (or better: lrom lirst till last experience), making it easy to determine what
is the complete lile ol the individual. lut larlit (1984) presented some luturistic
thought experiments that challenge the notion ol personal identity. A person at a
specilic moment has a mind that is composed ol e.g. memories, beliels, desires and
character traits.
lut what happens to a person during e.g. teleportation, when mind and body
are destroyed at one place and recreated at another7 what il a mind can be
multiplied in two exact copies, lor example when the teleportation lails, the


experience receive a stronger right to live than sentient beings who lack such personal identity over
time. 1he latter are replaceable, according to 8inger. 1he harm ol death is dependent on having a sense
ol the luture and a capability ol seeing onesell as existing over time. 8ell-aware persons who see
themselves as continuing selves existing over time are less replaceable, according to 8inger. loth
8inger's prelerences account and kegan's subject-ol-a-lile account can be supported by a theory that
values psychological connectedness.
tnica| ccnsistcncy anJ Anima| ua|ity
100
original mind and body are not destroyed but still a second mind and body are
created at the other place7 what about imperlect copying a mind m into a slightly
dillerent mind m'7 what about mind swapping: putting mind m1 (that originally
belonged to body l1) into body l2 and mind m2 into body l17
1
what about
multiple personalities, two minds in the same body7 what about splitting a mind
into two minds7 what about lusing two minds into one7 Or what about gradual
changes ol minds and bodies into completely dillerent minds and bodies, where
mind m1' in body l1' has elements lrom both m1 and m2 and both l1 and l27
ln those hypothetical cases, it becomes dillicult to make distinctions between
dillerent individuals. we should abandon the all-or-nothing relationship ol
personal identity. One radical option would be to consider a continuum ol
dillerent individuals, one lor each momentaneous mind at each moment ol time.
lowever, larlit (1984) pointed out that an individual can be delined or described
by a psychological connectedness and continuity between dillerent
momentaneous minds at dillerent moments in time (see also mcmahan, 2002).
compare it with a rope, composed ol dillerent strands, each strand having a
dillerent length. 1here is no strand that extends lrom one end ol the rope to the
other, but still the rope has a connectivity in terms ol connectedness and
continuity. 1wo points ol the rope are connected il there is a strand that runs lrom
the one to the other point. 1he more such strands between the two points, the
higher the connectedness. 1wo points ol the rope are continuously linked il there
are intermediate points such that each two neighboring points are connected
(even when the two endpoints are not mutually connected, when there is no
strand running between the endpoints). 1he strands ol the rope are the analog ol
properties ol the mind (e.g. memories, opinions and character traits). ln this
analogy, one end ol the rope corresponds with the beginning ol lile, the other with
the end. At the end ol your lile, you might not remember anything lrom the
beginning ol your lile, but there is a continuity: an interlinked chain ol memories
shared by intermediate momentaneous minds. 1he problem ol personal identity is
similar to the problem ol how to deline a rope and how to distinguish one rope
lrom another.
low to deal with this problem ol personal identity behind the veil ol
ignorance7 lehind the veil, an impartial observer sees the huge set ol all minds at
all moments. At each moment, a mind has a unique momentaneous well-being.
One option is that the impartial observer groups the set ol momentaneous minds
in subsets, each subset relerring to the complete lile ol one individual. Alter


1
8ee e.g. williams' lamous thought experiment ol torture (williams, 190).
mpartia|ity anJ pricritarianism
101
putting all momentaneous minds in subsets, the impartial observer looks at a
subset that now corresponds to the lile ol one individual (as delined by the
impartial observer). 1his subset is composed ol all momentaneous minds ol that
individual, each element having a momentaneous well-being that the individual
will experience. lt is this subset that the impartial observer gives a value, the value
ol lile, which corresponds with how much sJhe prelers to experience all
experiences ol all the momentaneous minds ol that individual (that subset).
ln most lamiliar cases, this grouping in subsets is easy and is restricted to
strong conditions (i.e. not all possible groupings are allowed). lut in the larlitian
situations, the grouping becomes complex and to a degree arbitrary. 1herelore, in
a later intermezzo l will present another approach how an impartial observer
behind the veil can solve this problem ol personal identity. 1hat new approach
will be more suitable (less arbitrary) to deal with luturistic larlitian situations ol
e.g. teleportation, mind copying and mind swapping.
All in all, value ol lile is the totality ol everything one prelers lrom behind the
veil ol ignorance, in the expectation to live the complete lile ol an individual over
time. lt is everything that would matter to you il you were a sentient being, living
its complete lile. 1he value ol lile is a complex lunction ol momentaneous
experienced well-being. As mentioned above, this momentaneous experienced
well-being is composed ol all the leelings that are the result ol (dis)satislaction ol
prelerences. 1he term 'experienced well-being' has two words, which means it
combines a mental state account (the subjective cxpcricnccs to like things) with a
prelerence satislaction account (the wc||-ucinq in terms ol what one wants). All
things that one likes and all things that one wants matter to the experienced well-
being. 1he value ol lile introduces normative elements: a weighting ol the
momentaneous experienced well-being and a rellection on this well-being lrom
behind a veil ol ignorance.
Value ol lile is very dillicult to measure. All we have is our empathy, our
scientilic knowledge and our imagination. we have to try placing ourselves in the
position ol others, by using empathy, by imagining that we could be the other
person, with all his or her needs and leelings. 1he 'emotional' method ol sampling
empathic leelings and the 'rational' method ol imagination behind the veil ol
ignorance are rules ol thumb to make educated guesses about the order ol the
values ol lile ol dillerent individuals. lmpathy and imagination are virtues to be
developed and already allow us to move quite lar.
As all sentient beings have subjective experiences ol their leelings and needs,
all sentient beings have a liletime well-being or a value ol lile lor themselves. 1he
model we are about to discuss therelore applies to all sentient beings. lt should
therelore also include mentally disabled humans and non-human animals. we
should not restrict this theory ol justice to only rational, sell-conscious beings.
tnica| ccnsistcncy anJ Anima| ua|ity
102
lence, a person should be interpreted as an individual who has personal
experiences. ln this interpretation, a person is equivalent to a sentient being.
Note that dillerent sentient beings, such as a lrog and a human, might have
strongly dillerent levels ol liletime well-being. 1here are lour reasons why a lrog
might have a much lower liletime well-being than a normal human. lirst, lrogs
likely have less needs and prelerences (e.g. less need lor accomplishments or
relationships) than most humans. 8econd, the intensity ol prelerence
(dis)satislaction might be lower in lrogs: a lrog might have less capacity than most
humans to experience pain and pleasure, due to a smaller brain with less
neurotransmitters and less receptors (see Vallentyne, 2006). 1ogether, these two
reasons imply a much lower momentaneous well-being lor the lrog. Next, lrogs
have a much shorter lilespan than those ol most humans. And as a lourth reason,
lrogs likely have less psychological connectivity between dillerent lile stages.
ln summary, two reasons imply a gap between the momentaneous well-being ol
a human and a lrog. ln terms ol liletime well-being, the gap is even bigger due to
two additional reasons that reler to the lilespan and the psychological
connectivity.
Vallentyne (2006) and loltug (200) discussed the lar reaching implications ol
an ethic ol redistributive (strict) egalitarianism when animals such as lrogs are
included, due to the vast dillerence in levels ol well-being. lgalitarianism becomes
very demanding lor humans, because in order to close the gap between lrogs and
humans, nearly all resources should go to lrogs (and many other non-human
animals). 1he theory that l propose has much less demanding consequences lor
humans due to two reasons.
lirst, l propose a prioritarian ethic instead ol an egalitarian one
(prioritarianism was also suggested by loltug (200), but as we will see, my
prioritarian ethic has some relevant benelits compared to his). compared to strict
egalitarianism, prioritarianism is more considerate to elliciency: the benelit lor
the worst-oll should not come at a cost ol much more liletime well-being ol the
better-oll.
And second, the above lour reasons not only indicate that lrogs have lower
actual levels ol liletime well-being, but also lower potential levels. 1his potential
level is the level ol liletime well-being that an individual would have when all
distributable goods (all means and resources on earth) are distributed to the
maximum benelit ol this individual. 1heir lower potential levels imply that, alter
distributing all resources on earth to a lrog, its liletime well-being will not
increase by the same amount as when all resources are distributed to a human. ln
other words, compared to lrogs, humans can be benelited much more by the same
amount ol resources. lumans are more ellicient than lrogs in translating means
mpartia|ity anJ pricritarianism
10!
and resources into liletime well-being. 1his higher elliciency is relevant in a
prioritarian (but not in an egalitarian) ethic.
compare it with the problem ol distributing an amount ol water between
dillerent glasses. 1he level ol water in a glass represents the actual level ol well-
being ol an individual, the volume ol the glass represents the potential level ol
that individual. A lrog is comparable to a small glass: pouring water into a small
glass is more dillicult than pouring it into a big glass, resulting in more waste lor
the small glass. 1he small glass more easily results in a spill (overllow). 1his waste
ol water decreases the elliciency ol a distribution ol water. Ol course, one could
increase the volumes ol the small glasses, just as one could (genetically) enhance
lrogs to increase their potential liletime well-being. lut then we do not have the
same glasses (lrogs) anymore.
4.! 1he maximin principle
ln the previous section, l discussed the notion ol liletime well-being (value ol lile).
lowever, nothing has yet been said about how to distribute these quantities. 1he
maximin distribution principle is a theory ol justice, lavored by }ohn kawls (191),
which can be derived lrom the thought experiment ol the veil ol ignorance. 1he
principle says that we should strive lor an increase or maximization ol the liletime
well-being ol the beings in the worst-oll position (the beings with a minimal
amount ol well-being). 1he locus is on the lowest values ol lile, trying to maximize
the lowest levels ol liletime well-being.
maximin can be derived lrom the veil ol ignorance by realizing that you could
be the individual in the worst-oll position. leeping that possibility in mind, you
may preler a world (or a moral law) where this lowest level is increased and
maximized. 1hat means you would preler a society where the lowest levels ol
liletime well-being are not so low, such that you no longer worry about getting
one ol these lowest levels. And importantly: inequality ol well-being is only
allowed il it is at the advantage ol the worst-oll positions. Other inequalities ol
well-being that do not match this condition are not accepted.
let's give an example with numbers. 8uppose there are two sentient beings, and
we can choose between dillerent situations. ln situation 1, sentient being A has a
value ol lile level 10, l has level 100. 8o there is a big inequality. lowever, this
situation is prelerable to situation 2, where A and l both have a value ol lile equal
to . 8ituation 1 is also better than situation ! where A has level and l has level
200, because in situation 1 the worst-oll being has a level ol 10 instead ol . Note
tnica| ccnsistcncy anJ Anima| ua|ity
104
that in situation !, the total sum ol well-being levels is 20, which is higher than
110 ol the lirst situation. maximin is therelore dillerent than sum-utilitarianism,
because it gives absolute priority to the lowest levels.
1he reason why someone would preler the lirst situation instead ol the third, is
that lrom behind the veil ol ignorance, not knowing whether sJhe will be A or l,
sJhe does not want to run the risk ol becoming the individual with the worst
outcome. lence, il we have to choose between situations 1, 2 and ! lrom behind
the veil ol ignorance, it is a choice between three games ol chance. which game do
we preler to play7 ln each game we have an equal probability ol becoming
individual A or l. lut il we have risk aversion (Arrow, 196, lratt, 1964), we do not
preler situations 2 and !, because in those situations we know that the worst-oll
position has level , whereas we could have had level 10 in situation 1. leople with
maximal risk aversion are real pessimists and always think as il they will become
the person in the worst-oll position. 1hey ask the question: what il l would be the
worst-oll person7 1hey would preler situation 1, even il it has a lower expectation
value, because in this situation they at least have a well-being ol 10. ln general,
risk aversion is the reluctance to accept a game ol chance with an uncertain
outcome rather than another game ol chance with a more certain, but possibly
lower expected outcome.
On the other hand, someone who is risk neutral would take the sum-utilitarist
choice by looking at the total expectation value ol well-being (the sum ol products
ol probabilities and levels ol well-being). ln situation 1, the expectation value is
Zx10-Zx100=. ln situations 2 and ! we have respectively Zx-Zx= and
Zx-Zx200=102,. 1he latter has the highest expectation value, so is prelerred by
the risk neutral sum-utilitarist (see e.g. larsanyi, 19!).
}ohn kawls' theory ol justice (191) incorporates the maximin principle. when
it comes to animal ethics, kichard kyder (2001) can be considered as the advocate
ol the maximin principle. lis theory ol 'painism' gives an absolute priority to the
so called maximum sullerer, the sentient being who is in most pain. 1his is clearly
the worst oll position. ln his theory ol animal rights, 1om kegan (198!) also
ollered two principles: the miniride principle and the worst-oll principle. kegan's
two principles can be interpreted in such a way that they can be unilied in the one
principle ol maximin.
1he miniride principle says: 8pecial consideration aside, when we must choose
between overriding the rights ol many who are innocent or the rights ol lew who
are innocent, and when each allected individual will be harmed in a prima lacie
comparable way, then we ought to choose to override the rights ol the lew in
prelerence to overriding the rights ol the many."(p.!0) 1o take an example,
suppose we have to choose between situation X where one individual would suller
and has a value ol lile equal to , whereas ten others would have well-being at
mpartia|ity anJ pricritarianism
10
level 10, and situation Y where the lirst individual has level 10 and the ten others
have all level . 1he harm done to each individual is the same (a drop ol well-being
ol levels). 1he miniride principle prelers situation X, and this is also what
maximin would say.
1he worst-oll principle says that: 8pecial considerations aside, when we must
decide to override the rights ol the many or the rights ol the lew who are
innocent, and when the harm laced by the lew would make them worse-oll than
any ol the many would be il the other option were chosen, then we ought to
override the rights ol the many." (p.!08) As an example, in situation X one
individual has well-being 2 whereas the other ten have well-being 10. 8ituation Y
is similar to the previous example: the lirst individual has 10 and the others have
. 1he worst-oll principle and the maximin principle both say that we have to
preler situation Y, because in situation X, the harm done to the lirst individual is a
drop ol 8 levels ol well-being. 1hat's worse oll than the other people in situation Y.
1he worst-oll principle strikes many people as counter-intuitive in some
extreme examples. what il instead ol harming ten people in situation Y, we
harmed a million people7 1he worst-oll principle lacks a kind ol elliciency. 1he
quasi-maximin principle that l am going to discuss in the next section, would be
more compatible with an intuitive judgment that to some degree elliciency is
important in distributing well-being.
4.4 1he quasi-maximin principle and prioritarianism
we saw that lrom behind the veil ol ignorance, we could arrive at two dillerent
theories ol justice, depending on our level ol risk aversion. 8omeone who has
maximal risk aversion prelers the maximin strategy. A risk neutral person prelers
the sum-utilitarist strategy. 1hese two strategies are but two options in a
continuum ol theories ol justice, because there is a continuum in the level ol risk
aversion. most people have a high but not maximal level ol risk aversion. 8o let's
take a look at another example. ln situation 1, person A had a well-being ol 10, l
had level 100. ln situation 4, we can increase the well-being ol A by a negligible
amount to level 10,01. ln order to do this, the level ol l has to drop a lot, to level
11. lt's as il we drive l to extreme poverty in order to increase the level ol the
extremely poor person A with a negligible amount. According to maximin, we
would preler situation 4, because 10,01 is higher than 10. lowever a person with
high but not maximal risk aversion would still preler situation 1. 1his person
tnica| ccnsistcncy anJ Anima| ua|ity
106
would adopt a quasi-maximin principle ol justice. lt is almost but not completely
maximin.
1here is another way to arrive at quasi-maximin. Our empathy is directed
towards the worst-oll individual, which is sentient being A in the above example.
lut il we have a low but not zero need lor elliciency, we would not preler situation
4. lt doesn't seem ellicient to drop l in order to increase A with just a tiny amount.
lt's too much a waste ol well-being.
1herelore, we have two reasons to preler situation 1: impartia|ity witn a niqn uut
nct maxima| |cvc| cj risk avcrsicn (nccJ jcr sajcty), anJ cmpatny witn a |cw uut nct zcrc
nccJ jcr cjjicicncy. 1hese two reasons cohere with each other and they are both
based on moral intuitions ol impartiality, salety, empathy and elliciency. 1he two
reasons correspond with a rational and an emotional approach, and with two
viewpoints: the rational approach looks at a situation lrom the outside, lrom an
impartial point ol view behind a veil ol ignorance. 1he emotional approach is more
down to earth: it looks at a situation lrom the inside, lrom the subjective
experience ol compassion with others. 1hese two coherent approaches give us
some justilication lor a quasi-maximin principle ol justice.
16

1he quasi-maximin (Qmm) principle lor a just distribution ol values ol lile.
maximize the values ol lile (liletime well-being levels) ol all sentient beings,
giving a strong priority on increasing the lowest values ol lile. l.e. maximize
the values ol lile ol the worst oll individuals, unless this is at the expense ol
much more well-being ol others.
1his Qmm-principle gives a high but not maximum priority to the worst-oll
individuals. lt is therelore a kind ol prioritarianism. ln prioritarianism, the well-
being ol an individual is weighted with a priority lunction. 1he lower someone's


16
lor some moral agents, these two approaches might be dillerent. lor example someone who has zero
risk aversion but a high empathic concern lor the worst-oll and a low need lor elliciency will get two
dillerent ethics. According to the impartial veil ol ignorance approach, this moral agent would be a
sum-utilitarian. According to hisJher moral intuitions ol empathy and elliciency, sJhe would be more
maximin-prioritarian. ll such a dichotomy occurs, the moral agent is allowed to take hisJher prelerred
approach to determine the level ol priority lor the worst-oll (this level is then democratically averaged
together with the prelerences ol all other moral agents, as is discussed in appendix 2, uemocratic
impartial prelerences ol moral agents"). lurthermore, l expect that most moral agents have some (non-
zero) level ol risk aversion, and most moral agents have a non-absolute need lor elliciency (a non-zero
priority lor the worst-oll). 8o even when both approaches might diller lor one moral agents, when we
look at the group ol all moral agents, we can expect that they might still easily reach a rather big
consensus on the non-zero level ol priority lor the worst-oll. looking at an individual, both approaches
might be mutually incoherent, but on the level ol the whole group, they might still be more coherent
with each other.
mpartia|ity anJ pricritarianism
10
well-being, the higher hisJher priority. As sum-utilitarianism maximizes the sum
ol well-being levels, prioritarianism maximizes the sum ol weighted well-being
levels. l reler to the mathematical section below lor more details. lut lirst, let's
discuss some applications ol this Qmm-theory.
4. Applications ol the quasi-maximin theory
4..1 kawls' theory ol justice
Although the comparison between values ol lile ol dillerent individuals in
dillerent situations is very dillicult, we can derive a set ol approximate rules ol
thumb that can move us closer to the Qmm-distribution ol value ol lile. ln his
theory ol justice, }ohn kawls derived three such principles (kawls, 191, 2001):
1) lquality ol basic liberties and rights.
2) lquality ol lair opportunity: il individuals have the same ambition and native
talents relevant lor a position that generates a benelit (e.g. a job), they should
have the same prospects ol success in competition lor that position (see also
Arneson, 2008).
!) lquality ol economic goods in terms ol the dillerence principle: the
distribution ol economic goods should be according to maximin. 1hat means that
economic inequalities should be in the greatest benelit ol the least advantaged
persons.
1hese kawlsian equality principles can be considered as rules ol thumb to
approach a Qmm-distribution ol well-being. let's lirst look at equality ol basic
liberties and rights. we only have to consider rights and liberties that clearly
allect the value ol lile. 1ake lor example the right to lree speech. ll l have a need
lor sharing ideas, l will leel lrustrated when l do not have the right to lree speech,
and this obstruction will lower my value ol lile. lowever, there are some speech
acts (e.g. hate speech or insults) that can lower the value ol lile ol other people
(the receivers). ln most cases, allowing these disdainlul speech acts will violate the
Qmm-principle. lirst, as kosenberg notes (kosenberg, 200!), someone uttering
disdainlul speech acts olten implies that this person has unmet needs. lnsults are a
tragic expression ol a person with an unmet need. ll your boss insults you by
saying that you are lazy, this most likely means that your boss leels lrustrated and
has an unmet need lor e.g. elliciency, and that he only lound a tragic way to
express himsell. Also hate speech and scapegoats indicate some unmet need (e.g.
lor social security or respect).
tnica| ccnsistcncy anJ Anima| ua|ity
108
let's try to apply our Qmm-model to this problem. As a starting point, we have
two persons. ln situation X, there is no lree speech. ly lack ol lurther details, and
by the symmetry between the persons, we have to assume that a priori (all else
equal) both persons have equal value ol lile, say level 100. 1his equality is an
important assumption in dealing with these kinds ol problems. ln situation Y,
there is lree speech, and as a consequence, person l insults person A. 1he value ol
lile ol person l increases to 101, but lor person A it decreases to 99. 8ituation Y
violates the Qmm-principle. 1o summarize: not all speech acts satisly the Qmm-
principle.
moving to the second kawlsian principle, where there is a scarcity ol social,
economic or political positions (education, jobs, elections,.), the equality ol lair
opportunity (and participation) can be derived lrom the original position. Only
someone who is more talented, motivated, trustworthy or experienced to do a job
that is benelicial to the least advantaged persons (or more generally a socially
benelicial job that helps approaching the Qmm-distribution ol liletime well-being),
should have a higher prospect to get that job. lence, equality ol lair opportunity is
a derivative ol the Qmm-principle.
1he third kawlsian principle (the dillerence principle) can also be easily
restated in the Qmm-lramework. lirst note that this latter kawlsian dillerence
principle relers to economic goods and not to the value ol lile (the liletime well-
being). lconomic goods (income, resources, wealth,.) only constitute a subset ol
lactors that contribute to the value ol lile. 1he Qmm-theory as described in this
section is more in line with the wellare based principles (like utilitarianism), and
hence also incorporates the distribution ol liberties, opportunities, capabilities
and all other lactors that contribute to the value ol lile.
ldeally, the economic goods should be distributed according to the rule that
realizes a quasi-maximin distribution ol liletime well-being. 1his means that lor
example disabled persons should get relatively more economic goods in order to
compensate lor their loss ol well-being, except when the transler ol economic
goods to these disabled persons cannot be done in a sulliciently ellicient way. ln
other words, when we are only capable ol increasing the well-being ol the disabled
person by a negligible amount by translerring huge amounts ol resources to these
disabled persons, we should not opt lor the transler.
lut as it is olten dillicult to determine the optimal distribution ol economic
goods, the economic goods can more easily be distributed according to kawls'
dillerence principle. 8o kawls' dillerence principle can be considered as an
approximation ol the Qmm-theory.
mpartia|ity anJ pricritarianism
109
4..2 kesponsibility and desert
8o lar lor kawls' dillerence principle. let us also take a look at resource-based (or
responsibility-based) and desert-based principles.
ln the resource-based principles ol justice (uworkin, 1981), one is concerned
about the importance ol personal responsibility. According to the Qmm-theory,
society should not keep on pouring resources down the drain, il worse-oll people
act very irresponsibly with these given resources (when they negligently squander
them) or il they are highly inellicient in translorming these resources into liletime
well-being (see cohen 1989, Arneson 1989, koemer 1996).
consider lirst the issue ol acting irresponsibly with given resources. 8ome lacts
that inlluence well-being (e.g. being born with talents or discovering new
resources by brute luck) are beyond someone's control or responsibility. lut, given
an amount ol resources, an individual has a personal choice and hence a personal
responsibility to transler these resources into well-being. what il sJhe makes
imprudent or irrational choices that squander resources7 Or what il in the hospital
we have to choose between helping two patients who are equally bad oll and who
can be equally benelited by a medical operation, the lirst one has a genetic disease,
the second had a car accident because she was a reckless driver7 luck
consequentialism (or responsibility-sensitive consequentialism) claims that the
part ol someone's well-being that is under responsibility ol the individual should
not matter in calculating the best distribution ol well-being. 8o how much should
responsibility and brute luck play a role in the distribution ol well-being7 let me
make three remarks on this.
lirst, it might be likely that there is no such thing as a lree will. leople might
make bad choices (e.g. reckless driving, gambling or being addicted), but they are
not responsible lor choosing brains that make them vulnerable lor those bad
choices, just as persons with genetic diseases are not responsible lor choosing the
bad genes. lence, we might overestimate the role ol personal responsibility. As
having a certain brain is a result ol brute luck, a lot (or all7) ol our personal
choices might in the end be beyond our control, beyond our responsibility.
8econd, in the hospital example (choosing between the reckless driver and the
person with the genetic disease), the choice who to help might inlluence the
distribution ol well-being. A choice to help reckless drivers (or other people who
make bad choices) might give wrong incentives to some people. lor example
people might become less dissuaded to make some bad choices. ln this sense,
personal responsibility plays only an instrumental role. 8imilarly, when lazy
workers or imprudent people experience a disadvantage due to their choice to be
lazy or imprudent, they should be helped, benelited or rewarded less compared to
the hard working and prudent people. 1he level ol benelits and rewards should be
tnica| ccnsistcncy anJ Anima| ua|ity
110
tuned to give the optimal incentives lor everyone to reach a Qmm-distribution ol
well-being.
1hird, lrom a liletime perspective, we have to take into account that the
reckless driver, the lazy worker or the drug addict already enjoyed a benelit in the
past (the pleasure ol driving recklessly, the pleasure ol relaxing at work, the
pleasure ol the drugs). 1his benelit in the past means that they have less right to a
benelit in the luture, compared to someone who had brute bad luck (all else
equal).
As a result ol these three considerations, l am tempted to minimize the
importance ol the distinction between well-being that is a result ol brute luck and
well-being that is a result ol personal choices. All types ol well-being are
important in Qmm-theory, and the dillerence between brute luck and personal
choice can only play an instrumental role in tuning incentives lor behavior.
Now consider the second issue, the problem that people might be highly
inellicient in translorming resources into value ol lile. Qmm-theory keeps track ol
the inelliciencies when distributing benelits. lor example consider a benelit that
generates a well-being ol units to a well-oll person who has initial well-being . ll
this person gets the benelit, his well-being will end up at the level 10. Now
consider a redistribution ol this benelit lrom the well-oll person to a worse-oll
person having initial well-being 1. As the worse-oll person is less ellicient in
translorming the benelit into well-being, she will only receive an extra ! units ol
well-being, ending up at level 4. lence we have to decide between option X=(10,1)
and option Y=(,4). Qmm-theory prelers option Y.
we can distinguish between two kinds ol inelliciencies in the translormation ol
resources into liletime well-being. lirst, there are the things that are beyond the
control ol the individual: an individual might have medical needs such that a lot ol
resources are required to generate a unit ol well-being. 1he second inelliciency
occurs in the development ol e.g. expensive tastes. laving an expensive taste
means that one needs a lot ol resources to satisly the taste and to increase the
well-being with one unit. lere we should make a distinction between modiliability
and satisliability ol prelerences. modiliability means that an individual can
inlluence the presence ol the prelerence: the individual has some power to switch
the prelerence on or oll. 1his modiliability should be distinguished lrom
satisliability: the power ol an individual to satisly a prelerence.
lxpensive tastes are not only inellicient, they are modiliable, and this property
ol modiliability is highly morally relevant. According to our Qmm-theory, as
people are responsible lor developing expensive tastes, they have a duty not to
develop those modiliable tastes, because those tastes generate extra inelliciencies.
lnstead ol pouring resources down the drain, therapy (e.g. meditation) can be a
cheap method to conquer those modiliable expensive tastes and addictions. And in
mpartia|ity anJ pricritarianism
111
order to dissuade people to develop expensive tastes, we should relrain lrom
redistributing resources to satisly expensive tastes.
Next to responsibility is the issue ol desert. ln the desert-based principles, one
wants to emphasize ellort (8adurski 198, milne 1986) or costs incurred in work
(uick 19, lamont 1994), or someone's contribution to society (miller 196, kiley
1989). lence, the notion ol desert that is used in Qmm-theory is based on two
aspects. lirst, it has a compensationist approach: compensate lor the ellorts, costs
or risks taken by an agent's past actions. 8econd, it can reler to virtuous actions
that contribute to the well-being ol others.
ccmpcnsaticn. According to the desert principle, we should distribute economic
goods corresponding to the virtue or deservingness ol a person (see e.g. lagan
1999). According to desert-based principles, things done in the past (e.g. someone
who worked hard yesterday) inlluence the just distribution ol current resources
(e.g. higher payment lor the one who worked hard yesterday). 1he Qmm-theory
uses a liletime perspective (a locus on liletime well-being), and this liletime
perspective allows to take into account an agent's past actions. lence, the liletime
approach allows lor a compensationist desert-based principle. lard work in the
past means that someone's momentaneous well-being in the past is low. 1his low
past momentaneous well-being can be compensated by a higher luture
momentaneous well-being to increase the liletime well-being.
ccntriuuticn. we can interpret virtuous work as work that contributes to the
society, and more specilically that promotes the Qmm-distribution ol well-being.
1he more someone contributes to Qmm, the more she should be rewarded in order
to support her choice lor Qmm. And the more her value ol lile decreases by doing
this important work (e.g. by doing hard, long, boring or dangerous work), the
more she should be compensated lor her loss ol well-being. 8o the more her value
ol lile decreases and the more her work contributes to Qmm, the more virtuous
and deserving she is.
8omeone who contributes more to the well-being ol the worst-oll persons,
should get prior access to more economic goods. lor example a nurse should
receive a higher income than a prolessional athlete, because the nurse's
contribution to the value ol lile ol the worst-oll individuals is higher. lree market
distributions ol economic wealth are not always compatible with the Qmm-theory.
ln a desert-based theory ol justice, one olten adds the 'greater gap principle'.
1he greater the gap between what someone deserves and what someone has, the
more priority should be given lor decreasing that gap. ln a sense, this is a
generalization ol prioritarianism as delined above, where priority should be given
to the most deserving person. 1he more deserving person is not always the worst-
oll person, but can also be the more virtuous person. 8o not only the well-being ol
an individual should matter (as in simple prioritarianism), but also someone's
tnica| ccnsistcncy anJ Anima| ua|ity
112
contribution to society (to approach the Qmm-distribution, i.e. to contribute to
the total weighted well-being ol all people in society) should be rewarded. And this
latter reward is only possible when it is not in conllict with the Qmm-theory itsell.
consider the lollowing example. 8uppose we have an ill person (well-being level
1), a poor physician (level 10) and a very rich and wealthy person (level 100). 1he
value ol lile distribution in situation X can be described with the three values
(1,10,100). Now, the rich person can give money to the physician so that the
physician is motivated to heal the ill person. we now get situation Y=(10,!0,!0).
8ituation Y is better than X according to Qmm. Note that the increase in well-
being ol the physician (!0 10 = 20) can be larger than the increase ol well-being
ol the ill person (which equals 9 in the example). we might compare this desert
principle with a negative leedback mechanism which acts as a stable attractor
when there are disturbances that push us away lrom the Qmm-equilibrium
position. 1his leedback mechanism actively pulls us back towards the state ol
Qmm, by stimulating (rewarding) people who contribute most to the total priority
weighted well-being ol society.
4..! luture orientation and restorative justice
ln the above section we saw that responsibility plays two roles in Qmm-theory.
lirst, the liletime perspective ol Qmm-theory allows us to take past actions into
account, in a way that some compensationist notion ol desert becomes important.
8econd, the Qmm-theory also looks at how we can praise or blame people to
inlluence their luture behavior. lence, the Qmm-theory has both a past and
lorward looking aspect.
1he lorward looking aspect has major implications lor the criminal justice
system that needs to be revised. ll the behavior ol a person is in strong violation ol
the Qmm-principle (and with other ethical principles to be discussed in next
chapters), then some rights ol that person should be taken away in order to
protect society lrom luture violations ol the Qmm-principle. lspecially when we
know that someone has malicious intentions, it is likely that this person will
violate the Qmm-principle in the luture. lmprisonment should be considered as a
kind ol quarantine to protect society lrom threats (such as murderers, pathogens)
that endanger a Qmm-distribution ol liletime well-being. moral responsibility lor
criminal behavior (i.e. behavior that deviates lrom the Qmm-principle) should be a
measure ol the likelihood that the person will do other crimes in the luture
(because e.g. hisJher brains are wired in a certain way that makes himJher more
susceptible to do crimes). 1he probability ol recidivism (i.e. the risk that someone
mpartia|ity anJ pricritarianism
11!
might perlorm actions in the luture that deviate lrom the Qmm-principle) should
be taken into account when liberties and rights are distributed.
1

8ome behavior such as stealing or lying would be permitted, however, il the
behavior is in correspondence with the Qmm-principle (lor example a poor thiel
who steals lrom the rich, a person who lies to protect someone's lile).
kestorative justice might be prelerred to retributive justice, because it might be
better lor the well-being ol both victims and perpetrators. most perpetrators
typically are victims themselves who tend to have a low well-being (they might
have strong leelings ol lrustration due to discrimination, lack ol education, lack ol
opportunities, or traumatic experiences in the past). 1he Qmm-principle,
combined with neuroscientilic evidence (about e.g. the lack ol lree will), implies
that the legal justice system should be more lorward looking (a restorative justice
that locuses on how to improve well-being and how to most elliciently change a
criminal's brain and behavior) instead ol backward looking (a retributive justice
that locuses on punishment, guilt and blame). 1he only backward looking part in
Qmm-theory has to do with compensations lor past actions, which relates to a
notion ol desert as we saw in the previous section. lnstead ol punishing people as
retributive justice, it is better to create circumstances in such a way that people
tend to behave more morally.
1o summarize, we see that the Qmm-theory combines and encompasses a lot ol
dillerent ideas: prioritarianism (keeping the balance between kawlsian maximin
and sum-utilitarianism) and desert-based, wellare-based and resource-based
theories.


1
lor example my right to use sell-delense against someone who is about to harm me depends on the
moral responsibility ol that person who is a threat lor me. 1he more responsible sJhe is, the more likely
sJhe will be a threat to others in the luture. consider a) a construction worker who stumbles by
accident and lalls right on me, b) a reckless car driver who is about to hit me by accident with the car
and c) a murderer who intends to kill me. ln all three cases, l can delend mysell, by doing something
that will harm the threat (the worker, driver or murderer). lut the level ol harm that l am allowed to
use depends on the level ol moral responsibility ol the threat. 1he more responsible, the more harm l
am allowed to cause to the threat in delending mysell (still avoiding any unnecessary harm). 1his
moral responsibility measures the likelihood that the threat will be a threat again in the luture. lt is
more likely that a murderer will commit a similar crime in the luture than that a car drive will again hit
someone, and the latter is still more likely than the probability ol the construction worker lalling again
by accident right on someone else. As a consequence, l am more entitled to kill the murderer in sell-
delense, than to kill the construction worker.
mpartia|ity anJ pricritarianism
11
contribute enough to the wellare lunction (i.e. when their procreation
lowers the wellare lunction)7 1he second paradox says that in some cases it
might even be good (in terms ol increasing the wellare lunction) to add an
individual who has a negative liletime well-being (a lile not worth living).
1o solve these paradoxes, we could add new principles to the prioritarian
theory. ln upcoming chapters, we will encounter such extra principles that
can deal with those two paradoxes. ln particular the sadistic conclusion can
be avoided with a mere means principle (chapter 6), and the mere addition
paradox can be avoided with a !-N principle (chapter 10.4).
2) lersonal identity: in section 4.2., we encountered some problems with the
notion ol personal identity over time, which has some consequences lor e.g.
the replaceability problem: is it permissible to kill someone and replace
himJher by a new individual who has the same momentaneous well-being7
1his replaceability problem can be avoided by looking at the liletime well-
being instead ol the momentaneous well-being, but that means we
encounter the issue ol personal identity and psychological continuity. ln
the appendix we will encounter a method to rewrite the liletime well-being

as an integral ol the momentaneous minds with a psychological


connectivity lunction that links those momentaneous minds with each
other. 1his psychological connectivity lunction has important implications
lor animal ethics, because not all sentient beings have similar levels ol
psychological connectivity: some animals (such as humans) have a strong
autobiographical sell and hence a strong psychological connectivity with
luture and past selves, others live merely in the here and now, as il they
merely exist ol dillerent, not-connected momentaneous minds. 1he latter
are more replaceable than the lormer (because one could say that the latter
are constantly replaced anyway).
!) uemocratic balancing ol prelerences: il we do the exercise ol the veil ol
ignorance, we might come to dillerent conclusions about the wellare
lunction. You and l might have dillerent levels ol risk aversion and hence
would preler to use dillerent concave lunctions j. And as liletime well-being
is not objectively interpersonally comparable, you and l might also have
dillerent estimates ol the levels ol liletime well-being

. lence, the wellare


lunction w that l would derive lrom behind the veil might be dillerent than
yours. who has the most correct wellare lunction7 Are my estimates better
than yours7 1he appendix describes a method how to democratically
balance our wellare lunctions. lach moral agent a behind the veil can
construct hisJher own wellare lunction

, which has a maximum value

(i.e. a best outcome according to this moral agent). we can take an


average ol weighted wellare lunctions:
tnica| ccnsistcncy anJ Anima| ua|ity
116


where the sum runs over all moral agents who do the exercise ol the veil ol
ignorance, and

is the number ol those moral agents. ln this way, each


moral agent contributes equally (democratically) to the average wellare
lunction.
4. 8ummary
ln this section, l derived the quasi-maximin prioritarian principle as a model lor
consequentialist theories ol justice. Along the road, we encountered multiple
problems: how to deline well-being, how to integrate well-being over a lilespan
(when persons change), how to compare well-being between persons and how to
distribute well-being between persons. 1he appendix deals with variable
populations (e.g. luture generations) and uncertain outcomes (lotteries).
Quasi-maximin is a theory close to maximin, but with a small tendency towards
sum-utilitarianism. 1he 'quasi' in Qmm-theory is derived in two dillerent ways:
lirst lrom impartiality (the veil ol ignorance) with a high but not maximal risk
aversion (a high prelerence lor security), and second lrom empathy (the equality
principle) with a small but not zero prelerence lor elliciency. Qmm is a special
lorm ol prioritarianism that is compatible with the moral virtue ol empathy and
the moral intuitions ol impartiality and elliciency.
Note that quasi-maximin, although a rational theory, already incorporates some
moral intuitions in some subtle ways. ln particular the level ol risk aversion and
the need lor elliciency cannot be derived lrom purely rational reasoning. 1hese
non trivial (not zero or one) values lor risk aversion and need lor elliciency result
into a prioritarian theory that best lits those intuitions (better than the extreme
theories ol sum-utilitarianism and maximin).
lowever, there are a lot ol other intuitive judgments that are in conllict with
and cannot be derived lrom the kawlsian veil ol ignorance. ln the next sections,
we will encounter some other moral intuitions that might overrule the Qmm-
principle. 1hese moral intuitions are important in the ethics ol care and the ethics
ol rights (lantian deontological ethics).


11
chapter lartiality and tolerated choice
equality
1he consequentialist theory ol prioritarian justice can be very demanding. keal
impartiality might imply that we need to sacrilice many ol our resources and
much ol our well-being in order to advance the worst-oll individuals. 1wo replies
can be given to this 'demandingness objection'. lirst, we should require that
governmental institutions and political laws are really impartial. 8econd, we note
that lor a moral agent some partiality can be tolerated under some conditions.
lspecially social or empathic beings have dilliculties being perlectly impartial. we
olten have dilliculties being impartial, because we have strong emotions towards
our relatives, lriends or co-living animals. 1his partiality might conllict with
consequentialist theories like Qmm-theory. lartiality can also be important lor
some moral patients. lmagine children growing up in a lamily ol perlectly
impartial parents. 1his will raise concerns about their emotional development and
well-being.
According to an ethics ol care (Noddings, 2002) we do not always have a duty to
take the impartial point ol view, because that would not respect interpersonal
relationships. we have stronger empathy with people that we know well, and
stronger personal involvement when we have a closer contact with someone.
1hese emotions (ol lriendships) inlluence our decision making, which is not
necessarily immoral.
1o make the theory ol justice less demanding, we can allow lor some partiality.
lartiality can trump the impartial theory ol justice described in the previous
section, under two conditions: 1) the violation ol impartiality should not be too
strong, and 2) the level ol partiality should be universalized according to the
lantian categorical imperative: we should want to live in a world where everyone
behaves with similar levels ol partiality. 1hese two conditions are related: il the
level ol partiality is too high, il the Qmm-theory is too much violated, we would
not want to live in a world where such levels ol partiality are universalized. 8o
tnica| ccnsistcncy anJ Anima| ua|ity
118
partiality can weakly overrule the Qmm-principle, and the principle ol
universalization (ol section 1.2) is crucial. lartiality is allowed to some degree as
long as we are willing to respect similar levels ol partiality ol everyone else.
.1 1olerated choice equality
1he inclusion ol a tolerated partiality in ethics generates an equality principle
which is dillerent lrom the prioritarian equality principle ol the previous section.
lor reasons to be discussed, l will call it 'tolerated choice equality'.
consider an example ol the burning house dilemma. ln animal rights
discussions people sometimes give such a dilemma, whereby we have to choose
between rescuing our child or a dog lrom a burning house (oary lrancione (2000)
also relerred to this dilemma in his book Your child or the dog"). Ol course the
meat eater expects that also animal rights activists would save their own child, so
they point at this kind ol partiality to justily speciesism. 1he argument can easily
be countered by changing the dilemma a bit: choose between your child and a
child with another skin color. Ol course, people are not necessarily racist when
they preler to save their own child. l will explain why not.
1here is indeed an cmcticna| incua|ity between children. lut suppose you were
at the burning house, and you chose to save the other child instead ol mine. ll l'd
tell you that my child has a higher moral status and a stronger right to live, due to
its skin color, l would be a racist. lut il l tolerate your choice to save the other
child instead ol my child, l would not be racist. 1he reason is that l consider you
and me to be morally equal, and the children in the house inherit this kind ol
equality. l tolerate your partiality, and therelore the children have inherited a
tolerated choice equality which is not in contradiction with the emotional
inequality that l leel. 1his new principle ol equality can be lormulated as lollows:
1he tolerated partiality principle. You are allowed to be partial as long as
you tolerate similar levels ol partiality lor everyone else, and il the
partiality is not based on lalse beliels or prejudices.
more specilically: when helping others, you are allowed to give (to some
level) priority to those with whom you leel a personal or emotional concern
or involvement, on the condition that you should tolerate the choice ol
other caregivers to give priority to whom they preler (their loved ones). 8o
you should tolerate the choice ol other helpers.
1olerated choice equality. ll 1) you want to help a person X (lor whom you
leel an emotional concern) and another helper wants to help person Y, il 2)
artia|ity anJ tc|cratcJ cncicc cua|ity
119
you consider the other helper as being equal to you, and il !) you tolerate
the choice ol the other helper to help Y, then persons X and Y inherit a
tolerated choice equality.
.2 1o whom applies the tolerated choice equality7
ln this section l try to answer the question who we need to take into account lor
this new principle ol equality7 when we look at consequantialist principles such as
Qmm-prioritarianism, it was sell-evident that all sentient beings should be taken
into account, because well-being is what matters lrom behind a veil ol ignorance.
8uppose in the burning house dilemma l had to choose between saving a child, a
dog or a car. ln principle, as with the veil ol ignorance, all entities in the universe
should be taken into account, including cars. lut il l saved the car, it could not be
tolerated, because the Qmm-principle will be violated lar too much. 1herelore, the
tolerated choice principle should only be applicable to sentient beings. lt 'inherits'
this criterion ol sentience lrom the Qmm-principle, and in the Qmm-principle it
was derived lrom the veil ol ignorance. cars are not sentient, so we cannot
inlluence their well-being, no matter what we do. Another reason why the
tolerated choice principle is applicable to all and only to sentient beings is that the
principle stems lrom leelings ol personal connection and empathy with others. Ol
course leeling empathy only makes sense towards sentient beings. we cannot leel
empathy with a car.
lt is important that this partial aid or care is motivated by leelings ol empathy
and concern. 1olerated choice equality therelore meets some criticism by
proponents ol an ethics ol care, who claim that impartiality is too 'cold'.
ooing back to the dilemma between saving a child or a dog, a true antispeciesist
should have to tolerate the choice ol someone who saves the dog. Note that a lot ol
people give more lood and medicines to their pet dog than to starving children,
but those choices are already tolerated to some degree. According to the Qmm-
principle, we would have to calculate the well-being ol all individuals involved,
whereby we have to take into account the lile expectancies and potential levels ol
well-being (some sentient beings have a richer and more complex emotional lile
than others). 8o it might be argued (lrom behind the veil ol ignorance) that saving
a mentally healthy human child would better correspond with the Qmm-principle
than saving a dog. Nevertheless, the principle ol tolerated choice says that some
slight violations ol the Qmm-principle should be tolerated. Otherwise we end up
with a too demanding impartial view that is in contradiction with some ol our
tnica| ccnsistcncy anJ Anima| ua|ity
120
strongest moral intuitions. You might save a mentally disabled child instead ol a
mentally healthy child, even il the disabled child will have an opportunity lor
well-being as high as a dog, i.e. lower than the healthy child.
let's give another dilemma to clarily some points. A trolley is moving at great
speed. On the main track you see someone you hold dear (e.g. your child or your
partner). You are standing next to a switch. You can turn the switch to save that
beloved person, but on the side track there is another person whom you do not
know. You have the choice between doing and allowing harm. ll you do nothing,
you allow your beloved person to be harmed. ll you turn the switch, your action
causes harm to someone. According to our theory, there is no morally relevant
distinction between doing versus allowing harm in this set-up. You are allowed to
turn the switch to save that beloved person.
what il on the side track there are two persons7 Are you allowed to be partial to
such a degree that you cause more harm by your action7 ll you turn the switch, l
can understand your choice, so l might tolerate it a little bit. lut only a little bit,
because your action delinitely violates the Qmm-principle. my intuition says that
we should not cause more harm by saving someone we hold dear. lut there is
another subtle slightly similar situation: imagine there are three tracks. ll you do
nothing, the trolley will take the main track and kill live unknown people. You can
turn the switch, so the trolley takes the second track and will kill someone you
hold dear. 1he third alternative is turning the switch lurther, sending the trolley
to the third track where it will kill two unknown people. ln this case, you are
allowed to turn the switch to save the live people on the main track. lut my
intuition says that you are allowed to send the trolley to the third track. l.e. you do
not have to choose the second option that saves most lives and best respects the
Qmm-principle.
ln section 6.1 l will elaborate more on these kinds ol dilemmas. l will put
lorward another principle that backs up the above intuition that we are allowed to
send the trolley to the third instead ol the second track, killing two people instead
ol one. And l will demonstrate that there are situations where we are not only
allowed to tolerate the partial choice ol a helper, but that we sncu|J tolerate it in
order to respect the helper.
what is relevant in this distinction between doing and allowing is the intention.
8uppose on the main track there are live people whom you don't know, and on the
side track there is one person whom you really hate. You always wanted to kill
that person, so now you see your chance to do so by turning the switch. 1he
trolley will take the side track and kill that hated person. 1he action itsell is
allowed because it is in agreement with the Qmm-principle. lut the intention is
wrong. 8o you would not be punished lor turning the switch, but you are a risk to
society by having malicious intentions. we can't trust you anymore to respect the
artia|ity anJ tc|cratcJ cncicc cua|ity
121
Qmm-theory in the luture. lmprisonment might be necessary, not because ol a
punishment, but because ol a protection. You have a duty to change your
malicious intentions.
l have given examples to indicate that some ol our moral intuitions say that the
tolerated partiality principle weakly overrules the Qmm-theory ol prioritarian
equality. 1his adaptation ol the theory is not inconsistent, it is better in line with
our intuitions, and therelore l consider it a better theory. ln chapter 6 we will
encounter moral intuitions that generate a principle that more strongly overrules
the Qmm-theory as well as the tolerated partiality principle. lelore we move to
that section, l brielly discuss a possible unilication ol tolerated choice equality and
equality ol opportunity.
.! 1olerated choice equality and equality ol opportunity
1he principle ol equality ol opportunity (Arneson, 2008) says that all people who
are equally qualilied should have equal prospects lor beneliting positions such as
jobs. }obs should go to the most qualilied (e.g. most talented and motivated)
persons and not to persons lor arbitrary, irrelevant reasons such as sex or race.
1olerated choice equality says that we are allowed to be partial to some degree.
ll a heterosexual man prelers to marry a woman instead ol a man, one might say
that this heterosexual man violates the equality ol opportunity between women
and (homosexual) men. 8o the heterosexual man is partial towards women, and we
tolerate such partiality. 1he heterosexual man is not sexist il he tolerates the
choice ol homosexual men to marry homosexual men. ll he says that no man is
allowed to marry a homosexual man, then it would be sexist.
Now consider a white employer who chooses a white job applicant instead ol a
black person. 1his choice might violate equality ol opportunity, and it might be a
racist kind ol discrimination. ls the employer willing to tolerate similar degrees ol
partiality ol everyone else7 1he employer might say that sJhe would tolerate the
choice ol another employer to hire a black person. lut this does not yet guarantee
tolerated choice equality, because in a competitive market the employer might
look at other employers as competitors, and hence as being unequal in some sense
(see the condition in the above lormulated principle ol tolerated choice equality:
you should consider the other helper as being equal to you). 1he employer might
have prejudices towards black people, thinking that black people are not good
workers. 8o the employer might be glad to know that other employers hire such
bad workers.
tnica| ccnsistcncy anJ Anima| ua|ity
122
1he above implies that in competitive environments, tolerated choice equality
cannot always be derived: il employers are competitive, an employer cannot
justily a partial choice (a partiality towards some job applicants) by tolerating the
choices ol the other employers. lor the heterosexual man, there was no
competition with homosexual men, so tolerating the choices ol those homosexual
men generates tolerated choice equality between men and women. 1he same goes
lor a man who is more sexually attracted to white women than to black women. le
prelers to marry a white woman, but il he tolerates the choices ol other white men
to marry black women, especially il he considers those other white men as his
equal (lor example il his brother wants to marry a black woman), he would not be
racist.
lt is allowed to violate equality ol opportunity, as long as there is a tolerated
partiality (tolerated choice equality). we can lurther explore the dillerence
between a racist employer who prelers to hire a white job applicant, and a non-
racist man who prelers to marry a white woman. 1he motivation becomes
important: what drives the employer to preler white people7 ll the employer is
alraid ol black people, sJhe is not necessarily racist. lut il sJhe has prejudices
towards black people, it becomes discrimination. 1he employer can overcome
those prejudices. 1he man who prelers white women can also overcome his
prelerence lor white women. lut this prelerence is much more dillicult to
overcome than the prejudices. changing opinions is easier than changing taste
prelerences. 1he tolerated partiality principle relers to personal or emotional
concerns, and such concerns are not easy to overcome.
1he above can be summarized in the lollowing combination ol tolerated choice
equality and equality ol opportunity.
1olerated choice equality ol opportunity. 8uppose persons A and l oller two
similar positions
1
and persons X and Y compete lor the position ollered by
A. A prelers to give the position to X. 1hen A respects the tolerated choice
equality ol opportunity il the lollowing conditions are satislied: 1) A
considers l as being equal (excluding competition between A and l), 2) A
would tolerate the choice ol l to give l's position to Y, and !) A's prelerence
lor X is not based on lalse beliels (prejudices) but on taste prelerences that
are not easy to overcome.




1
lere, the ollered positions can be interpreted as jobs, permissions to marry someone,.

12!
chapter 6 lasic right equality
6.1 moral dilemmas and strong moral intuitions
ln a previous chapter (!.!) we encountered some trolley dilemmas. 1hose
dilemmas were an interesting tool to demonstrate the importance ol deontological
judgments. ln the lirst dilemma, relerred to as the switch dilemma, a trolley is
going to kill live people on the main track. lowever, you can hit a switch so that
the trolley takes a side track, to save those live people. Lnlortunately, on this side
track there is one person. 1he structure ol the dilemma is: doing nothing results in
the death ol live people, acting (pulling the switch) results in the death ol one
person. Our theory ol prioritarian justice states that one person dead is better than
live people dead, so we should turn the switch.
lowever, in another dilemma, relerred to as the bridge dilemma (similar to
situation c in the previous chapter !.!), again a trolley is about to hit live people.
You can push a heavy man lrom a bridge in lront ol the trolley in order to block
the trolley. 1he heavy person will die, but the live people on the track will be
saved. A lot ol people have the intuition that pushing the heavy man is not allowed
(lauser et al., 2008).
ln a third dilemma, called the hospital dilemma, live patients in the hospital
need new organs in order to survive. lowever, no organs are available anymore. ls
it allowed to kill a visitor (against his will) and use his kidneys, liver, heart and
spleen lor transplantation to save the live people7 lere as well, most people are
deontologists: they are very reluctant to allow such actions, even il they could
save more lives.
what are the distinctions between those dilemmas's that can explain the
dillerent judgments7 As oreene (2001, 2004) pointed out, there is a dillerence
between up-close-and-personal situations (pushing the heavy man), and more
distancedJdetached situations (pulling a switch). 1his results in an emotional
inequality that we can tolerate (see previous section about the tolerated choice
tnica| ccnsistcncy anJ Anima| ua|ity
124
principle). lut there is more. what il we did not have to push the heavy man, but
simply push a button that will topple the heavy man lrom the bridge7 my intuition
says that even then, action is not permitted.
6.1.1 A lirst approach: uncertainty aversion
let us look at this bridge trolley dilemma lrom behind the veil ol ignorance.
8uppose you don't know who you will be: you can be any ol the six persons
involved (the heavy man on the bridge and the live people on the main track). You
can now decide between two possible worlds. ln the lirst, the heavy man will not
be pushed, in the second he will. which world would you preler7 ll you are really
sure that the plan to block the trolley by the heavy man will work, you would
rationally speaking rather be in the second world, because your chances ol
survival are live times higher. Only il you were the heavy man, you would die. ln
the lirst world, you would die il you are one ol the live people on the track. lut
now suppose, as in real lile situations, you are actually not sure that the plan ol
blocking the trolley will work. lerhaps the trolley is too last and the heavy man
not heavy enough to stop it7 1hen one or more people on the track would die as
well.
we already mentioned that lrom behind the veil ol ignorance, we might have
risk aversion that results in a quasi-maximin strategy. lut risk aversion implies
that we know the probabilities ol survival. ln this case however, we don't even
know the probability ol the plan to work. we don't know the chance on survival. lt
is a situation, not ol risk, but ol uncertainty (or ambiguity).
ln order to understand the ellects ol uncertainty, let's consider the example ol
the lllsberg paradox (lllsberg, 1961). An urn contains 60 balls. You know that
there are six dillerent colors ol balls, and that there are ten green balls. 1hat's the
only inlormation you have. You can now choose between two games ol chance. ln
the lirst game you win when you draw a green ball. ln the second game you win il
you draw a blue ball. leople who have strong uncertainty aversion preler the lirst
game, because then they at least know their probability to win (1J6). ln the second
game, they only know that their probability is somewhere between 0 (il there are
no blue balls) and J6 (il there are only green and blue balls). 8o people can not
only have risk aversion, they can have uncertainty aversion as well.
looking back at the trolley dilemma, we also have a choice between two games
ol chance. ln the lirst game (the world where the heavy man is not pushed), you
have a probability to survive (to win) equal to 1J6. ln the second game, you don't
know your probability ol winning. You have uncertainty about probabilities, and il
you have an aversion lor such uncertainties, you'd preler the lirst game.
8asic riqnt cua|ity
12
we see that the veil ol ignorance already comes pretty close to a lot ol our
moral intuitions. lirst, it values impartiality and well-being, as in a
consequentialist theory. 8econd, having a high but not maximum risk aversion, we
arrive at a prioritarian justice, which is coherent with our empathy and small but
not zero need lor elliciency. 1hird, having uncertainty aversion, we arrive at
moral judgments that correspond with some deontological moral intuitions. ln a
later chapter on the predation problem (see lart lll), we encounter another
implication ol uncertainty aversion lrom behind the veil ol ignorance, which
might be able to explain our tolerance towards predation.
we might ask ourselves the question how much risk aversion and uncertainty
aversion we should (or are allowed to) have. 1he veil ol ignorance does not
provide an answer. lut we could introduce a second veil. lehind the lirst veil, you
are a moral agent who does not know which being in the real world sJhe will be.
lehind the second veil, you do not know what kind ol moral agent sitting behind
the lirst veil you will be. 8o imagine that you are behind a second veil: you know
that you will soon be a rational (moral) being that is about to perlorm a thought
experiment ol the veil ol ignorance. lut at this moment you don't know how much
risk and uncertainty aversion you will have. You do know that most rational
beings (moral agents) have some risk and uncertainty aversions (because based on
psychological studies, most humans have these aversions). 8o likely you will also
get a high level ol risk and uncertainty aversion when you are behind the lirst veil.
lrom behind this lirst veil ol ignorance, having uncertainty aversion, you might
preler a situation where the Qmm-principle might be violated il the alternative
would be a situation ol uncertainty. lut not all has been said yet. Our moral
intuitions say that we are not allowed to push the heavy man or sacrilice a visitor
in the hospital lor transplantation, even il we can be very sure that the plan ol
saving live other people (the people on the track or the patients in the hospital)
will work. 8o let's look lor a universalized ethical principle that clearly expresses
these moral intuitions. l will lirst criticize some tentative accounts encountered in
the literature.
6.1.2 1entative ethical principles
ln a previous chapter we encountered some possible explanations lor the
dillerences in moral judgments in the trolley dilemmas. 8ome people (loorse,
1984, larris, 2000, lostow, 1989, waldmann and uieterich, 200) proposed that
there is a morally relevant distinction between sending a trolley to the victim
(which is done in the switch dilemma) and sending a victim to the trolley (which is
tnica| ccnsistcncy anJ Anima| ua|ity
126
done in the bridge dilemma by pushing the heavy man). lut we demonstrated that
this was a moral illusion.
Also in relation to the hospital dilemma, people make a distinction between death
by an existing threat (e.g. an organ disease) and introducing a new threat (killing a
visitor with a knile). lut such dillerences appear to be artilicial constructions.
consider the lollowing dilemma. live persons are on a moving platlorm on the
rails. ll you do nothing, the trolley will crush the platlorm and kill those people.
lut you can move the platlorm away lrom the rails in order to save the live. lut
this move will push another person (who is next to the platlorm) to an electric
cable. 1his one person will consequently die by electrocution. l believe that action
is allowed. lut here we see that lirst a new threat is introduced (the electric cable),
and second the victim is pushed towards this threat (the cable is not moved
towards the victim).
8ome people (e.g. lamm, 1989) say that there are morally relevant dillerences
between the causal chains in the switch and the bridge dilemmas. ln the bridge,
the action lirst results into threatening and harming the heavy man, and alter that
the live people are saved. ln the switch dilemma, the action simultaneously saves
the live on the main track and threatens the one on the side track. 1he harm done
to the one person on the side track occurs later in the causal chain, compared to
the harm done to the heavy man. 8o the structure ol the causal chains is dillerent.
1he problem with this approach is that we can invent dilemmas such as situation l
encountered in a previous chapter (!.!), where it becomes complicated to see what
the morally relevant aspects ol the causal chain are. And neither does it seem
really relevant when harm is done in a causal chain. A 'causal myopia' might also
be a moral illusion just like we demonstrated that an intervention myopia was an
illusion.
8ome people (keibetanz, 1998, mclntyre, 2001, lischer & kavizza, 1992, 8haw,
2006) reler to intentions and the uoctrine ol uouble lllect to justily the
dillerences between the switch and the bridge dilemmas. 1he doctrine says that
there is a moral dillerence between the intentional harm as a means versus the
loreseen harm as side-ellect (Quinn, 1989b). One critique is that this doctrine
relers to the moral dillerence between an intentional harm as a means and a
loreseen harm as a side-ellect. lowever, the interpretation ol this dillerence
becomes very dillicult in e.g. the loop trolley dilemma (discussed in section !.!). 8o
the interpretation ol this doctrine, and its application to trolley dilemmas, is not
clear. ln the appendix, this point is discussed a bit lurther. A second critique is that
actions can be permissible even when agents have bad intentions. lor example
when person A hates person l who is on the side track, and person A turns a
switch that sends a runaway trolley to person l in order to kill him, this act is
8asic riqnt cua|ity
12
permissible il turning the switch implies saving the people on the main track.
lerson A's intentions and moral character are bad, but my intuition says that the
act is good.
1

All the above tentative explanations and principles could be applied to animal
ethics, even il they are moral illusions. lut l preler a clear criterion that
distinguishes between the dilemmas, a criterion that can be translated in
something morally relevant, such as a right. ll we could reler to such a right, then
we might arrive at a new principle ol equality, where everyone has an equal claim
to this right. 1he right not to be killed will not do, as in both trolley dilemmas an
innocent person will be killed when the agent acts (when he pulls the switch or
pushes the heavy man). As we have seen in a previous chapter, another special
right is able do the job.
6.2 1he basic right and the mere means principle
1he right that solves the problem ol the dillerence between the switch and the
bridge dilemmas is the deontological or basic right not to be used as merely a
means to someone else's ends. we can see that the heavy man will be used as a
trolley blocker (human shield), and that the visitor in the hospital will be used as
an organ donor. lut the person on the side track will not be used as merely a
means to save the others. One does not need that person in order to pull the
switch and save the live.
1his basic right does not lollow lrom the veil ol ignorance, although it can be
compatible with it in situations ol uncertainty mentioned above. Neither does the


1
1his intuition is consistent with the universalization criterion that locuses on rules instead ol acts
(You may jc||cw tnc ru|c that everyone may lollow in similar situations."). 1he criterion says you may
do an action il you can lind a rule that justilies the action, il the rule is compatible with the ethical
system and il you can want to see this rule universalized. 8uppose l kill a person on the side track. l am
allowed to do this action il l can lind a justilying universalized rule. A rule that will not work, is: 1urn
switches in order to kill people you hate." ll l only have such a rule, l am not allowed to act. lut l can
lind another rule, such as 1urn switches in order to satisly the prioritarian theory, as long as no other
principles ol the ethical system are violated." l can do the same act (turning the switch) by lollowing
this rule. 1his universalized rule can justily turning the switch, even il in reality l happen to have a bad
intention. ln other words: a (bad) intention is not important in judgments ol permissibility ol an action
il there is a justilying rule lor that action that does not reler to the (bad) intention.
tnica| ccnsistcncy anJ Anima| ua|ity
128
basic right lollow lrom a leeling ol empathy. 1he basic right has dillerent origins,
and is highly coherent with many moral concepts and moral intuitions.
1) 1he basic right stems lrom a leeling underlying rcspcct, which is considered a
moral virtue to be developed. 1reating someone as merely a means is not
respectlul.
2) 1he basic right is related to the notion ol intrinsic va|uc, which is to be
distinguished lrom instrumental value. we give something intrinsic value when
that thing is important (valuable) beyond its use value.
!) 1he basic right is related to the notion ol Jiqnity and resembles a version ol
the lantian categorical imperative: Act in such a way that you treat humanity,
whether in your own person or in the person ol any other, never merely as a
means to an end, but always at the same time as an end." (lant, 18) According to
lant, humans have dignity. lut as l am going to demonstrate, there are good
reasons why not only humans but also other sentient beings have dignity and
should get a basic right.
4) 1he principle ol the basic right is a universalized ethical principle that is
consistent with our mcra| intuiticns in a |ct cj mcra| Ji|cmmas. l will list ten ol them.
a) 1he trolley dilemma: we are not allowed to push a heavy man lrom a
bridge, to lall in lront ol a runaway trolley, in order to block the trolley that
is about to kill live people on its track. we should not use a person as human
shield (1homson, 198).
b) Organ transplantation: we are not allowed to sacrilice a person against
his will, using his organs in order to save live patients in the hospital who
will die without new organs.
c) luman cannibalism: survivors in a lileboat should not sacrilice and eat
another person in order lor them to stay alive. we should not use a person as
lood.
d) lnvoluntary experimentation: we are not allowed to perlorm
experiments on a person in order to lind a therapy that will save many
people. we should not use a person as laboratory equipment.
e) 1error bombing: we are not allowed to kill a lew innocent civilians in
order to demoralize the enemy, win the war and save more lives.
l) 1orture interrogation: we are not allowed to torture a person in order to
gain inlormation about a bomb that will kill many people (especially when
the person is an innocent eyewitness who discovered the location ol the
bomb but is threatened by terrorists not to reveal the location, or when the
person is the innocent child ol the terrorist and the terrorist will only reveal
the location when his child is tortured).
8asic riqnt cua|ity
129
g) llackmail murder: we are not allowed to kill an important person il a
terrorist threatens to kill live hostages instead. we should not use a person
as ransom.
2

h) Nude photography: we are not allowed to take and sell nude pictures ol
a person against hisJher will in order to satisly thousands ol porn
consumers.
i) oang rape: a woman should not be lorced to have sex with lrustrated
men. we should not use someone as sex toy, even when the increase in total
well-being ol the rapists would more than compensate lor the loss ol well-
being ol the victim.
j) luman zoos: we should not lock up a strange looking person in a circus
or zoo lor the entertainment ol many other people.
k) 8capegoat: we should not prosecute an innocent individual, even il such
a prosecution would stop a riot that will kill many people.
!

l) 1he sadistic conclusion: we should not allow the birth ol someone who
will have a lile not worth living, even il the (weighted) average well-being
were negative (i.e. even il the wellare lunction would increase by
introducing a lile that is not worth living, see appendix 2, loss aversion").
many more situations can be given, such as trallicking (buying and selling
humans), slavery or gladiator lights. All these practices have something special in
common: a person is used as a means (as a human shield, trolley blocker,
experimental object, deterrent, inlormation source, ransom, sex object, toy,
painting, scapegoat,.) against his or her will. many people have the intuition that
the above practices are impermissible, even il the overall consequences in terms ol
lives and well-being would be better. 1he more examples we can give ol situations
where an action is not allowed when the victim is used as merely a means, the less
likely it is that our intuitive moral judgment ol impermissibility is a moral illusion.
1he coherence between those dilemmas gives credibility to our moral intuitions
and our corresponding universalized ethical principle.
) 1he basic right is also related to a restricted kind ol prcpcrtarian |iucrtarianism.
1his kind ol libertarianism is based on ucJi|y autcncmy, as a special property right
over one's own body. ln contrast with most libertarian theories, this restricted


2
1his example is similar to the thought experiment proposed by williams (8mart & williams, 19!,
p.9-100): should }im kill one lndian il relusing to kill this lndian implies that armed men will kill
twenty lndians instead7
!
1his example is similar to the dilemma ol 'lraming the innocent man' (mccloskey, 196): a race riot
(angry white people retaliating and killing black people) can be stopped by quickly arresting an
innocent black person, bearing lalse witness and punishing him in order to quiet down the situation.
tnica| ccnsistcncy anJ Anima| ua|ity
1!0
version allows some space lor distributive justice and avoids the conclusion that
taxation is comparable to slavery. 1his will be discussed in the next section (6.!).
6) 1he basic right is a lormulation ol the mcrc mcans princip|c. lt can be
interpreted as a specilication ol the Jcctrinc cj Jcuu|c cjjcct (mclntyre, 2011). 1his
doctrine ol double ellect (uul) relers to an action that has a good and a bad ellect.
One ol its crucial conditions states that such a double ellect action is not allowed il
the bad ellect is intended as a means to the good ellect (or il it is intended as an
end in itsell). 1he problem with the uul is that in some moral dilemmas it is not
always clear what it means to intend (rather than loresee) a harm as a means to an
end. 1he basic right (mere means) principle avoids the intention-loreseeing
distinction and instead specilies the means-end distinction. As we will see, this is
done by a more algorithmic procedure: the counterlactual question whether the
end could be achieved il the victim was not present.
) As l will discuss below (section 6.6), we can extend the mere means principle
in a way that it is not only immoral to usc but also to ccnsiJcr someone as merely a
means. 1his extension can explain other deontological intuitions such as the
dillerence between Jcinq vcrsus a||cwinq (loward-8nyder, 2011), the permissibility
ol partia|ity in impcrjcct Jutics cj ucncjiccncc (leauchamp & childress, 2011), and the
asymmctry cj prccrcaticna| Jutics (Narveson, 196, mulgan, 2006, mcmahan, 2009). lt
is as il this extended mere means principle unilies a lot ol deontological principles
and intuitions.

ln section !.4, l already demonstrated that the deontological basic right is not
necessarily a moral illusion. ll we now look at the above points, we see a strong
coherence ol the basic right principle with moral virtues (respect), moral
intuitions (in at least ten dillerent dilemmas, as well as in situations ol imperlect
duties and procreational duties), moral concepts (intrinsic value, bodily autonomy,
libertarian property) and deontological principles (double ellect, doing versus
allowing). 1his strong coherence indicates that the intuitions underlying the basic
right principle are not moral illusions. 1he basic right principle is not arbitrary,
artilicial or larletched. 1he principle can be clarilied, as l will do in the next
section, so we avoid luzziness as well. And looking at the totally dillerent
situations where it applies (the abovementioned ten dilemmas), we see that it is
not really context dependent. Another argument to see why it is not context
dependent is that the basic right is something that individuals always have,
independent lrom the situation. (compare it with the lact that in the mller-lyer
illusion a length is something that line segments intrinsically have, independent
lrom their environment). 1his is something dillerent than the tentative solutions
presented in the previous section (and chapter !.!), such as the 'protophysical'
explanations.
8asic riqnt cua|ity
1!1
Ol course, the above is not solid prool, the basic right might still stem lrom
moral illusions. Note that not everyone has the intuitions underlying this basic
right. lut these intuitions seem to be culturally independent. lt has more to do
with dillerent brain (mal)lunctionings (oreene et al., 2001). most people (most
moral agents) have these intuitions, and they are likely inborn. leople who lack
those intuitions might still be able to derive a consistent ethic, which will more
resemble a utilitarianJconsequentialist ethic. lor those utilitarians, animal
equality has to be applied as well (e.g. 8inger, 19). lere, however, l lollow and
respect those 'basic right'-intuitions that the majority ol moral agents appear to
have.

looking at the lormulation ol the basic right - use someone as merely a means
lor someone else's ends - we have to answer three questions:
1) what do we mean by use as 'merely a means'7
2) what do we mean with 'ends'7
!) who is the 'someone'7 l.e. who gets the basic right7
1he lirst question will be answered in the next section. Questions 2 and ! are
related and will be dealt with in the subsequent section (6.4).
6.! when is the basic right violated7
1he mere means principle linds its roots in lant's categorical imperative (lant,
18): never treat a person merely as a means to an end, but always at the same
time as an end. Yet, this landmark principle in deontological ethics lives on being
(re)interpreted and discussed until today (lor recent work, see e.g. 8canlon, 2008,
ch. !, larlit, 2011, ch. 9, lerstein, 2009).
when do we use someone as merely a means7 8lavery, human trallicking, rape,
cannibalism, involuntary organ donations, involuntary human experiments and
pushing a heavy man lrom the bridge in order to stop a trolley are all examples ol
basic right violations ol humans. what have these situations in common7 And how
to distinguish these examples lrom actions that do not violate the basic right7 l.g.
using a baker to get some bread, using an employee, sending your children to
school against their will, imprisoning a criminal or killing a person on a side track
in order to save live people on the main track. 1hese actions are not immoral, and
therelore should not be classilied as basic right violations.
tnica| ccnsistcncy anJ Anima| ua|ity
1!2
6.!.1 1wo words, two conditions
1he 'mere means' principle that generates the basic right, contains two words.
lence, two conditions need to be satislied. 1he lirst condition says that an agent
(the user) lorces (in the broadest sense) the victim to do or undergo something
against the will or interests ol the victim (lor example the victim does not want
the treatment). 1his is the 'mere' part. 1he second condition says that the
presence ol the victim is required in order to reach an end ol the user or someone
else.
4
1his is the 'means' part.


1he lirst condition, the 'mere' part ol the mere means principle, is something a
(rule
6
) utilitarian or consequentialist can agree with: doing or undergoing
something against your will generally lowers your well-being and can be
considered as a harm. lt is the second condition that gives the mere means
principle its deontological llavor. looking at the above examples, we see that the
presence ol the victim is required in order lor the plan to work. ll the innocent
civilian was not present in the terror bombing situation, the enemy would not
become demoralized. ll the important person was not present in the blackmail
situation, you could not kill him and use his death as a ransom to lree the
hostages. ll the lat man was not on the bridge, you could not push him and use
him as a shield to block the trolley. On the other hand, consider the switch trolley
dilemma: most people claim that when a runaway trolley is about to kill live
people on the main track, we are allowed to turn a switch in order to redirect the
trolley onto a sidetrack, where the trolley will kill one person. Although killing
this one person likely goes against her will, she is not used as a means, because her


4
ll the victim undergoes something against hisJher will lor an end ol the victim, sJhe is not used as
merely a means lor someone else's ends, and the mere means principle is not violated. lor example
keeping a scared patient in a hospital lor hisJher own sake is not impermissible. 1his is particularly
true in wildlile rescue centers, where injured or ill animals are kept against their will in order to help
them. laternalism might sometimes be immoral, but not in the wildlile rescue center, because it does
not violate the mere means principle and it promotes well-being.

1hese two conditions come close to the two conditions in lognar & lerstein (2010 p.1). 1) A person
treats another person mcrc|y as a means il it is reasonable lor her to believe that something she has
done or is doing to the other person renders that person unable to consent to her treating him as a
means to her aim." 2) A person treats another person as a means il she intentionally does something
to the other's body or mind in order to realize one ol her ends and she intends the other's body or mind
to contribute to her end's realization." 1his condition relers to the required presence ol the body.
6
A rule utilitarian looks lor those rules that, when they would be consistently respected in all similar
situations, would in general generate most well-being. A rule utilitarian prelers to stick to the rule even
il in a particular situation a violation ol the rule would promote well-being.
8asic riqnt cua|ity
1!!
presence is not required in order to turn the switch and save the live people on
the main track.
1he same goes lor other situations where a victim could not possibly consent"
(to use lorsgaard's expression (lorsgaard, 1996, p.1!8)). A rule utilitarian can
agree with her: il the rule (the maxim) ol our action precludes the possibility ol
the victim's consent (the victim could not rationally will to be treated that way),
we are not allowed to treat the victim that way. lorsgaard gives the examples ol
deception and coercion. A rule utilitarian might preler to stick to the rule do not
lie", because such a rule generally promotes well-being. lut in contrast to
lorsgaard's view, l think the mere means principle is only violated when the liar
wants the presence ol the deceived person in order to reach someone else's end.
Another example is imprisonment: violating someone's liberty without consent.
lmprisoning a murderer does not violate his basic right, because the presence ol
the criminal was not necessary in order to reach the end (a sale society). On the
contrary, his absence was prelerred. lmprisoning him might be the best strategy
to reach the quasi-maximin principle, even il we deprive his liberty. lut using this
murderer lor lorced labor violates his basic right.
1he next two sections explain the two conditions in more detail.
condition 1: the agent's behavior violates the interests ol the victim.
when the victim has a will, this condition says that the victim does not want the
agent's behavior

. lor autonomous beings, this condition relers to autonomy and


consent, but there is a whole spectrum ol possible interpretations ol autonomy. At
the two extremes, there are stronger (narrower) and weaker (broader)
interpretations ol the 'mere' part in the mere means principle.
wnat is ccnscnt
1he strongest interpretation says that the victim is used as mcrc|y a means when
sJhe is not able to give rational, inlormed consent. l reler to leauchamp and
childress (2001, ch.4) lor a discussion on consent, but it is clear that according to
this interpretation, the mere means principle is only applicable to rational beings:
beings who are able to understand relevant inlormation and give lree consent.

Ol course the agent's behavior has to be related to the use ol the victim. ll l buy bread lrom my
neighbor, l use my neighbor as a baker. ll l annoy my neighbor when l put my music loud, l do not use
my neighbor as merely a means, because the loud music is not related to buying bread. lowever, il my
intention is to annoy my neighbor with loud music in order to coerce him to bake bread lor me, it will
become a use as merely a means.
tnica| ccnsistcncy anJ Anima| ua|ity
1!4
1his is the traditional interpretation ol lant (18), lorsgaard (1996) and many
others.
1he weakest interpretation says that the victim has to do or undergo something
that sJhe does not want.
8
ln this interpretation, not wanting something means:
having negative emotions about it (or having a negative attitude towards it).
lositive and negative emotions indicate that a being has subjective prelerences or
interests. lor example, when pain generates a loss ol well-being, it indicates that
the individual wants to avoid bodily injury. lear indicates a need lor salety, and
similar needs or interests lie behind other emotions. 1he mere means principle
now becomes applicable to all sentient beings, i.e. beings who have developed (and
not yet permanently lost) the capacity to experience positive and negative leelings
that indicate the satislaction ol prelerences. 1he advantage ol this weaker
(broader) interpretation is that the mere means principle is also applicable to
mentally disabled (a-rational) humans. 1his corresponds with the intuition ol
many people.
ln the next section on who gets the basic right, l discuss the possibility ol an
even weaker interpretation, leaving the notion ol consent behind, and locusing
exclusively on interests. 1his is a very broad interpretation, because now non-
sentient beings with interests (e.g. living beings such as plants) can be victimized.
wny ccnscnt
According to a consequentialist wellare ethic, consent is important and counts in
the utility calculus. lut the consequentialist does not see a dillerence between the
consent ol someone whose presence is required, versus the consent ol someone
else. 8o the dillicult question becomes: why is the consent ol the person whose
presence is required so much more important than the consent ol the person
whose presence is not required7 1he deontologist has dilliculty answering this
question. Ol course, sJhe can reler to the coherence ol hisJher moral intuitions in
the situations given in a previous section. Alternatively, sJhe can give a rationale
such as: il presence becomes important, autonomy dictates that consent becomes
especially important. lut as the next example demonstrates: it is not the general
(lack ol) consent ol the victim that is important. Only the consent about the
presence ol (the behavior ol) the user has a special status.
An cxamp|c jrcm cccncmics: tnc pccr uakcr


8
we should understand this in a broad sense, which includes not wanting deception. A deceived victim
might not actually experience negative emotions, but il this victim does not want to be deceived that
way, it counts as a violation ol consent.
8asic riqnt cua|ity
1!
1he condition that the victim does not consent with the agent's behavior is
important. consider a poor person who decides to work in a bakery. le hates
getting up early in the morning to bake some bread, but his poverty gave him no
choice except bake or die. ll l buy his bread, l am using him: his body is necessary
to make the bread. lut although he hates baking bread, the poor baker does not
have a negative attitude towards my behavior. ln other words: my behavior did
not cause his poverty. ll my behavior was not present, the baker would still be
poor. 1herelore, l am not using him as merely a means. On the other hand, il l
threaten or lorce someone to work in a bakery, it becomes slavery and l am
causally responsible lor his bad situation. 1he agent causes a violation ol the rights
ol the victim, il the presence ol the agent is a necessary condition ol the harm.
1imcjramc cj tnc aqcnt`s ucnavicr
A tricky question concerns the boundaries ol someone's behavior. consider a slave
owner who claims that his slaves are better-oll as slaves than they would be in the
wild, because as indigenous people in the wild they would lace predators, diseases,
drought, hunger and other nasty things. 1he owner protects the slaves and gives
them lood. 8o it might be true that a slave would preler a lile as a slave over a lile
in the wild. lence the slave prelers the total behavior ol the owner over the
complete absence ol the owner, il absence means a miserable lile in the wild.
According to a broad interpretation ol the behavior, this slave is not used as
merely a means.
lut based on my moral intuitions, l preler a narrow interpretation that locuses
at a particular behavior at a particular time. According to this interpretation, the
slave is used as merely a means as soon as the owner does a particular thing that
the slave does not want, lor example whipping the slave (even il the slave prelers
the total lile ol a slave with whipping over the alternative lile in the wild). 1he
same goes lor the practice ol breeding slaves: even il a slave would preler the lile
ol a slave over the absence ol a lile (the slave would not have been born il the
owner did not breed slaves), it does not mean that the slave is not used as merely a
means when sJhe is whipped.
1he latter resembles a situation ol livestock larming: what il the lile ol a cow
raised in a humane livestock larm is better than no lile at all and better than a lile
in the wild, but the cow is still slaughtered lor meat7 lor the cow, the procedure ol
breeding, raising and slaughtering, considered as a whole, might be prelerable to
not being born at all or being born in the wild (and e.g. being eaten by a predator
at an early age). lut we should not look at the procedure as a whole: it is the act ol
slaughtering itsell that violates the mere means principle. 8laughtering (lor meat)
is a single act that the cow does not want and where presence ol the body ol the
cow is necessary. According to the non-consequentialist (non-wellarist) mere
tnica| ccnsistcncy anJ Anima| ua|ity
1!6
means principle it is better that a cow is not born at all than that a cow with a lile
worth living is used as merely a means when she is slaughtered. 1his is the same
logic as with the whipping ol human slaves who have a lile worth living. we do not
have a duty to breed and raise happy cows (see the section on the asymmetry ol
procreational duties below), but once we cause the birth ol a happy cow, we
should not violate its basic right. we should not slaughter and eat the cow.
9

lt might be the case that other moral agents have another intuition than l have,
that they want to take a broad interpretation ol the mere means principle, judging
the morality ol a use in terms ol the total behavior instead ol a particular behavior
at a particular time. 1hose moral agents might conclude that some kinds ol slavery
and meat consumption are permissible: when the lives ol the slaves and animals
are worth living and when the alternative would be that those slaves and animals
were not born. 1o deal with this dillerence in moral intuitions, a democratic
decision procedure amongst all moral agents (everyone who is capable ol
understanding this moral problem) might be a way out.
condition 2: the agent wants the presence ol the victim's body
1he second condition ol the mere means principle can also reler to a mental state
ol the agent, namely what the agent wants. 1his subjective mental state is
connected to an objective lact: the presence ol the victim that is causally required
lor the end ol the agent. 1his can be tested by a counterlactual thought
experiment: does the agent's plan still work il the victim was not present7 ll not,
then the victim is used as a means.
lut what exactly should be present7 what belongs to the victim that should be
present7 looking at the situations presented above, we note something peculiar: it
is the victim's pnysica| ucJy that should be present. 1he body is used as a means, il,
lor example, the bodily integrity is violated (e.g. meat production, experiments,
organ transplantation, bodily manipulation), il there is a sexual act with the body
(e.g. rape, harassment), il the body is lorced to do something (e.g. slavery), il the
body is lorced to be somewhere (e.g. in a cage), il the body is photographed or
viewed (e.g. nude photography without consent, violations ol bodily privacy) or il
the body has an economic price (e.g. trallicking).


9
mcmahan (2008) also discussed this issue ol humane larming, arguing that we should not kill a happy
cow, even il the lile ol the cow being raised and killed might be better than no lile at all. 1his 'logic ol
the larder' (purchasing animal products is good because it can increase the number ol animals whose
lives are worth living) was also criticized by matheny & chan (200). One ol their claims is that animal
larms prevent positive lives ol wild animals.
8asic riqnt cua|ity
1!
wnat is tnc ucJy
ll the victim's body plays a central role, we have to ask the question: what is the
body7 One rather artilicial delinition ol a body is: the composition ol all living cells
with the same uNA that are connected to each other. 1he artiliciality makes this
delinition less suitable in ethics. we can also ask the question what about artilicial
limbs or tools that extend the body7 1hose extensions are strictly speaking not
part ol the body because the person does not have an internal representation ol
those extensions. laving an internal representation might be a morally relevant
condition lor something to belong to the body ol a person. lrom the early stages ol
development, a subject creates an internal representation ol hisJher body: sJhe
learns what is part ol her body and what is not.
lowever, there are dillerent kinds ol internal representations: l can say that
this arm belongs to me, because l can autonomously move it, or because l can leel
it. ll l have sensations (il l can leel lor example pressure, temperature and pain) in
something, that thing belongs to my body. A body can be delined as those things ol
which someone has sensations or internal representations (think about the
representations in the motor cortex and somatosensory cortex).
ll internal representations - and especially sensations - are important: what
about paralyzed or anesthetized limbs7 As the mere means principle is related to
the notion ol bodily autonomy, we can say that those limbs belong to someone's
body il that person still believes they belongs to hisJher own body. ln other words,
we should respect what the individual believes is part ol hisJher own body. 1his
beliel can either be a conscious cognitive state, or can be the internal
representation itsell.
lut there is more. what about your gut bacteria7 Or internal parasites7 8uppose
you are inlected with a (rather harmless) parasite. l want to cut you open in order
to do important experiments on the parasite. Or l want to kill you to use your gut
bacteria lor some important purpose. lven il strictly speaking those bacteria and
parasites do not belong to your body, l still violate your basic right, because l
transgress a bodily boundary. we could say that, broadly understood, everything
that can only be accessed by transgressing something that has sensations, belongs
to someone's body. You have sensations in your belly, so l cannot cut your belly
open.
Another way to look at the problem ol the use ol parasites and gut bacteria is
the observation that those things would not be present il the victim's body was
not present. ln other words, il l want to use your gut bacteria, the presence ol your
body (delined as those things ol which you have sensations or internal
representations) is required. And il you do not want to be treated (cut open) that
tnica| ccnsistcncy anJ Anima| ua|ity
1!8
way, the conditions ol the mere means principle are met. l use your body
indirectly without your permission, il l cut you open to reach your gut bacteria.
1he strange thing about the mere means principle is that it points to the
importance ol the body, but it is not (yet) able to clarily what exactly belongs to
the body. Although more has to be said on this, l will not discuss it lurther here.
let me conclude by mentioning that a vague boundary ol what belongs to
someone's body does not necessarily present a problem in the construction ol a
non-arbitrary consistent ethic, because the strength ol the basic right (see section
6.) might also have a gradation. we could couple this gradation with the
gradation ol how strongly something belongs to someone's body. ll a thing
delinitely belongs to someone's body, the mere means principle would be strongly
violated when that body part is used without consent. lut il it doubtlully belongs
to someone's body, the violation ol the mere means principle should not count so
gravely.
1he relerence to the body also leaves the mere means principle with another
very basic question.
wny tnc ucJy
what is so important about someone's body, to give it a privileged status7 lrom a
theory ol property rights, we could say that the body is the only thing that a being
owns completely. 1he body lalls under the absolute competence ol a person. 1he
deontological mere means principle, with its locus on the body, corresponds with
a restricted kind ol propertarian libertarianism. ln the libertarian theory ol
property rights, agents lully own themselves and can acquire property rights in
external things (Vallentyne, 2012). lropertarian libertarianism states that private
property is the sole source ol legitimate authority. lts non-aggression principle is
restricted to violations ol private property. 1hese property rights are non-
negotiable (Nozick, 194).
1he mere means principle lits in a propertarian libertarianism, where people
have a lull property right over their own bodies. leople have a lull bodily
autonomy. 1he mere means principle does not imply that external things are
owned to the same degree as bodies are owned. Owning a body is much more
important than owning an external object.
ln slavery and trallicking, the bodies ol the victims are treated as someone
else's property in the legal or economic sense. 1hey are merchandise. 1his
property status is not respectlul, because only the victims themselves possess
their own bodies. 1he victims do not have to be aware ol this property status. lor
example selling babies is immoral, even when the babies do not understand the
notion ol private property. According to animal rights activist oary lrancione
(2000), we should also abolish the property status ol animals. 8o we should not be
8asic riqnt cua|ity
1!9
allowed to buy and sell animals (e.g. buying a pet lrom a breeder), even il those
animals (like babies or mentally disabled humans) cannot be aware ol their
property status.
An cxamp|c jrcm cccncmics: payinq taxcs vcrsus jcrccJ |aucr
According to Nozick's libertarianism, raising taxes is in some way comparable to
slavery: the state appropriates a part ol the work and time ol persons, without
their consent (Nozick, 194, ch.). lut il a libertarian restricts absolute property
rights to only the body, taxation is no longer impermissible. kaising taxes is
possible (even il the presence ol the tax payer is necessary to raise the tax, and
even il the tax payer did not give permission), because money is not completely
owned by a person. money is not a direct product ol only the body ol the worker.
1he worker not only uses hisJher own body, but sJhe also uses something
external. A larmer's manual labor belongs to the larmer. lut the mere means
principle allows lor an assertion that the soil used by the larmer is not completely
owned by the larmer. 1he soil, and everything else that is external to the bodies ol
persons, belongs to society. 8o the state can say to the larmer that il the larmer
wants to use something external to his body, the state (society) has a right to
interlere to the benelit ol society. lence, a part ol the harvest can be given to
society, in order to benelit the total good ol society or the well-being ol the worst-
oll individuals, i.e. lor distributive justice. ll you don't want to give away a part ol
your harvest, line, but then you are not allowed to use something that is external
to your body and that belongs to society.
ln this sense, we can clearly distinguish taxation lrom slavery. lorcing someone
to do labor is not allowed, because in that case the body ol the slave plays a central
role. lt is not allowed to lorce someone against his will to use something external
to make a product. lut il persons themselves want to use something external to
make products, society has a right to interlere by taxation.
ll external things can be owned completely by persons, as in propertarian
libertarianism, utilitarians have no grounds at all to improve well-being, and there
is no space lor distributive justice. lut il absolute property is restricted to the
body, as in the mere means principle, utilitarians can still use taxation to improve
well-being and a (kawlsian) system ol distributive justice is still possible to some
degree. lence, the only playing lield lor consequentialists (such as utilitarians,
egalitarians, prioritarians) who want to improve or equalize well-being exists
when people want to use something external to their bodies.
1he deontological mere means principle takes a position between
utilitarianJegalitarianJprioritarian consequentialism and propertarian
libertarianism. 1he latter says that people can have an absolute property right
over everything, the lormer says that there are no absolute property rights, not
tnica| ccnsistcncy anJ Anima| ua|ity
140
even over the own body. According to a consequentialist, people can not only be
lorced to pay taxes, but can also be lorced to let their own bodies be used. we have
a duty to help others in need, by paying taxes to help the poor. lut the utilitarian
can go lurther by claiming that we also have a duty and should (even without our
permission) donate our blood or a kidney. 1he mere means deontologist can
respond that such donations are morally good but not obligatory or enlorceable.
1hey are 'supererogatory' (good but not obligatory) because our blood and kidneys
are completely owned by us as parts ol our bodies so we can decide what happens
to them.
6.!.2 conclusion
1he above two conditions give us a lairly precise, clear and nuanced picture when
a basic right is violated. ll the agent causes harm to a victim, violating his interests
in a way that the victim does not want, and il the presence ol this victim's body
was required in order to reach an end, then the victim is used as merely a means
lor someone else's ends. we see a dual role ol the presence ol the bodies ol the
agent and the victim: the presence ol the body ol the agent is a necessary
condition in the causation ol the harm, and the presence ol the body ol the victim
is a necessary condition in achieving the goal.
ln the next section l will argue who gets the basic right. 1here are a lot ol
beings, each with dillerent levels ol complexity and interests. 8o giving them all
an equal claim lor this basic right will be dillicult. 1here is a gradation in
complexity and interests, and there is also a gradation in someone's ends. could it
be possible to make a coherent picture by coupling those two gradations7 we will
see that the questions who gets the basic right7" and what are the ends7" are
related to each other.
6.4 who gets the basic right7
looking at the consequentialist Qmm-principle, it was easy to see to whom the
principle applied: all beings who have a well-being, i.e. all sentient beings. l also
argued that the principle ol tolerated choice equality naturally applies to all
sentient beings, because we can only leel empathic concern lor sentient beings. 8o
we might think that the basic right principle also applies to all and only to sentient
8asic riqnt cua|ity
141
beings. lut we have seen that this principle is not derived lrom the veil ol
ignorance, and it is not based on empathy.
1he two conditions ol the mere means principle indicate two criteria lor
granting someone a basic right. lach ol the two criteria has a broad and a narrow
lormulation. 1he lirst criterion relers to the presence ol the body, so the being
should have a body. more narrowly lormulated: the being should have an internal
representation ol hisJher own body (it should know where its body ends and the
environment begins). 1he second criterion relers to the interests ol a being, so the
being should have interests. more narrowly lormulated, the being should be able
to want something. ln this interpretation, the being should be sentient in the
sense that it has a well-being composed ol positive and negative leelings related to
(dis)satislaction ol prelerences. 1hose leelings and emotions indicate what a being
wants. 1his criterion can be narrowed lurther by requiring higher mental
capacities lor autonomy or rationality.
As the basic right is based on respect, there is a second way to solve the
question who gets the basic right. we can ask who or what earns respect7 my
guess is that respect is connected to something complex and vulnerable. 1here are
dillerent complex and vulnerable things in the universe, such as living beings and
sentient beings. 1hese beings are characterized by having complex interests. cars
or stones do not have complex interests, because they don't even act to protect
their interests. 1hey can have an interest not to be broken, but that is a trivial
interest. we might say that complexity in interests is related to respect. And as we
have seen, respect means that we should not violate someone's basic right. Now,
rights are nothing but devices to protect interests. 8o it is not larletched to couple
the notion ol interests with the basic right. low can we do this in a natural way7
lirst, we observe that there is a gradation ol complexity in terms ol a gradation
ol interests (needs). koughly speaking we have ncn-|ivinq cujccts with only trivial
interests and low complexity. ivinq ucinqs have complex interests (to eat, to
live,.) and they have a high complexity (uNA, metabolism,.). lut some living
beings can perceive their environments, or respond to their environments in even
more complex ways, because they have nervous systems that allow them to have
inner, neural representations ol their bodies and environments. Although they
are unconscious (like robots), these sensorineural, perceptive or rcspcnsivc ucinqs,
such as invertebrate animals, have even more complex interests and they have
complex reactions towards them.
lut some responsive beings have more: a central nervous system that generates
a perceptual consciousness. 1hey are subjectively aware ol their environments and
bodies. 1he representations ol their environments and bodies are accompanied
with 'qualia' (lyrne, 2010), the subjective, private, direct, conscious experiences.
1ogether with qualia, a sentient being has a locus or special attention towards an
tnica| ccnsistcncy anJ Anima| ua|ity
142
object (kamachandran & lubbard, 2001). lor example: the leeling ol touch in my
lingertips only happens when l locus on my lingertips. }ust belore l paid attention
to this leeling ol touch, l was not aware ol it. 1here was an unconscious neural
activity (no anaesthesia), comparable to what responsive beings might experience.
lut only alter l locused on my lingertips, it became a conscious experience or
'quale' ol touch. 1his locus or attention is important in the conscious experience,
and it might be possible to see this in the behavior ol some animals, because the
locus decreases the awareness ol other things. lor example, a cat locusing at his
prey is no longer paying attention to other things. Or a lish (e.g. a trout) injected
with a venom becomes preoccupied with the pain, so that it pays no heed to a
threat coming towards him (ll8A, 2009). 1hese are indicators that those animals
have qualia, because they are analogous to our behavior when we have qualia and
locus. Now, qualia are olten neutral. l don't leel an urge to avoid touching books.
1he touch ol a book has no inlluence on my will. lut other qualia are allective in
nature, they are evaluated as being positive or negative. lor example, the leeling
ol a needle in my linger generates a quale that l wish to avoid. 1his quale ol pain
generates an urge in me. 1hose allective or evaluated qualia are the positive or
negative leelings and emotions such as pain, lear, distress or joy. 1his is where
well-being comes into play. 1hese leelings are related to interests or needs, they
are nothing else but subjective experiences ol (un)satislied interests. lear
indicates that the need lor salety is not satislied, pain indicates that the interest ol
bodily integrity is violated, lrustration may indicate a need lor lreedom.
kesponsive beings who have evaluated (allective) qualia are called scnticnt ucinqs.
1hey are subjectively aware ol their interests, so they not only have interests, they
not only react to them in complex ways, but they can also subjectively leel them.
1hese are the beings that have a subjective well-being, so things subjectively
matter to them. 1hey want things. kesponsive beings with only unconscious
experiences or neutral qualia, have no well-being, because the well-being is
composed ol evaluated qualia that are positive (joy,...) or negative (pain,
lrustration,.). 1hese beings do not want anything.
linally there are the raticna| ucinqs. 1hese are sentient beings with a sell-
consciousness and rational agency. 1hey not only have complex interests, they not
only react to them, they not only leel them, but they know and understand them.
1hese beings have the most complex emotional lives, with a luture perspective,
dreams and projects. 1hese rational beings not only want things, they are also able
to give inlormed consent.
1he above distinction between rational, sentient, responsive and living beings
can oller us some extra degrees ol lreedom to construct a consistent ethic that
best lits our moral intuitions. we clearly have a gradation ol complexity ol beings.
Now, looking at the delinition ol the basic right, it relers to the use as merely a
8asic riqnt cua|ity
14!
means to someone else's ends. lut the ends also have a gradation. 1here is a
dillerence between luxury and vital needs. 8o it would be very natural to couple
the gradation ol complexity in interests to the gradation ol the ends. let's look at
this gradation in ends in more detail, lrom luxury needs to survival ends.
uxury: these are needs that have a positive contribution to someone's well-
being when satislied, but these needs are created by society. we can create
circumstances where these needs no longer need to be satislied in order to have
an increase in well-being. luxurious needs are volatile, relative and variable.
lxamples are lashion, social status symbols and needs created by commercial
advertisements.
8asic nccJs: these are needs not required in order to stay healthy and alive, that
have a positive contribution to someone's well-being, are stable and not
determined by society. lxamples are social contact, knowledge and recreation.
vita| nccJs: these are needs that need to be satislied in order to stay alive and
healthy, such as medicines and health care (e.g. new organs lor patients with an
organ lailure).
5urviva| cnJs: these are vital needs that are not only important lor individuals,
but lor biodiversity as well (e.g. survival ol species). lxamples are lood, water, air,
sexual activity (procreation) and motion. ln part ! (chapter 10) l will discuss the
predation problem, whereby we will see that there is a morally relevant
distinction between survival ends and merely vital needs. Vital needs are
characterized by one criterion: necessity. 8urvival ends, on the other hand, are
characterized by three criteria: natural, normal and necessary. Natural means that
the behavior is directly developed by evolution (genetic mutations and natural
selection), and as biodiversity is delined by everything that directly evolved lrom
evolution, natural behavior contributes to biodiversity. Natural plus normal means
that the behavior is natural and happens a lot, and therelore contributes a lot to
biodiversity. Natural plus normal plus necessary means that much biodiversity
will be lost when the behavior stops. lating lood is natural, normal and necessary.
Organ transplantations or medical experiments are necessary, but not natural or
normal. 1herelore, lor a patient in the hospital, new organs or medicines can be a
vital need but not a survival end. ln summary, survival ends are in some sense
stronger than vital needs. 1he dillerence between survival ends and vital needs is
related to the moral value ol biodiversity, which is threatened il survival ends are
not satislied.
1he lollowing ligure represents the coupling ol two gradations: complexity in
interests and ends. what we see is that our approach contains the lantian idea
that rational beings are never to be used as merely a means. lut we extend this
basic right to other beings. uoing this makes our theory more coherent with some
moral intuitions. 1he lirst intuition says that mentally disabled humans (non-
tnica| ccnsistcncy anJ Anima| ua|ity
144
rational beings) are not to be used lor vital, basic and luxury needs. 1he second
intuition is that it is sell-evident to couple the basic right with the notion ol
interests, because rights are devices to protect interests. 1he third intuition is that
it is sell-evident to couple the complexity in interests with respect lor that being,
and to interpret respect in terms ol the basic right not to be used as merely a
means. 1he lourth intuition says that it is sell-evident to couple gradations with
each other, and the lormulation ol the basic right in terms ol use as means lor
ends serves perlectly lor such a coupling. 1his coupling immediately solves the
question ol who gets the basic right.


ligure 9. 1he coupling between ends and complexity. An X means that the
being has a right not to be used as merely means lor the respective ends. lor
example: it is not allowed to kill and use a living being lor luxury needs.
kational beings are never to be used.

looking at the above ligure, we get lour ethical principles.
10

1) All non-responsive living beings (e.g. plants and living cells) have an equal
claim to the basic right not to be used a merely a means lor our luxurious needs.
1his is a rellection ol a biocentric (1aylor, 1989) or deep ecology ethics, which
implies sobriety, no commercial advertisements and no status consumption.


10
lt is actually a continuum ol principles, because not only are there lour types ol beings, but a
spectrum ol beings with gradually increasing complexity. 1here is also a spectrum ol ends, because
there are no sharp boundaries between e.g. luxury and basic needs. we can couple these two spectra.
8asic riqnt cua|ity
14
2) All non-sentient responsive beings (e.g. invertebrates) have an equal claim to
the basic right not to be used as merely a means lor luxury and basic needs. we are
allowed to use them lor vital needs (e.g. experiments). lating animal products
(lrom both sentient and non-sentient animals) is not a vital need lor us, because
we can live healthy with a well-planned vegan diet (AuA, 2009). 8o eating animal
products is not allowed when it is not a vital need or a survival end.
!) All non-rational sentient beings (e.g. vertebrate non-human animals and
mentally disabled humans) have an equal claim to the basic right not to be used as
merely a means lor vital, basic and luxury needs. lxperimenting on animals or
using them lor organ (xeno)transplantation would not be allowed. lut eating
animals is allowed when it is a survival end, as we will see in the section on
predation in part !. lredators (and some indigenous people) are allowed to eat
meat, because they became dependent (by evolution) on other animals in order to
survive. lt's a survival end, because biodiversity will be lost il all predation was
prohibited. Ol course they are only allowed to eat animals until leasible
alternatives lor them are lound.
11

we have to add that sentient beings are beings who developed the capacity to
leel and have not yet permanently lost this capacity. 1his is relevant, because we
are not allowed to use sentient beings when they are asleep or temporarily
unconscious. lven when they can temporarily not leel anything, it is not
respectlul to use them as merely a means.
4) All rational beings (mentally healthy human adults and children) have an
equal claim to the basic right never to be used as merely a means, lor no ends at
all. lating rational beings is never allowed, not even lor survival ends. we should
protect rational beings lrom predators il we can.
Ol course violations ol the basic right are allowed when the Qmm-principle is
very strongly violated. 1he basic right is not absolute, because our moral intuition
says that it would be inellicient to let thousands ol people die simply because we
don't want to violate the basic right ol one individual. we have a small but non-
zero need lor elliciency, just like we had a small but non-zero need lor elliciency
in the context ol the Qmm-principle. 1his brings to the next section.


11
ln some situations, killing and eating mentally disabled humans might equaliy be permissible
(otherwise it would be speciesist). lut the tolerated partiality principle says that it is equally
permissible to preler eating non-human animals instead ol those disabled humans.
tnica| ccnsistcncy anJ Anima| ua|ity
146
6. low strong is the basic right7
ll the mere means principle is an absolute principle, it would correspond with a
basic right ol inlinite strength: a constraint that can never be passed. lut a lot ol
people have the intuition that the mere means principle should not be absolute.
looking at the bridge trolley dilemma, we can say that the basic right is stronger
than at least live times the right to live. lut in the end, a lot ol people have a non-
zero need lor elliciency in well-being: the death ol millions ol people might
surpass the basic right ol one individual, because the loss ol utility (well-being)
becomes too big. 1his means that the moral lorce (the strength) ol the basic right
is lower than the moral lorce ol a huge amount ol well-being.
As in principlism (leauchamp & childress, 2001), some intuitive balancing
between the mere means principle and the consequentialist principle (Qmm-
prioritarianism) is required. lt is a balancing ol 'moral lorces', comparable to the
lorces in physics: gravity is much weaker than electromagnetism, so gravity can
surpass electromagnetism only when gravitational masses are very big and
electric charges are relatively small.
well-being can sometimes surpass the basic right, or in other words: the ends
can sometimes overrule and justily the means. lence, a lirst advantage ol a linite
strength (a non-absolute principle where the basic right has a linite weight) is that
our intuitive need lor elliciency can still be met. lut the strength ol the mere
means principle can also depend on some other variables: how much harm is
caused to the victim7 low strongly does the victim want to avoid the treatment7
low much ol the treatment is disliked by the victim7 low much consent does the
victim give to the treatment7 low strongly does something belong to someone's
body7 And what are the mental capacities lor autonomy ol the individual7 1hese
variables should be included in the intuitive balancing.
As a consequence, a second advantage ol a basic right with a linite strength is
that it allows a coupling between dillerent gradations: the strength ol the basic
right can be correlated with how much harm is caused, how much the victim does
not want the treatment, how strong something belongs to the victim's body and
how directly the body is used.
linally, a third advantage ol a linite strength is that it allows lor a
consequentialism in deontological rights: a world where the basic right ol one
individual is violated is better than a world where the basic rights ol two
individuals are violated in a similar way. ll the strength ol the basic right would be
inlinite (absolute), we do not have this property (as two times inlinity equals
inlinity).
8asic riqnt cua|ity
14
8o we have three advantages ol a non-absolute mere means principle:
compatibility with a need lor elliciency, coupling with gradations and
consequentialism ol basic rights violations. One disadvantage is that a non-
absolute principle requires an intuitive balancing. 1his intuitive balancing should
be done by all moral agents in a democratic way
12
(see appendix 2, section
uemocratic impartial prelerences ol moral agents"): we should take a democratic
average ol the moral lorce ol the basic right. 1his can also be done mathematically,
see the next intermezzo: the wellare lunction can include k-parameters that
measure the violations ol the basic right. 1he average is democratic in the sense ol
being unweighted (taking an unweighted average ol the k-parameters in the
wellare lunction): all prelerences (intuitive balancing ol the moral lorces) ol all
moral agents count equally.
}ust as physics is not inconsistent when electromagnetism counteracts gravity,
so is ethics not necessarily inconsistent when the mere means principle
counteracts the Qmm-principle. lnconsistency might occur when the balancing
between these two principles is arbitrarily applied in dillerent situations. As il
gravity arbitrarily gains strength even when masses remain equal. 1he strength ol
lorces in physics should conlorm to universal laws.
An example ol an inconsistent balancing ol moral lorces occurs in situations ol
discrimination, where the moral lorce ol the basic right ol one individual is
estimated to be stronger than the basic right ol another individual who should
have an equal moral status. As we will see in the next chapter, speciesism is a kind
ol discrimination that is a moral illusion. 8uch moral illusions can create biases in
the balancing ol moral lorces by moral agents. moral agents who are vulnerable to
moral illusions might have inconsistent estimates ol the strength ol the basic right
ol dillerent individuals. 1his is important in e.g. discussions on the use ol sentient
beings lor medical experiments: animal researchers, as moral agents, might have
an illusory bias towards using non-human animals, as il the basic right ol those
animals is weak compared to well-being (ol humans). lowever, they would have
very dillerent estimates lor the basic right ol some mentally disabled humans:


12
Note that experiments demonstrated that a moral agent can have dillering intuitive estimates ol the
strength ol the basic right, depending on some irrelevant circumstances and cognitive biases: the
inlluence ol induced leelings ol disgust and humor (oreene, 2008), the lraming ol the description ol a
situation (letrinovich & O'Neill, 1996, 8innott-Armstrong, 2008, lanteri et al., 2008, kay & lolyoak,
2010) or the order in which dilemmas are presented (liao et al. 2011, 8chwitzgebel & cushman, 2012, ui
Nucci, 2012). 1herelore, a moral agent would have to agree that hisJher intuitive estimate lies in a
certain range, so sJhe should be very llexible and tolerant towards changes ol the strength ol the basic
right within this range.
tnica| ccnsistcncy anJ Anima| ua|ity
148
those humans have a much stronger basic right although there is no consistent
justilication lor that dillerence in estimated strength between those humans and
animals. lence, the democratic averaging ol the moral lorce ol the basic right is
only valid il the moral agents do not have moral illusions such as speciesism.
1he basic right principle not only overrules the Qmm-prioritarian principle in a
lot ol situations, but also overrules the tolerated partiality principle (tolerated
choice equality) in all situations. 1he tolerated partiality principle is too weak and
can never trump the basic right. 1o see this, take another look at the burning
house dilemma: your child or an unknown child7 ll you save your child, you are
not using the other child as merely a means. Now let's look to a dilemma that is
structurally very similar to the burning house dilemma. 8uppose you are a surgeon
and in the hospital is your child and an unknown child. Your child needs a spleen,
the other a liver in order to survive. lor the moment, you can keep them both
alive lor some days by administering a drug. ll you do nothing, both will die, as in
the burning house. lowever, you could stop giving the drug to the other child, so
that child dies. 1hen you could use his spleen to save your child. ln this situation,
people are very reluctant to say that the surgeon is allowed to let the other child
die in order to use his organs. 8o, the ethical principle that you may preler to save
your child lrom the llames in the burning house does not imply that you are also
allowed to save your child in the hospital, by killing (or letting die) another child
in order to use his organs lor a transplantation.
ln the appendix 2 l will derive a mathematical expression that measures
dillerent moral lorces generated by the Qmm-prioritarian principle and the basic
right principle.
6.6 1he extended mere means principle
1his section discusses an interesting relation between the basic right (mere
means) principle and the principle ol tolerated partiality. 1he mere means
principle can be extended, lrom using someone as merely a means to considering
someone as such. 1his extension can help us to explain the dillerences between
doing and allowing as well as positive and negative duties. l will demonstrate that
making these dillerences, using the extended mere means principle also generates
an argument to justily the tolerated partiality principle.
8asic riqnt cua|ity
149
6.6.1 uoing versus allowing
many people have the intuitive moral judgment that doing harm is worse than
allowing a similar level ol harm (lagan, 1989, p. 94). lushing a child in the water to
kill him is worse than not saving a drowning child. lowever, this olten heard
drowning child example is not a real moral dilemma: it is not a choice between
pushing versus not saving. 8o this example is misleading. A better example would
be the lollowing, 'switch trolley dilemma'. A runaway trolley is about to kill one
person on the main track. You can turn a switch and send the trolley to a side
track, where it will kill another person. lt is a dilemma, because you now lace a
choice between actively turning the switch versus doing nothing. ln this dilemma
it becomes less obvious that turning the switch and killing the person on the side
track is worse than allowing the person on the main track to die. A lot ol people
say that it is permissible to turn the switch (especially il the person on the main
track is your child) (lauser et al. 2008).
Now imagine there were three people on the main track, and the person on the
side track is someone l know. l do not want to kill this person on the side track, so l
let the three people on the main track die. You could say that l had a duty to turn
the switch, because one dead person is better than three dead people. lut il you
would say that to me, you would ccnsiJcr me as merely a means to the ends ol the
three people. You would not literally use me as merely a means, but according to
your judgment, my presence was required to save the three people, and l would
have to do something (turning the switch) l do not want.
ll you are not allowed to judge me lor not turning the switch, it appcars as ij
allowing the three people to die is not worse than killing one person. lence, il we
extend the mere means principle, lrom using someone as merely a means to
considering someone as such, we have coherence with a deontological rule ol
doing versus allowing. 1he extended mere means principle generates an apparcnt
!1

dillerence between doing and allowing. 1his dillerence corresponds with a
counterlactual account ol the doctrine ol doing versus allowing (see loward-
8nyder, 2011), which says that the presence or absence ol the agent is morally


1!
Note that the extended mere means principle merely says how l am not allowed to judge or consider
you. 1his does not imply that you do not have certain duties. You still might have a duty to turn the
switch to save the three people. 1hat duty is compatible with my duty not to judge you il you do not
turn the switch. lven il l am not allowed to consider you in a certain way, not much lollows lrom this
how you are allowed to act. we have to distinguish primary duties (how to act) lrom secondary duties
(how to judge actions). Nevertheless, we could say that my secondary duty (not to judge you) counts as
a justilication lor your (lack ol) primary duty.
tnica| ccnsistcncy anJ Anima| ua|ity
10
relevant: a 'doing' requires the presence ol the agent. ll an upshot would not have
occurred il the agent had been absent lrom the scene, the agent was not positively
relevant to the upshot (see lagan, 1989, p. 94).
1he dillerence between doing and allowing also corresponds with a dillerence
between positive and negative duties. A positive duty is a duty ol benelicence,
where the presence ol the agent (the helper) is required in order to benelit
someone. A burning house dilemma exhibits positive duties: the helper can only
save someone and cannot cause harm to someone (when the helper does nothing,
sJhe allows harm). A negative duty ol non-malelicence (the no-harm principle)
does not require the presence ol the agent: il the agent is not present, the no-harm
principle is trivially satislied because the agent cannot cause harm when sJhe is
absent.
ln situations ol negative duty, we can judge someone lor violating his duty ol
non-malelicence, without considering himJher as merely a means. lowever, il you
do not want to help someone, and il l claim that you violate the duty ol
benelicence, l would consider you as merely a means. 1herelore, violations ol
positive duties are considered less bad (more tolerable) than violations ol negative
duties. ln the next section, l explain why partiality not only is, but actually sncu|J
be more tolerated in positive duties than in negative duties.
6.6.2 1olerated partiality and imperlect duties
ln the previous chapter , l discussed the principle ol tolerated partiality, which
violates the consequentialist prioritarian principle. lut also the mere means
principle (the basic right) violates the consequentialist principle. 1here is a subtle
connection between the partiality principle and the extended mere means
principle.
lositive duties are imperlect duties, in the sense that, while we are not required
to live up to them at all times, these duties are deserving ol admiration. lelping
others is an imperlect duty, because there is a whole range ol possible levels ol
assistance that one could give. lerlect duties on the other hand can and should be
respected at all times (lor example the duty not to use someone as merely a
means).
looking at the intuitions ol a lot ol people, we appear to tolerate partiality in
positive, imperlect duties, but we are not so tolerant towards partiality in
negative, perlect duties. 1he reason why partiality is, and in lact should be,
tolerated in imperlect duties has to do again with the mere means principle in its
extended version: do not consider someone as merely a means.
tnica| ccnsistcncy anJ Anima| ua|ity
12
send the trolley to the third track, l am not allowed to judge you by saying that
you should have send the trolley to the second track. 1hat is because such a
judgment would still imply me considering you as merely a means (because your
presence was necessary to turn the trolley away lrom the live people on the main
track, and killing your beloved one is not what you like). 1herelore, the extended
mere means principle implies that you are allowed to cause more harm than you
could have avoided. You are allowed to kill two unknown people instead ol one
beloved person, violating the consequentialist prioritarian principle, only il your
action is the result ol saving even more people (i.e. saving the live people on the
main track).
1o conclude: the extended mere means principle can explain what kinds ol
partiality we sncu|J tolerate. Not tolerating some partiality would imply ccnsiJcrinq
someone as merely a means, which is immoral.
14

6.6.! 1he asymmetry ol procreational duties
1he 'asymmetry ol procreational duties' (Narveson, 196, mulgan, 2006, mcmahan,
2009) says that we do not have an obligation to give birth to happy children (out ol
the interests ol those children), but we do have an obligation nct to give birth to
children when we know that the lives ol those children will be miserable. 1hink
about the problem ol parents knowing they have a genetic delect which means
that their potential child will be seriously handicapped. Or think about animals
raised in the livestock industry. 1hose animals are bred lor their high
productivity, which olten means that they suller lrom serious physical problems
(e.g. big udders, lower immunity, delormations).
As we have seen in a previous chapter on population ethics, the wellare
lunction derived behind the veil ol ignorance shows a threshold ol well-being lor
high population sizes: il a newborn sentient being gets a liletime well-being below
some power-averaged value, it lowers the wellare lunction. l argued that a deontic
permission still allows lor the procreation ol such beings. lut what il a newborn


14
lerhaps one could extend the tolerated partiality principle to include special duties towards people
with whom one has special relationships (e.g. special duties towards lriends, own children,.). 1he
rationale might be something like this: il l have a special relationship with someone l hold dear, l want
the presence ol that individual in my lile. kemember that one ol the two conditions ol the basic right
principle is that the presence ol the other person is wanted. 8o il l have a prelerence lor the presence
ol the other person whom l hold dear, l run the risk ol using himJher as merely a means. ln order to
avoid this risk, it can be compensated with special duties ol assistance. 1hat means l am not only
allowed to help my lriend, but l also have to some degree a special duty to help him.
8asic riqnt cua|ity
1!
sentient being will get a liletime well-being above the threshold, such that the
wellare lunction increases7 uo the parents have a duty to procreate in this case7
1he extended mere means principle says that we cannot judge parents who do not
want to procreate, even though their luture children would increase the wellare
lunction. 1hat is because we cannot consider those parents as merely a means (as
breeding machines), doing something that they do not want. ln other words: the
asymmetry ol procreational duties is coherent with our deontic extended mere
means principle. Other possible solutions to explain the asymmetry (as in e.g.
mcmahan, 2009) are not needed.
ln summary, there are two relevant levels ol liletime well-being and two
dillerent principles that make procreation permissible (i.e. neither a duty nor a
prohibition).
1) ll a potential being would have a liletime well-being above some positive
threshold such that the wellare lunction would increase il the potential being gets
born, we do not have a duty to procreate, because we cannot consider a woman as
merely a means to increase the wellare lunction. 1he mere means principle
implies that the woman is allowed to be partial towards her own prelerences,
because the woman can decide what happens to her body.
2) ll a potential being would have a liletime well-being below that positive
threshold but still above zero, i.e. il the potential being would still have a lile
worth living, then we delinitely do not have a duty to procreate (because that
would lower the wellare lunction), but we have a deontic permission to procreate.
we may procreate and give birth to that potential being il we want to. 1he
biodiversity principle (the !-N-principle, to be discussed below in section 10.4)
implies that procreation is still allowed.
!) ll a potential being would have a liletime well-being below zero, i.e. a lile not
worth living, we have a duty not to procreate.
1



1
Although l am not so sure about this third rule. lt might be the case that a lot ol animal species
(especially the species with a reproductive strategy ol so called r-selection) give birth to short
miserable lives that are not worth living: a majority ol those animals starve or are preyed upon and die
shortly alter they come into existence (see e.g. lorta, 2010c). ll all those species are not allowed to
procreate, a lot ol biodiversity will get lost. l suggest we take much more ellort and do scientilic
research to increase the liletime well-being ol those animals by e.g. redesigning nature.
tnica| ccnsistcncy anJ Anima| ua|ity
14
6. Application: the least harm principle and
vegetarianism
1he basic right principle has a lot ol implications. lt might solve an important
objection against vegetarianism. uavis (200!) argued that an omnivore diet, killing
and eating big grazing animals (e.g. cows) would cause less harm to sentient beings
compared to a vegan diet. A vegan needs a crop lield, so it might be possible to
count how many animals (e.g. mice) die by accident using this crop lield. 8uppose
that live mice are accidently killed in the harvesting process to produce the same
amount ol nutrients as the omnivore's diet where only one cow is used. 1hen the
omnivore causes less harm than the vegan.
whether a vegan diet causes more sullering or death is a scientilic question that
is strongly debated due to lack ol good data. At least matheny (200!) and lamey
(200) criticized the argument ol uavis. lut here we can somehow avoid this issue,
by introducing the basic right. 1he omnivore uses the cow as merely means, so the
basic right ol the cow is violated. On the other hand the lield mice are accidently
killed, so they are not used as merely a means. ll the basic right trumps the right
to live, a vegan diet remains more ethical.
we can compare this with driving a car. lt is true that car trallic sometimes
accidently kills children. Now imagine (hypothetically) that we invent a new lorm
ol transportation, some kind ol teleportation device. lowever, this device can only
work il you kill a person and use his body to drive the teleportation device. Are we
allowed to kill and use an innocent person as merely a means, in order to save
more children lrom car trallic7 l expect that most people have the intuition that
using a person lor the teleportation is not allowed.


tnica| ccnsistcncy anJ Anima| ua|ity
16
ligure 11: weak and strong overruling ol the Qmm-theory

All three principles contain a notion ol equality. Also the universalist principle
discussed in the lirst part ol this dissertation contains a notion ol equality. 1his
leads us to lour principles ol equality, one lormal and three material principles.
1) lormal equality ol universalist impartiality: treat all equals in all equal
situations equally. 1his is a lormal principle because it does not say how we should
treat everyone. 1his lormal principle is applied to the other three, material
principles ol normative ethics. lor example in the Qmm-theory, this lormal
equality results in an important symmetry property ol the mathematical
lormulation: the liletime well-being ol individuals is interchangeable.
2) lrioritarian wellare equality. 1his equality means that governments,
prolessional health care and economic and legislative structures should be
impartial and should strive towards a just distribution ol well-being, according to
the Qmm-principle. lt is a material equality principle, because it gives content to
how well-being should be distributed. lormal equality ol impartiality says that a
unit ol well-being counts equally lor all sentient beings in all similar positions. l.e.
the identity ol individuals is not important when it comes to distributing well-
being. lut the material equality adds substance, by claiming that priority should
be given to the individuals in the worst-oll positions. As a result ol this priority, il
total liletime well-being is constant between situations, the situation which has
the most equal distribution ol liletime well-being is the best.
1
lence,
prioritarianism lies in between sum-utilitarianism (maximizing well-being) and
egalitarianism (equalizing well-being).
!) 1olerated choice equality: even though we would save our own child in the
burning house dilemma, we would tolerate the choice ol someone else who has
saved the other child. ln general we should tolerate small levels ol partiality,
especially when personal relationships are involved. 1his partiality in personal
relationships is central in an ethic ol care.
4) lasic right equality: all beings with a same level ol complexity in interests,
should have an equal claim to a basic right not to be treated as merely a means to
someone else's end.


1
1his was seen on the expression ol the wellare lunction: =.(1-), with the population lactor,
the average liletime well-being and an inequality metric.
5ummary cj part twc
1
with these lour principles
2
we can get a lairly nuanced picture ol animal
equality, as we will see in the next part ol this dissertation. 1he reason why we get
a nuanced picture is that the equality principles correspond with our moral
intuitions, and are not in contradiction with at least two other kinds ol inequality.
1) lnequality ol outcomes. 1his is strict egalitarianism, which strives lor
complete equality in well-being. 1his is not undesirable. lt violates a strong moral
intuition that says that inequality in well-being is permissible il it is at the
advantage ol the worst-oll individuals. lt is better to have two persons with levels
ol well-being 10 and 20, than levels ol well-being both equal to 1.
2) lmotional inequality. we are allowed to give some prelerence to those
individuals we hold dear. we do not have a duty to be impartial in our personal
relationships. we do not have to toss a coin in a burning house dilemma, il we
have to choose between saving our own child and saving another child.
lmotional inequality is not in contradiction with e.g. the universal declaration
ol human rights, which says that all human beings are born lree and equal in
dignity and rights. lut in the third part ol this dissertation, we will argue that it is
a kind ol discrimination to limit this equality principle to only humans. we will see
that it is possible to extend the moral community (the 'circle ol equals') to all
sentient beings, claiming that all sentient beings are equal in the above lour
senses. 1his extension is necessary il we want to stick close to our strongest moral
intuitions in a consistent way. 8uch an extension would make our theory more
compatible with our strongest intuitions, compared with the current inconsistent
ethics ol our speciesist society.
.1 lquality and veganism
ll we extend the material principles ol equality to animals, then we see that
animals in the current livestock and lishery industries are maltreated in three
ways.
lirst, the consumption ol animal products is likely a violation ol the Qmm-
principle: it is impossible to imagine that humans, il they were not allowed to eat


2
ln the next chapter l will discuss a lilth equality principle to solve the predation problem. 1his lilth
principle is a principle ol behavioral lairness: il a zebra is allowed to eat lor survival, then so is a lion.
more generally: everyone has an equal right to a behavior that is both natural, normal and necessary (a
behavior that contributes to biodiversity). 8o we end up with live principles ol equality.
tnica| ccnsistcncy anJ Anima| ua|ity
18
animal products, would be worse-oll than animals bred in lactory larms and
slaughtered in slaughterhouses. 1he pleasure ol the taste ol animal products
(meat, milk, lish,.) does not outweigh the sullering ol those animals. lrom behind
the veil ol ignorance, you cannot preler a world where eating cheese is allowed. ln
such a world, you have a non-zero probability to end up being a dairy cow with a
low value ol lile equal to say ! (because you suller in the livestock industry, and
you have an early death). You also might end up being a human who is able to
enjoy the taste ol cheese (he has a value ol lile equal to say 11). On the other hand,
in a vegan agriculture, this person can no longer enjoy eating cheese, so his value
ol lile decreases a tiny bit, to say 10. lut there will be no dairy cows in lower
positions, so you would not have a probability to end up worse. leing risk averse,
you would preler the vegan world, because then you do not have a probability to
get a value ol lile equal to !. ll a cow is born, the impartial observer behind the veil
prelers the cow to have a well-being ol e.g. instead ol a lower well-being ol ! in
the livestock industry. Quasi-maximin prioritarianism would therelore imply
veganism.
Also the principle ol tolerated partiality is violated in the livestock industry. ll
we tolerate the choice ol a dairy larmer to use the milk ol a cow in order to
increase the well-being ol a human who loves cheese, then we should also tolerate
someone who makes the opposite choice, such as breeding women and using their
breast milk to make cheese to give to animals who like breast milk cheese. 1his,
however, we would never tolerate.
1hird, the use ol animals and animal products is a violation ol the basic rights ol
animals, because these animals are used as merely a means. 1he bodily integrity ol
dairy cows is violated (by artilicial insemination, lorced milk production and early
death) and they are treated as property.
1hese three dillerent criticisms ol the livestock and lishery industries should
not be conlused with each other. 8peciesism causes serious violations ol three
ethical principles ol equality, based on justice, empathy and respect. when applied
to animal ethics, the equality principles give a complete picture ol equality that
extends approaches in the literature. lor example, lrancione (2000) only locused
at the basic right (the property status ol animals). 1his is a 'negative' approach, in
the sense that it only says what we are not allowed to do (related to negative
rights ol not being treated in some ways). 1he prioritarian theory ol justice also
ollers a 'positive' ethics, because it says something about our duty to help others
(and a corresponding positive right to be helped).
1he three material principles ol equality do not stand on their own. 1hey have
to be combined with a universalist imperative. As we have seen, this principle ol
universalism implies lour universalizations, two ol them are particularly
important: universalization with respect to the moral agents and with respect to
5ummary cj part twc
19
the moral patients. 1he lirst universalization with respect to moral agents will be
lurther discussed in the next section, where it is related to non-ideal theory (i.e.
situations without universal compliance). Alter that, the second universalization
with respect to moral patients will be discussed. As we will see, this second
universalization is related to antidiscrimination and the absence ol certain
hierarchic dualisms.
.2 ldeal and non-ideal theory: applying the universalist
imperative
1he universalist imperative says that we should not directly lollow e.g. the
prioritarian quasi-maximin principle on our own. 1his universalist imperative is a
bit related to the golden rule. we can state it in dillerent ways. lor example
according to the lantian categorical imperative: Act only according to that maxim
(moral rule or guiding principle) whereby you can, at the same time, will that it
should become a universal law. Or: abide by those principles which we would like
everyone to abide by. Or: give the good example and lollow the rule that every
moral being (everyone who is capable, rational and inlormed) should have to
lollow, even il no-one else does so.
!

1his universalist imperative rellects an unconditional commitment and we
should, il need be, swim up against the stream. we should abide by those rules
which are universalizable, which means that il every moral agent (who is capable
and inlormed) should lollow those rules and consequently apply them, there will
be no undesirable consequences that violate one ol the above principles ol
equality.
when choosing a rule-based action (an action based on a maxim or a guiding
principle), we should ask ourselves: what are the consequences il everyone (who is
well inlormed and able to do that action) does a similar action or lollows that rule7
ln other words: what are the consequences in an ideal utopian world with
universal compliance to the rule7 ll the consequences are good (il they satisly the
three material principles ol equality), then we should do that action or lollow that
rule, even il others don't.


!
1his relerence to rules turns the theory in a rule consequentialism" instead ol an act
consequentialism" (looker, 2011).
tnica| ccnsistcncy anJ Anima| ua|ity
160
ll we want to do an action, but we cannot lind an underlying ethical guiding
principle or rule lor that action that can be universalized to all similar situations,
we should not do that action. Actions are only permissible il you can lind a
justilying universalized rule that is consistent with the ethical system (i.e. a rule
that does not violate a principle ol the system). lor example: il l want to take the
train this morning, the rule lveryone has the right to take this train this
morning" cannot be applied to all persons, because that would result in an
overcrowded train. lut l can lind another guiding principle that guides my action
to take the train and that can be generalized: lveryone has the right to take the
train when there is still some place available on the train, when a lair distribution
ol train rides is possible and when there is no-one lelt who wants to occupy the
lree place and has an equally strong or stronger right or reason to take that train".
8o l can justily my use ol the train by relerring to this second principle.
ln other words: l am allowed to do an action (or inaction) only il l can lind a rule
or guiding principle that can be used to guide the action, given that 1) it is okay lor
me il this rule gets universalized (the non-arbitrariness condition) and 2) the rule
is compatible with all other principles ol the ethical system (the consistency
condition). when l can't lind a consistent, universalized guiding principle that
justilies the action, the action (or inaction
4
) is not allowed.
lt might be possible that the universalist imperative does not give an exclusive
answer to the question what guiding principles we should act upon. ll we see that
two dillerent kinds ol actions or rules are compatible with the universalist
imperative, i.e. il universal compliance to guiding principles A and l give the same
good consequences, then we break the tie by a reality check. ln reality, i.e. in a less
ideal world without universal compliance, not everyone will lollow that specilic
action or guiding principle. 8o we should look at the consequences ol our rules and
actions in the current, real non-ideal world (without universal compliance). ll
guiding principle A would have prelerable outcomes in the current non-ideal
world than principle l, we should act according to guiding principle A.
8o we start with an ideal theory: deriving those guiding rules that, with
universal compliance (amongst all people who are able to lollow the rules),


4
1he inaction relers to e.g. not helping someone in need. when l don't help someone, l should come up
with a rule that explains my not helping. ll l can't lind such a rule that l am willing to see universalized,
l have the duty to help. ll my guiding rule is simply l never help" or something like l never help when
l don't leel like helping", l will not be willing to do the universalization, because that would mean no-
one might help me when l am in need. ll my rule is l don't help at moments when l recently already
helped a lot ol others", l am willing to universalize this rule, so then it is okay not to help at that
moment.
5ummary cj part twc
161
generate the best results according to our principles ol equality. ll there is a tie
between those derived rules, we can select the best ol those rules by looking at
non-ideal theory, i.e. by looking at the consequences il there is no universal
compliance (and in particular: il there is as much compliance as in the current real
world

). Non-ideal situations will serve as tie-breakers.


6

1he prisoner's dilemma in game theory can illustrate non-ideal theory. 1he
dilemma laces a choice between cooperation and delection. ll both players in the
prisoner's dilemma game cooperate, they generate the best outcomes (according
to the prioritarian Qmm-theory). lowever, il one ol the two players delects, the
cooperator loses and the gains lor the delector increase. ll both players delect,
they generate a suboptimal outcome. 1he lollowing table presents the possible
outcomes (levels ol well-being) ol a prisoner's dilemma lor the two players (bold
type values lor one player, italic values lor the other).
Player 2
Cooperate Defect
Player 1 Cooperate 3,3 5,0
Defect 0,5 1,1

1he iterated prisoner's dilemma allows lor multiple, successive rounds. ldeally,
the best outcome lor both players in an iterated game is mutual cooperation. lut il
a player who always cooperates encounters a delector, the cooperator loses. lt is
shown that in a non-ideal world, with delectors, olten the best strategy in an
iterated prisoner's dilemma is 'tit-lor-tat' or 'equivalent retaliation' (largreaves-
leap & Varoulakis, 2004, p.191). According to this strategy, the lirst move (in the
lirst round) is always cooperation. 1he second move (in the second round) is the
same as the other player's move in the previous round. ll the other player delects,
you retaliate in the next round by delecting. ll the other player cooperates again,
you are lorgiving and cooperate in the next round.

1here are many degrees ol non-compliance, so there will be many dillerent non-ideal theories. 1he
prelerred non-ideal theory is the one that is applicable to the current real world, i.e. the one derived
lrom the current level ol non-compliance.
6
larlit (2011) discussed a similar solution to the ideal world objections" where a universalized rule
might have bad consequences in non-ideal situations lacking universal compliance. According to larlit,
we should adopt new, conditional rules. lor example: lollow the rules whose being lollowed by
everyone would make things
go best, unless some other people have not lollowed these rules, in which case do whatever, given the
acts ol others, would make things go best." (p.262). 1his principle needs lurther relinements that l will
not discuss here.
tnica| ccnsistcncy anJ Anima| ua|ity
162
8o a player has at least two good rules in the ideal world: 'always cooperate' and
'tit-lor-tat'. ll everyone lollows tit-lor-tat, the result will be continuous
cooperation. 1o break the tie between these two strategies, note that tit-lor-tat is
better than 'always cooperate' in a non-ideal world. 8o the player can lollow tit-
lor-tat

(prelerably with some level ol lorgiveness


8
).
ln real lile situations, there are olten more than two players who can cooperate
or delect. As the universalist imperative (do what everyone should do) is related to
the important idea ol giving the good example, l believe that in multiple player
situations it is good to tend as much as possible towards cooperation, because that
strategy is most visibly the strategy ol 'the good example'.
let's look at some political animal rights issues to discuss the importance ol this
universalist imperative.
.2.1 1he argument ol lutility
A lot ol meat eating people object that il they became vegetarians or vegans, the
impact on the lood market and the livestock sector will be negligible: not a single
cow will be spared. lowever, the rule says that il everyone became vegan, then
the end situation will be one without a livestock industry, which is better
according to the three principles ol equality. 8o therelore any individual has a
moral duty to give the good example and become vegan.

Adding an exception, a rule like always cooperate" is specilied into a rule like always cooperate,
unless others don't (then lollow tit-lor-tat)". A criticism ol rule consequentialism is that you can always
lurther specily a rule, such that in the end you end up with an inlinitely specilied rule, which is
equivalent to act consequentialism (8mart & williams, 19!). kule consequentialism collapses into act
consequentialism as long as there is room lor adding exceptions to the rule (uo X unless Y"). l think
that the approach ol specilying rules (using exceptions) is valid and permissible. lt is allowed to move
closer to act utilitarianism, on the important (non-trivial) condition that one does so on a path ol
universalized rules, i.e. that one always relers to (specilied) rules that everyone should lollow. 1he
more specilied, the more complicated a rule becomes, and complicated rules have disadvantages.
where you stop along this path is up to you. As long as you can lind a (specilied) rule that permits your
action alter universalization, your action is allowed.
8
lt might happen that the other player delects by mistake, ending up in a vicious circle or an unending
death spiral" ol mutual delections. 1o avoid this, a good player should sometimes be a bit more
lorgiving, by occasionally cooperating, even when the other player delects. ll the other player plays tit-
lor-tat (with lorgiving) as well, both players can escape the circle ol delection.
5ummary cj part twc
16!
.2.2 1it-lor-what7
8uppose someone kills and eats ten small animals (chickens), unless l kill a big
animal (a cow) and give it to him. minimizing violations ol basic rights and well-
being implies that it is better to use one big animal as merely a means, than to use
ten small animals as merely a means. 8o l should kill a big animal and give the
meat to that person7 1his becomes a subtle issue. what il l take the conditional
rule: uo not kill animals, unless others kill small animals lor consumption and
you can reduce their killing by killing a big animal yoursell and sell the animal's
products to those people"7 Lniversalizing this rule in an ideal world will generate
the best outcome, because no animal will be killed. 1he unconditional rule uo not
kill animals" also generates the best outcome in an ideal world. lut in a non-ideal
world, the conditional rule will be better.
9

1he argument to kill a big animal is similar to an argument given by domestic
lur larmers: ll we don't produce lur, then people will buy lur lrom countries with
weaker animal wellare laws. l can produce cheaper and animal-lriendlier lur that
will outcompete the lur lrom those horrible loreign lur larms. As a consequence,
those lur larms have to lower their production. lence, my production ol lur will
decrease the total animal sullering in the world." Ol course there will be other
political strategies to decrease animal sullering (e.g. negotiations, import
restrictions), but lor the sake ol the argument, suppose that those domestic
larmers are right. 1hose domestic lur larmers work in a non-ideal world (non-
compliance ol loreign lur larms). lence, they could use the conditional rule not to
produce lur, unless it outcompetes worse lur production.
lowever, l do not believe that the conditional rule should be lollowed. 1here
are two replies to this, one lrom a deontological (mere means) perspective, the
other lrom a game theoretic perspective. 1he deontological consideration goes as
lollows: ll l kill that big animal, then the animal will be used as merely a means in
twc ways: as consumption product by the other person (the other person uses the
meat ol the animal) and as ransom or a medium ol exchange by me (l use the
animal in order to save the lives ol the other small animals). we can say that this
double use, and especially the new use as medium ol exchange, is never permitted,
not even in non-ideal situations.


9
williams (8mart & williams, 19!, p.9) ollered a similar thought experiment to counter
utilitarianism: should oeorge accept a job at a laboratory lor chemical warlare il relusing the job
implies that another person takes the job and will do the unethical research with lar greater zeal7
tnica| ccnsistcncy anJ Anima| ua|ity
164
kelated to this is the issue ol the (non-)consequentialism ol deontic principles:
do we have a duty to minimize basic rights violations by violating someone's basic
right7 we can say that il we violate the basic right ol the big animal in order to
stop ten basic rights violations ol the ten small animals, the lormer basic right
violation (ol the big animal) counts heavier than a latter basic right violation (ol a
smaller animal) just as a latter basic right violation counts heavier than a right to
live.
10

As a second reply, the next table presents the outcomes lrom a game-theoretic
point ol view. 1he values represent the number ol animals that stay alive. ll both
players cooperate, the eleven animals (the one big and ten small animals) will live.
lut player 2 wants to kill ten small animals. ll player 1 reluses to sacrilice a big
animal (i.e. il sJhe cooperates), then ten small animals will die and the big animal
lives. ll player 1 kills a big animal, perhaps player 2 no longer kills the ten small
animals. ll both players 'delect', then all eleven animals will die.

Player 2
Cooperate Defect
Player 1 Cooperate 11 1
Defect 10 0

As with the above mentioned iterated prisoner's dilemma game, player 1 can
play the strategy ol tit-lor-tat. lut in contrast to the prisoner's dilemma, the new
game has the worst outcome when both players delect. lence, the retaliation
strategy ol player 1 is tricky. ll player two does not cooperate, we end up with the
worst outcome. lut il player 2 cooperates (does not kill ten small animals), then
tit-lor-tat requires that player 1 also cooperates in the next round (does not kill a
big animal).
1he example ol the lur larmer clearly demonstrates where this game strategy
leads us. ll the domestic lur larmer succeeds in out-competing the loreign lur
production, then the domestic larmer has to stop his lur production. lut then the
loreign larmers might get new market space. 1he loreign production increases
again, the loreigners delect, and the domestic lur larmer will delect again by
producing lur, resulting in a decrease ol the loreign lur production. 1his results in


10
ln appendix 2 lntermezzo: combining the basic right with the prioritarian theory", a mathematical
equation is given with a basic rights term

. we can say that

when individual i (e.g. the


big animal) is used in situation X as merely a means to stop the basic right violation ol individual j (e.g.
a small animal) that would have occurred in situation Y.
5ummary cj part twc
16
a continuous, high lrequency cycle ol quick changes between delection (domestic
lur production) and cooperation (no domestic lur production).
1here are two ways out ol this cycle: (almost) always delect or (almost) always
cooperate. 1he lirst way is a choice ol continuous delection. lnstead ol a tit-lor-tat
strategy, it becomes a tit-lor-what strategy. 1his strategy lollows a rule: always
delect il you uc|icvc the other person would delect when you cooperate. llayer 1
might uc|icvc that player 2 will delect again and again once player 1 cooperates. ln
that case, player one might decide to continue delecting, no matter what player 2
does. 1he domestic lur larmer continues his lur production, even when the loreign
production is out-competed.
lowever, this strategy ol continuous delection has two problems. lirst, how
can player 1 know that player 2 will cooperate even when player 1 would
cooperate7 1he beliel ol player 1 that mutual cooperation is impossible, can never
be disproven as long as player 1 keeps on delecting: player 1 does not even give
player 2 the opportunity to demonstrate his unconditional cooperation. ll player 2
would decide to cooperate even when player 1 cooperates, player 1 will never be
able to know this il sJhe always delects because ol a lalse beliel.
A second danger ol such tit-lor-what strategies is that outsiders cannot easily
inler the rule or true motives ol those domestic lur larmers: the rule depends on
what player 1 believes about player 2, instead ol what player 2 did. lut outsiders
cannot get reliable access to what player 1 believes. 1he lur larmers might lie
when using the above argument: their true intentions might just be to sell lur, not
to light lor animal rights. kelated to this is the importance ol giving the good
example. l believe a lot ol people would have dilliculties in seeing the good
example behind the strategy ol trying to stop lur production by producing lur
yoursell. 1herelore, in such lur larmers' games, l believe it is better to apply the
rule to always cooperate.
ln conclusion, in non-ideal situations we should 1) not introduce a new use as
merely a means (e.g. a use as ransom), 2) not lollow a tit-lor-tat strategy il it result
in a high lrequency cycle ol cooperation and delection and !) not lollow a tit-lor-
what strategy (continuous delection) because ol the risk ol lalse beliels and the
lack ol clarity ol intentions behind the delection.
.2.! lrohibition laws
consider a prohibition law: the government will punish anyone who eats, buys or
sells meat. lmagine that il our government enlorces this law, a black market ol
animal products will be generated. 1hese products are smuggled illegally lrom a
loreign country, where the rights ol animals are violated on a much larger scale
tnica| ccnsistcncy anJ Anima| ua|ity
166
than the rights violations in our domestic livestock industry belore the
prohibition. ln other words: imagine that animal products are subject to the same
'iron law ol prohibition', like alcohol products. 1his iron law says that, in many
countries, prohibiting alcohol production and trade will result in more alcohol
abuse. making alcohol illegal will generate worse consequences. Ol course, there is
a dillerence between illegal alcohol production and trade on the one hand, and
illegal livestock production and trade on the other. likely, the livestock
prohibition will not be subject to the iron law ol prohibition, because some basic
conditions that led to the iron law ol prohibition lor alcohol are not met in the
prohibition ol animal products. A government will be able to lorbid the trade and
production ol animal products, just as with human products. uue to characteristics
ol livestock industry, illegal livestock production and trade is much easier to lind
than illegal alcohol distillers and bootleggers. lut lor the sake ol the argument:
imagine that prohibition ol animal products would make matters worse in terms
ol animal rights violations. what should we do then7 8hould we give in to a kind ol
blackmail il meat eaters say that prohibition will result in worse animal rights
violations overall7 ll they say that making meat illegal results in a temptation to
eat more meat and a worsening ol conditions lor the animals (e.g. smaller cages,
no more government control)7
ll no-one trades, produces and consumes animal products, no basic rights will
be violated, so therelore l should not buy, sell, produce or consume animal
products. 1his rule applies to the ideal world ol universal compliance, and, as it
gives the best results in this ideal world, we should stick to this rule in a non-ideal
world as well. lut what about prohibiting and punishing others who produce or
consume animal products7 what about a government policy to make meat illegal7
ln the ideal world, prohibition and non-prohibition would be equally good,
because there will be no-one to be punished. 8o prohibition and non-prohibition
would result in the same consequences in the ideal world. we have a tie between
two rules: prohibition and non-prohibition. 1he best rule should now be derived
by a reality check. 1he move to a non-ideal world will be a tie-breaker. what il, as
in reality, not everyone will lollow the rule to abstain lrom animal products7 ll it
would occur that a prohibition in a non-ideal world would result in more animal
rights violations, compared to a non-prohibition, then non-prohibition should be
prelerred. ln other words: il animal products would be subject to the iron law ol
prohibition (as with alcohol), il prohibiting and punishing the production and
trade in animal products would result in worse animal rights violations (which is
likely not the case), then prohibition and punishment would not be a
government's duty. ln this case, our only duty will be to abstain lrom buying and
selling animal products ourselves, but we should not prohibit and punish others.
5ummary cj part twc
16
.2.4 8ell-delense against culpable attackers and innocent threats
Non-ideal theory also deals with situations ol sell-delense against attackers who
violate a moral rule. 1o discuss this issue, let us consider a trolley problem that
represents a rather broad picture ol sell-delense against both culpable and
innocent threats.
lmagine some people start driving a trolley, and this trolley is then heading
towards a number ol potential victims on the main track. ll the people in the
trolley have the intention to hit the victims, and il there is no sulliciently strong
justilication to hit the victims, then those people in the trolley are culpable
attackers. ln general, a culpable attacker is someone who consciously wants to do
an action that violates a moral rule, in particular an action that decreases the
wellare lunction or the more general moral weight (that contains the wellare
lunction plus some additional terms that represent violations ol the mere means
principle). lt might also be the case that the people in the trolley are unaware ol
the victims on the main track (i.e. they are misinlormed), are coerced to start the
trolley (i.e. they act under duress), or are innocent in some other way (i.e. they are
insane or hypnotized). ln those cases, the people in the trolley are innocent
threats.
lurthermore, imagine that the potential victims can save themselves by turning
a switch that sends the trolley to a side track. 1his side track ends in a ravine, so
turning the switch will result in harming the people sitting in the trolley. lut on
this side track, there may be a number ol innocent bystanders who also might be
harmed when the switch is turned.
lrom a moral point ol view, l make no distinction between innocent bystanders
on the side track and innocent threats in the trolley. 8o in general we have a
number ol N
v
potential victims on the main track, N
c
culpable attackers in the
trolley and N

innocent people in the trolley and on the side track. 1he level ol
damage per person that might belall each ol the potential victims il they don't
delend themselves is J
v
.
11
lence, the total damage ol all the victims when they
don't delend themselves is N
v
J
v
. 8imilarly, il the victims delend themselves by
turning the switch, they cause total damage N

* and N
c
J
c
* to respectively the
innocent people and the culpable attackers. (1he * relers to the situation where
the victims act in order to delend themselves.) 1o make it more general, we can


11
looking at the wellare lunction, this damage per person might be written as

,
with

the loss ol liletime well-being ol a victim. when a victim is used as merely a means, the
damage becomes much bigger, because J
v
includes the parameter r
v
that measures the victim's basic
right violation.
tnica| ccnsistcncy anJ Anima| ua|ity
168
also consider a damage N
v
J
v
* that the victims might still receive even when they
delend themselves (lor example when turning the switch would be harmlul lor the
delending victims as well), and a damage N

that the innocent people get when


the victims do not delend themselves (lor example when the innocent people on
the side track are blown away by the trolley passing by at lull speed on the close-
by main track, or when the innocent people in the trolley get hurt when the
trolley hits the victims on the main track).
1he question now is: when is it allowed lor the potential victims to delend
themselves, respecting a proportionality condition on sell-delense7 1here are
three proportionality constraints that the delending victims should respect. lirst,
the most obvious constraint: in their delense, the potential victims should take the
option that avoids any unnecessary harm. ll the same results could be achieved
with a lesser harm, then they should opt lor the delensive action that causes the
lesser harm.
8econd, note that the culpable attackers, by consciously wanting to violate a
moral rule (i.e. consciously wanting to decrease the wellare lunction or moral
weight), place themselves in a sense outside ol morality. ln that case, the
delending victims should avoid decreasing the restricted wellare lunction or
moral weight. 1his restriction means that the harms (the loss ol liletime well-
being) sullered by the culpable attackers as a result ol the delensive action ol the
potential victims, are not included in the equation. ln other words, in the
restricted wellare lunction the levels ol liletime well-being ol the culpable
attackers is constant and equal to the levels they would have when they did not
violate the moral rule. lence, a delensive action by the potential victims is
allowed il the lollowing inequality constraint is satislied:


1his means that the total harm ol the delensive action (excluding the culpable
attackers) should be lower than the harm sullered by the victims and innocent
people when the victims don't delend themselves. lere, again, the harm is
measured by the decrease in wellare lunction or moral weight. 1he harms sullered
by the culpable attackers are not included in this inequality constraint.
12



12
1he permissibility ol sell-delense is determined by the choice between the wellare lunction
w(x

,x
V
,x
c
) where the potential victims (who have a well-being x
V
) do not delend themselves, and
w*(x

*,x
V
*,x
c
) where the potential victims delend themselves. Note that the culpable attackers have the
constant liletime well-being x
c
instead ol x
c
and x
c
*. 1his constant liletime well-being is the level the
culpable attackers would have when they did not attack, i.e. when they did not violate a moral rule.
8imilarly, in the wellare lunction ol the psychological connectedness description, we lix the value ol

, i.e. the integrated well-being ol a person who has at time t the culpable intention to violate a

5ummary cj part twc
169
what il we suppose that the culpable attackers should be treated exactly as
innocent threats7 lmagine there are live culpable attackers (N
c
=) and only one
potential victim (N
v
=1). 1he liletime well-being ol the live attackers might trump
the liletime well-being ol the one potential victim. 1his would mean that il the lile
ol the victim is at stake, the delensive action ol the victim should not result in the
death ol more than one ol the attackers. 1his seems counter-intuitive. ll you are
attacked by live killers, you are allowed to kill all ol them in sell-delense, il killing
them is the least harmlul option you have in your delense.
As an example ol animal ethics, consider a person being attacked by live
predators. ll the only option ol sell-delense is to kill everyone ol those live
predators, then my moral intuition says that the attacked person is allowed do kill
all live predators, even when the death ol live predators might be worse (in terms
ol loss ol liletime well-being) than the death ol one prey.
1!

1hat is why the harms caused to the culpable attackers should not be included
in the wellare lunction or the moral weight. lut the culpable attackers are not
completely placed outside ol morality. 1rue, the culpable attackers consciously
violate a moral rule against harming others. 1he total damage they cause is
u=N
v
J
v
-N

. Are the victims in their delense allowed to do anything with the


culpable attackers7 No, there is a third proportionality constraint that they should
respect, given by the lollowing inequality


ln other words: to make the permissibility ol sell-delense in line with our moral
intuitions, we can state that each single culpable attacker who consciously wants
to cause a total level ol damage u is liable to that amount ol damage in the
delensive action by the victims. lence, the maximum permissible level ol damage
J
c
* that the victims can cause to each one ol the culpable attackers is u.
1his third proportionality constraint does not lollow lrom the maximization ol
the wellare lunction or the moral weight. lt is a constraint that is applicable only
to non-ideal situations, where some people consciously want to violate the wellare
lunction. 8o we should lollow the lollowing conditional rule: maximize the


moral rule, when that person did not violate that rule (as indicated by the quotation mark). 1he extra
constraint on sell-delense in this lormulation becomes more complicated. One option is lor example:

, i.e. the damage to the culpable person (the


dillerence ol the integrated well-being between the no-attack situation and the delense situation) is
less than the time-maximum ol the sum ol damages ol all victims (the dillerence between the no-attack
situation and the no-delense situation).
1!
ln chapter chapter 10 l will argue that predators are still allowed to hunt. 1his does not contradict
the permissibility ol prey to delend themselves.
tnica| ccnsistcncy anJ Anima| ua|ity
10
wellare lunction, unless others (the culpable attackers) consciously don't. ln that
case, maximize the restricted wellare lunction (excluding the harms sullered by
the culpable attackers) and respect the proportionality constraint on the harms ol
the culpable attackers." ll everyone complies with this rule (the ideal situation),
the wellare lunction is obviously maximized. lut compared to the unconditional
rule maximize the wellare lunction", this conditional rule works better in non-
ideal situations: it better lits with our moral intuitions about sell-delense.
.2. 8ummary
ln summary, we start by looking at an ideal theory. ln a Ltopian world with
universal compliance, we could derive the lollowing rule that generates the best
outcome relative to our theories ol equality: don't consume, trade or produce
animal products. 1his rule should also be applicable to a non-ideal world. On the
other hand, ideal theory could not make a decision between the rules ol
prohibition versus non-prohibition. 1hen we have to look at a non-ideal world
lacking universal compliance. ln such a world, under certain specilic conditions
(resulting in an iron law ol prohibition), non-prohibition (non-punishment) would
be prelerable. 1hese conditions might be met lor alcohol production and trade, but
likely not lor animal products. Non-ideal theory is also relevant to derive
proportionality constraints in delensive action.
8o lar lor universalization with respect to the agents. ln the next section, l will
shed more light on the second kind ol universalization, with respect to the
patients. 1his relates to discrimination as a violation ol the lormal principle ol
equality. lut discrimination is also related to ideologies ol hierarchic dualisms.
lelore l argue in the next chapter that speciesism is a discriminatory hierarchic
dualism, l will brielly discuss the properties ol hierarchic dualisms.
.! lormal equality, discrimination and hierarchic
dualism
lormal equality is related to the notion ol discrimination. let's deline
discrimination as causing a disadvantage to an individual (or a group), based on a
value-laden distinction between individuals (or groups), where the distinction is
not justilied or relers to properties ol the individuals (or groups) that are not
deemed morally relevant in that situation.
5ummary cj part twc
11
ln other words: a person A discriminates l against c, il A believes (and acts on
the beliel) that l has lower value than c (meaning c sncu|J have more rights,
advantages or opportunities than l), where this value dillerence has no
justilication or is derived lrom properties ol l and c that are not morally relevant
or are not an acceptable motive lor the decisions and behavior ol A.
1he question is: what are morally (ir)relevant properties or criteria7 ln part ! l
will argue that being human is not a relevant property, and sentience is. lut lor
now, il we place it in the Qmm-lramework, the answer is simple: morally relevant
properties are all properties that are related to improving the value ol lile ol all
individuals who have a well-being, in line with the Qmm-principle. we argued that
desert based and resource based principles lollow lrom Qmm. 8o morally relevant
properties are amongst other things: desert (contribution to the value ol lile ol
others), ellort, incurred costs and personal responsibility.
we now move lrom the Qmm-principle to the principle ol tolerated choice. ln
the burning house, we would save our own child instead ol an unknown child.
Now, the tricky point is that your child is not more deserving or responsible,
simply because it is your child. Actually, the lact that it is your child is not
important in the light ol Qmm-theory. Ol course, il you lose your child, your value
ol lile will be allected. lut don't lorget that the parents ol the other child in the
burning house will also leel sad when their child dies. 1he death ol your child is as
bad as the death ol the other child, il we look at Qmm-theory.
8o do we have discrimination7 ln some sense yes: there is an emotional
inequality in our behavior towards dillerent children. Yet, there can still be some
subtle lorm ol equality present, which is the tolerated choice equality. lt is related
to the words 'value-laden' in the delinition ol discrimination. what do we mean by
this7 8uppose the parent ol the other child passes the burning house, and saves his
child. 1here are two ways how you can react. You can say that what that person
did was immoral, because your child has more intrinsic value, a higher moral
status or a stronger moral right to live. Or you can say that, although you regret
that your child died, you accept and tolerate the choice ol that person to save his
child. ln the latter case, you and the other parent are in some sense equal, and
therelore your children inherit a tolerated choice equality, although there is an
emotional inequality lrom your point ol view.
As we have seen, there is one subtlety with tolerated choice equality: what il a
white employer reluses to give work to a black person7 ll the judgment ol the
employer is based on prejudice, the employer makes a value-laden distinction
between white and black employees, which is racist discrimination.
lence, tolerated choice equality should be distinguished lrom a moral value-
laden inequality, which results lorm a discriminatory ideology such as racism,
sexism or speciecism (kyder, 19). 1hese kind ol ideologies are hierarchical
tnica| ccnsistcncy anJ Anima| ua|ity
12
dualisms (llumwood, 199!) between an upper side (the oppressors) and a lower
side (the oppressed). lierarchical dualisms can be characterized by one or more ol
the lollowing properties (this is a small extension ol the theory ol llumwood).
1) 1he lower side is radically excluded lrom the upper side, by believing that
there is a deep gap between the two sides. Any overlapping between the two sides
is denied.
2) 1he lower side is negatively delined: the oppressed lack the properties which
are used by the oppressors to justily the oppression.
!) 1he lower side is homogenized, individual dillerences between people lrom
the lower side are denied, by use ol e.g. stereotyping.
4) 1he lower side is marginalized: the oppressors do not show care and
empathy. 1he personalities and needs ol the oppressed are denied or scorned.
) 1he lower side is unjustly criminalized, they are the scapegoats. 1he
innocence ol the lower side is denied.
6) 1he lower side is instrumentalized (objectilied), they are used as tools, as
means to the ends ol the upper side. 1he intrinsic value ol the lower side is denied.
1he lirst three characteristics are psychological mechanisms to sustain and
justily unequal treatment. 1hese mechanisms result in violations ol the tolerated
choice principle. characteristics 4 and are violations ol the Qmm-principle. 1he
sixth is a violation ol the basic right principle. 8o we see that our three material
principles ol equality are related to ideologies ol hierarchical dualisms. when one
or more ol these characteristics are present, there is a value-laden dillerence, and
we can speak ol immoral discrimination.
ln the next part ol this dissertation, we move lrom normative ethics to applied
ethics, in particular to applied animal ethics. An animal ethic gives a lundamental
critique on the ideology ol speciesism. looking at the above six characteristics ol
hierarchical dualisms, all ol them are present in our current speciest society, just
like they are present in racist and sexist societies. Although this is not yet prool
that speciesism is a kind ol immoral discrimination comparable to racism and
sexism, it strongly enlorces that idea. 1he prool will be given in the next chapter.



lart ! Animal ethics

tnica| ccnsistcncy anJ Anima| ua|ity




1
chapter 8 8peciesism as a moral illusion
8.1 1he current situation: patho-anthropocentrism
ln this chapter l argue that speciesism is based on a moral illusion. 1he species
boundary is not morally relevant. 8entience is the morally relevant criterion. Our
current speciesist society is discriminating non-human sentient beings (animals).
Anthropocentrism is the ideology that ascribes a central moral value to all
humans (individuals belonging to the species ucmc sapicns). lowever, not all
entities with human uNA have an equally high moral status. ln a lot ol countries,
abortion is legal
1
(not murder), and human embryos are also sometimes used in
stem cell research and therapy. An olten heard justilication lor this use ol embryos
(as merely a means, because those embryos die), is that those individuals have not
yet developed a complex central nervous system that gives them the capacity to
leel and be conscious. Apart lrom conservative religious people, most people are in
lavor ol legal abortion and embryonic stem cell research and therapy. As a lot ol
other people, l do not consider a lertilized human egg cell as a human being. 1rue,
it has the complete genome ol a human being, but by that criterion a skin cell
would also be a human being. At most, the lertilized egg cell can (under the right
circumstances) develop into a human being.
lt appears that there is some luzziness when exactly we call a being a human
being. A lot ol people are in lavor ol using non-sentient embryonic humanlike
beings as merely a means in scientilic research to help other people. 8o, the
criterion ol sentience already drips in the ideology ol anthropocentrism. leelings
are relevant as well, so we might rather call the current ideology 'patho-


1
lowever, the legality ol abortion (especially in cases ol rape) does not yet imply that human letuses
and embryos have a low moral status. One could argue that the letus is not allowed to use the pregnant
woman as merely a means. 1he pregnant woman has autonomy over her body.
5pccicsism as a mcra| i||usicn
1
1he small lines (arrowheads) represent the morally irrelevant properties, such
as specilic genes, physical appearance (such as having a tail), or having the
capacity to get lertile ollspring with someone. 1hose things do not matter, just like
having a white skin color or having a penis doesn't matter lor moral status. lut
according to racists, skin color does inlluence their intuitive judgment that black
people have a lower moral status. }ust like those morally irrelevant criteria, the
arrowheads in the ligure are geometrically irrelevant as well. 1hey do not
determine the lengths ol the two horizontal line segments. As we have seen, this is
the notion ol context independence.
we can now make the analogy with discrimination: ll you judge that the two
horizontal lines in the mller-lyer illusion are dissimilar whereas in reality they
are not, then you discriminate. ll the lines are in reality dissimilar, then judging
them to be dillerent in length is not discrimination.
we have seen that not everyone is susceptible to the mller-lyer illusion (some
indigenous people have no dillerential judgment when they grow up in
environments without straight lines ol tables and staircases), and we have seen
how the underlying psychological mechanism works (!u adaptation ol a 2u
image). we also know something about the psychological mechanisms behind
discrimination. lt is based on an in-group-out-group bias (1ajlel, 1981, whitley &
lite, 2010). Although in-group-out-group value judgments occur intuitively,
several studies (lurzban et al., 2001, cosmides et al., 200!) demonstrated that the
choice ol in-group-out-group (e.g. based on race) is not inborn, but is culturally
dependent and can be inlluenced by changing cultures. 8peciesism is also
culturally dependant. ln some cultures (e.g. }ainism) and in a big part ol the animal
rights movement, people do not (or no longer) have the prejudicial judgment that
the moral status ol humans is higher than other animals. 1he intuitive judgment is
not universal and not inborn. lut people growing up in a speciesist society
assimilate this ideology until they get this discriminating moral intuition. 1he
same happened with people growing up in racist societies. 1hey olten perceive
their in-group-out-group distinction as being natural, but it is not.
8imilarly, people growing up in an environment with houses and tables, olten
see straight edges, and therelore they assimilate optical intuitions about lengths ol
lines. 1he Jispcsiticn lor such an assimilation process is natural (inborn), but the
rcsu|t is not. ln-group-out-group thinking is natural, inborn and universal as well,
but the result (which group is the in-group), is not.
1his means that ideologies such as white-dominant racism, male-dominant
sexism or human-dominant speciesism are strongly culturally determined. 1hese
ideologies are not universal, and perhaps the underlying intuitions behind those
ideologies are more llexible and can change more rapidly than some ol our
'deeper' moral instincts. 1he dividing line between the ingroup and the outgroup
tnica| ccnsistcncy anJ Anima| ua|ity
18
can be inlluenced by society and is vulnerable to change. lt remains to be seen
whether intuitions behind e.g. the mere means principle are equally llexible and
inlluenced by culture. Although there remains some experimental controversy
(8achdeva et al., 2011), l slightly expect that those intuitions behind e.g. the mere
means and Qmm principles are more universally 'hard wired' in our brains. 8ome
evidence lor this 'universal moral grammar' hypothesis can be lound in mikhail
(2000, 200), lauser et al. (2008) and O'Neill & letrinovich (1998).
8.! low do we know whether speciesism is a moral
illusion7
ln the muller-lyer illusion we had reliable instruments to demonstrate that it is an
illusion: we could use a measuring stick or something to cover the small lines. ln
ethics, our reliable instruments are valid arguments, so we need arguments based
on ethical principles that lorm a coherent system and are compatible with our
strongest moral intuitions. 1hese two requirements, coherence and compatibility,
are very important. ln the optical illusion, we used a coherent theory ol geometry,
which is compatible with two very strong intuitions: translation invariance and
context independence. ll we want to argue that speciesism is really discrimination,
we need tools ol similar power.
And we have those tools. we will present no less than ten arguments: live
arguments against the species boundary (to demonstrate that the criterion ucmc
sapicns is not morally relevant), and live other arguments to show that sentience is
really important. 1hose ten arguments are also based on moral intuitions, some ol
them quite strong. And they lorm a coherent theory: the arguments mutually
support each other. lor the speciesist it would be very dillicult to attack this
system.
1he live arguments against the species boundary can be compared with the
principle ol context independence in the mller-lyer illusion. 1he live other
arguments in lavor ol sentience can be compared with the principle ol translation
invariance. 8o, as in the mller-lyer ligure, we have one intuition (the human-
animal value dillerence in ethics or the length dillerence in geometry) which is in
contradiction with several other intuitions (e.g. the importance ol impartiality in
ethics and the context independence in geometry). And as in the optical illusion,
in the moral illusion we have two options. lirst, we could abandon all ten
arguments and their underlying strong moral intuitions. 1his would save our
5pccicsism as a mcra| i||usicn
19
intuitive judgment about the human-animal value distinction. Or, second, we
could admit that this value distinction is an illusion, and we can save the stronger
moral intuitions. l believe that the combination ol the latter ten intuitions
(underlying the ten arguments) is stronger than the one intuition about the value
ol humans versus animals. 8o the easiest thing to do is to acknowledge that this
human-animal value distinction is a moral illusion, similar to the optical illusion in
geometry. 1his acknowledgment is lurthermore acceptable il we keep in mind that
the in-group-out-group distinction is - just as the mller-lyer illusion - not inborn
but culturally dependent. And the lact that even alter realizing it is an illusion, the
intuition persists, does not justily this intuition. 1he mller-lyer intuition, too,
was 'cognitively impenetrable' (lylyshyn, 1999): our intuition keeps on saying the
one line appears to be longer than the other, even alter we have measured them.
lelore we give the ten arguments, l suggest we lirst have a look at the lollowing
ligure.


ligure 1!: mental capacities ol responsive beings

lach vertical bar in the above ligure represents a responsive being (an animal
with a nervous system). we can (lor simplicity's sake) consider three mental
capacities: sentience, sell-consciousness and moral consciousness. 1hese capacities
are represented as grey horizontal bars, because the thresholds are olten vague.
when does a being exactly have a moral consciousness7 Although the threshold is
vague, we can clearly see that only humans (but not all ol them!) possess a moral
tnica| ccnsistcncy anJ Anima| ua|ity
180
consciousness. Non-human great apes (but not all ol them!) possess sell-
consciousness. And quite a lot ol animals possess sentience.
looking at this ligure, how do we decide the moral community7 which vertical
bars (individuals) do we take into account7 kacists only considered white people
(and perhaps not all white people). 8peciesists could consider all ucmc sapicns. lut
what about great apes and potential human-animal hybrids7 (8ee next section to
learn more about such hybrids.)
Another possibility is to take only the vertical bars that reach above some
threshold. l.g. lant (18), kawls (191), cohen (199), 8cruton (1998) and many
other anti-animal rights philosophers took only those beings who have a moral
consciousness. 8o they did not know what to do with the humans below the
threshold ol moral agency. A lot ol anti-animal rights philosophers reler to such
mental capacities that mentally disabled persons lack. lor people who are
concerned about the rights ol those disabled persons, this approach is olten
ollensive.
lut l suppose that the argument lrom marginal cases is valid (uombrowski,
199), meaning that mentally disabled human orphans have a high moral status
and basic rights (as they have in modern societies). ln other words: criteria such as
higher mental capacities, language, sell-consciousness, moral consciousness or
social bonds are already excluded and are morally irrelevant in the contexts l am
thinking about, i.e. when it comes to our treatment ol animals lor lood or
experiments.
1he great ape project (8inger, 199!) might take sell-consciousness as the
relevant threshold. lut even proponents ol this project exclude some seriously
mentally disabled humans. l will argue that we'd better take the threshold ol
sentience. 1his includes almost all mentally disabled humans. Only those human
beings who are merely alive or responsive but not sentient, would be excluded.
(lut as we have seen in the chapter about the basic right, also responsive and
living beings should be given some weaker version ol the basic right.)
ln the next section l give live arguments against the species boundary, and live
other arguments in lavor ol the sentience criterion. 8ome important remarks are
in order belore l present the arguments.
1) with moral status, l am relerring to an agent-independent moral status,
which means agent-dependent relational aspects are not (and cannot be) the basis
ol the human-animal distinction in our current society. consider the burning-
house dilemma: l have to choose between saving my child or the dog. l preler to
save my child, because l leel a stronger connection or relationship with my child
than with the dog. l also leel a stronger connection with my child than with yours,
so again l would preler to save my child. 8ome partiality might be allowed, as long
as we tolerate similar levels ol partiality ol other moral agents. l would tolerate
5pccicsism as a mcra| i||usicn
181
your choice to save your child instead ol mine. lut il l believe that my child has a
higher moral status independent lrom my relationship with my child, l would not
tolerate your choice to save someone with a lower moral status, and this will
become a kind ol discrimination.
1he agent-relative relational moral status is important in an ethic ol care, but
cannot explain the huge gap in moral status between humans and non-human
animals in our current society. 1he partiality rellected in this gap is intolerably
big, because we would not tolerate similar partiality in the other direction, where
animals would have the status that humans have now, and vice versa. we would
not tolerate that a non-human animal would kill and eat a human just lor taste.
8imilarly, l would tolerate your choice to save your child instead ol mine lrom the
burning house, but l would delinitely not tolerate that you kill my child lor the
gustatory pleasures ol your child. Your special relation with your child does not
allow you to do the latter.
2) l am supposing a moral individualism as delined by mcmahan (200) and
kachels (1990): the agent-independent moral status ol an individual is uniquely
determined by hisJher own particular characteristics. lence, hisJher group
memberships are irrelevant. ln lact, this assumption is an analogue ol the
principle ol context independence in the mller-lyer ligure: the length ol a line
segment is uniquely determined by an intrinsic property and its belonging to the
group ol line segments with outward pointing arrowheads should have no
inlluence. moral individualism takes an eraser to gum out the irrelevant elements
such as group memberships.
!) when l reler to species, l reler to the biological (scientilic) notion ol a
species, and when l reler to humans, l reler to the species ucmc sapicns. 1his might
seem obvious, but l don't want to target a straw man: people who delend
speciesism reler to 'humans' and 'species', and l don't know ol any other notion
that makes sense ol these terms apart lrom the biological notion. 8ome
philosophers delend speciesism, not by relying on the scientilic notion ol species,
but rather by using a 'lolk' notion ol 'human beings' as a (natural) kind (e.g.
chappell, 2011). lowever, those philosophers are not clear on whether the group
ol human beings (according to the lolk notion) equals the group ol ucmc sapicns
(according to the biological notion). ll those two groups are identical, there are
two options: either this equality is a mere coincidence, or there is a (causal)
explanation ol why these two groups coincide. 1he lormer case is very unlikely,
the latter case means that the lolk notion can be reduced to the scientilic notion
(or perhaps the scientilic notion is based on the lolk notion, which means that
biologists would be guided by lolk intuitions). ll the two groups are not identical,
those philosophers should clarily what which human beings are not ucmc sapicns
or which ucmc sapicns are not human beings.
tnica| ccnsistcncy anJ Anima| ua|ity
182
ln the section on essentialism and heuristics, l will elaborate more on lolk
notions ol natural kinds.
4) l do not assume that all speciesists (people believing in the status gap
between humans and animals) believe that species is the only morally relevant
criterion lor moral status. ln lact, speciesists who delend stem cell research or
abortion likely include sentience as necessary criterion, next to species
membership. According to them, non-sentient humans such as lertilized embryos
have a lower moral status. And most speciesists also support animal wellare laws,
which implies again that sentience has some importance.
8.4 live arguments against the species boundary
1he arguments that l will present, are based on a moral intuition: in order to avoid
the risk ol opportunism in our ethics, we should try to avoid adding arbitrary,
artilicial, larletched or luzzy elements to our ethical system. 1he arguments are
strongly based on the biological sciences. As kachels (1990) and lull (1986)
demonstrated, especially the uarwinian paradigm undermines some metaphysical
beliels about 'humanity', 'human nature' or 'the human kind'.
1) 1he biological species boundary is arbitrary. 1here are two kinds ol
arbitrariness: a vertical and a horizontal one. 1he vertical arbitrariness asks the
question why we should select 'species' lrom the list ol biological categories7 l
belong to the kingdom ol animals, the phylum ol chordates and vertebrates, the
class ol mammals, the inlraclass ol eutheria, the order ol primates, the suborder ol
dry-nosed primates, the inlraorder ol simians, the superlamily ol lominoidea, the
lamily ol great apes, the genus ucmc, the species ucmc sapicns, the subspecies ucmc
sapicns sapicns and the ethnic group ol whites. 1here are dillerent genetic
allinities. lt is arbitrary to pick out the species. One could reply that individuals
within a species are characterized by similarities (in terms ol e.g. common genetic
heritage, physiology or behavior), but the same goes lor the other biological
categories. 1here are multiple similarities, and they come in degrees. lt remains
arbitrary to pick out some similarities as being important, and also to pick out a
specilic degree ol those similarities as being important.
Next to vertical arbitrariness, horizontal arbitrariness asks that, il we select
species as the relevant category, why should we take one species instead ol
another7 Also the choice lor a specilic species amongst the many species is
arbitrary.
5pccicsism as a mcra| i||usicn
18!
Note again that the existence ol mentally disabled humans demonstrates that
there is no exact correlation between mental capacities and species. Note also that
relerring to a jc|k concept ol 'human being' as a kinJ (as proposed by e.g. chappell,
2011) instead ol a specilic biological conception currently lavored by science,
would not avoid the arbitrariness. Not only is the notion ol a 'kind' ambiguous (is
it a natural property or a construct7), each clarilication ol this notion results in
the observation that there are many possible 'kinds'. ll 'human being' is a kind,
then lor example 'primate', 'two-legged being' or 'caucasian' can be kinds as well.
lt is not clear why the kind ol human beings is an exceptional kind, compared to
the many other possible kinds that one can think ol.
Also other discriminations have a double arbitrariness. lor example in religion,
there are the groups ol lenedictines, koman catholics, catholics, christians,
Abrahamists,. why should people in Northern-lreland pick the third category in
this row7 And within this category, why should they pick catholics instead ol
lrotestants7 1he same applies to ethnicity: it is arbitrary to pick the second
category in the ranking ol a) people lrom Antwerp, b) the llemish people, c)
lelgians and d) luropeans. And it is arbitrary to preler the llemish people to the
walloon people in lelgium.
2) 1he biological delinition ol species is very complicated and too artilicial and
larletched to be used in a moral system. One ol the many delinitions ol species
relers to the possibility ol interbreeding and getting lertile ollspring. lut why
should this possibility be relevant7 lt is too larletched to say that a being has a
moral status il its close relatives (parents) could have gotten lertile ollspring with
some other morally relevant beings. (l reler to its close relatives because the
individual itsell could be inlertile.) lt is unlair that an individual gets rights
because his parents are able to do something with others. lt is unjust to take a
principle where non-human animals simply have bad luck having the wrong
parents.
kelated to this is the issue ol so called ring species such as the arus gulls, the
nsatina salamanders or the oreenish warbler. 8uch ring species consist ol
dillerent populations, whereby A can get lertile ollspring with l, l with c, c with
u, but u not anymore with A. }ust as populations ol ring species are spatially
related to each other, we can say that all species in nature are temporally related
in a similar way. king species 'are only showing us in the spatial dimension
something that must always happen in the time dimension.' (uawkins, 2004,
p.!0!). look at the phylogenetic tree. A modern ucmc sapicns could have lertile
ollspring with an ancestor, that ancestor with an older ancestor, and so moving up
a branch ol the phylogenetic tree until we reach a common ancestor ol both ucmc
sapicns and another species. 1hen we move down the branch ol that other species.
8o there is a chain ol populations connecting our species to any other species. 1he
tnica| ccnsistcncy anJ Anima| ua|ity
184
clue is that the higher moral status ol A (a ucmc sapicns) compared to u (an
individual ol another species) strongly depends on the lact that l and c are dead.
lormulated this way, it becomes clear that such dependency on the accidental
non-existence ol individuals cannot be morally relevant.
Again one might object that speciesism does not reler to the biological notion ol
species, but to the lolk concept ol 'human being' as a kind. lut this escape
maneuver does not work either: the lolk concept is perhaps even more
complicated than the biological concept ol a species. 1he lolk concept ol 'human
being' might be based on a pattern recognition: when we are conlronted with an
object, we can spontaneously recognize the pattern and see a human being. lut il
such a pattern recognition is based on an algorithm, it is a highly complicated
algorithm. 1he lolk notion looks trivial, but it is not.
!) 1here is a potential luzzy boundary: it is not unlikely that a human-
chimpanzee hybrid (humanzee or chuman) can be born. Nearly 10/ ol mammal
species can lorm interspecies hybrids. ln the wild and in zoos, there exist lion-
leopards, lion-tigers, camel-lamas, dolphin-killer whales, sheep-goats, grizzly
bear-polar bears and oll course the well known horse-donkeys (mules). ll these are
possible, and il the genetic distance between humans and chimpanzees is not
larger than the distance between those interbreeding species, it is possible that
humanzees can be born. what would the moral status ol this hybrid human be7
1here is an arbitrariness here as well. And what il a neanderthal (ucmc
ncanJcrtna|cnsis) would still exist7 would we give him the basic right7 And what
about other ancestors such as the Austra|cpnitccus and the ucmc naui|is7 And there
is more arbitrariness when we look at the possibility ol human-animal chimeras. A
chimera is an individual composed ol genetically distinct cells that originate lrom
human and animal zygotes. 1he body cells ol chimeras can range lrom 100/
human to 100/ non-human. where to draw the line ol humanity7 what would the
moral status ol such chimeric individuals be7 And we could also genetically modily
humans and animals. All ol this blurs the line between humans and non-human
animals. 8cience will not be able to propose criteria to determine whether beings
such as neanderthals, hybrids, chimeras and genetically manipulated beings
should be called 'human', just as scientists are not able to determine whether
grains ol sand should be called a heap. 1his luzziness is a philosophical issue, not a
scientilic one.
8imilar luzzy boundaries occur in other kinds ol discriminations: sexism is
laced with dillerent kinds ol transsexuals and intersexuals (who have genital
ambiguity or mixed chromosomal genotypes) and racism is laced with dillerent
kinds ol mixed races such as mulattoes. 1his also makes it very complicated to
deline sex and race.
5pccicsism as a mcra| i||usicn
18
4) 8pecies boundary relers to genes or appearance, and these are not morally
relevant, because racism and sexism are also olten based on genes or appearances
and antiracismJantisexism states that such a basis is not morally relevant. ll we
say that skin color (or the genes that generate skin color) is not morally relevant,
we should apply this rule consistently (universally) and state that no relerence to
appearance or genes is morally relevant when it comes down to someone's basic
rights, as long as we do not have an argument that some appearances or genes are
exceptional. A racist should be able to explain why skin color is morally relevant
but e.g. hair color isn't, and il he can't explain it, then he should treat skin color as
hair color. Otherwise we open the door lor opportunism. 8o we should
universalize the rule that genes and appearance are morally irrelevant lor cvcrycnc
in a|| situations related to basic rights violations. Also, there is no 'interest gene'
connected to all and only to humans, there is no gene that makes a being to have
interests.
!

) lelonging to a certain species instead ol another is not something that we
can choose, it is not something we achieved, it is beyond our responsibility.
lelonging to a certain species is also not related to subjective needs and
prelerences. lence, we should not be rewarded lor belonging to a species. we do
not deserve special treatment by having some specilic genes. oiving a higher
moral status to beings who did not choose to be born that way is in violation ol the
merit principle. ll we are to be rewarded, it is not merely because we are born in
some way rather than another, but because we either have a certain responsibility
lor an action (lor example we did an ellort or we contributed to something
valuable) or we have needs and are able to subjectively experience and preler
things (lor example we have a well-being and leel our needs). moral advantages
(rights, resources, opportunities,.) should be given to someone who deserves it or
someone who needs it. lut someone's species is not related to merit
(responsibility) nor need. On the other hand, as we will see below, sentience is
related to having subjective needs and prelerences, so sentience is a reason to give
someone a moral advantage.

Note that the above live arguments are very similar to the principle ol context
independence ol the mller-lyer optical illusion. ln the mller-lyer illusion, the
irrelevant context (the arrowheads) was characterized by arbitrariness,
artiliciality and luzziness. As the species distinction has those same


!
kecently, liao (2010) developed a 'genetic basis lor moral agency', but that approach was criticized by
orau (2010). l would add that this genetic basis account is as larletched as genetics is complex.
tnica| ccnsistcncy anJ Anima| ua|ity
186
characteristics, we can say that the species are part ol an irrelevant context. 1he
lirst, second and third arguments above are similar to the (vertical and horizontal)
arbitrariness, artiliciality and luzziness ol introducing a geometrical rule that says
that lour-legged ligures with outward pointing arrowheads decrease the length ol
line segments.
Also in the lourth argument, bodily appearance ol a being is, just like the
arrowheads, some external lactor, and we have to universalize the rule that no
external elements are important. ln the mller-lyer illusion, this is the
universalized rule that context is never important lor determining a length. ln
daily lile, we olten (unconsciously) use the rule ol context independence (you can
simply ask an architect who makes a drawing ol a house). lt would be inconsistent
to always use this rule, except in the case ol the mller-lyer ligure, because there
is nothing really special about this ligure. 8o we should apply context
independence consistently. lt would be strange that exactly in the mller-lyer
ligure context independence would not apply. 1hat is why antiracists and
antisexists should apply antidiscrimination consistently and hence become
antispeciesists as well.
8. live arguments in lavor ol sentience
ln the previous part ol this dissertation, l already addressed some arguments why
sentience is morally relevant. ln this section l summarize them again, and give a
lew more arguments. One ol the reasons why there are dillerent arguments lor
sentience is that there are dillerent moral virtues (empathy, impartiality, respect)
and dillerent normative systems. uillerent arguments lor sentience stem lrom
these dillerent normative ethical systems and moral virtues. All the arguments
have the same structure: starting with two assumptions (one lactual and one value
statement) one can derive that sentience is morally relevant.
1) wellare ethics (consequentialism) and lairness ethics (contractualism):
lact: Our own well-being matters to us.
Value: lmpartiality is important. 1he thought experiment ol the veil ol
ignorance (kawls, 191) is a nice tool to check impartiality. }ohn kawls only limited
his theory to rational beings. lut this thought experiment can be made more
impartial (more consistent) when applied to all entities in the universe, as was
proposed by kowlands (199, 1998). lmagine that you might be any object or entity
in the universe, but you don't know who or what you might be. You could be a
non-sentient thing without well-being, or a sentient being. low would you like to
5pccicsism as a mcra| i||usicn
18
be treated7 ll you were non-sentient, this question would not matter to you,
because nothing done to you will inlluence your well-being. You would not
experience or preler anything. 8o being sentient will imply a dillerent treatment,
because well-being matters to you
4
.
2) Virtue ethics and ethics ol care:
lact: we can leel empathy with all and only with sentient beings (beings who
can leel and have a well-being).
Value: ueveloping the virtue ol empathy (compassion) is important.
!) kights ethics (deontologism):
lact: A sentient being is a being that has interests and can subjectively leel its
interests. leelings are nothing but allective conscious mental states that indicate
that needs or interests are satislied or not. lor example pain indicates that bodily
integrity is not satislied, lear indicates that salety is not satislied.
Value: lrotection ol interests by respecting rights is important. lt is not
larletched to see a connection between rights, interests and leelings: leelings
detect interests, interests are protected by rights. 1his is at least less larletched
than making a connection between e.g. rights and having certain genes, belonging
to a certain biological group, or getting lertile ollspring.
4) Other ethics:
lact: mental capacities such as consciousness are something very complex and
vulnerable in the universe.
Value: we should protect and respect entities that have vulnerable and complex
mental capacities. laving a consciousness is at least something much more
remarkable than having the genes ol an arbitrary species. ll a sentient being
becomes a non-sentient being, he loses something valuable and does not gain
something in return. On the other hand, il a white person becomes a black person,
he loses one skin color but gains another, il a man becomes a woman, he loses one
sexual organ but gains another, il a human becomes a non-human animal, he loses
some physical properties and genes, but gains other.
) 1he argument lrom marginal cases (uombrowski, 199):
lact: lerceptual consciousness (sentience) is the only mental capacity that
mentally disabled humans share with other humans.


4
leing human is not what would matter to you. 1o see this, ask yoursell the question what you would
preler: you remain a human being but will be in a persistent coma without consciousness, or you turn
into an animal but keep your mental capacities lor well-being. l would preler the latter, which means
that l value mental capacities such as sentience more than biological categories such as species.
8imilarly, being male and being white is not what matters to me.
tnica| ccnsistcncy anJ Anima| ua|ity
188
Value: Our intuition says that mentally disabled persons are to be respected
because ol some inherent, mental capacity that they posses. 1he real reason why
we help them is because they can suller, they have interests, they can be harmed.
Other reasons, such as indirect rights or a slippery slope argument made by
carruthers (1992), are in a sense disrespectlul towards those individuals, because
they deny their intrinsic value (see the heuristics argument in section 8.8). Neither
is 'being alive' a sullicient criterion lor giving mentally disabled persons rights,
because human egg cells and embryos are also alive but they have a dillerent
moral status.

1he above live arguments cohere with each other and indicate that sentience is
a basis lor moral concern and moral status. lt is not larletched to see a connection
between rights, interests and sentience. 1his is at least less larletched than making
a connection between e.g. rights and the possibility ol getting lertile ollspring.
1his set ol live arguments is related to the translation invariance in the mller-
lyer illusion. }ust like length is an inherent property ol a line segment, these
arguments reler to a characteristic value ol sentience. 1he lirst two arguments,
which reler to impartiality and empathy, can be related to the idea that a ruler can
be seen as a device to make our length judgments impartial (objective instead ol
subjective). As the ruler is a device to shilt (translate) lrom one position to
another, empathy and the veil ol ignorance are (emotional and rational) devices
that also help us to 'translate' ourselves into the positions ol other sentient beings
and measure how rich their emotional lives are (how important things are lor
them). lence, impartiality in ethics is the analogue ol translation invariance in
geometry. lthicists should develop compassion as a virtue, just as geometers
should value the accuracy ol rulers.
One might argue that the notion ol sentience also has luzzy boundaries, just like
the notion ol a species, as we have discussed above. when is a being sentient7
what about invertebrates, plants,.7
1his is lirst ol all a matter ol lact (science). As we've seen, scientists do not have
and will never lind indicators to determine at what point a being (a hybrid, a
chimera, an ancestor or a genetically modilied person) should be called human.
lut scientists already do have quite a lot ol indicators to test whether a being is
sentient (see next chapter). And they will likely discover new indicators when they
gain more knowledge about how consciousness works. 1he species boundary has
an inherent luzziness, the sentience boundary is rather a matter ol scientilic
uncertainty. A being cannot be both sentient and non-sentient at the same time. A
5pccicsism as a mcra| i||usicn
189
being cannot both leel and not leel something at the same time. As with
computers: either the 'sentience program' works (is switched on), or it doesn't.


lut a hybrid is hall human and hall non-human at the same time. A species
always has an inherently arbitrary cut-oll point. 1he boundary ol a species is
continuous, because a lot ol properties that characterize a species are continuous
properties. lence, it is always arbitrary to select a point on this continuum. 1he
cut-oll point lor sentience, on the other hand, is at the value ol zero (i.e. the point
where all positive and negative leelings become absent, where the 'leelings
program' in the brains does not run). 8uch a zero point is (at least in theory) well-
delined lor sentience, but it is not well-delined lor species. look at all the
ancestors ol humans and ask the question: what exactly needs to be absent in
order lor an individual (an ancestor) to stop being a human being7 1his question
cannot be answered in a non-arbitrary way.
8econd, in our culture, non-human animals already have some moral status:
look at the animal wellare laws. 1hese laws reler to the wellare (sentience) ol
animals, so we are already able to use this criterion, even when there is still some
scientilic uncertainty about e.g. invertebrates.
1hird, in human rights ethics as well there is scientilic uncertainty about
sentience: consider the discussion on abortion and stem cell research. 1here is
scientilic evidence that lertilized human egg cells are not (yet) sentient, so they
have a lower moral status according to many people. lere also we are able to deal
with this scientilic uncertainty.
lourth, even il there is an inherent gradation in the levels ol sentience (lrom
simple to complex emotions), it is not really a threat to the theory, because it
makes sense to couple the gradation ol sentience to a gradation ol moral status
(see the chapter on the basic right, section 6.4). All beings with a developed,
complex, lunctioning central nervous system, all beings with a level ol sentience
equal or higher than those ol (most) vertebrates, developed human letuses or
mentally disabled humans, have a very high moral status.
lilth, il the sentience boundary is luzzy, why add a second luzzy boundary, the
species7 ln ethics we should strive to avoid as much luzzy notions as possible
(otherwise we risk opportunism). we should delete the most arbitrary ol the luzzy
boundaries: the species.
kelerring to the mller-lyer illusion, we see that line segments have a
continuous gradation (lrom short to long), but that is dillerent lrom the gradation

lowever, see the discussion about lractional number ol minds (lractional consciousness) in appendix
2 lntermezzo: a more complex lormulation to solve the replaceability problem".
tnica| ccnsistcncy anJ Anima| ua|ity
190
ol the angles ol the arrowheads. 1he lormer is an inherent property ol line
segments, the second is something external (contextual).

ln summary, our current society has a patho-anthropocentric ethic. lt takes two
criteria into account: sentience and species. lut it is better to drop the latter
criterion, because the species boundary cannot determine or inlluence someone's
moral status, just like arrowheads cannot inlluence the length ol a line segment.
1here is a very simple thought experiment that demonstrates that sentience,
not species, is what matters. lmagine that tomorrow you will either remain a
human being, but you will permanently lose consciousness, or you become a non-
human sentient animal who is able to leel joy and other positive emotions. which
choice would you preler7 l would preler the latter option.
8.6 8peciesism and cognitive impenetrability
As mentioned belore, one characteristic ol optical illusions is its cognitive
impenetrability (lylyshyn, 1999): even alter measuring the lengths ol the mller-
lyer illusion, they still appear to be dillerent. 1he question is whether speciesism
has a similar kind ol cognitive impenetrability: do our spontaneous, intuitive
judgments regarding the moral status ol humans versus animals still rellect some
speciesism alter we learn about the above arguments against speciesism7
Although this question is dillicult to answer, l am inclined to say yes, based on lour
reasons.
lirst, in discussions with meat eaters, a lot ol people remain speciesist alter
learning about the above arguments. 1hose people give inconsistent counter-
arguments (lallacies) to justily speciesism. 1his looks like a moral dumblounding
(laidt, 2001, 2012), where people have strong intuitive moral judgments but lail to
express a rational principle to explain their intuitive reactions.
8econd, even some animal rights activists exhibit some speciesist language that
rellects essentialistic thinking (see next section). lor example in discussions those
animal rights activists olten reler to the notion ol humans. 1his might indicate
that those antispeciesist animal rights activists have dilliculties in overcoming the
moral illusion. my personal experience conlirms this: l am an animal rights
activist, but l am aware that l still have some intuitive speciesist judgments about
the moral status ol humans and animals. lt takes some cognitive ellort to
overcome those intuitions, just as it takes a cognitive ellort lor utilitarians to
5pccicsism as a mcra| i||usicn
191
overcome the strong emotional intuition that we should not push a lat man lrom a
bridge in order to stop a runaway trolley (oreene et al. 2004, oreene, 2008).
1hird, it seems that a lot ol antispeciesist animal rights activists have an
emotionally dillerent response towards eating human corpses versus non-human
corpses. lating dead human bodies (even il no human was killed and no human
rights were violated) is accompanied with a strong leeling ol moral disgust. A lot
ol animal rights activists do not have a similar strong leeling ol disgust when it
comes to eating dead animals (e.g. lrom road kill). lt is unsure whether this
dillerence in the leeling ol disgust is related to a dillerence in moral status and
whether it rellects a cognitive impenetrability ol speciesism, but it might be an
indicator.
lourth, perhaps the best scientilic evidence that speciesism is to some degree
cognitively impenetrable, comes lrom studies on implicit associations (oreenwald
et al., 199, 1998, uevine, 2001). 1he lmplicit Association 1est (lA1) is an
experiment to measure spontaneous implicit attitudes that people have towards
e.g. races. 1he reaction speed is measured when experimental subjects have to
associate pairs ol concepts. 1hose concepts can reler to races (e.g. laces ol black
and white people) and values (e.g. positive and negative words like 'joy' and
'pain'). According to lA1 studies, a lot ol people have implicit racist attitudes,
although those people have explicit antiracist attitudes: they can explicitly state
that they are against racism and that they value black and white people equally,
although they have shorter reaction times when they have to associate black
people with negative values. lmplicit prejudice and stereotyping might explain
this dillerence between explicit and implicit attitudes (uevine, 2001).
Although an lA1-test about speciesism is not yet perlormed, l expect that those
lA1-studies about racism and sexism can be extrapolated to speciesism. what
these lA1-studies show, is a kind ol cognitive impenetrability: even il a racist
learns everything about racist prejudices and stereotyping, even il sJhe recognizes
how immoral racism is (that race is arbitrary, artilicial and not morally relevant),
hisJher implicit negative attitudes towards other races do not simply disappear.
8. lsychological background theories: human prejudices
and essentialism
One more thing needs explaining: what is the mechanism behind the moral
illusion ol speciesism7 ln the mller-lyer optical illusion, the coherent intuitions
tnica| ccnsistcncy anJ Anima| ua|ity
192
ol context independence and translation invariance are brought into a 'wide
rellective equilibrium' (kawls, 191) by introducing background theories about the
underlying psychological mechanism. we know that the mller-lyer illusion is
created by our brains, in order to adapt to !u-vision. Our optical system has a bug,
it's stuck when looking at a 2u-image that rellects elements ol !u-perspective,
such as the mller-lyer image. we have seen that our brains use a kind ol heuristic
(attribute substitution) to estimate lengths, using !u-interpretations ol e.g.
staircases.
Also in the case ol speciesism we have some well-established psychological
knowledge about prejudice, stereotyping, the inlluence ol language and words,.
(see e.g. llous, 200!, lor some mechanisms behind prejudices towards animals). let
us summarize the psychological background theories that turn antispeciesism into
a wide rellective equilibrium.
n-qrcup-cut-qrcup uias
lsychologists studied the mechanisms behind optical illusions such as the mller-
lyer illusion (lurves & lotto, 2002). Also a lot ol research has been done on the
psychology ol discrimination, locusing on e.g. stereotyping and prejudice (whitley
& lite, 2010). ln-group-out-group discrimination is based on a cognitive bias: in-
group-out-group bias or in-group lavoritism (1ajlel, 1981, whitley & lite, 2010)
relers to a pattern ol lavoring one's in-group members over out-group members.
1his bias contains elements such as out-group homogeneity (Quattrone & }ones,
1980, kubin & ladea, 2012), a pattern ol underestimating the dillerences between
out-group members. Also in the case ol speciesism we have some well-established
psychological knowledge about prejudice, stereotyping and the inlluence ol
language and words (see e.g. llous, 200!).
sscntia|ism
Our brains appear to be trained in essentialist thinking to categorize groups. 1he
lirst three arguments l presented against the species boundary indicate that there
really is no essence related to a species (see also lull, 1986). lssentialism means
that there are characteristics that all elements ol a specilic set (e.g. a species)
possess and elements ol other sets don't possess. All elements ol that set can be
accurately described and delined by those characteristics. 1hat specilic set
therelore has a unique delinition.
lssentialism in biology is rooted (or rellected) in ancient philosophical thinking
(e.g. llatonism), as well as major religions. ln those religions it is believed that
there is something special to all and only to humans: all humans, and only
humans, have an eternal soul, or are created in the image ol ood. lut since uarwin,
the scientilic consensus says that there is nothing special about a species. lt is just
5pccicsism as a mcra| i||usicn
19!
an arbitrary abstract classilication with its limitations and dilliculties (kachels,
1990). 8imilarly, a racist thinks ol races or ethnic groups as being essentialized
natural groups, even though it is now well known that there really is no essence
related to an ethnic group or race.
lven more: several studies give explanations lor this phenomena that people
rapidly (but incorrectly) tend to categorize entities in terms ol essentialized
groups (oil-white, 2001). Our intuitions are not always in line with science.
According to oelman (200!) and many other psychologists, children and adults
intuitively describe biological entities in essentialist terms. leople (lrom dillerent
cultures and backgrounds) automatically think that biological categories have
invisible essences (lloom, 2010). As we have seen in the section on discrimination
in part two ol this dissertation, the psychological mechanisms ol hierarchic
dualisms also tend do work with essentialistic concepts: e.g. the big gap between
the upper and lower side, the homogenization ol the lower side,.
looking at the literature, it is remarkable how many people delending
speciesism are essentialists, by relerring to personhood or humanity (the human
species) as a 'kind', having a 'substantial nature'. (8ee e.g. chappell, 2011, cohen,
2001, linnis, 199, p.48, lee & oeorge, 2008, 8canlon, 1998, p.186, 8cruton, 2000. 8ee
mcmahan, 200 and 1anner, 2006, lor an extensive critique ol the 'argument ol
kinds'.)
1his subtle mechanism is also rellected in our language. lt is amazing how olten
one encounters human-centric notions in our culture without even noticing. look
again at the title ol this section: what is the word 'human' doing there7 why not
'primate'7 Or look at delinitions ol discrimination: how olten are these delinitions
already lrom the start restricted to humans (or persons, where a person
automatically means a human)7 ll people might restrict the delinition ol
discrimination to our species, then a racist is allowed to restrict this delinition to
whites. lt is better to start lrom a really impartial delinition, as we have done in
the chapter about discrimination (section .!).
look at discussions between speciesists and vegans. low olten do speciesist
people respond with: lut humans."7 '1he human' does not exist. ll it exists, then
'the primate' or 'the mammal' would also exist. why does no-one mention them7
Antispeciesists who grew up in a speciesist society, really have to 'deprogram'
themselves. Olten animal rights activists still use some essentialistic language.
Once you are completely deprogrammed, you start to see how strange this
constant relerring to 'humans' really is. lt sounds like someone is constantly
relerring to 'dry-nosed primates'. ll you hear someone saying 'humans and
animals', it sounds as crazy as 'primates and animals'. 1he reader is invited to read
any book on animal ethics, and to replace everywhere 'humans' into e.g. 'primates'
or 'placental mammals'.
tnica| ccnsistcncy anJ Anima| ua|ity
194
8o il in a discussion about animal rights people respond by saying something
like: lut most humans have rationality and the capacity ol moral thinking", the
very same statement would be true lor the lamily ol great apes or perhaps also
primates. most great apes also have rationality (just count them: more than 6
billion great apes have high levels ol rationality).
6
And il species has an essence (ol
say rationality), why should the class ol mammals not have an essence either7 lt is
equally possible to look at mammals as a kind. ll mentally disabled persons get
special rights because they belong to the kind ol humans with a rational nature
(most humans have rationality), then it is also lair to say that humans do not get
special rights, because they belong to the kind ol mammals, lacking a rational
nature (most mammals are not rational beings). Our language is strongly biased
towards one group (the species ol humans), by presenting this group as having a
substantial nature or a kind. meanwhile, it neglects the other possible groups,
natures or kinds.
uoes language simply rellect our essentialist thinking, or is our essentialist
thinking amplilied by our language7 ln any case, the lact that a speciesist tends to
think ol species as essentialized groups does not imply that there is an essence to a
species.
let's brielly reler back to the mller-lyer illusion. we can say that straight lines
have an important essence, as they are primitive geometrical objects or can be
delined in a unique, simple way (e.g. zero curvature, shortest distance between
points). lut it would be strange to speak ol an important essence related to all line
segments having outward directing arrowheads. 1hese ligures do not lorm an
important category ol geometrical objects, because they are vaguely and
arbitrarily delined. lt requires a lot ol inlormation to correctly deline such
geometrical objects, just as it requires a lot ol inlormation to correctly deline a
species such as ucmc sapicns. And many elements in those delinitions will appear
highly arbitrary. we built a geometry on lines, we do not built a geometry on
special ligures with arrowheads. 8imilarly, we should built an ethic on morally
relevant criteria (e.g. well-being), not on highly complex categories such as
species.


6
lnterestingly, alter hearing this statement, people olten have an automatic response that most ol the
great apes do not have rationality: only humans have rationality, whereas orang-utans, gorillas,
chimpanzees and bonobos do not. 1hat is lour against one. 1his is other evidence that people tend to
think in terms ol species (or genera, because there are dillerent species ol gorillas) instead ol
individuals.
5pccicsism as a mcra| i||usicn
19
1here is more to say: essentialism is a very clear example ol a heuristic that uses
attribute substitution (see 8unstein, 200). 1he next section gives an extensive
discussion on speciesism as a prime example ol a moral heuristic.


8.8 8peciesism as a moral heuristic
8

lver since kyder (191) introduced the term speciesism - a prejudicial
discrimination on the basis ol species membership - more than 40 years ago, it has
attracted a great deal ol controversy
9
. 1his chapter combines the philosophical
rellections on speciesism with a recent development in moral psychology, namely
moral heuristics (8unstein, 200).
uo the test: ask any person what justilies our current use ol animals lor
experiments, lood, clothing or entertainment. chances are high that you will hear
an answer that sooner or later relers to a distinction between humans and non-
human animals. Next, you can ask them what it is about humans that other
animals lack and that justilies a dillerent treatment ol humans and animals. most
people will answer this question, so it is a common beliel that this is a meaninglul
question. Now, again the chances are very high that the answer will reler to a
mental capacity that most humans have and animals lack: sell-consciousness,
creativity, rational rellection, the ability to speak, understand ethics, sign social
contracts, have a sense ol justice, and many others. 1he list ol authors and
philosophers who have delended speciesism by relerring to such mental capacities
is long (see e.g. carruthers, 1992, cohen, 199, 8cruton, 1998).
1he antispeciesist now comes up with the 'argument lrom marginal cases' (see
e.g. uombrowski, 199, wilson, 2001), which might be better (more neutrally)
termed 'argument lrom atypical humans'. Atypical humans reler to a minority
group ol ucmc sapicns who lack mental capacities such as rationality. 1he
argument says that such atypical humans exist, and giving those atypical humans
a moral status comparable to typical humans would be inconsistent il the mental
capacity is a necessary condition lor moral status.

1he section is based on lruers (201!a).


8
1his section is based on lruers (201!), 5pccicsism as a Vcra| ucuristic, lhilosophia.
9
lor some recent discussions in the literature, see lernstein (2004), chappell (2011), orau (2010), lorta
(2010b), lee and oeorge (2008), liao (2010), mcmahan (200), Nobis (2004) and 1anner (2009).
tnica| ccnsistcncy anJ Anima| ua|ity
196
conlronted with this argument lrom atypical humans, some people delending
speciesism attempt to extend or reline their criteria in the hope ol including all
atypical humans (and still excluding all non-human animals). 1hey reler to the
potentiality ol developing a certain mental capacity in the luture, the possibility
that they themselves might later become mentally handicapped, or the presence
ol interpersonal relationships between those atypical humans and typical humans.
lowever, it is striking that those attempts are too olten doomed to lailure
(some authors who have delended the argument lrom atypical humans against
such attempts include: uombrowski (199), luther (200), mcmahan (200) and
1anner (2006, 2009)). 1he antispeciesist can persist by relerring to more
extraordinary atypical humans who lall outside the scope ol those extended and
relined criteria. 1he abovementioned criteria are invalid when applied to lor
example an incurable, seriously mentally handicapped young orphan. 8uch human
beings exist (in lact, l happen to be a loster parent ol such a Vietnamese boy). l am
not aware ol any proposed set ol mental capacities plus relinements that allows
the inclusion ol such humans in the moral realm, and at the same time excludes all
non-human animals.
ln the many conversations that l have had, my opponents who delended
speciesism olten gave one linal response: a simple allirmation that those
extraordinary atypical humans are still humans and therelore should be
protected. 8trikingly, people giving such a response are olten not aware ol the
circularity in this reasoning. And what is more: using the argument lrom atypical
humans in such conversations olten triggered reactions varying lrom indignation
to overt outrage.
8.8.1 1he heuristics hypothesis
lor animal rights advocates, the above sounds very lamiliar. 1he hypothesis that l
want to put lorward is that this common speciesist thinking is based on a
heuristic. leuristics are intuitive, ellicient rules ol thumb applied when lacing
complex problems (lahneman & 8hane, 2002). As will become important in our
discussion ol speciesism, these heuristics work by a process called 'attribute
substitution': our brains (unconsciously) substitute a computationally complex
target attribute lor a heuristic attribute that is easier to calculate or detect. ln
recent literature, as a spin-oll ol the work ol lahneman and 1versky (1982), the
study ol moral heuristics has gained some inlluence (8unstein, 200, 8innott-
Armstrong, Young & cushman, 2010). ln general, a heuristic works pretty well in
most cases, but as 8unstein argued, in certain, atypical situations, moral heuristics
might 'mislire' and create erroneous intuitive judgments. l am going to argue that
5pccicsism as a mcra| i||usicn
19
this misliring ol the heuristic is exactly the case in situations with atypical
humans. ln lact, speciesism is a very clear example ol the mechanism ol attribute
substitution. ll the speciesism heuristic hypothesis is true, it can explain why a lot
ol people are 'blind' to the argument lrom atypical humans, why a lot ol people do
not seem to be aware that they deny the rights ol mentally disabled humans when
pointing at some complex mental capacities, and why the speciesism intuition is so
obstinate.
A lot ol people have the conviction that moral status depends on a complex
mental capacity, such as rationality. 1his mental property is the so-called target
attribute ol a being. lut the problem ol this target attribute is that it is dillicult to
detect. ll we encounter a being, how can we quickly decide whether or not she has
the relevant mental property7 Our brains have lound a solution: they
unconsciously substitute the target attribute lor a heuristic attribute that is easier
to detect. 1his heuristic attribute is based on something our brains are good at:
pattern recognition (margolis, 198). lor example, looking at ligures, we can very
quickly interpret a ligure as the letter A, without being able to explain what
exactly characterizes a letter A. computers are not (yet) able to detect a letter so
quickly. 8imilarly, looking at objects, we can very quickly determine whether it is
a human, even il no-one is able to clarily what set ol elements, conditions and
characteristics delines a human being. we look at an individual and immediately
see the pattern (lace, behavior, etc.) that corresponds to a human, because our
brains are trained that way. Now, looking at the set ol objects that have the target
attribute ol rationality on the one hand, and the set ol objects that have the
heuristic attribute ol a human being on the other hand, we see a strong overlap
between these two sets, with a low percentage ol exceptions. 1he exceptions are
the atypical humans. most beings that have the 'human pattern' also have the
target attribute. 8o our brains use the species criterion (our human recognition
capacity) as a heuristic. when an object looks like a human, when it has the
characteristic pattern ol a human, intuition says that the object has the target
attribute mental capacity as well. 1his 'speciesism' or 'looks-like-a-human'
heuristic works pretty well in most cases, but not in the atypical cases. ll
speciesism is a heuristic, it explains why antispeciesist people so olten reler to the
argument lrom atypical cases.
1he speciesism heuristic becomes particularly clear in the recent work ol,
among others, chappell (2011), who relers to a 'lolk' notion ol human species
(which - as l interpret it - is based on our pattern recognition skills) to determine
who counts as a person: ln normal cases, we have already identilied a creature as
a person belore we start looking lor it to manilest the personal properties, indeed
this pre-identilication is part ol what makes it possible lor us to see and interpret
the creature as a person in the lirst place. And that pre-identilication typically
tnica| ccnsistcncy anJ Anima| ua|ity
198
runs on biological lines." (chappell, 2011, p.1). 1he pre-identilication is nothing
but the attribute substitution: our brains immediately and unconsciously
substitute the target attribute (a property ol personhood - or what chappell and
others might have in mind: a complex pattern ol mental properties that constitute
personhood) lor a heuristic attribute (that runs on biological lines").
lven though we know that heuristics can sometimes result in erroneous
intuitions or judgments, it does not imply that we are better oll without heuristics.
8unstein (200) and rule utilitarians (see discussion in 8haw, 1999, pp.14-10)
argued that without those heuristics or rules ol thumb we might make more
mistakes. 1he question l address in this section is whether using the speciesism
heuristic is permissible, uselul or dangerous. uo we have to keep it, improve it or
throw it away because it makes some errors in atypical cases7
ln the lollowing sections l discuss the strongest pros and cons ol keeping the
speciesism heuristic in atypical cases. Alterwards, l argue that - even il it is not
irrational or inconsistent to stick to the speciesist heuristic - it is better (more
respectlul towards atypical humans) to take another heuristic which uses
sentience instead ol rationality as its target attribute.
8.8.2 1ime and knowledge constraints
One advantage ol heuristics is that these are rules ol thumb that can help us make
quick decisions in situations with time and knowledge constraints. compare
heuristics with trallic laws. 1he target attribute in trallic would be a rule such as:
Always drive as to maximize well-being" or maximize elliciency and minimize
accidents." 1his target attribute rule is too dillicult to lollow, so it is substituted
lor simpler heuristic rules, such as: Always stop in lront ol a red trallic light." lut
in atypical cases, when there really is no other trallic around, there is no harm in
ignoring a red light. most people would say that introducing a new trallic law:
8top at red lights except when crossroads are sale," would make matters worse,
because we cannot be sure enough whether crossroads are sale. lerhaps we are
not smart or alert enough to judge the salety. lerhaps we are tempted to judge
salety to our own advantage. lerhaps we are biased and we ignore red lights even
when the situation is not sale.
8o, it is olten conceded that strongly holding on to heuristics is a good strategy.
uoes the same apply lor the speciesism heuristic7 1he dillerence between trallic
situations and situations related to treatment ol atypical humans is that in the
latter we do have time and (scientilic) knowledge to inlluence our decisions. 1rue,
in emergency situations, the analogy with trallic might be valid. ll you see some
creature drowning, and you have to be quick to decide to rescue that being, it
5pccicsism as a mcra| i||usicn
199
would be ellortlul, time-consuming and unreliable to look lirst lor the mental
capacities ol that drowning being. ll you see it is a human, you will show a direct
response and jump in the water to save this human. And in most cases, your
judgment will be correct: in most cases, the drowning human will be a rational
sell-conscious being who deserves to be saved. weighing the probability that it is a
mentally handicapped human against the cost lor you to rescue the human, would
still make you conclude that it is better to stick to the speciesism ('looks-like-a-
human') heuristic.
lut when we have to decide how to treat mentally disabled persons, whether
we should clothe them, leed them or use them in experiments, we do have time
and access to inlormation about their mental states. ln these cases, other
heuristics than the 'looks like a human' heuristic might be more accurate, in the
sense that these new heuristics also cover all rational beings, but include lewer
non-rational beings. lor example, we could look at results ol communication or
lQ-tests, adaptive behavior or neurological lunctioning. And scientists might come
up with more accurate and laster techniques to see what the mental capacities ol a
mentally disabled human are. leople with average intelligence might be
vulnerable to bias and erroneous judgments about the mental capacities ol beings,
but scientists and judges might be able to make sulliciently wise judgments.
8o where do we stand7 uo we leel comlortable with the idea that having
lundamental rights would depend on our subjective state ol knowledge7 lt is
absurd to claim that mentally disabled humans have rights merely because we are
at this moment 'too stupid' to work with more accurate heuristics. we, and at least
scientists, philosophers, and judges, are intelligent enough to determine which
human being is certainly not a rational being, and l am not aware ol historical or
psychological evidence that suggests that using the target attribute directly or
using more accurate heuristic attributes instead ol the speciesist heuristic
attribute results in real violations ol the rights ol rational beings. 8o, in most cases
we do have time and we have already developed ellicient ways to detect mental
capacities. lut one might object: how reliable are those scientilic tests7 And a
more lundamental question is how reliable should these tests and relined
heuristics be7 Answering that question eventually becomes a matter ol taste, ol
gut-leelings. Although my intuition says that time and knowledge constraints are
not sullicient reasons to stick to the speciesism heuristic, we should accept that
this discussion remains unresolved and that it is not yet irrational or inconsistent
to stick to the speciesism heuristic due to the above concerns about our limitations
ol knowledge.
tnica| ccnsistcncy anJ Anima| ua|ity
200
8.8.! lear ol a slippery slope
lollowing carruthers's slippery slope argument (carruthers, 1992), one could
argue that it might be better to retain the speciesism heuristic. 8everal
philosophers have criticized the slippery slope objection ol carruthers
(uombrowski, 199, 1anner, 2009), but we can look at this argument lrom the
heuristics perspective.
leople might worry that not retaining the speciesism heuristic, i.e. using the
target attribute ol rationality directly instead ol the heuristic attribute, might
result in more serious errors overall. more rights ol real rights holders (i.e. truly
rational people) might be violated, because the target attribute is dillicult to
detect or there is no sharp distinction between having and not having the target
attribute.
A lirst reason that people might give to justily this view is that rationality and
other mental capacities are a matter ol degree. 8o we have a 'sorites' problem ol
where to draw the line. when removing grains lrom a heap, when does the heap
become a non-heap7 when removing mental leatures, when does a person lose its
rationality7 ll we cannot answer this question, we risk making erroneous
judgments about the rationality ol some persons.
8econd, one might point to the lact that our cognitive biases can unconsciously
skew our judgments. ln situations involving possible atypical humans, we might be
vulnerable to bias and erroneously judge the situation to our own benelit (we
might too easily start to think that a specilic human is non-rational, and treat him
or her as a non-rational being to our own benelit).
1hese two observations combined will put us on a dangerous slippery slope,
where we will move towards real violations ol the rights ol rational people. lut
also two objections to this slippery-slope argument can be raised. 1he lirst is that,
il there is indeed an unavoidable spectrum ol mental capacities, we might be able
to couple this to a spectrum ol rights. 8ome people with higher mental capacities
could be given more or stronger rights claims.
lowever, this lirst objection might not run so smoothly. when it comes to
lundamental rights, some might preler to stick to the binary view: either one has
an absolute claim to this right, or one does not have the right at all. 8uch a binary
view cannot be coupled in a non-arbitrary way to the supposed spectrum ol
mental capacities.
A second counter-argument to the slippery-slope argument is that we are
already able to deal with such slippery slopes. consider situations where we have
to decide whether mentally handicapped humans have a right to vote or a right to
marriage. uillerent countries have dillerent ways ol dealing with the right to vote
lor mentally handicapped humans. ln some countries, a judge or medical
5pccicsism as a mcra| i||usicn
201
practitioner will decide whether a mentally handicapped person has this right. ln
others, the person needs to undergo a psychological test or needs to be under a
protective measure such as a guardianship (lor the situation in luropean
countries, consult lkA (2010)). whatever solution a country prelers, there seems
to be a general lack ol worry that this exclusion ol mentally handicapped persons
lrom the right to vote would put us on a slippery slope towards broad violations ol
the right ol rational humans to vote.
lowever, this second counter-argument is based on a presupposed analogy
between a right to vote and a more lundamental right such as the right not to be
harmed or the right not to be used as merely means to some else's ends. One might
object that we should be more concerned about slipping down a slippery slope
when the slope involves a lundamental right. A second possible objection to this
second counter-argument is that the demarcation line between people who can
have the right to vote and those who do not might be easier to draw than the
demarcation line between rational and non-rational people. lt might be easier to
check whether someone is able to vote (i.e. by doing a communication test), than
to check whether someone is able to reason or is sell-conscious. 1he analogy
between spectra ol lundamental rights and spectra ol political rights might be too
weak.
1he above discussion indicates that, as with the argument ol time and
knowledge constraints, things are not yet completely resolved. lt might come
down to a kind ol uncertainty aversion: il people have a strong lear lor slipping
down the slope ol lundamental rights when we are conlronted with atypical
humans, they have a strong uncertainty aversion. 1hey are worried about the
question: what might happen to my rights and the rights ol my loved ones, il we
stop giving lundamental rights to mentally handicapped humans7"
1he major problem that l have with this slippery-slope delense ol the
speciesism heuristic is that it seems disrespectlul to claim that the moral status ol
mentally handicapped people merely depends on our uncertainty aversion or our
supposed inability to put barriers on a slippery slope, instead ol it depending lor
example on the real interests and leelings ol those mentally handicapped people.
Are we so sure that we will slip down the slope when we look lor more accurate
heuristics than the speciesism heuristic, say a heuristic based on some
psychological tests7 1his question has no easy answer, but at least to me,
relinements ol the heuristic rule (to make it better lit with the target attribute
such as rationality) do not seem to be impossible, nor do they seem to be so
dangerous lor the rational people. 1hey are dangerous lor a-rational, atypical
humans.
Also, l am doubtlul that those people who delend the speciesism heuristic due
to an uncertainty aversion, have a consistently strong uncertainty aversion in
tnica| ccnsistcncy anJ Anima| ua|ity
202
other situations in their lives. lt seems strange to me that merely avoiding a
slippery slope is the real motivation lor people to take such care ol mentally
disabled orphans. ll that were the real motivation, we could expect that one would
have a very high level ol uncertainty aversion and lear ol slippery slopes. lut such
a high level ol uncertainty aversion seems incompatible with the way we deny
some atypical humans a right to vote (it is unlikely that judges are really unable to
make wise decisions about who is able to vote), and with the treatment ol animals
in lactory larms. we do not seem to worry at all about slippery slopes or the
potential negative inlluence on our rights when we treat thousands ol sentient
animals the way we do in lactory larms. lut, admittedly, it is dillicult to test such
apparent inconsistencies in people's uncertainty averse attitudes towards slippery
slopes.
8.8.4 1he emotional cost ol excluding atypical humans
moral heuristics are olten strongly internalized rules, which means that rule
violations are olten accompanied with strong emotions ol indignation, guilt or
moral disgust. 1he abovementioned reactions ol people (sometimes overt outrage
when they are conlronted with the argument lrom marginal cases) indicate that
the speciesism heuristic and the rights ol atypical humans are also strongly
emotionally charged. leople have empathic concerns lor the mentally disabled.
lven il the emotions that people leel towards atypical humans would be irrational
il all that mattered was a property (mental capacity) that those atypical humans
lack, we observe that violating the heuristic will result in an emotional cost, and
this cost is not to be underestimated. 8ometimes it might be rational to stick to
irrational leelings.
compare it with lear ol heights. lmagine that most people had a strong lear ol
heights. uo they react irrationally7 Not necessarily: these people might claim that,
il they conquered their lear in situations where they could not lall (e.g. when they
are salely attached), they might react less learlully in more dangerous situations
where lear is required or advantageous. 1hey might know about themselves that
they will be tempted to make erroneous judgments in dangerous situations. And
they also know that it takes some ellort (e.g. some costly therapy or locussed
meditation) to overcome their lear ol heights. 8o these people have a heuristic:
always avoid tall buildings. 1he costs ol overcoming their lear might be greater
than the cost ol avoiding tall buildings, so even when they could not lall lrom the
tall building, they do not necessarily react irrationally by keeping the heuristic.
1hese people have weighed all the costs and benelits, including the emotional
ones.
5pccicsism as a mcra| i||usicn
20!
1he same could be said about the leeling ol indignation that one experiences
when looking at human rights violations, even il the human is mentally
handicapped and is lacking the relevant mental capacities. 1he emotional cost ol
reacting in a more detached or neutral way towards those atypical humans might
be greater than the benelits that one could obtain lrom violating their rights.
lowever, this weighing ol the emotional cost against the potential benelits is
olten very dillicult. One might object that the benelits lor real rights holders ol
violating the rights ol atypical humans should not be underestimated either. 1here
might be health advantages in using atypical humans in e.g. medical experiments.
Atypical humans (such as mentally disabled orphans) are olten better research
models lor rational humans, compared with non-human animals, because the
atypical humans are genetically and physiologically closer to the rational persons.
8o their use in medical experiments might give better results than il non-human
animals are used as models lor rational persons. 1he atypical humans could also be
used lor organ transplantations and blood translusions, to help rational persons in
need. ll we do not sacrilice non-rational, atypical humans, then rational humans
might die. ls that not more serious than the abovementioned emotional cost7 lt is
not easy to decide this issue.
consider again the trallic laws. most people leel repugnance when they drive
through a red light, because they have internalized an important rule. lut are we
not allowed to drive through red lights in emergency situations (e.g. when there is
a child in the middle ol the crossroad, in the distance there is a car coming, and we
could only bring the child to salety by ignoring the red light)7 low stubbornly do
we have to stick to the heuristic trallic rules (always stop at red lights") in such
situations7 As with some ol the previous questions that l raised, we have to admit
that these questions do not have easy answers.
lor me, the 'emotional cost' delense ol the speciesism heuristic seems to be too
weak at this moment, but l again have to admit that my judgments might be
biased, that the emotional cost is not to be underestimated either or that using
atypical humans in experiments (instead ol using only non-human animals) would
not be a sullicient improvement lor the health ol rational beings.
ln summary, the above three delenses lor the speciesism heuristic (time and
knowledge constraints, slippery slope, and emotional cost), remain largely
undecided and are neither clearly irrational nor inconsistent. ln the next and linal
section, l will explore my main objection to the speciesism heuristic, even il this
speciesism heuristic was not applied in an irrational or inconsistent manner.
tnica| ccnsistcncy anJ Anima| ua|ity
204
8.8. 1he importance ol sentience
1he speciesism heuristic was based on the assumption that the target attribute is
some higher mental capacity such as rationality. ln my view, this seriously
underestimates the importance ol another mental capacity that even most animals
have: sentience. 1he importance ol sentience can be seen by asking the questions:
what is the real reason why people help mentally handicapped humans in
institutes7 what really drives those health care workers to take care ol atypical
humans7 l do not believe that they are willing to accept that the only reason why
they take such great care ol the mentally disabled is that they lear a slippery slope
or the emotional costs ol overcoming a heuristic.
1he idea that some humans have rights merely because our heuristic mislires
seems incompatible with the moral intuitions ol many people. lor example, it is
more plausible that the persons who take care ol handicapped people in lact
respond to the needs ol these people. 1hey are not concerned with a higher
mental capacity as the morally relevant target attribute. 1hese health care
workers have empathy, and they are happy when they see that the mentally
disabled humans leel pleasure or joy in something. 1hey want to avoid their
sullering. 1herelore, according to these carers, sentience is one ol the most
important target attributes, and sentience is likely to be the most important
motivator lor them to help these mentally handicapped people. ll the empathy
and moral intuitions ol a health care worker towards atypical humans are clear
expressions ol an undercurrent in the common morality ol our culture, which l
believe they are, we can say that sentience is important in our common morality.
lut its importance is underestimated due to the dominance ol the speciesism
heuristic.
1he problem with the speciesism heuristic is that it claims that rights ol
atypical humans are only indirect results ol misliring heuristics and that these
humans in lact do not deserve rights because they are not rational beings. lut it is
highly disrespectlul towards atypical humans to say that they only have an
indirect moral status, that they in lact do not deserve rights but that we intuitively
give them rights merely because our speciesism heuristic mislires.
consider the mainstay ol the speciesism heuristic: the ability to see a dillerence
between humans and non-human animals. without such a clear observable
distinction, the heuristic attribute would not be that uselul. lut as biology now
shows, this human-animal dillerence is not absolute or essential. ln the past, there
were human ancestors with more and more non-human (non-rational) properties
as we look lurther into the past. 1here is in lact a whole continuum ol ancestors,
moving down the evolutionary branch, till we meet a common ancestor ol, say,
humans and pigs. Also, it might not be genetically impossible lor human-
5pccicsism as a mcra| i||usicn
20
chimpanzee hybrids to be born. 8uch hybrids are inlertile ollspring ol a human
and a chimpanzee parent, which means that each cell contains the uNA ol both
humans and chimpanzees. Or what about human-animal chimeras, beings who
consist partly ol human body cells, partly ol non-human body cells7 Or what about
genetically modilied humanlike beings7 lt is hard to believe that we really leel
comlortable with the thought that, il the ancestors, hybrids, chimeras or
genetically modilied humanlike beings were alive among us, our speciesism
heuristic loses its strength and we would drastically alter our ethics and our
treatment ol atypical humans. Although these examples are hypothetical, they
should give us some discomlort.
lt is awkward to claim that mentally disabled humans are just lucky that we
have pattern recognition skills and that lor us in the current situation it is easier
to see distinctions between humans and non-humans than to see distinctions
between rational and non-rational beings. uisabled humans have basic rights, but
not because they are just lucky that the borderline humanJnon-humans do not
exist yet or do not exist anymore. And neither are they just lucky that the
borderline rationalJnon-rational beings do exist. ll there were no borderline cases
ol rationalJnon-rational humans, il we were clearly able to make a demarcation
between rational and non-rational humans, the slippery slope argument would
completely lail, because the slope would contain a really big gap, and this gap is a
good place to stop any lurther slipping down the slope.
10

All the above questions and rellections should take us to the conclusion that
sentience is more important than rationality. 8entience is a better target attribute
lrom a moral point ol view, lor two main reasons. 1he lirst reason is based on
impartiality. kowlands (199, 1998) and Van den lerg (2011) derived the sentience
criterion through a contractarian 'veil ol ignorance' thought experiment. 1his


10
mentally disabled humans would only have an indirect or dependent moral status il their moral
status depends on the existence ol human beings with intermediate levels ol rationality and the non-
existence ol human-animal hybrid beings. lence, a dependent moral status violates the intuition ol
independence. 1his can be compared with the discussion on independence in the prioritarian wellare
ethic (see appendix 2 lroblematic properties ol number-dampened prioritarianism"). whereas l could
tolerate a violation ol independence in the prioritarian wellare ethic (where the level ol priority lor
someone's well-being can depend on the (non)existence ol other beings), l do not tolerate a violation ol
independence when it comes to such a lundamental aspect as someone's moral status. As we have seen,
the violation ol independence in the wellare ethic is justilied on the grounds ol coherence between two
strong intuitions: the prelerences ol an impartial observer behind the veil ol ignorance and the
avoidance ol the repugnant conclusion together make a strong case to allow violations ol
independence. 1he violation ol independence in the case ol granting moral status cannot be justilied
because a similar strong coherence is lacking.
tnica| ccnsistcncy anJ Anima| ua|ity
206
thought experiment tests our impartiality by lorcing us to take the positions ol
others, as il behind a veil ol ignorance we do not know whose lile we are going to
live. 1he veil needs to be as thick as possible in order to respect maximum
impartiality. 1his means that we have to include the positions ol all non-humans,
non-rational beings and non-sentient beings. ll l put mysell in the position ol a
non-sentient being, things would not matter to me, because l would not have any
subjective experiences or consciousness. ll l were a non-rational but sentient
being, things would matter to me, because l would still have a sense ol well-being.
1his relerence to prelerences and what would matter to the subject is also the
basis ol a utilitarian-consequentialist vindication ol sentience (8inger, 19). ln
summary, when we value our own well-being (what matters to us) and we value
impartiality as in consequentialist or contractarian ethics, then the well-being ol
all sentient beings should be valued.
1his impartiality argument lor the sentience criterion is also coherent with a
second argument in lavor ol sentience: the virtue ol compassion (8lote, 2001). ln
almost all major religious and philosophical traditions compassion is considered as
one ol the greatest ol virtues. lt is based on a leeling ol empathy lor the sullering
ol others and a desire to act on that emotion. compassion is directly related to
sentience because we are able to leel empathy with (non-rational) sentient beings,
but not with non-sentient beings who cannot suller. compassion also plays a key
role in an ethics ol care (oilligan, 1982). compassion is what drives health care
workers to help mentally disabled people.
lmpartiality (supported by some interpretations ol consequentialist and
contractarian ethics) and compassion (supported by some interpretations ol virtue
ethics and ethics ol care) both point to the importance ol sentience. 8uch a
coherent justilication lor sentience is lacking lor the higher mental capacities such
as rationality. lrom an impartial perspective, having a sense ol well-being is not
restricted to rational beings alone. And lrom a virtues perspective, there is no
moral virtue that restricts attention to rational beings alone
11
. 1here are two
arguments to justily rationality, but these arguments can also justily sentience.
1he lirst argument to justily the criterion ol rationality is by pointing out that a
coupling between rights and rationality is not lar-letched: rationality can be
delined as the ability to understand and respect rights and interests. lut neither is
the coupling between rights and sentience lar-letched: rights protect interests and
leelings detect interests. lor example, pain indicates a violation ol bodily integrity


11
Although one can argue that virtues like honesty and lair-mindedness indirectly reler to a notion ol
rationality. lowever, the virtue ol compassion directly and strongly relers to sullering and sentience.
5pccicsism as a mcra| i||usicn
20
and lear detects the interest ol salety. 8entient beings value their own interests
due to their positive and negative leelings. 1herelore, it is not lar-letched to
couple rights to sentience. lrom a rights perspective, both rationality and
sentience are equally valid.
12

A second way to justily the criterion ol rationality is by relerring to intuitions
that some people have towards rational beings. 8ome people simply intuit that
rational beings have a higher moral status than other beings. lowever, the
sentience criterion is also coherent with our intuitions about helping mentally
disabled humans or pet animals such as dogs, and intuitions about prelerring
animal wellare laws. An exclusive locus on rationality in ethics cannot explain
those attitudes towards non-rational beings.
1!

As with other mental capacities, sentience is dillicult to detect. Ol course, when
conlronted with an individual being, we can always try to do some tests to see
whether it is sentient. ll we take sentience as the most important target attribute,
and il sentience is dillicult to detect, we can look at a suitable heuristic attribute.
leuristics might be uselul in many cases, so we should not throw away all such
rules ol thumb. looking at our current scientilic knowledge about sentience (see
next chapter), we can take the biological group ol vertebrates as the
corresponding heuristic attribute: vertebrates are also easy to recognize, and
science indicates that there is a strong overlap between the group ol vertebrates
and the group ol sentient beings (see e.g. orillin (2001) and ll8A (2009) lor
sentience in lish).
l suggest that - in contrast to our rather 'lixed' attitude towards the speciesist
heuristic - we can and should have a more 'llexible' attitude towards the
vertebrate heuristic. when possible (when scientists have accurate ways ol
determining sentience), it might be best to dispense with all heuristics, including
the vertebrate heuristic. lor example, we might also have to include some large
crustaceans and molluscs such as squid, because they also might be sentient. And
presumably some atypical vertebrate animals are non-sentient.
lt would not be disrespectlul towards those non-sentient animals il we do not
stick to the vertebrate heuristic, i.e. il we were to give those specilic non-sentient
individuals a lower moral status than sentient vertebrates. 1he reason why the
speciesist heuristic was disrespectlul towards non-rational humans is exactly


12
1he coupling between rights and species is really lar-letched, because we cannot see a connection
between rights and genes or between rights and the ability to beget lertile ollspring.
1!
ln common morality, the attitudes towards dillerent non-rational beings are not consistent.
consistency can be improved by uplilting the moral status ol non-human vertebrate animals, due to
their sentience.
tnica| ccnsistcncy anJ Anima| ua|ity
208
because those humans are sentient: they have a sense ol well-being and leelings
that express interests. 1he disrespect does not lie in the use ol a heuristic, but in
the choice ol the target attribute. choosing a target attribute that excludes
sentient but non-rational humans is disrespectlul, even when these atypical
humans are saved by the 'misliring' ol a speciesist heuristic.
1o conclude, people give too much credence to the arguments in lavor ol the
speciesism heuristic, or to arguments in lavor ol speciesism in general. 1he reason
why they give too much credence to those arguments is perhaps because those
people use animals on a huge scale. lor example, they decide three times a day to
eat animal products. 1hey have lriends and lamily who use animals in similar
ways. 1hey know that their parents and grandparents used animals in similar
ways. 1hey see 1V commercials that promote meat, see animal circuses in their
hometowns, and see no-one (or just a lew 'extremists') complaining, etc. lor those
who consume animals on a daily basis, a lot is at stake (especially lor their sell-
image), so we can expect that this creates a real bias towards justilications ol
speciesism and the use ol animals. 1hose people are less willing to accept the
moral importance ol sentience, and the extension ol rights to all sentient beings.
8.9 8ummary
looking at the above, we now have a lairly strong coherent picture that implies
that discrimination such as speciesism is a moral illusion. lts vertical and
horizontal arbitrariness, its artiliciality, its violation ol impartiality, its cognitive
impenetrability, its relation to cognitive biases such as essentialism and heuristics,
and the lact that ideologies such as white-dominant racism, male-dominant
sexism or human-dominant speciesism are strongly culturally determined, all
corroborate the conclusion that discrimination is an illusion.
l presented a set ol live arguments why the species boundary is irrelevant, and
another set ol live arguments why sentience is relevant. 1he lirst set ol arguments
(against the species boundary and essentialist thinking) is analogous to the
principle ol context-independence in the optical illusion. 1he second set
corresponds with the principle ol translation invariance in geometry.
All these arguments cohere with each other: we have a situation ol narrow
rellective equilibrium (uaniels, 199) where strong intuitions and principles
mutually support each other, and according to this narrow rellective equilibrium,
speciesism is an illusion.
5pccicsism as a mcra| i||usicn
209
lut there is more: speciesism is also in conllict with a wide rellective
equilibrium (see uaniels, 199 lor the notion ol wide rellective equilibrium). 1his
wide rellective equilibrium not only contains moral intuitions and ethical
principles, but also contains (scientilic) background knowledge about e.g.
psychology and cultural anthropology. we know that not everyone is susceptible
to speciesism (it depends on culture and education), and more importantly: we do
have insights in the psychological mechanisms behind speciesism: heuristics
(attribute substitution) and essentialist thinking (using language with prejudices,
stereotyping,.). compare this with the mller-lyer illusion: we know that not
everyone is susceptible to this illusion (8egall et al. 196!), and we do have insights
in the optical mechanisms behind this illusion (a heuristic ol automatic
perspective corrections lrom !u to 2u).
ln geometry we have a very coherent picture that is much stronger than this
one optical intuition about the dillerences in lengths ol the mller-lyer ligure. l
therelore believe that the whole antispeciesist picture is coherent to such a high
degree that it is much stronger than that one moral intuition about the moral
status gap between humans and non-human animals. what else would the
speciesist need in order to be convinced7 1hat the speciesist intuition is
cognitively impenetrable or that essentialist thinking happens automatically, are
not sullicient reasons to say that the species boundary is morally relevant.
At one point the analogy between the mller-lyer illusion and the speciesism
illusion goes wrong: the huge dillerence in our treatment ol pigs versus mentally
disabled humans can only mean that there is a huge ellect ol belonging to the
human species. On the other hand, the mller-lyer illusion is rather subtle, it does
not create huge dillerences in length judgments.
Nevertheless, l expect that this new way ol looking towards speciesism can shed
a light on why the speciesist intuition is so pervasive and dillicult to change. As
optical illusions, the speciesism illusion is cognitively impenetrable. 1he analogy
with optical illusions might help us to argue lor a more consistent ethical theory
based on equality between all sentient beings.

tnica| ccnsistcncy anJ Anima| ua|ity



211
chapter 9 1he sentience problem
l already mentioned that there is a gradation in mental capacities. 8ome beings
have a richer, more complex emotional lile than others. lut there is also the
scientilic question: which beings are sentient7 lt is a scientilic question, because
being sentient is a matter ol lact, and science is about discovering lacts. 8cience
can determine criteria lor sentience. 1he sentience problem then consists in
determining those criteria, and testing animals to see whether they satisly those
criteria. 1he latter raises real ethical concerns. lelow l discuss the scientilic and
ethical problems ol sentience.
9.1 1he scientilic problem
1here are lour criteria to see whether a living being is sentient.
1) 1nc aJaptivc rc|c cj jcc|inqs. lain and other leelings can oller an evolutionary
advantage to living beings. leelings ol pain can result in avoiding some behavior,
learning, protection or sell care. lut we have to be aware that the organ that
generates leelings (the brain), might consume a lot ol energy. 8o not every living
being will invest in such an organ. lor living beings who cannot move in complex
ways, having leelings is useless. llants cannot show the light, llight or lreeze
responses, so lear is not uselul lor them.
2) Anatcmica| uasis. we know that pain is related to specilic neurons
(nociceptors or pain receptors) and other anatomical properties (e.g. the central
nervous system). Also other leelings, such as lear, are related to specilic parts in
the central nervous system (e.g. the limbic system and the amygdala).
!) 8cnavicr. leelings are olten associated with specilic behavior (e.g. the light,
llight or lreeze response ol lear). lealing wounds, scratching, loss ol sexual
interests, vocalizations, body movements, lacial expressions and many other
tnica| ccnsistcncy anJ Anima| ua|ity
212
things might indicate that the living being leels pain. lspecially when the behavior
persists lor a long time, is repeated quite some time alter the event, or changes in
more complex ways, we can exclude the possibility that the behavior is just an
automatic, instinctive or rellexive response.
4) nysic|cqy. when there are other changes in the body (e.g. laster respiration
rate, heartbeat, blood pressure, eye movements,.) then we might expect emotions
to be present. Also the ellect ol some chemicals (endomorphines and analgesics)
on the behavior might indicate sentience.
with these lour indicators, we can test whether an animal is sentient. According
to the current scientilic consensus (masson, 199, orillin, 2001, lekoll, 200) most
likely vertebrate animals with a lunctioning central nervous system are sentient
and can leel pain, lear and distress. lerhaps some squids and large crustaceans are
also able to subjectively leel something.
let's take a concrete group ol vertebrate animals: lish. 1he lollowing citations
represent the current scientilic consensus (ll8A, 2009):
1here is scientilic evidence to support the assumption that some lish species
have brain structures potentially capable ol experiencing pain and lear. 1he
balance ol evidence indicates that some lish species have the capacity to
experience pain. [.] kesponses ol lish, ol some species and under certain
situations, suggest that they are able to experience lear. [.] lrom studies ol
sensory systems, brain structure and lunctionality, pain, lear and distress there is
some evidence lor the neural components ol sentience in some species ol lish. Our
knowledge and understanding ol manilestations ol sentience in lish, however, are
limited. [.] lrom studies ol sensory systems, brain structure and lunctionality,
pain, lear and distress there is some evidence lor the neural components ol
sentience in some species ol lish. Our knowledge and understanding ol
manilestations ol sentience in lish, however, are limited. [.] 1he stress physiology
in lish is directly comparable to that ol higher vertebrates."
we see that the statements are very prudent, because there are so many lish
species, and only a lew are studied (the most lamous studied species are rainbow
trout). 1he lish that were studied indicated signs ol sentience. 1he lollowing is a
list ol 11 criteria to test the presence ol a pain system.
1. 1here should be pain receptor cells present.
2. 1here should be a nociceptive neural pathway lrom the tissue to a
higher brain structure.
!. ln this brain structure, there should be specialized processing systems
that are active when the tissue is damaged.
4. 1here should be specialized transmitter substances along the neural
pathway.
1nc scnticncc prcu|cm
21!
. ln the specialized brain part, there should be endogenous opioids and
opioid receptors.
6. 1here should be electrophysiological responses to cuts and bruises.
. 1here should be a suspension ol normal activity associated with noxious
stimuli. lor example eating and sexual activity should stop.
8. 1here should be behavioral chance and avoidance in the short term
(moving body parts, scratching, avoiding the threat).
9. 1here should be learned avoidance ol places in the long term.
10. 1here should be a measurable inlluence ol analgesics in reducing
responses.
11. 1here can be ellects ol chronic stress (e.g. a mallunctioning immune
system alter long exposures to pain).
All these criteria are lullilled lor a rainbow trout. 1rout have nociceptors and a
neural pathway lrom these pain receptors to their brains. 8cientists have injected
a bee venom in the lip ol a trout. 1he trout stopped eating, and relused to eat lor
quite some time. le started scratching his lip in the sand. 1he trout showed less
avoidance when new learlul objects were placed in the water, as il the trout was so
concerned with his pain that he became less aware ol his environment. lut alter
the trout was injected with some analgesics, he stopped scratching, became more
aware ol his environment and swam away lrom learlul objects (8neddon et al.
200!a, 200!b).
consistency in judgments implies that we have to give those lish at least a
strong benelit ol the doubt. ll a mentally disabled human (who was not able to
talk) showed similar reactions in similar experiments, we would judge this human
to be sentient. 1herelore, we should have the same judgments towards those lish.
when it comes to invertebrates, things get more dillicult, because not all ol the
above criteria are satislied. 8ome insects simply continue eating alter they lose a
leg. 8entient beings would likely stop eating, no matter how hungry they are. we
could give insects the benelit ol the doubt, by not eating them.
llants are even less likely to be sentient, as none ol the criteria are lullilled.
8ome plants are able to react to a threat (e.g. produce poisons to protect
themselves against herbivores), warn other neighboring plants, and communicate
in rather sophisticated ways. 8ome plants might even have a mechanism lor sell-
recognition (larban, 2009). Although this is nothing yet compared to
(sell)consciousness, those plants have roots that can recognize whether other
roots belong to the same plant or not. 1hese are enough reasons to grant plants
also some basic right, as we have seen in the chapter on the basic right (section
6.4). lut we should not yet conclude that those plants are sentient. Our immune
system and computers also have very complex patterns ol communication and
sell-recognition, but that does not yet make them conscious systems. lt might be
tnica| ccnsistcncy anJ Anima| ua|ity
214
possible that plants have a yet unknown system that makes them sentient. lut
such a system is not yet lound, and lor the same matter, it might be possible that
the hairs on my head are sentient beings. lerhaps they also have a special system7
ll we would give plants the benelit ol the doubt, placing them at the same level
as real sentient animals, then we have to realize that we have to eat plants in order
to survive. lating plants is a survival end, and il plants become equal to all other
sentient beings, then we are also allowed to eat beings who are delinitely sentient.
we cannot run that risk based on the current very weak evidence, so we need
much more evidence belore we should say that plants are sentient.
Although we do not know yet how sentience is generated by the brains, l
believe that at the most basic level, having an allective leeling is binary: either you
have it or you don't. Once a being has at least one allective leeling that rellects a
need, it is sentient (with respect to that need) and it has a well-being. 1he above 11
criteria are nothing but rules ol thumb to determine lrom the outside whether
someone experiences pain.
1he reason why l believe in the binary nature ol sentience is that l cannot
imagine mysell to have a hall leeling. lither there is light on the stage (even il it is
a dim light), or there isn't. 1his distinguishes sentience lrom other mental
capacities such as rationality. kationality is more dillicult to deline, and it has less
this on or oll characteristic. 1here are multiple aspects behind rationality (you
need e.g. sell-consciousness, some imagination and a memory), and sometimes not
all aspects are present.
compare it with letters on a paper. leing sentience is like having a stain ol ink
on a paper: either there is a drop ol ink on the paper (even il it is a tiny drop), or
there isn't. kationality is like having the letter A on a paper: this requires much
more aspects. lt is not always easy to see whether the ink stains lorm a letter A or
not. lrom a scientilic perspective, it is in theory possible to see the presence ol ink
stains: one only needs a tool to see the stains (e.g. a good microscope). ll such a
tool does not yet exist, it does not mean that the presence ol ink stains is not a
matter ol lact. lowever, determining whether the stains lorm a letter A always
involves an element ol (subjective) interpretation.
1



1
1he analogy between mental states and ink stains can be explored lurther. lor example il ink stains
are connected, they belong to the same pattern (e.g. the same letter), just as dillerent leelings can be
connected to belong to a same person. 1here can be luzzy boundaries between ink stains, just as there
can be luzzy boundaries between mental states, generating the problem ol psychological
connectedness explored in section 4.2.4.
1nc scnticncc prcu|cm
21
9.2 1he ethical problem
we have seen criteria to determine whether a being is sentient, whether it can
experience pain, lear or distress. l mentioned an experiment that caused pain to
rainbow trout. 1he problem is: animal experiments are olten violations ol the
basic right, because the being is used as merely a means (experimental object). 8o
that would mean we would not be allowed to do tests to determine whether an
animal is able to leel. 1hat means we could not so easily determine whether the
animal is able to leel. lerhaps we might never know, even il it would be
scientilically possible to know.
8o, are those tests morally permissible7 we can argue that they are, because we
could say that we do not really violate the basic right ol such an animal when our
goal is to determine whether it has the basic right. ln other words: the animal is
used as merely a means to an end, and the end is determining whether the animal
has a basic right. 1his end is not a survival end, neither is it a vital need, a basic
need or a luxury need. 1he end is ol a totally dillerent moral category.
Actually, this idea is not larletched. we already have a similar approach
towards humans. what do physicians do when they have a patient who does not
seem to react to certain impulses7 8ometimes the physician tests the patient, to
see whether she still reacts to painlul stimuli. 1his is a test that might cause pain,
very similar to the tests with the rainbow trout. lt is not disrespectlul towards the
patient.

tnica| ccnsistcncy anJ Anima| ua|ity




21
chapter 10 1he predation problem
ln this chapter we arrive at the last big problem ol animal ethics. lt is a very
serious one, as it might punch a big hole in a consistent animal rights ethics. 1he
problem receives some attention lrom time to time (e.g. cohen & kegan, 2001,
cowen, 200!, lbert & machan, 2012, lverett, 2001, link, 200, lorta, 2010,
8apontzis, 1984, 8immons, 2009), but the challenge that this problem poses is
however underestimated by both animal rights ethicists and critics. lt is strange
why animal rights critics do not toss this problem about more regularly in
discussions, il their goal is to show that an animal ethics is inconsistent. 1o
present the problem as clearly as possible, let's start with the lollowing two
scenarios.
!) 1nc prcJaticn Ji|cmma. A lioness is going to attack a zebra in order to leed her
two hungry whelps. You are sitting in a car, looking at the scene. You can easily
save the zebra, simply by turning on the engine ol your car, chasing the lioness
away. 8hould you save the zebra7
.) 1nc transp|antaticn Ji|cmma. ln a hospital two children need new organs, but
no organs are available. A surgeon is about to kill a visitor against his will, in order
to use his organs to save the two patients. ls he allowed to do so, or should we
interlere and stop him7
A lot ol people, including most animal rights activists
1
, say that we do not have
a duty to protect zebra lrom lions, but we do have a duty to protect the visitor
lrom the surgeon. 1he killing ol the visitor against his will should be prohibited.


1
1his claim is based on a personal (unpublished) survey that l did with more than !0 animal rights
activists. 90/ ol them would not condone organ (xeno)transplantation, whereas only / would not
condone animals preying on animals and 14/ would not condone animals preying on humans. Also,
another 24/ ol respondents were undecided in the latter case. 1he preying on animals and organ
(xeno)transplantation dilemmas showed only 0-6/ undecided responses. 1his demonstrates that the
problem ol animals preying on humans is the most dillicult dilemma lor animal rights activists, and
that in the two predation dilemmas (preying on animals and on humans), animal rights activists are
more speciesist than in the two organ (xeno)transplantation dilemmas (using either humans or pigs as
organ donors). lowever, the order ol presentation ol the dilemmas also inlluences the judgments.
tnica| ccnsistcncy anJ Anima| ua|ity
218
Now, il we adopt an antispeciesist ethic, we have to be able to switch the positions
between animals and humans. lence, what il, instead ol a zebra, there was a
lonely human child being attacked7 Or what il instead ol killing a visitor, the
surgeon would kill a non-human animal (e.g. a pig) lor xenotransplantation (and
suppose that xenotransplantation works)7
lor an ethic based on animal equality, there are in lact two problems ol
predation and transplantation.
1) what is the morally relevant dillerence between predation and
transplantation7 Alter all, in both cases (meat consumption and organ
transplantation) a sentient being (lioness and surgeon) kills another sentient being
(zebra, human or pig) without permission and uses parts ol its body (muscle tissue
and organ tissue) by taking these body parts up in the bodies ol vulnerable
sentient beings (whelps and child patients) in order lor them to survive. 1he
analogy could not be clearer. Yet, most people, including delenders ol animal
rights, condone the predation but condemn the transplantation (antispeciests also
condemn xenotransplantation). 1he problem arises because our moral intuitions
say that in one lile-and-death-dilemma action is permissible, whereas in the other
dilemma it is not.
2) ln the predation situation, the animal rights activist would protect the
human child, but most activists would not protect the zebra. uoes this mean they
are still speciesist7
1hese two problems (the morally relevant dillerence between predation and
transplantation, and the dillerence between a zebra and a human as prey) will be
called the Jijjcrcncc prcu|cm and the prcy prcu|cm respectively. ln my opinion, these
two problems combined are the weakest spot in a consistent antispeciesist animal
rights ethic. 1hey are underestimated by both animal rights advocates and critics.
1he predation problem can be a strong weapon in the hands ol a speciesist
(such as cohen in cohen & kegan, 2001, p!0), because sJhe could easily solve both
problems with one stroke, by claiming that there is a morally relevant dillerence
between humans and non-human animals. According to speciesism, all individuals
belonging to the human species have a higher moral status than everything else.
ln that case, everyone is allowed to use animals lor xenotransplantation and
predation. lut lions are not allowed to use humans. we have a duty to protect
humans, but no duty to protect zebras. ly solving both problems with this one
simple criterion, it seems that the speciesist ethic is coherent. lut is it7
As we have seen in a previous chapter, the species boundary cannot be morally
relevant. 1he live arguments against this criterion show that the speciesist
solution is not coherent alter all. lt is at least as arbitrary and artilicial.
what l will do in this section is present solutions to both the dillerence and the
prey problems that are consistent with an antispeciesist theory ol animal equality.
1nc prcJaticn prcu|cm
219
uoing this we can avoid arbitrary elements in our ethic, and the resulting ethic lits
with our strongest moral intuitions.
1he principle ol tolerated choice equality might solve the prey problem,
whereas the triple-N-principle (relerring to three criteria: normal, natural and
necessary) and a 'lairness' principle might make a morally relevant distinction
between predation and transplantation. l demonstrate that the triple-N-principle
is in agreement with moral intuitions that a lot ol people (both speciesists and
animal rights activists) have, and that it is in correspondence with the moral value
ol biodiversity. 1he tolerated choice, lairness and triple-N principles might make
our antispeciesist ethic much more coherent than the speciesist one.
10.1 lnvalid solutions to the prey problem
8tarting with the prey problem, l lirst brielly mention a lew invalid solutions,
given by some animal rights ethicists.
8ome people claim that humans are not part ol the natural diet ol lions (e.g.
8immons, 2009). 8o lions are only allowed to eat what is part ol their natural diet.
One problem with this argument is that it poses a strange, arbitrary distinction
between humans and non-humans. lsn't it strange that ol all species on earth only
ucmc sapicns never are or have been natural prey7 more importantly: when is
something part ol someone's natural diet7 ll a lion wants to hunt a human, is the
human not part ol his prelerences, and hence part ol his diet7 As rabbits were
introduced in 1asmania some 200 years ago, they can be considered as not being
part ol the natural diet ol a 1asmanian devil. 8o we now have a duty to protect
rabbits lrom 1asmanian devils7
Another claim is that uuties ol assistance exist only insolar as potential
beneliciaries require assistance in order to llourish according to their nature"
(lverett, 2001 p) lt is believed that we don't have a duty ol assistance towards
zebra, because they can llourish without our assistance. On the other hand,
children cannot llourish without our assistance, so we have a duty to protect
children. 1he problem with this approach is that it seems to make an arbitrary
distinction between llourishing according to one's nature or not. what il the zebra
is injured7 uoesn't he require assistance then7 And what does llourish according
to their nature" really mean7 lt resembles some kind ol essentialist thinking that
we countered in a previous chapter.
tnica| ccnsistcncy anJ Anima| ua|ity
220
10.2 A hypothetical solution to the prey problem
when asking animal rights activists whether they would save the human, they said
they would, although they would not necessarily save the zebra. 1hey had real
dilliculties with those scenarios and were tempted to be what they thought was
speciesist, and consequently claimed that true antispeciesism was impossible to
achieve.
1he solution to the prey problem that l would propose, is based on tolerated
partiality (or tolerated choice equality). 8o the prey problem is a situation where
this equality comes in handy. lt is a principle that helps us make our moral
intuitions (at least a bit more) compatible with antispeciesism.
As in the burning house dilemma, we are allowed to save the sentient beings
with whom we leel a strong connection or empathy. 1hus we should tolerate a
choice to protect the lion instead ol the zebra or to protect the child instead ol the
lion. we have the right to be partial, as long as we respect equal partiality lrom
others. ll someone saves the human, sJhe should tolerate the choice ol someone
else saving the zebra instead ol the human. 1he tolerated choice equality allows us
to reconcile two prelerences ol the antispeciesist: on the one hand sJhe would (in
most cases) save a human rather than an animal, but on the other hand sJhe
doesn't want to be speciesist. we are not speciesist il we tolerate the choices ol
someone who protects the zebra or the lion (i.e. someone who did not protect the
human). lveryone is lree to choose whether to protect the zebra, the human or
the lion.
lmpathy is crucial in the tolerated choice equality, it is the driving lorce to help
vulnerable beings. ll someone leels more empathy and connection with one being
than with another, this justilies his prelerence lor saving the prelerred being. we
should tolerate hisJher choice, because empathy is a moral virtue and sJhe acted
with empathy as long as sJhe did not hate or disdain the other being.
lor speciesists this tolerance ol saving the zebra or the lion instead ol the
human is likely the hardest nut to crack in the whole theory ol animal rights. lt
might be in rather strong contradiction with one ol their moral intuitions. ln
practice however, they should not be so concerned, because asking what they
would do, most animal rights activists responded that they would save the human
anyway.
1here is one more thing. As we have seen in the chapter about the basic right,
we could introduce a criterion that relers to a higher moral status related to some
mental capacities like sell-consciousness, moral agency or rationality. 8o we might
have a stronger duty to protect those sentient beings who possess those special
mental capacities, or who will develop them, or who have close relatives (e.g.
1nc prcJaticn prcu|cm
221
parents) with such special mental capacities. 1he only thing a speciesist then has
to accept is tolerated choice equality between seriously mentally disabled human
orphans and non-human animals such as zebra. ll we say we have a duty to protect
those disabled orphans, whereas we do not have a duty to protect non-human
animals because all humans have a higher moral status than non-humans, then we
become too partial. lt is a kind ol speciesism, and like racism or sexism it is a kind
ol partiality that we cannot tolerate.
ln summary, we do not have a duty to delend the prey lrom predators, but we
are allowed to delend the prey il we leel an emotional need to do so. (we can add
that we are only allowed to delend the prey il this does not result in severe
ecological damage.) ln this permission to delend the prey, we have a right to be
partial to some degree (not too much), as long as we respect similar levels ol
partiality ol others. And il the prey has some special mental capacities, we might
have a stronger duty to protect it.
10.! lnvalid solutions to the dillerence problem
Next, we move to the dillerence problem. Again, l lirst brielly discuss a lew invalid
proposals encountered in the literature to solve the problem ol the dillerence
between predation and organ transplantation.
1) leasibility: stopping transplantations is leasible, stopping predation is not.
leter 8inger gave a practical argument: lt is impossible to intervene in nature to
protect all prey animals (8inger, 1990, p226). lowever, a single intervention, such
as saving a zebra in lront ol you, is leasible. 1he leasibility argument is related to
the demandingness objection against consequentialist wellare ethics, which can be
shown to be impotent in some way (8obel, 200).
2) moral agency: surgeons are moral agents, lions are not. kegan's answer
(kegan, 198!) is that carnivorous animals (e.g. lions) don't have moral reasoning
capacities, they are not moral agents. As amoral beings, they don't have duties ol
non-malelicence and benelicence. lurthermore, according to this view, we as
moral beings do not have a duty ol benelicence towards a victim when the agent
(aggressor) is an amoral being.
tnica| ccnsistcncy anJ Anima| ua|ity
222
1he next table presents the duties ol interlerence that we have when a moral or
amoral agent (aggressor) attacks a moral or amoral victim.
2


Victim
Moral being Amoral being
Agent
Moral being Obligation Obligation
Amoral being No obligation No obligation

1he table shows that we only have obligations ol benelicence towards the
victim when the aggressor is a moral being. lut this moral agency account laces
some counter-intuitive problems. lirst, what il amoral predators attack moral
humans7 kegan is not clear about our duties ol saving those humans, but even he
points at an intuition that at least it cannot be wrong to interlere and save the
human.
!
8econd, a lot ol animal rights ethicists believe that we do have a duty to
protect sentient beings lrom amoral threats such as lalling rocks. 1hird, what il
some animals were moral or gain moral consciousness7 At this moment no non-
human animals are real moral agents, although dolphins and great apes might
come close: they have a sell-consciousness, a high (social) intelligence and
premoral sentiments (8hermer, 2004, p. 16). And dolphins kill and eat sentient lish.
A strong ethical theory should be able to deal with counterlactual situations: what
il dolphins gained moral agency7 uo we then have a duty to interlere and save the
lish7 lourth, what il some humans needed meat to survive7 my intuition tells me
that we do not have a duty to stop moral humans or moral dolphins lrom hunting
il those humans and dolphins need meat in order to survive. Alter all, it seems
strange why non-rational (amoral) predators should have an unlair advantage
against rational (moral) predators: the lormer can hunt, the latter can't. (lor
lurther criticism ol kegan's argument, see also }amieson, 1990 and lox, 1999
p.16!.)


2
8uppose that interlerence (benelicence towards the victim) always harms the agent. lor example
interlering in predation harms the predator by limiting its lood supply.
!
As discussed in lbert & machan (2012), kegan's theory would imply that it is prohibited to save the
human prey. lut elsewhere, kegan claimed that 'it cannot be wrong to do what will harm the child
[who has come into possession ol a loaded revolver and has begun to lire it at us], even though the
child is innocent and so does no wrong.' (kegan 198! p. 29!). lt is not clear why our obligation to
interlere in this situation with the armed child would not extend to the situation ol the predator.
1nc prcJaticn prcu|cm
22!
!) libertarianism. kelated to the moral agency account, is the theory ol
libertarianism (lbert & machan, 2012). 1his theory says that moral agents have a
duty not to harm others, but they do not have a duty to protect others (e.g. protect
prey). ln other words: amoral beings do not have rights ol benelicence, a right to
be protected lrom harm. 8o libertarians say that we do not have a duty ol
benelicence towards a victim when the victim is an amoral being. lut all moral
agents still have a duty ol non-malelicence towards moral and amoral sentient
beings. 1he next table presents our duties ol interlerence to help victims lrom
aggressors (agents). 1his libertarian theory is in some way a 'transposition' ol
kegan's account.

Victim
Moral being Amoral being
Agent
Moral being Obligation No obligation
Amoral being Obligation No obligation

1his libertarian theory is also plagued with some problems. lirst, mentally
disabled humans are amoral beings and hence have no right to be protected.
4
1his
seems counter-intuitive to me. 8econd, hypothetical carnivorous moral agents
(e.g. moral dolphins) would have a duty ol non-malelicence, so they should not
hunt. Although we do not have an obligation to stop the dolphins lrom hunting,
the moral dolphins themselves should now abide the rights ol sentient lish.
8imilarly, moral humans have a duty not to move when insects are sentient
(because moving around harms those sentient insects and animal rights
libertarians say that moral agents have a duty ol non-malelicence towards all
sentient beings).


4
lbert & machan discussed an extension ol this strict libertarian ethic, where we have special duties
towards some amoral beings, based on special relationships we have with them (e.g. parental
relationships). lut is the relationship that l have with a mentally disabled crpnan more special than the
relations that l have with amoral animals7 1he problems with this extension is: 1) what counts as a
sulliciently special relationship in order lor us to have a special duty7 And 2) how to avoid arbitrariness
(e.g. relerence to arbitrary biological classilications such as species) in this extended libertarian
theory7 A non-arbitrary extension that includes the rights ol mentally disabled orphans to be saved
lrom harm, results in a complete animal rights ethic, where we have a duty to help amoral animals as
well.
tnica| ccnsistcncy anJ Anima| ua|ity
224
4) oroup protection: lions lorm a group, patients in the hospital don't.
lxtinction ol a group is worse than the death ol a number ol single individuals. lut
lirst, a group (e.g. a species) is an abstract set ol individuals. uoes an abstract set in
itsell have interests7 why give moral status to this abstract set7 8econd, a group is
an arbitrary set. what group should we take7 8hould we consider the species ol
lions (antncra |cc) as a group7 why not take a specilic population ol lions as a
group7 Or lor that matter the genus ol antncra, the lamily ol lelidae, the order ol
carnivora, the class ol mammalian or the phylum ol chordata7 And il we have
siblings (or twins) in the hospital who both need organs, aren't these siblings a
group in some sense7 1hey have some unique properties in common. 1hird, the
group argument is artilicial. consider the lonely lion problem: what il there is only
one lion lelt in nature7 ls this one lion still a group, or a single individual7 ls he
allowed to hunt7 uoes the right to hunt depend on the presence ol other similar
individuals7
) lllness: the patients in the hospital are ill, the zebras aren't. lut what does
'being ill' mean7 loth the patients and the whelps are leeling very sick when they
don't get something. 1he symptoms might be quite similar. loth will die
eventually. lor a predator, meat can be considered as a medicine to stay healthy.
8econd, this criterion doesn't seem to be quite empathic towards the ill persons.
why should you lose the right to use someone (violate his rights) il you are ill, and
are you allowed to use someone il you are not ill7
6) lxistence: lions would not even exist il they were not allowed to hunt,
surgeons and patients in the hospital were born, even il lorced transplantation
was not allowed. lut suppose that we have a boy who has a genetic delect, and this
boy needs a new organ in order to survive. Also, the mother ol this boy had the
same genetic delect, and !0 years ago they did a successlul organ transplantation
to save this woman. 1he woman survived, and gave birth to a son with a similar
genetic delect and thus a similar disease. ln this case the boy's very existence
depends on the transplantation lor his mother. 8hould we now make an exception
and tolerate the coerced sacrilice ol someone lor organ transplantation to save
this boy7
) lcological disasters: preventing predation would result in ecological
catastrophes, whereas prohibiting transplantation would not. 1his argument
actually makes some sense.

ll we would consistently intervene in nature to



1his point was already made by 8inger in 19!: lions play a role in the ecology ol their habitat, and
we cannot be sure what the long-term consequences would be il we were to prevent them lrom killing
gazelles". many other philosophers made this point (e.g. 8immons, 2009).
1nc prcJaticn prcu|cm
22
prevent predation, then a lot ol predator populations might die ol starvation,
which would have ecological ellects on prey species (e.g. overpopulation,
increased competition and spread ol diseases). lt is very dillicult to calculate the
overall ellects on animal death and sullering, because ecological interactions can
be very complex. lrom a precautionary principle we might say that at this
moment it is better not to consistently intervene in predator-prey interactions in
nature. 1his means we do not have a duty to intervene. Ol course, once (in a lar-
away luture) ecologists would be able to calculate that the extinction ol predator
species would be good on the whole, then we should go lor predator extinction.
8) Lncertainty aversion. 1his is related to the previous proposal ol ecological
disasters. lt can be derived lrom the thought experiment ol impartiality (kawls'
veil ol ignorance), as described in the section about prioritarian ethics. we have
seen that lrom behind the veil ol ignorance, one's level ol risk aversion was
important, resulting in a theory ol quasi-maximin prioritarianism. l also
mentioned that next to risk aversion, uncertainty (or ambiguity) aversion is
important as well. we have seen that the solution ol the trolley problem (pushing
a heavy man lrom a bridge) could depend on one's uncertainty aversion behind
the veil ol ignorance. 1o explain uncertainty aversion, l presented lllsberg's
paradox (lllsberg, 1961), and made a connection between this paradox and the
trolley problem. what l will demonstrate now, is that the very same idea ol
lllsberg's paradox might solve the dillerence problem.
1he lollowing version ol lllsberg's paradox will be uselul in this discussion. An
urn contains three balls ol two dillerent colors: green and red. You win the game
when you draw the green ball. You can choose between two games ol chance. ln
the lirst game, you know that one ball is green and the others are red. lence, your
probability to win is exactly 1J!. ln the second game, you only know that at least
one ball is red. Now your probability to win is between 0 (when all balls happen to
be red) and 2J! (when the two unknown balls happen to be green). which ol these
two games do you preler to play7 ll you have - like many people - uncertainty
aversion, you'd preler to play the lirst game, because in that game you at least
know your chances to win.
we can simplily the predation problem to see the analogy with the above
lllsberg paradox. 8uppose lrom behind the veil ol ignorance you know that you
will be born as one ol three sentient beings: one predator or two prey animals. You
now that the predator needs two prey in order to survive. You can now decide
whether predation is allowed or not. You have to choose between two games ol
chance. ln the lirst game, a world with predation, you know that the predator is
going to kill and eat the two prey. 1hese two prey lose, they are the red balls, and
red means dead. You have 1J! chance to be born as this predator and survive. 8o
your chance to win is 1J!. ln the second game, a world without predation, you
tnica| ccnsistcncy anJ Anima| ua|ity
226
know that the predator will die lrom starvation (he is the red ball). lut what
happens with the two prey7 ll there are enough resources, they can both survive
(they are both green balls). lut likely there are not enough resources lor both. 8o
they might start to light until one or both ol them die. Or they might overexploit
the resources so that one or both ol them eventually die lrom starvation. You
don't know what will happen, and most ol all: you don't know the probabilities lor
them to survive. Your chance to win is something between 2J! (il there are
enough resources lor both prey) and 0 (il everyone dies). Lncertainty aversion
implies that you would preler the world with predation.
1he dillerence with the organ transplantation goes as lollows. 1here are two
patients and one visitor. You can be one ol them. ll you choose a world where
lorced organ transplantation is not allowed, you will win when you are the visitor,
because he will survive (he is the green ball). lowever, il you choose a world with
transplantation, the visitor is sacriliced against his will, so he will be the red ball.
ll the bodies ol both patients accept the organs, they can both survive. lut il
one or two ol the transplantations are not successlul, one or two ol the patients
die. You do not know the success rate, so your probability to win will be between 0
(everyone dies) and 2J! (both patients survive). Lncertainty aversion implies that
you would preler the world without transplantation.
A lot ol people have uncertainty aversion, and this might already justily the
choice ol a world with predation and without transplantation. Lncertainty
aversion is able to make a dillerence between predation and transplantation. lut
even il we respect uncertainty aversion as a cognitive bias, this uncertainty
aversion account still laces some problems. lirst, it implies that moral duties
depend on our current, contingent, subjective state ol knowledge. ll knowledge
about ecosystem lunctioning increases, we might need to change our judgments
about predation, and our duties towards prey.
8econd, in a lot ol situations, the uncertainty aversion gets 'skewed'. consider a
more realistic predation problem behind a veil ol ignorance. 8uppose one lion eats
hundred zebras in order to stay alive. 8o there are 101 individuals: one lion and
hundred zebras. You can be born as any ol those animals. You can again choose
between two worlds. ln a world with predation, the lion eats all the zebras, so you
have probability 1J101 to be born as the lion and survive. ln the world without
predation, the lion delinitely dies, but as a consequence ol increased competition
and ecological overshoot, some ol the hundred zebras might die or start lighting
lor scarce resources and kill each other. You now have an uncertain probability
between 0 and 100J101 to be born as a zebra that is able to stay alive. lt's a game ol
chance, and you have to choose between a certain probability 1J101 versus an
uncertain probability between 0 and 100J101 to win (to survive). lven il the
1nc prcJaticn prcu|cm
22
second game has uncertainty, a lot ol people would still preler to play that game,
because 1J101 is close to 0 and hence the uncertainty is 'skewed'.
1hird, there might be people who think that even without uncertainty, we still
do not have a duty to interlere in predation. ll it seems counter-intuitive that our
attitude towards predation, and the late ol predators, depends on our subjective
state ol knowledge, we have to look lor a more principle-based ethic that allows
predation. 8o in the next section l want to explore a justilication ol predation that
might satisly the needs ol those people. 1he principle that l propose can be
expressed as a !-N-principle, related to the intrinsic value ol biodiversity.
10.4 A lirst hypothetical solution to the dillerence
problem: the !-N-principle
when it comes to justilication lor predation, a lot ol people also give answers that
are typically used by meat eaters: they say it is 'necessary', 'normal' or 'natural' lor
the lion to hunt. Note that these are the three N's that melanie }oy pointed at in
her discussion ol carnism, the olten hidden ideology ol most meat eaters in our
culture (}oy, 2001, 2009). 8o even some animal rights activists are still tempted to
use the same 'naturalistic arguments' that meat eaters (carnists) use.
what l am going to do now is 1) clarily the criteria 'necessary', 'natural' and
'normal' to make the principle more accurate, 2) demonstrate that none ol those
three criteria are separately valid (none are sullicient to violate rights), !) show
that the combination ol all three criteria might be in line with our moral intuition
(predation is natural, normal anJ necessary and therelore allowed, transplantation
is not), 4) argue that there is a connection between the three N-conditions and
biodiversity, and ) argue why biodiversity can be morally relevant to allow rights
to be violated. ll this strategy works, we not only have a consistent animal ethic
that is compatible with the moral intuitions ol most activists and other people (e.g.
with the intuition that we should not intervene in predation), we also have a
principle which is in essence based on criteria that carnists already use. 8o the
speciesistJcarnist would have more dilliculties in replying that this !-N-principle
is not a good solution.
lirst, we have to clarily the three criteria. 1he condition 'necessary' indicates a
sulliciently strong vital need lor an individual (e.g. lood) or a group ol individuals
(e.g. procreation). 'Normal' means lor simplicity that something occurs olten. ll
apples olten lall lrom trees, it is normal. 1he criterion 'natural' points at
tnica| ccnsistcncy anJ Anima| ua|ity
228
everything that is created by evolution. lvolution is the aimless process ol genetic
mutation and natural selection. conscious, intentional inventions by intelligent
beings are not considered natural, although the intelligence ol those beings is
natural (arose by evolution). A process is natural il it originates lrom natural
evolution instead ol being created by artilicial means (such as conscious
inventions).
8econd, l argue that none ol the three N criteria separately are sullicient
conditions. Necessity is not a sullicient criterion: lor the patient in the hospital it
is necessary to get a new organ in order to survive. 8o necessity alone does not
distinguish between predation and transplantation. Yet, in ethics, it seems
intuitively clear that necessity has some moral signilicance.
what about natural7 Quite olten a carnist argues that eating meat is natural lor
humans, and therelore it is allowed. lut there is some danger in applying the
criterion ol naturalness. ls rape natural7 Alter all: by natural processes (evolution)
men have developed an body part that makes them capable ol raping women,
quite a lot ol male animals (e.g. cats) rape lemale animals in nature, also humans
rape each other, our ancestors most likely raped each other lor thousands ol years,
and it is even very likely that we owe our very existence to the lact that some ol
our ancestors once raped a woman. 8o what else do we need to say that something
is natural7 8imply put: rape is natural (and normal in the animal world), but not
necessary lor men. 1herelore it is an unnecessary violation ol rights, and hence
immoral, even il rape was as natural as one could think ol.
One could say that the patients in the hospital got ill by natural processes. lut
this does not make organ transplantations natural. 1ransplantation was a
conscious invention. lt did not evolve by a blind process ol mutation and natural
selection, it is delinitely not instinctive behavior.
1he ancestors ol carnivores on the other hand did not consciously think about a
problem ol nutrient shortage, intentionally look lor a solution, experiment a bit,
discover meat as a solution, whereupon they adapted and suddenly became
dependent on meat. lredation is therelore natural because it originated by a
process ol evolution (mutation and natural selection).
lor carnivores and omnivores (like humans), we could say that meat
consumption is natural, because we have developed - by blind evolution - a
digestive system that makes it possible to eat meat. lating meat also olten
happens instinctively, which is not the case lor transplantations. lut lor humans,
meat consumption is, just like rape, not necessary (AuA, 2009). As it is an
unnecessary violation ol rights, it should not be allowed, even il it was as natural
as one could think ol. we should eat vegan instead. And linally, note that intensive
livestock larming is a lar cry lrom naturalness. 8o animal products lrom larming
are neither necessary nor natural.
1nc prcJaticn prcu|cm
229
And what about normality7 ll rape happens a lot, that does not yet justily rape.
ll killing happens a lot in wartime, it does not justily murder.
All three criteria considered separately are not valid as moral arguments. Also
combinations ol two ol the three criteria are not sullicient. ll rape is natural and
happens quite a lot, it is still immoral because it is not necessary. ll organ
transplantation is necessary and happens quite a lot in some distant country, it is
still immoral because it is not natural. And il predation is natural and necessary lor
the predator, but not normal, then l guess we are tempted to stop this predation. ll
every being on earth was vegan and then suddenly by a process ol natural
selection a small group ol animals appeared who needed to kill and eat other
animals in order to survive, animal rights activists would likely intervene (il they
could). 1hey would not allow those predators to kill others.
8o let's put all three criteria together: suppose something is normal, natural anJ
necessary. what il rape was not only natural, but also normal and necessary lor
humans7 ln other words: what il all human populations would go extinct il they
were not allowed to rape7 lxtinction on a massive scale. 1hen that's some
argument. Although l don't know what most people would say, we might tolerate
rape in that case.
l believe that predation by natural predators is allowed because it is normal,
natural and necessary, and transplantation is not allowed because it is not normal
and will never be natural. 1he !-N-conditions put together might solve the
dillerence problem. 8o let's now try to lormulate the above
normal-natural-necessary criterion in a more exact and complete !-N-principle:
1he !-N-principle. ll (a) a sulliciently large group ol sentient beings became
by (b) an evolutionary process (c) dependent on the violations ol rights ol
other sentient beings lor their survival, they are allowed to violate those
rights lor that purpose. (lut we are also allowed to protect ourselves and to
protect prey il we are inclined to do so, and we have a duty to intervene in
predation once leasible alternatives such as healthy non-animal lood lor the
predators are lound. ln practice, we should give dogs vegan lood.)
1his principle clearly relers to (a) normality, (b) naturalness and (c) necessity.
Ol course there can be luzzy boundaries between normal and not-normal, natural
and not-natural, necessary and not-necessary. 8ome things can be very normal,
natural and necessary, others less so. 1he idea is that these luzzy boundaries
create a gradation and that this gradation can be coupled with the gradation ol
rights violations: some actions are strong violations, others less so. An example ol
such a coupling ol gradations was presented in ligure 9, section 6.4).
tnica| ccnsistcncy anJ Anima| ua|ity
2!0
10. 1he value ol biodiversity
10..1 coupling the !-N-principle to biodiversity
1he above !-N-principle makes a distinction between predation and
transplantation, and this is in line with the moral intuitions. lut what is the moral
relevance ol those three criteria7 consider the no-harm principle. 1his principle
corresponds with our moral intuitions, but there is more: there exists a natural
property called well-being. ll we give intrinsic value to this natural property, then
the intuitions behind the no-harm principle are coherent with this value ol well-
being.
could we do the same lor the !-N-principle7 uoes there exist a natural property
that we can value and that is coherent with the !-N-principle7 1he answer is
allirmative: l suggest that the !-N- principle is connected to the moral relevance ol
biodiversity.
lelore we discuss the moral value ol biodiversity, it is important to give a
delinition ol biodiversity that will be suitable lor our purposes. liodiversity
consists ol all variation in lile lorms, entities and processes that are the direct
result ol natural evolution, where natural evolution is generated by random
genetic mutations. 1his delinition ol biodiversity corresponds more or less with
the common notion, as it includes genetic variation, species variation and
ecosystem variation. lut the notion ol biodiversity in the scientilic literature does
not always make it clear whether or not genetic modilication counts as
biodiversity enhancement. my delinition explicitly excludes intentional,
intelligent interventions such as genetic modilication. Our intelligence is a direct
result ol evolution, but the product ol this intelligence, i.e. the intentional creation
ol new genes, lile lorms or ecosystems, does not contribute to biodiversity,
because biodiversity was delined in terms ol only the Jircct results ol a blind
process ol evolution, excluding indirect results. 1he importance ol this exclusion
ol genetic modilication will be discussed in a later section.
8o, let us deline biodiversity as everything that directly originated lrom
evolution. ll now we would suppose that biodiversity is morally very important,
then naturalness is relevant.
6
Or in other words: il a process (behavior or property)


6
1his is similar to ooodin's green theory ol value" (ooodin, 1992). ooodin claimed that things created
by natural processes possess a higher value than things created by artilicial processes. 1he intrinsic
value ol biodiversity plays a central role in many holistic environmental ethics (lenson, 2001, kolston,
1988).
1nc prcJaticn prcu|cm
2!1
is natural, it contributes to biodiversity by delinition. ll a process is natural and
normal, it contributes a lot to biodiversity. And il a process is natural, normal and
necessary, biodiversity would drastically decrease il that process no longer exists.
A drastic decrease ol biodiversity can be considered worse than violations ol
rights, and this makes the connection between the !-N-principle and the value ol
biodiversity. 1he connection is only valid il all three N-criteria are present.
Applied to predation: the existence ol all predators in the world strongly
contributes to biodiversity, and this biodiversity lrom predation has a moral value
that strongly trumps the basic rights ol individuals (e.g. the right not to be used as
merely means), and also trumps utilitarian calculations ol well-being.
1here is a subtle issue relating the three N criteria to biodiversity. uo acts (such
as the behavior ol predation) itsell contribute to biodiversity7 Or is it rather the
existence ol predators that contributes to biodiversity (even il the predators
manage to survive on vegan lood and stop predation)7 ln other words: is
biodiversity the variation in dillerent entities (lile lorms, genes,.) or does it
include variation in processes or behaviors as well7 ll processes are not included,
the !-N-principle and biodiversity principle are slightly divergent, because the !-
N-principle relers to processes (types ol behavior).
l tend to think il processes contribute to biodiversity, they only do so slightly.
As a result, a world where lions live but survive on vegan lood is better than a
world where they live and hunt prey. 1he sullering ol the prey is worse than the
loss ol process biodiversity (the loss ol the behavior ol predation).
Another potential dillerence between the !-N-principle and the value ol
biodiversity lies in the answer to the question: should predators stop eating once
they have procreated7 As soon as predators have procreated and have viable
ollspring, the predator population can survive. One could say that biodiversity
does not decrease when all predator parents die. lence, when a predator has
viable ollspring, meat is still necessary lor the live ol this individual, but not lor
the conservation ol biodiversity. One the other hand, one could say that the
existence ol predator parents does contribute to biodiversity, as il the group ol
predator parents lorm a separate population. 1he number ol predators will
drastically decrease when all predator parents die. And this decrease can be
counted as a drastic decrease ol biodiversity.
1hese kinds ol considerations, the values ol process biodiversity and parent
biodiversity, are a matter ol intuitive estimations. moral agents can come to a
democratic agreement on how much value we should give to process and parent
biodiversity and to what degree those values trump the value ol well-being.
1here is one more assumption required to make the above connection between
the !-N-principle and biodiversity tight: duties and moral principles should be
universalized (compare with the universal law lormulation ol the categorical
tnica| ccnsistcncy anJ Anima| ua|ity
2!2
imperative ol lant). we do not have a duty to protect the prey, even il we could in
a particular situation, because il we did say we have a duty in this situation, then
we should want this rule to be universalized to all predators. 1hen we should want
predation to be prohibited at all times, anywhere.

lut predation is necessary lor


predators, so this universalization implies that we should want all predators to go
extinct. lredation is also normal, so a lot ol predators would go extinct alter
universal prohibition. And that would be a tremendous loss ol biodiversity. 1his
excludes (excuses) us lrom an intervention duty.
10..2 lntrinsic or instrumental value ol biodiversity7
why is biodiversity important7 what kind ol value does it have7 Ol course
biodiversity could have instrumental value in the sense that it could contribute to
the well-being ol sentient beings (in terms ol ecosystem services). lut il a world
without predators and with a lower biodiversity would have a higher overall well-
being, an instrumental value ol biodiversity is not sullicient to allow predation.
ll our intuitions about the permissibility ol predation track instrumental value,
they become strongly dependent on empirical lacts and scientilic discoveries.
8uppose scientists lind new ways to remove some predators without causing
ecological instability or other harmlul side ellects, hence increasing the
aggregated well-being but lowering biodiversity. ll in this hypothetical situation
we would still have the moral intuition that those predators are allowed to hunt, it
means that a mere instrumental value ol biodiversity is not sullicient in the
justilication ol predation. ln that case, the biodiversity solution to the predation
problem only works il biodiversity has intrinsic value or non-empirical value.
Non-empirical value means that the value does not depend on empirical lacts.
lnstrumental values are always empirical values: il something has instrumental
value, its value depends on contingent, empirical lacts ol how uselul it is lor
something else.

Lniversalization does not mean that only the actually living moral beings have the duty. 1he current
number ol moral beings might be too low and their skills too limited to stop all predation everywhere.
lence, intervention by the current moral agents might not endanger biodiversity. lowever, a duty
should not depend on an arbitrary state ol the world (e.g. the current state with the current number ol
moral agents). A duty should also be universalized in all hypothetical situations, including situations
where there are enough moral agents with enough skills to prevent predation everywhere.
1nc prcJaticn prcu|cm
2!!
10..! low valuable is biodiversity7
1o solve the predation problem, the value ol biodiversity should trump the value
ol aggregated well-being as well as the basic right. 1he problem is: il biodiversity
and aggregated well-being can be measured at all, they will be measured in
dillerent, incomparable units. 1his does not pose a problem il biodiversity would
always trump well-being, i.e. il any amount ol biodiversity would always be more
valuable than any amount ol well-being. lut this seems counter-intuitive: saving
the presence ol one single gene does not justily a huge loss ol well-being.
lerhaps the intrinsic values ol well-being and biodiversity are objective
properties and can be compared in the same way as we can compare two dillerent
physical lorces. lut what il you are not such a moral realist who believes in
objective moral lacts, but you still want to respect the intuition that predation is
permissible7 ln that case, you should recognize that only moral agents are the
sources ol intrinsic values. 8o we, as moral agents, give intrinsic value to an
amount ol biodiversity and an amount ol aggregated well-being. And then it is up
to us to decide which one ol those values is the strongest. lt will be an intuitive
judgment, and moral agents can democratically come to some mutual agreement
on the strengths ol those values.
1his intuitive balancing ol values may seem ad hoc, but note that a wellare
ethic is already vulnerable to the same problem (see appendix 2). lirst, well-being
is not objectively interpersonally comparable. my qualia ol happiness are
measured in dillerent units as your qualia ol happiness, similar to the way that
seeing red may be dillerent lor dillerent persons. And second, even il we can
measure everyone's well-being in the same unit, there is no objective way to
aggregate well-being: should we take the sum, the average or a weighted average
ol everyone's well-being7 low do we balance total well-being against a lair
distribution ol well-being (i.e. balance elliciency against equality)7 8hould we be
utilitarian, egalitarian or prioritarian7
1he most reasonable thing to do in a consequentialist wellare ethic, is to use
our moral intuitions to make judgments and compare and aggregate everyone's
well-being. lt seems that intuitive balancing is unavoidable, also in wellare ethics,
but the lack ol objectivity should not undermine the whole idea ol wellare ethics.
consequentialist wellare ethicisis might still come to a democratic agreement on
how to measure, value and balance everyone's well-being. 1he same goes lor the
inclusion ol new intrinsic values, such as biodiversity.
tnica| ccnsistcncy anJ Anima| ua|ity
2!4
10..4 An analogy between biodiversity and well-being
ln the previous sections, l argued lor an intrinsic value ol biodiversity, based on
moral intuitions about natural behaviors. 1his results in a rather narrow rellective
equilibrium (uaniels, 199), where the principle ol biodiversity is coherent with
moral intuitions. lut we can move to a wider rellective equilibrium by including
some background theories. ln itsell, a background theory cannot justily a moral
principle, but it can count as some supporting evidence. 1he background theory
presented in this section is an analogy between biodiversity and well-being.
8o in what sense can biodiversity (everything that is the direct product ol
evolutionary processes) be understood to have intrinsic value7 1he intrinsic value
ol biodiversity could be compared with the intrinsic value ol well-being. A person
or sentient being has two moral values: heJshe is irreplaceable and heJshe has a
well-being which has intrinsic value. we could say that ecosystems, too, are
irreplaceable and have a biodiversity that has intrinsic value.
8
let's explore this
analogy between well-being and biodiversity a bit lurther. 1hese analogies do not
justily the intrinsic value ol biodiversity, but they can be considered as supporting
evidence to make the case lor the value ol biodiversity a bit more coherent and a
bit less lar-letched.
1) looking at sentient beings, we see that they tend to increase their well-being.
1hat is because these beings have multiple needs, and they are looking lor
strategies to satisly their needs as much as possible (trade-olls, resource scarcity
and incompatible strategies limit their growth ol well-being, though). Now,
looking at ecosystems, we see that they tend to increase their biodiversity. 1hat is
because these ecosystems consist ol procreating living beings, and they are subject
to genetic variation (natural selection by resource scarcity limits the growth ol
biodiversity, though).
2) loth well-being and biodiversity are a collection ol dillerent things: pleasure,
lriendship, absence ol pain and reading a good book all contribute to well-being,
just like genes, biotic landscapes, ecological processes, species and genera all
contribute to biodiversity. loth well-being and biodiversity are natural properties
that are dillicult to calculate and express in one number, but we are able to see
increases and decreases.


8
Note that - in contrast with persons - ecosystems do not have clear boundaries, so it might be
problematic to speak ol the irreplaceability ol ecosystems. Only the whole larth has a clear boundary
as an ecosystem. lurthermore, as discussed in the appendix 2 ueriving the wellare lunction behind
the veil ol ignorance", the problem ol irreplaceability ol sentient beings can be solved without a need
to introduce an 'irreplaceability value'.
1nc prcJaticn prcu|cm
2!
Note that the intrinsic value ol biodiversity should be distinguished lrom a
problematic intrinsic value ol species. As species are abstract biological categories,
the notion ol species value laces serious problems. lirst, the delinition ol a species
is complex and the moral relevance ol a delinition based on lertility ol ollspring is
not clear. 8econd, looking at ring species and hybrids, species can have luzzy
boundaries. oiving intrinsic value to species might be as bizarre as giving value to
cheeks: hitting your lelt cheek lowers your well-being, just as killing a species
lowers biodiversity, but that does not mean that cheeks have intrinsic value.
!) lt is good to increase well-being. ls it possible to increase biodiversity in
nature by introducing new species through e.g. genetic engineering7 1he answer
is no: genetic engineering contributes to the variation ol lile lorms (as long as it
does not result in increased competition and extinction ol species), but it is not a
Jircct
-
result ol evolution. oenetically modilied species are consciously created by
intelligent beings. 1hey are not the product ol a blind process ol genetic mutation
and natural selection. Only the variation that is the direct consequence ol natural
evolution counts as biodiversity.
we can compare this with a problem in wellare ethics: the 'experience machine'
(Nozick, 194). lmagine a virtual reality machine that gives a lot ol pleasure when
your brains are plugged into the machine. lven il pleasure experiences (all
positive leelings) increase, most people leel reluctant to plug into the machine,
because they might have a need lor authenticity (being in the real world instead ol
a virtual reality), or they might have a need to actually do something (getting
pleasure through activity instead ol through merely experiencing things).
}ust as biodiversity is composed ol the variation ol all lile lorms and processes
that are the direct result ol natural evolution, we can say that well-being is
composed ol the variation ol all positive leelings and emotions that are the result
ol prelerence (need) satislaction. ln this sense, the pleasures experienced in the
experience machine do not contribute to well-being, unless the individual wants
these pleasure experiences (i.e. il the individual has prelerences lor these
experiences in the machine).
10

1he only possible strategy to increase someone's well-being, is by eliminating
obstacles that prevent prelerence satislaction (i.e. eliminating barriers that
enlorce trade-olls, or eliminating scarcities). 8imilarly, the only possible strategy
to increase biodiversity is by eliminating ecosystem pressures that increase


9
lt is an indirect result, because the intelligence ol the scientists is a result ol evolution.
10
As we have seen in section 4.2.2, this combines mental state accounts with prelerence satislaction
accounts ol well-being (see 8haw 1999, chapter 2).
tnica| ccnsistcncy anJ Anima| ua|ity
2!6
competition over scarce resources. lntroducing drugs (or an experience machine)
to increase happy leelings is similar to introducing new species (or genetic
engineering) to increase variation in lile lorms: they do nct contribute to valuable
well-being and valuable biodiversity.
All in all: well-being is lor a sentient being what biodiversity is lor a natural
ecosystem. we should not lower someone's well-being without good reason, and
we also should not lower biodiversity without good reason. Neither should we be
willing to have a much lower biodiversity. 1he lollowing table summarizes the
analogy between biodiversity and well-being.

Well-being Biodiversity
Natural property of sentient beings Natural property of ecosystems
Tendency to increase Tendency to increase
Constraints: trade-offs and (resource)
scarcity limit growth of well-being
Natural selection: competition and
(resource) scarcity limit growth of
biodiversity
Variation of (positive minus negative)
feelings and emotions
Variation of living organisms and
processes
Result of preference satisfaction Direct result of evolution
No artiIicial means` to increase well-
being using an experience machine
No artificial means to increase
biodiversity using genetic engineering

ln summary, the above discussion allows us to introduce the lollowing
principle:
1he intrinsic value ol biodiversity. we should protect biodiversity and allow
behavior that contributes to biodiversity, whereby biodiversity is delined as
all variation in lile lorms and processes (behaviors) that are the direct result
ol natural evolution (i.e. being generated by random genetic mutations).
liodiversity lor ecosystems is analogous to well-being lor sentient beings:
both are intrinsically valuable properties ol an entity (ecosystem, sentient
being) that is unique and irreplaceable.
1he above 'biodiversity principle' or 'triple-N-principle' has some resemblance
with kegan's amorality criterion. lut it is not about the amorality ol the lion, but
the amorality ol nature and evolution. 1his amoral natureJbiodiversity criterion is
a new articulation ol some shared moral intuitions and attitudes ol animal rights
1nc prcJaticn prcu|cm
2!
activists. 1hey are olten worried that interlering in nature is a kind ol human
arrogance. lerhaps this might be a bit similar to the cultural relativists who claim
that imposing our human rights on other cultures is arrogant.
11
lt is not the lact
that the lion is an amoral agent, which grants him the right to kill and eat others.
1he point is that the lion is part ol a big thing, which we will call nature or the
'Other'. 1his 'Other', however, has a completely dillerent morality than ours. lt is
'amoral', which means: beyond the morality ol our moral world. condoning
predation is a kind ol respect lor this Other. we are not responsible lor the cruelty
that evolved within the world ol the Other.
ll a lion decides to hunt a zebra, that is part ol the amoral world. lut il l decide
to use sentient beings lor transplantations, it is part ol our moral world. As
predation contributes to biodiversity whereas transplantation doesn't, this makes
a distinction between predation and transplantation. laving said all this, l do
believe that predation is a very serious moral problem, and perhaps some
(utilitarian inclined) ethicists are right that in the end we should look lor ways to
intervene and decrease the vast amounts ol sullering in the wild, including the
sullering caused by predators. we should not be alraid ol intervening, we should
not be alraid ol being too arrogant, because wild nature is really arrogantly cruel.
At least we should not underestimate the cruelty ol nature, and at least we should
openly discuss this issue ol decreasing wild animal sullering.
10.6 8ome lurther tests lor the !-N principle
let us test this biodiversity principle (or triple N principle) with some examples.
let us check whether the above hypothesis is compatible with our moral
intuitions. ll it is compatible with our intuitions, the animal rights ethic extended
with the above principle becomes more coherent.
As a lirst example, suppose someone becomes ill and needs new medicines that
have to be tested on sentient beings. 1he experimentation did not originate lrom
an amoral process ol blind evolution, it originated lrom moral agents. 1herelore,
an animal rights activist should be against experimentations on sentient beings
(without their consent).


11
1his is not an argument pro cultural relativism. many animal rights activists are not cultural
relativists.
tnica| ccnsistcncy anJ Anima| ua|ity
2!8
A second example is the killing ol insects by accident. 8uppose that we discover
that ants are sentient beings. what will happen to our ethics7 uo we have a duty
not to move, in order to save the ants7 ln other words, does the ant not only have
a right not to be used as merely a means, but also a right not to be killed by us
walking around7 lt is clear that the behavior ol walking around originated by a
blind evolutionary process, and that walking around is very vital lor large animals
like us. ll we did not walk around, we would not even be here alive today.
According to the hypothesis, we are then still allowed to walk around and
accidentally kill ants. 1his conlorms to our moral intuitions. On the other hand,
road kill, the killing ol sentient beings by cars, is morally wrong, because cars are
not natural and driving a car does not contribute to biodiversity.
what about killing an annoying lly on purpose7 ll we would discover that llies
are really sentient beings, then our hypothesis says that we should not kill the lly,
because we, as aggressors, did not have a vital need that was in danger. lven il
killing llies is natural and normal. And we should delinitely not use llies as merely
a means to our ends. 8o the lly would have a right not to be used as merely a
means, and also the right not to be killed on purpose lor non-vital needs.
Another example is procreation, in particular the birth ol animals which will
have a lower liletime well-being than ours. lmagine that all animals have a lower
lile expectancy and lower capacities lor well-being (lower emotional richness)
than humans. 1hose animals are like (mentally) disabled humans. lrom behind a
veil ol ignorance, you can choose between two worlds. ln the lirst world, animals
are born with a lower (but still positive) value ol lile. ln the second world, there
are only humans with a high value ol lile. You might preler the second world,
because in that world you have probability 1 ol being a human with a high value ol
lile. ln the lirst world you risk being born as an animal with a lower value ol lile.
1his thought experiment would imply that it is good to make all non-human
animals inlertile, so that those animals go extinct.
A similar problem arises with the birth ol animals that will have a high critical
resource consumption level (i.e. the positive, non-zero level ol resource
consumption at which well-being ol that individual equals zero). 1otal well-being
would be maximized il only the beings with lowest critical resource consumption
levels would procreate. 1he resources needed lor one individual with a high
critical resource consumption level can better be spent on more individuals who
have the lowest critical consumption levels. 1he latter individuals can generate
more wellare with the same amount ol resources (see 8hiell, 200).
1hings might even be worse lor r-selection species who have a very high
reproduction rate but a very low individual survival rate (lorta, 2010c). most ol
those animals have a very short lile and an early death. ll they develop sentience,
1nc prcJaticn prcu|cm
2!9
they have very lew opportunities lor positive experiences. lerhaps they have lives
that are not worth living.
1he conclusion that those animal species that do not contribute enough to the
aggregated well-being (i.e. the wellare lunction as described in appendix 2), are no
longer allowed to procreate, goes against our moral intuitions. ll we say that
biodiversity has moral value and that procreation is normal, natural and
necessary, those animals are still allowed to procreate. As with predation, we do
not have a duty to stop procreation. lut we are allowed to intervene in
procreation to some degree. lor example il parents know that their luture child
will be disabled (il the potential child would not contribute to the wellare
lunction), then (early) abortion would be allowed.
kelated to the above problem is the issue ol genetic enhancement. uo we have a
duty to genetically enhance species who would otherwise not contribute enough
to the wellare lunction7 8hould we genetically modily lrogs to increase their
potential levels ol liletime well-being7 Or genetically modily lions such that they
no longer need meat7 changing genes on purpose is not a natural behavior. lence,
those new genes do not contribute to biodiversity. ll we would replace all lrogs
with enhanced lrogs, some biodiversity will get lost because the genes that
characterize the unenhanced lrogs will disappear lrom the gene pool and the new
genes do not contribute to biodiversity. 1he question is whether this loss ol
biodiversity (the loss ol some specilic genes) trumps the increase ol the wellare
lunction. 1his becomes an intuitive balancing ol two competing values which can
be approached in a democratic way (see appendix 2).
A linal example is the situation where a child holding a gun is about to kill
another child. 1his child has no moral agency, but still we have the duty to
intervene and protect the second child. 1hat is because the lirst child does not
have a vital need to kill the other child. Also, killing with a gun did not evolve by
an evolutionary process, so it is not natural.
1he above examples indicate that our triple-N-principle is coherent with a lot
ol our moral intuitions. As a summary, the lollowing table gives an overview ol
solutions lor live ol the abovementioned challenges to a consequentialist wellare
ethic.
1. lredation: carnivores are allowed
12
to hunt, kill and eat many prey
animals, even il they harm sentient prey.
2. motion: a human (or another big animal) is allowed to move around and
kill (by accident) many insects, even il insects were sentient.


12
ln the sense that we (moral agents) do not have a duty to interlere in predation to save the prey.
tnica| ccnsistcncy anJ Anima| ua|ity
240
!. Organ transplantation: a surgeon is not allowed to sacrilice a victim
without inlormed consent, even il patients are dying when they do not
receive new organs.
4. medical experimentation: a researcher is not allowed to sacrilice
someone without inlormed consent, even il the developed medicines
could save many patients in the luture.
. lrocreation: all animals are allowed to procreate
1!
, even il those animal
species do not contribute enough to wellare lunction.


Intuitive
moral
judgments
Consequentialist
welfare ethics
Mere means
principle
Principle X
1. Predation Allowed Not allowed Not allowed Allowed
2. Motion Allowed Not allowed - (undecided) Allowed
3. Organ
transplantation
Not
allowed
Required Not allowed -
(undecided)
4. Medical
Experimentation
Not
allowed
Required Not allowed -
(undecided)
5. Procreation Allowed Not allowed - (undecided) Allowed

A common property ol the above problems is the ncccssity cj a scricus wc|jarc |css.
1he behavior (predation, transplantation,.) is necessary lor:
1. existing beings to stay alive (the predators, patients and big animals in
problems 1 to 4),
2. potential beings to have a lile
14
(problem ), or
!. populations to survive (problem ).
1he second column in the above table presents the intuitive moral judgments
about the allowance ol the live types ol behavior. 1his column is the opposite ol
the third column, which gives the results according to consequentialist wellare
ethics. 1o reconcile the wellare ethics with the moral intuitions, we can develop a
two-step approach.


1!
1his does not imply that they are allowed to have as many ollspring as they want, because we have to
avoid overpopulation. lor example having more than ! children would be problematic il this rule
would be universalized: the resulting exponential population growth will hit the boundaries ol the
planet.
14
lrocreation is necessary lor a potential being to get a lile, in the sense that without procreation, the
potential being could never come into existence.
1nc prcJaticn prcu|cm
241
1he lirst step (column lour) introduces the deontological mere means lrinciple
(discussed in section 6.2), which changes a lew consequentialist judgments (organ
transplantation and medical experimentation). 1his principle adds some
impermissibilities that trump the consequentialist principle, so alter this step,
none ol the live behaviors is allowed. lut that is not sullicient to match all
judgments with the second column in the table.
1herelore, in a second step, a new principle X has to be introduced (column
live). 1his principle X trumps both the mere means and the consequentialist
principles. Alter this second step, the (im)permissibility ol the live behaviors
matches the intuitions. we have seen that the !-N-principle, which corresponds
with the value ol biodiversity, serves as a good principle X, because it changes the
judgments in the three remaining problems: predation, motion and procreation.
1here are dillerent ways to measure biodiversity. 8cientists can propose dillerent
biodiversity metrics. lut each metric shows a drastic decrease when predation,
motion and procreation stops. lence, whatever biodiversity metric one chooses, it
is compatible with the moral intuitions ol a deontological naturalistic ethic that
permits those types ol behavior.
1he intrinsic value and the resulting !-N-principle lit in a wide rellective
equilibrium (uaniels, 199): they are coherent with strong moral intuitions in
three dillerent cases (predation, motion and procreation), they are coherent with
a notion ol naturalness (which a lot ol people care about), and they are coherent
with some background theories ol biodiversity (some properties ol biodiversity
make the above mentioned analogy between biodiversity and well-being sensible,
the latter having intrinsic value in consequentialist wellare ethics).
1he next section introduces a second possible principle X. Also this principle
lits in a wide rellective equilibrium: it is coherent with the same three moral
intuitions, and coherent with a notion ol lairness.
10. A second hypothetical solution to the dillerence
problem: behavioral lairness
1he above !-N-principle relers to some seemingly arbitrary criteria ol naturalness
and normality, and to some seemingly mysterious intrinsic value ol biodiversity (it
is an intrinsic value that a moral agent gives to a property ol a non-sentient
entity). lor those who leel uncomlortable with the above solution, there is a
second promising solution to the dillerence problem that avoids these relerences
tnica| ccnsistcncy anJ Anima| ua|ity
242
to nature and biodiversity. 1his second solution is based on a notion ol lairness. ll
the lion is not allowed to eat the zebra, then the lion could say that as a matter ol
lairness, the zebra is not allowed to eat either. 8o the principle claims that A is
allowed to do X with l, il l also does X and is allowed to do X. ll the zebra (l) is
allowed to eat (do X), then the lion (A) is allowed to eat as well. 8o il the zebra eats
something, and il the zebra would say that she is allowed to eat, then the lion is
allowed to eat as well, even when the zebra is the lood. 8imilarly: il a (sentient)
insect is moving around and is allowed to move around, then l am allowed to move
around as well, even il that harms the insect by accident.
1his is in contrast with organ transplantations and medical experiments. l
cannot use you as merely means lor organ transplantation or medical
experiments, because you don't perlorm medical experiments and
transplantations yoursell. 8o my claim that l am allowed to use you becomes
invalid, because you are innocent when it comes to medical experiments. 1he
zebra and the insect, on the other hand, were not innocent when it comes to
respectively eating and moving.
1he tricky point is: what is the behavior X exactly7 1he zebra could say to the
lion: You are not allowed to eat me." 1hen X could reler to 'eat this zebra'.
lairness requires that the lion could say to the zebra: 1hen you are not allowed to
eat yoursell either." lut that's line lor the zebra: she was already lollowing the
rule not to eat hersell. 8imilarly, il X meant 'eat zebra', the zebra could lile with
the rule that it is not allowed to eat zebras. 1he same goes lor X equal to 'eat
animals'. 1he idea is that the behavior X should not reler to specilic individuals or
groups ol individuals. X should only reler to a behavior, such as 'eat' or 'do medical
experiments'. ll the lion is not allowed to eat, then neither is the zebra, and the
zebra could not live with that rule.
lut then another problem lurks. 1o what kind ol behavior should X reler7 You
could say: You are not allowed to use me in experiments, because l'm innocent: l
don't use anyone in experiments." my reply could be: 8ure, but you do use plants
lor lood, so the true X means 'use someone or something'", and according to that
view, you are guilty. lut that X would be too general: it would imply that all kinds
ol uses are not allowed. 1herelore, X should be the most accurate and specilic
description ol a behavior.
1he most accurate description ol the behavior ol the lion would be: 'eat'. lut il
you are allowed to eat (e.g. eat plants), then l would be allowed to eat you. ln
particular, sometimes you do eat plants just lor taste. ll a being is allowed to eat
something lor taste (instead ol survival), then so do l, and il that being happens to
taste good, l am allowed to eat her7 l am allowed to eat you il you taste good and il
you eat something lor taste7
1nc prcJaticn prcu|cm
24!
1o avoid this conclusion, we have to reler to the intention or the purpose ol the
behavior. ln particular, we can keep the necessity criterion ol the above !-N-
principle. 1herelore, X should be more specilic, such as 'eat lor survival'. l can
survive without eating you, so eating you would not be lor survival.
lt is not always clear to see whether someone eats lor survival or pleasure, but il
someone eats a sentient being, causing harm to the sentient being, and il there are
healthy alternatives available, the intention to eat lor survival will be llawed.
8o lar, we have a promising lairness principle which roughly sounds as: ll you
are allowed to do a specilic behavior lor survival, then so am l allowed to do the
same type ol behavior lor survival." lut this principle still needs some lurther
relinements and clarilications when we look at more hypothetical cases.
lirst: what il plants were sentient beings7 llants don't kill and eat other living
beings. uoes this mean that plants are innocent and that herbivores (including we)
should not eat them7 Ol course, plants do consume chemical resources and energy.
8o we have to state that consuming chemical resources and energy" is the same
type ol behavior as eating".
8econd, what il a sentient being consciously decided: ln consuming chemical
resources, l will only consume non-sentient beings."7 ls it lair to eat this sentient
being7 1his sentient being is in a sense lucky that she can survive on consuming
non-sentient beings. A lion, on the other hand, is not that lucky: he needs sentient
beings. 8o the lion is allowed to eat this sentient being. lowever, il the lion was
lucky in the same way (il he could survive by eating non-sentient beings), then he
should restrict his consumption to non-sentient beings as well.
A third objection is based on the common-sense judgment that doing
experiments on (non-sentient) plants is allowed. lut now a mad scientist could
say: ll you are allowed to do experiments, on plants, l am allowed to do
experiments as well, on you!" 1o avoid this, we make a distinction between moral
and amoral beings. 1he lairness claim cannot be put lorward against the victim,
when the victim is a moral being. Note that amoral humans (babies and mentally
handicapped humans) and non-human animals do not perlorm experiments on
plants, so they should not be used by the mad scientist either.
ln summary, we get the lollowing principle.
1he principle ol behavioral lairness. An agent is allowed to do a specilic type
ol behavior that causes harm to victims
1
, il (1) the behavior is necessary, (2)


1
lere, causing harm should be understood in a broad sense. l.g. lowering the total wellare lunction is
a cause ol non-personal harm where the victims can be considered as the total population. Also using
someone as merely a means is a cause ol harm.
tnica| ccnsistcncy anJ Anima| ua|ity
244
the harm is minimal (the agent does not have an alternative that causes less
harm), (!) the victims are amoral agents, (4) the victims perlorm the same
type ol behavior and () the victims are allowed to do that behavior. ll the
agent has a better option (that causes less harm), then sJhe should choose
that option.
1his lairness principle means that il A is allowed to do something, then so is l,
under certain conditions. 1he reader can verily that this lairness principle is able
to withstand the tests mentioned in the previous section: impermissibly using
sentient beings against their will in experiments, permissibly getting ollspring
who insulliciently contribute to the wellare lunction
16
, permissibly killing insects
by accident when making a movement, impermissibly killing an annoying lly on
purpose, and permissibly eating plants even il plants are sentient.

10.8 8ummary
we have seen three principles, the lairness principle, the triple-N-principle (or
biodiversity principle) and the tolerated choice equality that solve the prey
problem and the dillerence problem, the two components ol the predation
problem. 1hose principles can help us derive a consistent and coherent animal
rights ethics that can be reconciled with moral intuitions shared by a lot ol people.
kegarding the !-N-principle, l lirst clarilied the meaning ol normal, natural and
necessity. 8econd, l demonstrated that none ol the three criteria separately are
sullicient. 1hird, l showed that the combination ol all three criteria can make a
distinction between predation and transplantation. Next, l made a connection
between the three N-criteria (as we have delined them) and biodiversity. lilth, l
explained in what sense biodiversity can be said to have moral value. And linally l
tested the !-N-principle in other situations, indicating that it corresponds with our
moral intuitions. 1he result is a principle that says that basic right violations are
only allowed when all three N-criteria are met.
1he three N justilications (normality, naturalness and necessity) that animal
rights activists seem to hold are specilic interpretations ol the same three


16
lrocreation is not a necessary need lor an existing individual: no-one will die without procreation.
lut it can be considered necessary lor potential luture beings.
1nc prcJaticn prcu|cm
24
justilications in the ideology ol carnism (}oy, 2009) and might give us a lurther
clues about the question why it is so dillicult to convince people to become
vegetarians or vegans. lecoming vegan should be easy in principle, because eating
animal products is not necessary lor survival. Yet, carnist people appear to have
strong emotional objections, and they use a lot ol naturalistic lallacies to justily
their behavior. lor example, in the L8 there are more people in lavor ol hunting
lor pleasure, than there are in lavor ol medical experiments with animals (lerzog,
2010). And in contrast to hunting, medical experiments are believed to save lives.
lt becomes clear that there are some hidden sensitivities that people including
animal activists might have. we mentioned the lear lor human arrogance and
the apparent similarities with cultural relativism and respect lor the Other. lut a
beliel in naturalness might be an important moral intuition ol a lot ol people.
1he !-N-principle can be justilied with a relerence to an intrinsic value ol
biodiversity. lowever, those who dislike such a mysterious intrinsic value could
rather adopt another principle to solve the predation problem: the behavioral
lairness principle. 1his principle says that a specilic behavior that harms a victim
(e.g. killing and eating a living being) is allowed only il the action is necessary lor
survival, il the victim (e.g. the prey) is an amoral being, il the victim is also guilty
ol a similar type ol action and il sJhe is allowed to do that action. ll the zebra eats
lor survival and is allowed to do so, then so also is the lion. lut il a harm (a loss ol
well-being) occurs to a sentient victim, and il there are healthy alternatives that
do not involve such harm, then ol course those alternatives should be chosen.
linally, we can combine the !-N-principle and the principle ol behavioral
lairness into a new equality principle: everyone has an equal right to a behavior
that is both natural, normal and necessary (i.e. a behavior that strongly
contributes to biodiversity).

tnica| ccnsistcncy anJ Anima| ua|ity




24
chapter 11 1he property problem and the
harvest problem
As animal products are not necessary lor us, we (humans) should become vegans.
lut veganism is not good enough. 1here is another issue: what about property
rights lor animals7 ln particular: are we allowed chasing away animals to clear a
lorest in order to built a house or extend a crop lield7 what about the birds and
the squirrels who built nests in those trees7 we destroy their houses and habitats
by cutting down the trees. lrom an antispeciesist point ol view, this is similar to
the destruction ol someone's property or stealing someone's land.
And what il we use a crop lield (to produce vegan lood)7 1he crop lield is our
property, but some other animals invade the lield and start eating our grains,
lruits and vegetables. Are we allowed to chase them away, even il this results in
more sullering by those animals due to an increased competition between those
animals lor scarce lood resources7 Are we allowed to kill them il they keep
returning to our crop lields7
And third, what il we kill those small animals by accident when we use
machines to harvest our crops or drive with big trucks7 8ome animals (small
rodents, birds,.) die, even when vegan lood is produced. At
www.animalvisuals.org, an estimate is given ol the number ol animals killed due to
harvesting. lor a vegan it comes down to roughly 2 animals killed per year. 1his is
still much less than the number ol animals killed on purpose and by accident in
the production ol animal products (meat requires slaughtering an animal but also
harvesting crops lor livestck leed), but it is not something we can dismiss so easily.
tnica| ccnsistcncy anJ Anima| ua|ity
248
11.1 labitat destruction
let us consider the lirst problem: the destruction ol wild animal habitats. lndeed,
we are not allowed to invade a loreign country, destroy the houses and chase the
local people away. lsn't it speciesist to allow the destruction ol animal habitats,
destroying the nests ol the birds and squirrels7 Or do animals have habitat rights7
A consistent ethical theory ol animal equality indeed implies that animals have
habitat rights similar to our property rights. lut this does not mean that we are
not allowed to use natural resources even when using those resources requires
invading someone's habitat. we are allowed to use habitat ol animals to some
degree, because il we are not allowed to use natural resources, then neither is no-
one allowed. ll an animal uses natural resources and a habitat, it means that
another animal is no longer able to use those resources. lut animals are allowed to
use natural resources, and hence so are humans. As habitat and natural resources
are scarce, there is always competition between dillerent sentient beings.
1o demonstrate that using an animal's habitat is not necessarily speciesist,
imagine that a mentally disabled human escaped lrom a care institution. 1his
human is as intelligent as a bird, and he decides to climb into a tree and build a
nest. As a tree is a bird's habitat, this tree becomes the habitat ol the disabled
human. Lnlortunately, we want to cut down that tree, to produce paper lor
important books, or to extend our cropland, or to build a house lor ourselves.
chasing away the disabled human can be considered as a harm to that human. 1he
mentally disabled human might get injured, or he might run to another tree that is
already occupied by another sentient being (say another escaped mentally
disabled human), increasing their competition lor scarce trees. 1his harm done to
the mentally disabled human is not a use as merely a means. chasing away the
human does not violate his basic right.
we are allowed to cause harm to someone, as long as the victim is not used as
merely a means, and as long as the quasi-maximin prioritarian principle ol justice
is not violated. ln practice, this latter condition means that 1) we should be very
carelul in cutting down trees and harvesting crops, 2) we should strongly decrease
our consumption ol natural resources in order not to invade too much in someone
else's habitat, !) we should stop the lurther destruction ol wild animal habitat
(increase the area ol nature reserves and stop the expansion ol human
settlements) and 4) we have a strong responsibility to help potential victims
(taking care ol wild animals by strongly increasing our support lor wildlile rescue
centers, giving lood aid to animals in need, protecting wild ecosystems).
uonaldson and lymlicka (2011) developed a political theory ol animal rights,
whereby the habitat ol wild animals should be considered as sovereign nations or
1nc prcpcrty prcu|cm
249
sovereign animal territory. we do not have the right to colonize and displace the
citizens (wild animals) ol these spaces. ll we move into animal territory, we should
compensate any harm done to wild animals, by helping wild animals in need. kisks
to wild animals can be compensated with benelits lor them. 1herelore, merely
veganism (abstaining lrom consuming animal products) is not good enough. A
vegan does not violate the basic right not to be used as merely a means, but sJhe
might still violate the quasi-maximin principle il sJhe is not carelul. lt can be
compared with invading another country: even il you are not killing humans lor
lood (you are anticannibalistic), you do cause harm when you steal someone's
land.
ll we are allowed to carelully chase away wildlile animals to some degree, and il
we want to avoid speciesism, we are also allowed to chase away those escaped tree
sitting mentally disabled humans. ln a sense, this means that mentally disabled
humans and wildlile animals have weaker property rights than rational beings like
you and me (i.e. mentally capable humans). lt can be argued that this dillerence in
the strength ol property rights can be derived lrom the quasi-maximin principle,
il we keep in mind that dillerences in mental capacities generate dillerences in
how property rights inlluence well-being. ln other words: some sentient beings
have special mental capacities, which means that their well-being is more strongly
inlluenced by how properties are distributed. 1his happens especially when a
sentient being has the capacity to understand the notion ol a property, to invest in
hisJher property (making himJher leel more emotionally attached to hisJher
property) and is able to cooperate with others in the search lor a lair distribution
ol property. 1hose people (e.g. rational beings) might have stronger property
rights compared to other sentient beings, because their well-being is more
strongly dependent on the distribution and treatment ol properties.
1herelore, there is a dillerence between destroying the house ol a rational
being and destroying the nest ol a bird by accident in the case ol cutting down a
tree. lirst, a more rational being might have a stronger emotional connection and
understanding ol property, and second, a house is dillerent lrom a nest in terms ol
ellort to construct it and in terms ol replaceability. 1herelore, it may be worse to
violate (mentally healthy) human property rights than (non-human) animal
habitat rights.
1his stronger property right lor rational beings can also justily why rational
beings are more strongly permitted to delend their own properties (e.g. delend
their crop lields against invaders). Yet, we should look lor animal lriendly, non-
lethal methods to avoid animals eating our crops, just as we have to look lor
human lriendly solutions when a group ol humans invade our cropland to steal
our lood. lilling them or poisoning them can only be a very last resort. And some
solidarity with animals might also be required to compensate lor some harm done
tnica| ccnsistcncy anJ Anima| ua|ity
20
to wild animals: we should be willing to produce lood to help hungry (wild)
animals. Again, this means we should more strongly support wildlile rescue
centers.
ln this context, we can also reler to a principle ol tolerated partiality. we might
leel more concern to respect the property rights ol mentally disabled humans
compared to the property rights ol birds, but we should tolerate someone who
takes more ellort to respect bird habitat above the property ol a mentally disabled
human.
11.2 Animals killed in harvest
lmagine there are two animals (e.g. small rodents) that are killed in agriculture to
produce vegan lood lor one human lor one year. Are we allowed to larm some land
il we know that per vegan person every year on average two vertebrate animals
are killed7 Or is vegan agriculture a violation ol the quasi-maximin prioritarian
principle7
1o shed some light on this issue, remember lirst ol all that in section 4.2 we saw
two lactors that inlluence someone's liletime well-being: 1) the richness ol
emotions (some sentient beings have more and stronger prelerences and can
experience higher levels ol momentaneous well-being il those prelerences are
satislied) and 2) psychological connectedness with someone's past and luture
(some sentient beings have richer and stronger autobiographical selves as well as
stronger prelerence towards the luture).
lmagine we have lour beings: a human person (with high mental capacities and
a strong psychological connectedness), a big animal (e.g. a cow) and two small
animals (e.g. rodents). 1here are three situations: 1) the human produces vegan
lood, at the cost ol endangering the two rodents (those two rodents might get
killed il they happen to be in the crop lield at the wrong time), 2) the human does
not produce vegan lood nor kills the cow, which means the human cannot eat and
the rodents and cow survive, and !) the human eats the cow, the rodents are not in
danger because the cropland is not harvested. which ol these situations is the best
lrom the point ol view ol prioritarian justice7
As the rodents likely have lower potential levels ol momentaneous well-being
and also have a weaker psychological connectedness with their past and luture
selves, they likely have lower levels ol integrated liletime well-being. looking at
the wellare lunction equation in appendix 2 lntermezzo: a more complex
lormulation to solve the replaceability problem", it is lar lrom obvious that
1nc prcpcrty prcu|cm
21
situation 2 is better than situation 1. Alter all, the liletime well-being ol the vegan
is strongly reduced when sJhe is no longer allowed to larm some land, and the
early death ol the rodents does not strongly decrease their integrated liletime
well-being il those rodents have less psychological connectedness.
what about situation !7 1his situation is delinitely the worst with respect to the
basic right principle (the basic right ol the cow is violated in situation !), but is it
also the worst with respect to liletime well-being7 1he wellare lunction does not
give a clear answer, except that one might think that the death ol one animal (the
cow in situation !) is better than the death ol two animals (the rodents in situation
1).
1he principle ol prioritarian justice says that we should not cause more harm in
such a way that the wellare lunction decreases. lut it is lar lrom obvious which ol
the above three choices is the best lrom the point ol view ol prioritarian justice. ln
that case, we can do two things.
lirst, we can reler to the principle ol tolerated partiality. 1hat principle was
lirst derived in situations where we help others, and the choice is between helping
A versus helping l. Now we are lacing a situation where we are harming others,
and the choice is between harming A versus harming l. 1he principle ol
prioritarianism says that we should minimize harm, where harm is now delined as
a decrease ol the wellare lunction. lut il it becomes dillicult to calculate whether
the wellare lunction decreases, we can say that we are allowed to be partial to
some degree, to preler the choice that is in our own best interest, i.e. situation 1.
8econd, we can introduce a heuristic rule ol thumb: harming an identiliable
victim is worse than harming a non-identiliable victim. 1he animals that die by
accident in agriculture (e.g. the two rodents) are non-identiliable victims. ll l eat a
vegan product, l cannot identily the rodents that died due to harvest. lt might be
possible that no animal was harmed when a vegan meal was produced. On the
other hand, il l consume an animal product (e.g. the meat ol the cow), l do know
that this product comes lrom an identiliable victim. A piece ol meat, an egg or a
drop ol milk comes lrom someone's body, so at least someone is harmed, and we
know who. ln that sense, situation ! is worse than situations 1, because the cow is
an identiliable victim that is killed whereas in situation 1 there is no identiliable
harm (it is not clear that the two rodents are actually killed). Also in situation 2,
there is an identiliable harm to the vegan human who has nothing to eat.
1herelore, situations 2 and ! are worse than situation 1, according to this heuristic
rule ol identiliable harm, because situations 2 and ! involve identiliable victims.
lurthermore, we can add that the harm done to non-identiliable victims can be
compensated by helping animals who need help. ll my behavior results in the
death ol one non-identiliable animal (i.e. l don't know who dies, l only know that
someone dies), l can compensate this harm by saving the lile ol an identiliable
tnica| ccnsistcncy anJ Anima| ua|ity
22
animal (i.e. a specilic animal who needs help). lven il harming identiliable victims
is not worse than harming a non-identiliable victims, there is still a dillerence: the
second harm can be compensated by helping others who we did not harm,
whereas the lirst harm can only be compensated by helping the harmed
identiliable victims themselves. As using resources (e.g. doing agriculture) and
emitting waste harms non-identiliable animals, we can compensate lor this harm
by sulliciently helping other animals.
ln summary: vegan agriculture might be permissible, even il non-human
animals die by harvesting the crops. what it means is that we should lirst ol all be
much more carelul in agriculture, mining and lorestry, trying to avoid harming
and killing animals (e.g. permaculture and zero tilling agriculture). 8econd, we
should lower our ecological lootprint and lower our use ol natural resources such
as cropland. And third, we should compensate harm done to wild animals by
strongly supporting wildlile rescue centers that help all kinds ol wild animals
(including small rodents and birds). merely eating vegan lood is not good enough
according to a consistent ethic ol animal equality.


2!
chapter 12 1he core argument lor veganism

most philosophers and ethicists still consume animal products. why do they not
come to the conclusion that veganism is a moral duty7 1his section presents an
argument lor veganism, using a lormal-axiomatic approach: all axioms (starting
points such as basic delinitions, moral assumptions and empirical lacts), as well as
the logical steps lrom those axioms to the conclusions, will be stated as explicitly
as possible. 1his axiomatic approach has three advantages.
lirst, it allows us to directly study the question: il philosophers and ethicists
want to continue eating animal products, which ol the assumptions are they not
accepting7 lveryone who wants to justily the consumption ol animal products
should be able to indicate at least one ol the axioms that sJhe rejects.
8econd, the approach sets a new lramework lor a review ol the literature on
animal rights, speciesism and vegetarianismJveganism. lthicists who delend
speciesism or meat consumption in the literature can be associated with specilic
axioms that they reject.
And linally, the axiomatic approach allows to lormulate the least restrictive
assumptions and delinitions. 1he presented argument will be as parsimonious as
possible, using minimal assumptions and delinitions necessary to reach the
conclusion. 1he scope ol the delinitions and moral assumptions will be as narrow
as possible, the empirical lacts will be as reliable as possible and the criteria or
conditions in the delinitions and moral assumptions will be as strong as possible,
making it more dillicult to reject or disbelieve these delinitions, lacts and
assumptions. ln this sense, it is a minimalist or core argument lor veganism.
1he argument uses roughly twenty delinitions, moral assumptions and
empirical lacts. Ol course shorter arguments lor veganism are possible. lor
example: we should not cause unnecessary harm, consumption ol animal
products causes unnecessary harm, therelore we should abstain lrom animal
products". Or: kespecting the golden rule (do unto others.) is a moral duty,
tnica| ccnsistcncy anJ Anima| ua|ity
24
omnivorism violates the golden rule, therelore veganism is a moral duty". lut
those arguments are too short: much more can and need to be said.
let us start with a weak lormulation ol a basic right.

ucjiniticn !: 1nc uasic riqnt is tnc riqnt nct tc uc intcnticna||y uscJ as mcrc|y a ucJi|y
mcans jcr tnc ncn-vita| cnJs cr nccJs cj ctncrs. A victim is uscJ as mcrc|y a ucJi|y mcans
wncn
!) tnc ucJy cj tnc victim is ncccssary tc acnicvc tnc cnJs cj tnc ctncrs (i.c. tnc cnJs
ccu|J nct uc rcacncJ wncn tnc ucJy is auscnt),
.) tnc victim Jccs nct want tnc trcatmcnt cj tnc cwn ucJy in tnat way (i.c. tnc victim
nas tc Jc cr unJcrqc scmctninq aqainst its wi||), anJ
1) tnc |css cj wc||-ucinq cj tnc victim wncn trcatcJ in tnat way is mucn niqncr tnan tnc
|css cj wc||-ucinq cj cacn cj tnc ctncr inJiviJua|s wncn tncir cnJs arc nct rcacncJ.
1his is a weaker lormulation than the basic right used by e.g. kegan (198!) and
lrancione (2000). lirst, it is restricted to non-vital ends. 1his means that the below
argument is not applicable to e.g. predators or indigenous people who need meat
in order to survive. lence, we are avoiding the predation problem (lbert &
machan, 2012, link, 200). No matter what we think about the predation problem,
the below argument should be immune to our opinions in survival cases.
8econd, it is a weak delinition because it is restricted to the use ol a body that
needs to be present in order to reach the ends. lence, the below argument is also
not applicable to e.g. the insect problem (are we allowed to kill insects by accident
il insects happened to be sentient beings7) and the harvest problem (what about
animals who die by accident in harvest7). lence, it avoids the discussion about the
least harm principle and vegetarianism in e.g. uavis (200!), lamey (200) and
matheny (200!). 1he insects and mice that are killed by accident are not used as
merely a bodily means, so their death does not constitute a violation ol the basic
right. ll this basic right is stronger than a right to lile (in particular the right not to
be killed), then we should still be vegan even when more animals are killed by
accident in vegan lood production than in livestock larming.
8imilarly, the conditions in the above delinition avoid another possible
objection to veganism: the 'logic ol the larder' (see e.g. matheny & chan, 200).
According to this argument, livestock larming might be permissible because
without livestock larming those animals would not even exist, and it might
perhaps be better to exist and live a lile on a larm, than not to be born at all. One
could similarly try to justily slavery - in particular the breeding ol slaves - on the
grounds that without such slavery those slaves would not even exist. lut those
slaves would still be used as merely a bodily means. ll the basic right trumps a
possible right to existence, slavery would still be immoral and the logic ol the
larder becomes invalid. 1he same applies to livestock larming.
ccrc arqumcnt jcr vcqanism
2
kegarding the third condition, the loss ol well-being ol the victim is an
'unnecessary sullering' when compared to the loss ol well-being ol the other
individuals. 1he loss ol luture well-being should be taken into account. when it
comes to the consumption ol animal products, the loss ol well-being ol the other
individuals (the people who consume animal products) should be a relative loss ol
well-being, i.e. a dillerence between animal products and available, tasty plant-
based alternatives. lresumably, this loss might be much smaller than what people
tend to think: people are olten unaware that there are e.g. vegan sausages that
taste almost as good as meat sausages (see e.g. Allen e.a. 2008), and well-planned
vegan diets might have health benelits that lurther reduce the loss ol well-being
(e.g. AuA, 2009).

Vcra| assumpticn !: At |cast a|| numans wnc arc scnticnt (navinq a ccnscicusncss,
jcc|inqs, prcjcrcnccs anJ a wc||-ucinq) sncu|J at |cast qct tnc uasic riqnt nct tc uc uscJ as
mcrc|y a ucJi|y mcans jcr scmccnc c|sc`s ncn-vita| cnJs. lurtncrmcrc, wc sncu|J qrant tnis
uasic riqnt tc scricus|y mcnta||y Jisau|cJ scnticnt numans (wnc |ack mcnta| capacitics sucn
as sc|j-ccnscicusncss cr mcra| rcascninq), uascJ cn c|car rcascns tnat arc intrinsic (i.c.
sc|c|y rcjcr tc prcpcrtics cj tnc inJiviJua|) anJ ncn-cmpirica| (i.c. sc|c|y kncwinq tnat tncy
arc scnticnt numans is cncuqn tc qrant tncm tnc uasic riqnt).
!

8entient humans should certainly not be used against their will lor our lood
(e.g. killing or conlining people in order to eat their bodily products), because we
do not need human meat to live a healthy lile and such an involuntary use lor lood
is not respectlul and causes an unjustiliable large loss ol well-being.
1he condition that the reason lor granting someone the basic right should be
intrinsic and non-empirical is lundamental in the argument lor veganism. lt is not
surprising that this condition is highly disputed in the literature. 8ome
philosophers simply deny some rights to mentally disabled humans due to their
lack ol mental capacities: desires (lrey, 1980), consciousness (carruthers, 1992),
the capacity to lree choice (machan, 2004), the capacity to comprehend rules ol
duty (cohen, 1986), being able to contribute to the social reproductive process ol
beings who are capable ol acting on reasons (ooldman, 2001) or some other
capacity. lowever, most ethicists want to keep the basic right lor most or all
humans, including the mentally disabled humans. ln delending the consumption
ol animal products, these ethicists olten come up with non-intrinsic reasons as the
basis lor granting the basic right to mentally disabled humans. 1he major problem


1
lxplicitly relerring to mentally disabled humans is ol course the well-known argument lrom marginal
cases (e.g. uombrowski, 199).
tnica| ccnsistcncy anJ Anima| ua|ity
26
with non-intrinsic or empirical reasons is that they underestimate the importance
ol respect lor moral patients such as disabled humans: what is the real reason why
those mentally disabled humans deserve respect and should be granted at least the
basic right7 ll the reason is merely empirical or extrinsic, there was always the risk
ol the situation being otherwise. 8entient humans shouldn't be just lucky that
circumstances are the way they are. 1heir basic right shouldn't depend on a happy
coincidence, but should be valid in all possible worlds. kestricting to intrinsic and
non-empirical reasons means that arbitrariness ol the situation is avoided as much
as possible. And avoiding arbitrariness is a sign ol respect lor the moral patients.
1herelore, those mentally disabled sentient humans should get the basic right
even il they were never mentally abled belore
2
, they had no potentiality to become
abled in the luture
!
, they were disabled due to a genetic mutation (their disability
is determined by their genetic make-up) such that they lack some ol the genes
necessary lor e.g. moral agency
4
, they were as dependent on humans lor their
well-being as non-human animals are

, they were (abandoned or disowned)


orphans who had no close kinship with other humans who want to give them
rights
6
, they looked superlicially dissimilar to abled humans

, they had a special


property that no other humans have
8
, they were consciously bred by someone else
to be used as merely a means and hence they owe their lives to someone else
9
, they
had to eat meat in order to survive or predators needed to eat them
10
, it was
considered by others as a custom, ritual or tradition (with symbolic meaning) to


2
8ome argue that mentally disabled humans should get the basic right because we could become
mentally disabled in the luture and then we would want our basic right to be respected (see e.g. wreen,
1984). lowever, we cannot become a mentally disabled human who was never mentally abled belore.
!
1his is an olten heard argument to grant rights to non-moral agents such as children who have a
potential to become moral agents in the luture (e.g. melden, 1980). 8till, some mentally disabled
sentient humans have as little potential as non-human animals.
4
8ee the genetic basis ol moral agency (liao, 2010).

8ee ounnarsson (2008).


6
Narveson (198) used an argument that rellects this condition ol kinship with other individuals.

According to Narveson (19), one ol the reasons why we give rights to mentally disabled humans is
because ol leelings ol sympathy on the basis ol superlicial similarities. 1his sympathy is merely
triggered by similarities, and should not be conlused with empathy. Also wreen (1984) uses the
argument that we identily ourselves with human non-persons.
8
1his relers to a possible reply to the super-chimp (lumar, 2008) or super-cat (wreen, 1984) examples:
a highly intelligent mutant super-cat would not get rights il rights are based on species normality
(what most members ol the species have). 1his seems counter-intuitive because this unique cat is
rational. 8o the reply goes that this super-cat must belong to another species than lc|is Jcmcstica (even
il it can still interbreed with other cats). lut then a same reasoning allows to conclude that a mentally
disabled human with an exceptional property is no longer a ucmc sapicns.
9
1his is the underlying rationale ol the logic ol the larder (see 8cruton, 2004, matheny & chan, 200).
10
1his counters the argument ol 'moral sociability as a precondition to justice' (larilan, 200). A
subject has no moral sociability il its right to lile is incompatible with the right to lile ol someone else.
ccrc arqumcnt jcr vcqanism
2
use them as merely a means
11
, using them as merely a means did not have negative
ellects on the morality and behavior ol moral agents towards other moral agents
12
,
using them would benelit us as much as using non-human animals as merely a
means
1!
, using them would be a better late (less sullering) lor those mentally
disabled humans than they would have had in the wild (in the absence ol other
humans)
14
, their use as a means prevented more harm to other humans
1
, their use
as a means allowed using some resource that could not be used otherwise
16
, the
majority ol humans had no issue with using these disabled humans
1
, the majority
ol humans had moral intuitions that excluded mentally disabled humans
18
or that
did not track species membership as a natural kind
19
(e.g. the majority had the
intuition that disabled humans belong to a dillerent natural kind than abled
humans), the majority ol humans were mentally disabled
20
, some humans were
carnivorous predators
21
, there was a clear dillerence between the abled and the
seriously disabled humans (i.e. when semi-mentally disabled humans would not
exist)
22
, or we were non-human moral agents
2!
.


11
8cruton (2006) and maclean (2010) emphasize symbolic meanings ol eating animal meat as well as
taboos about e.g. eating human corpses to justily a distinction between humans and animals.
12
1his relers to the argument ol indirect duties or duties towards onesell, used by e.g. lant (18, part
ll, paras 16 and 1) and carruthers (1992).
1!
Narveson (198) tried to avoid the conclusion that use ol mentally disabled humans is permissible by
claiming that their use would not be as benelicial lor us alter all.
14
1his relers to a condition proposed by larilan (200) lor non-human species, but hereby translated to
mentally disabled humans.
1
1his relers to the least harm argument against vegetarianism proposed by uavis (200!), but hereby
applied to mentally disabled humans.
16
1his relers to the argument against vegetarianismJveganism that livestock larming allows us to use
resources such as grazing land that otherwise remain unavailable lor direct consumption.
1
8ee e.g. Young (1984), lor whom the morality ol killing X depends on others who have an interest in
X's continued existence. lut also 8cruton (2006) relers to the sentiments ol others about the way we
are allowed to treat someone. 1he impermissibility ol using mentally disabled humans merely due to us
being disturbed by that idea is like the prohibition ol eating e.g. human cultured meat, plants that have
a symbolic (e.g. religious) meaning or alcoholic beverages that are considered taboo in some cultures.
1hese prohibitions have nothing to do with rights violations.
18
l.g. ooldman (2001) relerred to moral intuitions that excluded animals.
19
l.g. levy (2004).
20
1his relers to the normality argument: moral agency is normal lor humans because most humans
poses moral agency. l.g. 1homas (2010).
21
1his relers to the predation argument: we are allowed to eat animals when some animals eat other
animals lor survival.
22
1his relers to the slippery slope argument (e.g. carruthers, 1992): il we start using mentally disabled
humans, we might end up using mentally abled humans, because there is a continuum lrom disability
to ability. 8ee also lruers (201!).
tnica| ccnsistcncy anJ Anima| ua|ity
28
1his long list captures a big part ol the animal ethics literature. 1he arguments
in lavor ol speciesism and meat consumption olten reler to extrinsic or empirical
lacts. consider the potentiality argument: il giving a disabled human the best lood,
protection and education, does it eventually become a rational, sell-conscious
being7 ll not, then the human has no potential. lut this potentiality is an empirical
issue: il potentiality is a necessary condition lor the basic right, and il scientists
discover that a mentally disabled sentient human cannot become a rational being
even with the best lood and protection, this disabled human would suddenly lose
its basic right. lrom the point ol view ol real respect lor this sentient human, this
is unacceptable. lurthermore, the potentiality argument is dependent on
empirical lacts lrom another perspective: suppose that scientists discover that
some plants, once you keep them alive long enough, say thousands ol years, will
eventually spontaneously develop rationality. lt seems counter-intuitive to
conclude that we are now no longer allowed to eat those potentially sell-conscious
plants.
Or consider the last possibility in the above list: what il we were non-human
moral agents7 ll we respect mentally disabled humans merely because we are
humans ourselves, then the moral status ol those disabled humans depend on an
extrinsic lact that could have been otherwise. lmagine that all mentally abled
humans were replaced by some non-human moral agents. 8hould these non-
human moral agents respect those remaining mentally disabled humans7 ll the
answer is yes, then respect lor those disabled humans cannot be merely a human
prejudice as in williams (2006). ll the answer is no, then what do we really mean
with respect lor disabled humans7
moral assumption 1 is restricted to at least sentient humans. 1his does not
mean that it is permissible to use other individuals - such as non-human sentient
beings - as merely a means. lor example also dogs and cats have this basic right
according to many people. lut keeping the argument as parsimonious as possible,
we do not have to assume in advance that those particular non-human sentient
beings should get the basic right.
lurthermore, the basic right is based on a no-harm principle, so it doesn't
necessarily include all our moral concerns towards non-sentient humans. Other
moral considerations which are not based on interests or rights - lor example a
moral taboo about eating human corpses (uiamond, 198) - might be included in
our ethical system without undermining the argument lor veganism. Yet, the


2!
1his relers to the human prejudice argument (williams, 2006): in the absence ol an impartial point ol
view, we could (as humans) be partial in lavor ol other humans, lrom our own particular (human) point
ol view.
ccrc arqumcnt jcr vcqanism
29
impermissibility ol killing and eating mentally disabled sentient humans should
primarily reler to the no-harm principle ol the basic right, il we believe that the
interests and rights ol those humans are ol primary importance.

ucjiniticn .: rcjuJicia| Jiscriminaticn cj inJiviJua| (cr qrcup) A rc|ativc tc 8 is a
systcmatica||y Jijjcrcnt trcatmcnt cj A anJ 8 (c.q. 8 qcts mcrc aJvantaqcs tnan A), wncrcuy
1) it is c|aimcJ tnat A nas a |cwcr mcra| status tnan 8 (c.q. tnat A nas |css intrinsic
va|uc cr wcakcr riqnts tnan 8) in tnc scnsc tnat cnc wcu|J nct tc|cratc swappinq
pcsiticns (trcatinq A as 8 anJ 8 as A), anJ
2) tncrc is nc justijicaticn cr tnc justijicaticn cj tnc prcvicus pcint rcjcrs tc mcra||y
irrc|cvant critcria (prcpcrtics tnat arc nct acccptau|c mctivcs tc trcat A anJ 8
Jijjcrcnt|y in tnc ccnccrncJ situaticn), wncrcas A anJ 8 uctn mcct tnc samc mcra||y
rc|cvant critcria tc trcat anJ va|uc tncm mcrc cua||y.
1he lirst condition is crucial il we want to avoid discrimination in a burning
house dilemma (clr. lrancione, 2000), where we have to choose between saving A
versus l. 8aving your own child instead ol a child with another skin color does not
(yet) mean that you are racist. Antidiscrimination does not imply that you should
llip a coin and give each child an equal 0/ survival probability. You are allowed
to be partial in lavor ol your child, as long as you tolerate me saving the other
child. uue to this kind ol toleration we can avoid an inconsistency between some
kind ol partiality and the antidiscrimination principle. Above all, a partiality
towards your own child saving your child instead ol another child or a dog in the
burning house does not mean that you are allowed to use other children or dogs
as merely a bodily means lor the non-vital ends ol your own child.
1he second condition is crucial il we want to avoid arbitrariness. lerhaps
williams (2006) disagrees with the importance ol this second condition when it
comes to the human prejudice.

Vcra| assumpticn .: wncn it ccmcs tc rcspcctinq tnc uasic riqnt, a critcricn cr prcpcrty
is mcra||y irrc|cvant tc a niqncr Jcqrcc ij mcrc cj tnc jc||cwinq ccnJiticns arc mct:
1) tnc prcpcrty is aruitrary (tncrc is nc ncn-circu|ar ru|c tnat sc|ccts tnc prcpcrty cut
cj a mu|tituJc cj simi|ar kinJs cj prcpcrtics), cr
2) tnc prcpcrty is nct intrinsic (it Jccs nct rcjcr sc|c|y tc tnc inJiviJua| pcsscssinq tnc
prcpcrty), cr
!) tnc prcpcrty is inncrcnt|y Jijjicu|t tc Jctcct, Jcjinc cr Jc|imit (tnc prcpcrty is ncn-
cmpirica| cr tncrc arc nc scicntijic critcria anJ mctncJs nct cvcn in princip|c tc
c|car|y scc wnctncr tnc prcpcrty is prcscnt).
Note that l allow lor a gradation in moral irrelevance, i.e. moral irrelevance is
not necessarily an all-or-nothing issue. ll a property meets all three ol the above
tnica| ccnsistcncy anJ Anima| ua|ity
260
conditions, it is extremely morally irrelevant. we will see that species membership
can be extremely morally irrelevant.
1he anti-arbitrariness condition states that il property X is morally relevant,
then also should be all properties Y and 2 that are similar to X. lxamples ol
criteria that are morally irrelevant because they are arbitrary are: physical
characteristics and appearances (e.g. skin color, behavior, gender), genetic
properties (e.g. race, ethnicity, genetic kinship), prelerences (e.g. sexual, political)
or belonging to an arbitrary group (prelerring one group when there is a
multitude ol groups in a complex hierarchy or taxonomy).
1hese properties are arbitrary because there is a multitude ol physical
characteristics, genetic properties, prelerences and groups that all appear similar
lrom the perspective ol the basic right. looking at the lormulation ol the basic
right, we do not see any relerence to such properties. 1he lormulation ol the basic
right relers to 'body', 'will' and 'well-being', but not to e.g. skin colors, genes or
groups. 1he basic right does not allow us to answer questions like: why this skin
color instead il that7 why this sexual prelerence instead ol that7 why this group
instead ol that7 ll these questions cannot be answered in a meaninglul way, the
properties are arbitrary because the dillerent properties (e.g. skin colors) are
similar. 8uch arbitrariness opens the door lor abusive opportunism. ll l can say
that skin color X is the morally relevant one, then you are allowed to say that color
Y is the relevant one.
many ol the abovementioned properties are also dillicult to deline or delimit:
there are gradations in skin color, there are intersexual people, there are mixtures
ol ethnicities.
Other examples ol irrelevant criteria are non-intrinsic properties, such as being
a descendant ol certain privileged parents or ancestors. One could say that all
rational agents and all their descendants (even when they are not rational
themselves) should get the basic right. 1his group includes all non-rational,
sentient humans, but lor those non-rational descendants ol rational agents the
property is merely extrinsic. 8omeone should not get the basic right merely
because sJhe is lucky to have the right parents. lurthermore, looking at the third
condition, the property ol rationality is dillicult to delimit. ln particular, looking at
our ancestors, we cannot point at an ancestor and say that this was the very lirst
ancestor who was rational.
lxamples ol criteria that are irrelevant because they are inherently dillicult to
detect, are supernatural (non-empirical) properties based on e.g. religious notions
such as having a soul or being created at the image ol ood. lt is inherently
impossible to scientilically (empirically) establish the truth ol those properties, so
they cannot be used to determine who gets the basic right. 1he same goes lor
notions like 'dignity', 'potential', 'nature', 'kind' and 'essence'.
ccrc arqumcnt jcr vcqanism
261
1he next three lacts demonstrate that species is an extremely morally
irrelevant property.

lact !: Ncxt tc tnc qrcup cj numans as a uic|cqica| spccics, tncrc arc many ctncr
uic|cqica| qrcups anJ c|assijicaticns. 1ncrc is a wnc|c ranqc cj qrcups at Jijjcrcnt |cvc|s:
pcpu|aticns (wnitc pccp|c), suuspccics (ucmc sapicns sapicns), spccics (ucmc sapicns),
qcnus (ucmc), jami|y (qrcat apcs), injracrJcr (simians), crJcr (primatcs), injrac|ass
(p|accnta|s), c|ass (mamma|s), suupny|um (vcrtcuratcs), pny|um (cncrJatcs), kinqJcm
(anima|s). t is aruitrary tc pick tnc spccics ucmc sapicns cut cj tnis |ist.
8tatements such as most ol the beings belonging to the species ucmc sapicns
are rational agents" are equally valid as most ol the beings belonging to the
inlraorder ol simians are rational agents". when it comes to such statistical
normality it remains arbitrary to reler to the species instead ol e.g. the inlraorder.
You and l are simians, just as we are humans.

lact .: 5cmc ways tc Jcjinc spccics arc usinq a ncn-intrinsic prcpcrty ncticn cj spccics,
c.q. intcrurccJinq cr Jcsccnt.
oranting the basic rights based on a biological delinition ol species is lar-
letched or artilicial. lor example, it is artilicial to claim that having close
relatives who could have had lertile ollspring with someone else" is morally
relevant lor the basic right, because it raises the question: what has that got to do
with it7" 8imilarly, descent is a non-intrinsic property. No-one has chosen to have
parents or ancestors who have or had certain privileged properties that are
considered as characteristically human (e.g. a seriously mentally disabled
individual who has parents with certain cognitive capacities).

lact 1: 1nc pcssiu|c cxistcncc cj numan-anima| nyuriJs, cnimcras cr qcnctica||y mcJijicJ
pccp|c, anJ tnc jcrmcr cxistcncc cj ccmmcn anccstcrs cj numans anJ ncn-numan anima|s,
arc uic|cqica| jacts tnat inJicatc tnat tnc qrcup cj numans cannct uc c|car|y Jc|incatcJ.
0nc|car (ccmp|cx anJ juzzy) ucunJarics rcmain.
According to evolutionary biology and genetics, scientists will never be able to
tell - not even in principle - when an individual is a human. ln other words: there
are no non-arbitrary empirical lacts that determine a sharp boundary between
humans and non-humans. 8omeone's moral status and basic right should not
depend on the accidental non-existence ol such unclear boundary cases. ln this
sense, evolutionary biology undermines speciesism (see mcmahan, 200, kachels,
1990). ll all human-animal intermediates (all our ancestors) still existed, the
notion ol numan rights will delinitely appear to be arbitrary.
most importantly: the above three biological lacts undermine essentialism. 1his
essentialism is pervasive in the anti-animal rights literature. ln lact, almost all ol
tnica| ccnsistcncy anJ Anima| ua|ity
262
the anti-animal rights literature can be divided in two big parts: 1) the denial ol
intrinsic, non-empirical reasons lor granting rights to mentally disabled humans,
as we have discussed above, and 2) the believe in essentialism.
ln the literature, essentialism can be recognized by expressions like 'rational
nature' (lee & oeorge, 2008), 'essential nature ol a living kind' or 'a specilic type ol
substantial nature' (lumar, 2008). 1his essentialism is the believe that a thing
either is or is not a human being, and that all and only humans share some
essence. Other philosophers who reler to 'essence', 'nature' or 'kind' are e.g.
chappell (2011), cohen (2001), linnis (199, p.48), lee (2004), 8canlon (1998, p.186)
and 8cruton (1998). 1he argument lrom kinds was criticized by e.g. 1anner (2006).
liology shows that no natural 'kind' or 'essence' is related to the species ucmc
sapicns. lt is just like the set ol letters 'l': we can recognize and read a letter l when
we see it (at least in most cases), but this set ol l's is very abstract and dillicult to
deline. what do the letters l, l and l have in common7 ll biological entities have
kinds, then so do letters. lut not only is a kind ol letter dillicult to deline, there
are also many other sets ol letters, such as the sets ol letters 'l' and 'l', as well as
unilied sets ol letters such as 'l, l and l'. ll it is meaninglul to say that the letter 'l'
is ol a dillerent kind than the letter 'O', then it is equally meaninglul to say that 'l
and l' is ol a dillerent kind than 'O and Q'. lntroducing hybrid letters such as 'l, l
and u' makes it even more complicated lor letter essentialism.
lor both letters and living beings, we can generate a whole hierarchy ol
possible classilications (taxonomies), and between the groups are luzzy and
complex boundaries (hybrids). lence, even attempts to reler to a 'narrowest
natural kind' (levy, 2004) are doomed to lailure. According to levy, one could
deline humans as the narrowest natural kind that encompasses all rational human
beings. 1his narrowest kind would then correspond with the complete species
ucmc sapicns. lut it remains arbitrary to take the narrowest kind instead ol e.g. the
broadest kind or the largest natural kind ol which the majority are rational
humans. constructions such as 'narrowest kind' are too artilicial and arbitrary lor
granting someone a basic right. And a narrowest natural kind still laces the
problem ol luzzy boundaries.
1he reader is invited to read through the anti-animal rights literature, and
translate the word 'human' into 'moral person', 'sentient being' or 'primate', as
well as 'species' into 'set', 'inlraorder' or 'kingdom'. 1his strategy helps to bring
unconscious or implicit essentialist thinking to the surlace. 1ake lor example an
expression that the only point ol view humans can take is the human point ol
view" (e.g. williams, 2006). 1he same could be said lor primates with their 'primate
point ol view'.
combining moral assumption 2 with lacts 1, 2 and !, we arrive at our lirst
intermediary conclusion.
ccrc arqumcnt jcr vcqanism
26!

ccnc|usicn !: 1nc critcricn 'ucinq numan` is mcra||y irrc|cvant tc an cxtrcmc Jcqrcc.
combining this conclusion with delinition 1 gives:

ccnc|usicn .: A Jijjcrcncc in trcatmcnt anJ cva|uaticn, uascJ cn spccics (tnc critcricn
'ucinq numan`), is a kinJ cj prcjuJicia| Jiscriminaticn.
Ol course, conclusion 2 does not yet give us a moral judgment. we have to
introduce a normative principle, such as: il you are allowed to discriminate (i.e.
arbitrarily choose your victims at will and violate their basic rights), then so am l,
and so is everyone, and this is something we cannot want. 1herelore, most ol us
would introduce a next moral assumption.

Vcra| assumpticn 1: prcjuJicia| Jiscriminaticns is ccrtain|y immcra| wncn it ccmcs tc
qrantinq anJ rcspcctinq tnc uasic riqnt.
uiscrimination is intrinsically immoral, which means that its immorality is
independent lrom empirical circumstances. 1he immorality should not depend on
e.g. the interconnectedness ol social reproduction in the contemporary world"
(ooldman, 2001, p64) or the accidental lact that only a lew people believe that
discrimination is e.g. 'natural' and therelore permissible.
conclusion 2 with moral assumption ! gives us:

ccnc|usicn !: 1nc critcricn 'ucinq numan` sncu|J nct uc uscJ jcr qrantinq riqnts.
1ogether with moral assumption 1 we get:

ccnc|usicn : 1nc sct cj ucinqs wnc rcccivc junJamcnta| riqnts - in particu|ar tnc riqnt
nct tc uc uscJ aqainst cnc`s wi|| as ncn-vita| jccJ - sncu|J nct cxp|icit|y rcjcr tc 'numans`.
Anctncr critcricn tnan 'ucinq numan` must uc uscJ. 1nat critcricn must uc mcra||y
rc|cvant, anJ a|| scnticnt mcnta||y Jisau|cJ numans sncu|J mcct tnat critcricn.
we know what criteria are morally irrelevant, but what criteria are relevant7

Vcra| assumpticn !: wncn it ccmcs tc rcspcctinq tnc uasic riqnt, a critcricn is ccrtain|y
mcra||y rc|cvant in rc|aticn tc an cua| trcatmcnt anJ mcra| cva|uaticn uctwccn
inJiviJua|s, ij it is an iJcntijiau|c cr mcasurau|c (i.c. cmpirica|) prcpcrty tnat:
!) wc ccu|J Jcrivc jrcm an impartia| (ncn-aruitrary) pcint cj vicw (tnc mcra| vicwpcint),
anJ
.) is c|car|y rc|atcJ (i.c. nct in a jar-jctcncJ way) witn tnc ncticn cj tnc uasic riqnt, anJ
1) jc||cws jrcm mcra| virtucs cr va|uau|c jcc|inqs (i.c. cmcticns tnat arc impcrtant in cur
mcra| Jccisicn makinq, sucn as cmcticns tnat mctivatc us tc nc|p ctncrs cr tc rcspcct
riqnts).
Again, this assumption gives three conditions, such that il a criterion meets all
three ol them, it should delinitely be considered as a relevant criterion. 1he next
tnica| ccnsistcncy anJ Anima| ua|ity
264
three moral assumptions demonstrate that a criterion ol sentience (having a
consciousness with a well-being and a will) is delinitely morally relevant.

Vcra| assumpticn : lrcm an impartia| pcint cj vicw, wc||-ucinq is ccrtain|y impcrtant.
cckinq at Jiscriminaticn, tnc Jijjcrcncc in trcatmcnt is mcra||y scricus in particu|ar wncn
tnc wc||-ucinq cj inJiviJua|s is ajjcctcJ. n aJJiticn tc tnis, jcc|inqs anJ cmcticns - pcsitivc
cr ncqativc ccnscicus scnsaticns - arc impcrtant uccausc jcc|inqs ajjcct wc||-ucinq.
A thought experiment can be used to check impartiality: imagine that you will
be born as something or someone, but you do not know who or what you will be.
1hen, as you value your well-being, you would want to take into account the well-
being ol all beings in a serious and impartial (non-arbitrary) way. Note that il you
were a sentient being, your well-being would be taken into account, whereas il you
were something without leelings and desires, you could not care about what
happened to you. You would not leel or notice anything and you would have no
desires about how you are treated. 1his thought experiment is an extension ol the
kawlsian veil ol ignorance, extended to include not only rational agents but all
entities (see Van ue Veer, 199, kowlands, 199, Van den lerg, 2011).
lmpartiality takes everyone and everything into account, but no matter what
we do, we automatically respect the prelerences, well-being and leelings ol those
things that lack prelerences, well-being and leelings. we cannot treat something
against its will il that thing has no will. we cannot inlluence the well-being ol
something that has no well-being. we cannot harm something that does not lind
anything important. ll something has no consciousness, it cannot be aware ol how
we treat it, let alone dislike its treatment.

Vcra| assumpticn e: A riqnt is a ccrtain way cj prctcctinq an intcrcst cr nccJ. 5cnticnt
(pcrccptua||y ccnscicus) ucinqs navc ccmp|cx intcrcsts anJ navc pcrccpticns cj tncsc
intcrcsts. csitivc anJ ncqativc jcc|inqs arc rc|atcJ tc wnat an inJiviJua| wants. ucncc,
tncrc is a ncn-jar-jctcncJ ccnnccticn uctwccn qcttinq riqnts anJ navinq nccJs, jcc|inqs anJ
a wc||-ucinq. n particu|ar, it is nct jar-jctcncJ tc qivc a scnticnt ucinq tnc riqnt nct tc uc
uscJ aqainst its wi|| as mcrc|y a ucJi|y mcans, uccausc a scnticnt ucinq nas an inJiviJua|
wi|| anJ a scnsc cj its cwn ucJy, anJ tnis riqnt Jircct|y rcjcrs tc tnc usc cj scmccnc`s ucJy
anJ wnat is wantcJ cr wi||cJ.
According to this assumption, the basic rights should apply to literally
everything and everyone, without arbitrary restrictions. lut we already
automatically respect the basic right ol a non-sentient entity that has no will and
no sense ol its body.

Vcra| assumpticn : ccnccrn, cmpatny anJ ccmpassicn arc va|uau|c jcc|inqs (mcra|
virtucs) tnat ajjcct cur mcra| ucnavicr. wc can jcc| ccnccrn jcr scnticnt ucinqs uccausc tncy
ccrc arqumcnt jcr vcqanism
26
navc intcrcsts anJ arc vu|ncrau|c. wc can jcc| cmpatny (ccmpassicn) witn scnticnt ucinqs
anJ Jcvc|cpinq cmpatny witn a|| scnticnt ucinqs (cxtcnJinq tnc circ|c cj ccmpassicn) is a
mcra| virtuc.
1his assumption gives an extra argument why sentience is important. One
could object that sentience might be dillicult to detect, deline or delimit. lowever,
sentience is in a sense an all-or-nothing issue: either there is a leeling (no matter
how weak), or there is not. lither something is wanted or not. ln history there was
a moment when the lirst conscious experience arose in a living being, just as there
is a lirst moment when, alter turning on my computer and moving the mouse, the
cursor on the screen starts to move. 8entience is dillicult to detect, but as we will
see, it is the very business ol scientists to lind out which being is conscious. lt is a
scientilic question and consciousness is strongly related to empirical processes.
lrom moral assumptions 4- we get:

ccnc|usicn e: '8cinq scnticnt` is a qccJ canJiJatc cj a mcra||y rc|cvant critcricn jcr tnc
uasic riqnt.
1his does not exclude other criteria lrom being relevant. Our minimalist, core
argument says that at least sentience is relevant. lut which being is sentient7 As
sentience is a natural property, we can look at science.

Vcra| assumpticn : j tncrc is sujjicicnt scicntijic cviJcncc tnat a ucinq is scnticnt, wc
must assumc tnat it is scnticnt. j tncrc is a c|car |ack cj scicntijic cviJcncc, wc cannct
assumc tnat tnc ucinq is scnticnt. 5cicntijic cviJcncc inc|uJcs anatcmica| cnaractcristics,
ucnavicr, pnysic|cqica| cnanqcs anJ cvc|uticnary aJaptivc mccnanisms tnat unJcr|ic
jcc|inqs anJ cmcticns.
lere we can reler to e.g. the cambridge ueclaration on consciousness (2012), as
well as current animal wellare laws and scientilic opinions lor use ol animals in
e.g. scientilic research (such as ll8A, 200, 2009). 1his demonstrates that the
question ol sentience lies in the realm ol science.

Vcra| assumpticn -: vcrtcuratcs witn a juncticna| ncrvcus systcm mcct sujjicicnt
scicntijic cviJcncc jcr ucinq scnticnt, in tnc samc way tnat at |cast scmc mcnta||y Jisau|cJ
pccp|c mcct sujjicicnt scicntijic cviJcncc.
1his is a moral assumption instead ol a lact, because it contains the word
'sullicient', which is normative.

Vcra| assumpticn !: |ants sncw a markcJ |ack cj scicntijic cviJcncc jcr ucinq scnticnt.
combining moral assumptions 8 to 10 gives:

tnica| ccnsistcncy anJ Anima| ua|ity
266
ccnc|usicn : vcrtcuratc anima|s (anJ scmc crustaccans anJ ccpna|cpcJs) arc scnticnt
ucinqs, p|ants mcst |ikc|y nct. or mcrc accuratc|y: tnc |ikc|inccJ tnat vcrtcuratcs witn
juncticna| ncrvcus systcms arc scnticnt is tnc samc as tnc prcuaui|ity tnat scmc mcnta||y
Jisau|cJ numan arc scnticnt, anJ is mucn niqncr tnan tnc prcuaui|ity tnat a p|ant is
scnticnt.
As the basic right relers to non-vital needs, we have to know whether the
consumption ol products lrom vertebrates is vital lor us.

lact !: wc can qrant tnc riqnt nct tc uc uscJ as mcrc|y a mcans (in particu|ar jcr jccJ
anJ c|ctninq) tc a|| vcrtcuratcs (anima|s witn a juncticna| ccntra| ncrvcus systcm), witncut
tnrcatcninq cur vita| nccJs.
Again, this lact is based on science. 8ee e.g. the position ol the Academy ol
Nutrition & uietetics (AuA, 2009): we do not need animal products in our diet to
live a healthy lile. Veganism is leasible in terms ol health. Not only leasibility in
terms ol physical health is required, but also in terms ol agricultural productivity
and ecological health. looking at studies ol land use, soil lertility, nutrient cycling,
resource use and pollution, we can conclude that a vegan agriculture is most likely
suitable to leed all humans without increasing our ecological impact (see e.g.
lairlie, 200, 8tehlest e.a., 2009, cll, 201!, Olewski, 2010).
1he underlying moral assumption ol the above lacts is ol course that on these
issues we should lollow the scientilic consensus opinion ol e.g. dietitians,
ecologists and agricultural scientists.
lrom conclusions 6 and and lact 4 we get:

ccnc|usicn : 8cinq scnticnt (navinq jcc|inqs) is a qccJ canJiJatc cj a mcra||y rc|cvant
critcricn jcr tnc riqnt nct tc uc uscJ uy us jcr jccJ cr c|ctninq, anJ uy tnis critcricn
mcnta||y Jisau|cJ scnticnt numans surc|y qct tnat riqnt.
ln addition to sentience there may be other morally relevant criteria, such as
being alive, having sell-awareness or having a moral sense. lowever, these criteria
cannot be related to the basic right in terms ol use lor lood or clothing, due to the
lollowing two lacts.

lact : wc nccJ p|ants as jccJ tc survivc. 5c wc cannct qrant tnc riqnt nct tc uc uscJ as
jccJ
.!
tc a|| |ivinq ucinqs (inc|uJinq p|ants), witncut tnrcatcninq cur vita| nccJs.



24
Note that this use is not a use as merely a means il plants do not possess a will and cannot be used
against their will. lence, this right is broader than the basic right in delinition 1.
ccrc arqumcnt jcr vcqanism
26
lact e: 1ncrc arc scnticnt mcnta||y Jisau|cJ pccp|c wnc navc nc mcra| cr sc|j-
ccnscicusncss cr mcra| scnsc.
As we have seen, the mere lact that these mentally disabled humans are
descendants ol (grand) parents who have such mental abilities, is not morally
relevant, because lineage should not allect having basic rights and privileges.

lact : Apart jrcm spccics, tncrc is nc kncwn iJcntijiau|c cr mcasurau|c prcpcrty tnat
qcncratcs a c|car anJ rc|cvant Jistincticn uctwccn mcnta||y Jisau|cJ scnticnt numans anJ
ctncr (inncccnt, ncn-aqqrcssivc) scnticnt vcrtcuratc anima|s.
lrom lacts - we get:

ccnc|usicn -: lcr tnc timc ucinq, wc Jc nct jinJ ctncr mcra||y rc|cvant critcria ncxt tc
'ucinq scnticnt` tnat !) arc mct uy scnticnt mcnta||y Jisau|cJ numans anJ .) Jc nct tnrcatcn
cur vita| nccJ jcr jccJ.
lrom conclusions , 8 and 9 lollows:

ccnc|usicn !: A|| scnticnt ucinqs (cspccia||y vcrtcuratcs witn juncticna| ncrvcus
systcms) navc tnc uasic riqnt nct tc uc uscJ as mcrc|y a ucJi|y mcans (i.c. aqainst tnc wi||)
jcr cur ncn-vita| cnJs, in particu|ar jcr jccJ anJ c|ctninq.
Ol course, the question remains whether animals are used as merely a means. l
will not describe the practices in livestock larming and lisheries. lt will sullice to
mention that they are treated against their will, they lose a lot ol well-being, and
ol course their bodies are used. 8o we have to assume the lollowing lact.

lact : Anima|s in |ivcstcck anJ jisncrics arc uscJ as mcrc|y a ucJi|y mcans jcr cur ncn-
vita| cnJs.
One linal moral assumption is necessary.

Vcra| assumpticn !!: wc must jc||cw tnc ru|c tnat cvcrycnc (wnc is capau|c) must jc||cw
in a|| simi|ar situaticns.
1he rule that is relevant in this context is: boycott animal products. lrom
conclusion 10, lact 8 and moral assumption 11, we get:

ccnc|usicn !!: As a ru|c, vcqanism is a mcra| Juty jcr ycu anJ mc.
1his concludes the core argument lor veganism. lt is based on consistency,
using a set ol assumptions. lerhaps we also need the assumption that not only
consistency is important, but that a consistent argument remains valid even when
there are other inconsistencies elsewhere in our ethics. 1here might still be
inconsistencies with respect to some moral dilemmas, such as situations where
someone is used as merely a means lor vital needs (e.g. the predation problem or
tnica| ccnsistcncy anJ Anima| ua|ity
268
using someone against his will lor organ transplantations or lile-saving
experiments). Or there might be inconsistencies or arbitrariness with respect to
e.g. moral taboos (e.g. not eating human corpses, not burying animals) or extra
rights (e.g. not killing non-sentient humans).
8o we have to assume that even il there are local inconsistencies in our ethics,
those inconsistencies cannot undermine the consistency ol another part ol the
ethical system. lthics should work with a paraconsistent logic, where one local
inconsistency does not yet blow up the whole system. lt is like solving a crossword
puzzle: ll a white box contains more than one letter due to two conllicting words,
there is a local inconsistency at this white box. lut this local inconsistency does
not invalidate the rest ol the crossword puzzle. we are not allowed to arbitrarily
lill in letters at will everywhere even il there happens to be a local inconsistency.


ccrc arqumcnt jcr vcqanism


tnica| ccnsistcncy anJ Anima| ua|ity
20
Argumentation scheme lor veganism


ccrc arqumcnt jcr vcqanism



tnica| ccnsistcncy anJ Anima| ua|ity




lart 4 lpilogue

tnica| ccnsistcncy anJ Anima| ua|ity



2
chapter 1! 1he moral hands
1!.1 1he moral hand ol normative ethics: live principles ol
a complete and coherent ethic
1he previous chapters discussed a lot ol ethical principles. As a summary, and to
demonstrate the compactness, completeness and coherence ol the ethic, let me
use a metaphor: the moral hand. lach ol the live lingers represents an ethical
principle.
-1he thumb: rule universalism (see section 1.2). You must lollow the rules that
everyone (who is capable, rational and inlormed) must lollow in all morally similar
situations. You may only lollow the rules that everyone (who is capable, rational
and inlormed) may lollow in all morally similar situations. lrejudicial
discrimination is immoral. we should give the good example, even il others don't.
}ust like we have to place the thumb against the other lingers in order to grasp an
object, we have to apply the principle ol universalism to the other lour basic
principles.
-1he lorelinger: justice and the value ol liletime well-being (see section 4.4).
lncrease the well-being (over a complete lile) ol all sentient beings alive in the
present and the luture, whereby improvements lor the worst-oll positions (the
worst sullerers, the beings who have the worst lives) have a strong priority.
liletime well-being is the value you would ascribe when you would live the
complete lile ol a sentient being, and is a lunction ol all positive and negative
leelings that are the result ol (dis)satislaction ol prelerences (everything wanted
by the being).
-1he middle linger: the mere means principle and the basic right to bodily
autonomy (see section 6.2). Never use (or consider) the body ol a sentient being as
merely a means to someone else's ends, because that violates the right to bodily
autonomy. 1he two words 'mere means' reler to two conditions, respectively: 1) il
in order to reach an end (e.g. saving someone) you lorce a sentient being to do or
tnica| ccnsistcncy anJ Anima| ua|ity
26
undergo something that the being does not want, and 2) il the body ol that
sentient being is necessary as a means lor that end, then you are not allowed to
treat that being in that way. A sentient being is a being who has developed the
capacity to want something by having positive and negative leelings, and who has
not yet permanently lost this capacity. 1hey have the experience ol having
prelerences. 1he middle linger is a bit longer than the lorelinger, and so the basic
right is a bit stronger than the liletime well-being (e.g. the right to live). 1he basic
right can only be violated when the lorelinger principle ol well-being is seriously
threatened.
-1he ring linger: naturalness and the value ol biodiversity (see section 10.4). ll a
behavior violates the lorelinger or middle linger principles, the behavior is still
allowed (but not obligatory) only il that behavior is both natural (a direct
consequence ol spontaneous evolution), normal (lrequent) and necessary
(important lor the survival ol sentient beings). }ust as liletime well-being is the
value ol a sentient being, biodiversity is the value ol an ecosystem and is a
lunction ol the variation ol lile lorms and processes that are a direct consequence
ol natural evolution. 1he valuable biodiversity would drastically decrease il a
behavior that is natural, normal and necessary was universally prohibited
(universally, because you have to put the thumb against the ring linger).
-1he little linger: tolerated partiality and the value ol personal relationships
(see section .1). }ust as the little linger can deviate a little bit lrom the other
lingers, a small level ol partiality is allowed. when helping others, you are allowed
to be a bit partial in lavor ol your loved ones, as long as you are prepared to
tolerate similar levels ol partiality ol everyone else (everyone, because you have to
put the thumb against the little linger). 1his principle could be extended to
include some special duties towards people with whom one has personal
relationships (e.g. special duties ol assistance towards lriends, own children,.).

1he ethic ol the normative moral hand is a pluralist-principlist ethic that
borrows elements lrom dillerent ethical theories. 1he thumb is related to a
lantian categorical imperative and to rule consequentialism (looker, 2011). 1he
lorelinger is related to a consequentialist wellare ethic such as utilitarianism
(8inger, 199!), painism (kyder, 2001) and contractualism (kowlands, 199). 1he
middle linger rellects a second lormulation ol the lantian categorical imperative,
although it relers in particular to bodily autonomy and is applied to all sentient
beings instead ol only rational beings. 1he ring linger is related to an
environmental ethic (kolston, 1988) and the little linger to an ethic ol care
(Noddings, 2002).
1hese live lingers might look like post hoc rationalizations ol moral intuitions
(see e.g. laidt, 2001, oreene, 2008), and in lact they are. lxcept that each ol the
1nc mcra| nanJs
2
live principles is backed up by a coherent set ol arguments, which makes it less ol
an arbitrary system.
1he thumb principle allows us to satisly a need lor consistency and impartiality
(non-arbitrariness), it allows us to deal with game-theoretic ideal and non-ideal
situations (with or without universal compliance, see section .2), it allows us to
tackle moral illusions, it lits with our intuition that one should give the good
example, and it helps to generate principles (other principles such as the ring
linger and little linger only work when the thumb principle is applied).
1he lorelinger can be derived by either a veil ol ignorance thought experiment
(assuming some levels ol risk aversion, loss aversion and uncertainty aversion, as
well as some aspects ol prospect theory, see appendix 2) or by the moral virtue ol
compassion (assuming a need lor a certain level ol well-being elliciency, as well as
some moral intuitions about variable populations and personal identity). 1he
liletime perspective is justilied by two coherent reasons: the intuition that persons
are not replaceable, and the intuition that there is a dillerence between
intrapersonal and interpersonal harm, i.e. a dillerence between imprudent
behavior (harming your luture sell) and immoral behavior (harming another
person).
1he middle linger is consistent with moral intuitions in many (at least ten)
dilemmas, with notions ol respect and intrinsic value, and with a propertarian
libertarian ethic ol bodily autonomy.
1he ring linger lits with moral intuitions in situations ol predation, procreation
and motion, and is compatible with moral intuitions about naturalness (as used in
e.g. a carnist ideology) and the value ol biodiversity in environmental ethics.
1he little linger is compatible with our intuitions about partiality in situations
ol imperlect, positive and procreational duties, and it lits in an extended mere
means principle ol the middle linger (section 6.6).
lurthermore, the principles are also made compatible with scientilic
background theories ol biology (e.g. gradual evolution) and psychology (e.g.
personal identity and mental capacities). 1hese background theories challenge
some common sense assumptions on e.g. boundaries between species, boundaries
between persons and boundaries between levels ol mental capacities. Our moral
hand is perlectly able to deal with non-sharp boundaries between species, persons
and mental capacities. lor example the mathematical expression ol the lorelinger
principle includes a connectivity lunction to deal with complex issues ol
psychological connectedness and personal identities (appendix 2 lntermezzo: a
more complex lormulation to solve the replaceability problem"), the basic right ol
the middle linger couples gradations ol mental capacities with gradations ol needs
(section 6.4), and the biodiversity principle ol the ring linger works well even
when there are no clear boundaries between species.
tnica| ccnsistcncy anJ Anima| ua|ity
28
1hese considerations challenge the objection that the ethic ol the live lingers is
merely a post hoc rationalization ol intuitions. lt may be a rationalization, but it is
also a coherent system in wide rellective equilibrium (uaniels, 199). lt is a
mystery why a rationalization ol moral intuitions and emotions (oreene, 2008) can
have such a level ol coherence.
1!.1.1 live principles ol equality
1he live lingers ol the moral hand ol normative ethics produce live principles ol
equality.
-1he thumb: the lormal principle ol impartiality and antidiscrimination. we
should treat all equals equally in all equal situations. we should not look at
arbitrary characteristics linked to individuals. 1his is a lormal principle, because it
does not say how we should treat someone.
1he other lour principles are material principles ol equality. 1hey have specilic
content and are generated when the thumb is applied to the lour lingers.
-1he lorelinger: prioritarian equality ol liletime well-being (the principle ol
priority lor the worst-oll). As a result ol this priority, we have an egalitarian
principle: il total liletime well-being is constant between dillerent situations, then
the situation which has the most equal distribution ol well-being is the best.
-1he middle linger: basic right equality. All sentient beings (with equal levels ol
morally relevant mental capacities lor well-being) get an equal claim to the basic
right not to be used as merely a means to someone else's ends.
-1he ring linger: naturalistic behavioral lairness. All natural beings (who
contribute equally to biodiversity) have an equal right to a behavior that is both
natural, normal and necessary (i.e. a behavior that contributes to biodiversity).
Natural beings are beings evolved by evolution. l.g. il a prey is allowed to eat in
order to survive, a predator is allowed to do so as well (even il it means eating the
prey). ll the natural, normal and necessary behavior involves several options, the
option that causes the least harm (the least loss ol well-being, the least violations
ol basic rights and the least loss ol biodiversity) should be chosen (e.g. il an
omnivore can survive by eating sentient animals as well as by eating non-sentient
beings, sJhe should not eat the sentient beings).
-1he little linger: tolerated choice equality. lveryone is allowed to be partial to
an equal degree that we can tolerate. ll you choose to help individual X instead ol
individual Y, and il you tolerate that someone else would choose to help Y instead
ol X, then X and Y have a tolerated choice equality (even il X is emotionally more
important lor you than Y).
1nc mcra| nanJs
29
1he lorelinger, middle linger, ring linger and little linger correspond with resp.
a wellare ethic, a rights ethic, an environmental ethic and an ethic ol care.
1!.1.2 Applications ol the live lingers
1!.1.2.1 1he lingers applied to the consumption ol animal products
1he live moral lingers can be applied to the production and consumption ol
animal products (meat, lish, eggs, dairy, leather, lur,.).
-1he lorelinger: compared to humans, livestock animals are in the worst-oll
position due to sullering and early death. 1he loss ol liletime well-being ol the
livestock animals is worse than the loss ol well-being that humans would
experience when they are no longer allowed to consume animal products.
livestock and lisheries violate the lorelinger principle ol well-being.
-1he middle linger: the consumption ol animal products almost always involves
the use ol animals as merely a means, hence violating the mere means principle ol
the middle linger.
-1he ring linger: animal products are not necessary lor humans, because well-
planned vegan diets are not unhealthy (according to the Academy ol Nutrition &
uietetics, AuA, 2009). liodiversity will not decrease when we would stop
consuming animal products (on the contrary, according to LN lAO the livestock
sector is likely the most important cause ol biodiversity loss). lence, the value ol
biodiversity cannot be invoked to justily the consumption ol animal products.
-1he little linger: we would never tolerate the degree ol partiality that is
required to justily livestock larming and lishing. lence, tolerated partiality cannot
be invoked to justily the consumption ol animal products.
lt lollows that veganism is ethically consistent, and the production and
consumption ol animal products are ethically inconsistent.
-1he thumb: give the good example, even when other people continue
consuming animal products. lrom this principle, it lollows that veganism is a
moral duty.
1!.1.2.2 1he lingers applied to the problem ol abortion
1he above section nicely demonstrated that all the live lingers are relevant in the
problem ol consumption ol animal products. most other ethical issues only require
one or two ol the live principles (e.g. only the thumb and lorelinger are sullicient
to argue lor gay marriage). lut there is another ethical issue that can only be
grasped with all live lingers: the problem ol abortion. 1his problem nicely
tnica| ccnsistcncy anJ Anima| ua|ity
280
illustrates how the live lingers work, so as an illustration let us apply the dillerent
lingers to the problem ol abortion.
-1he lorelinger: early abortion is allowed, when the letus is not yet sentient.
Once a letus has developed the capacity to leel, it becomes a sentient being.
Aborting this sentient being will result in a very low liletime well-being (value ol
lile) due to the short lilespan ol the aborted letus. 1he letus is in the worst-oll
position, so should get a strong priority lor an increase ol liletime well-being. 1hat
means that late abortion (when the letus is sentient) is not allowed.
1

-1he middle linger: il a woman does not want the pregnancy, and as her body is
necessary lor the letus to survive (and il the woman is not responsible lor this
dependency, e.g. when she was raped), we can say that the letus (unconsciously)
uses the pregnant mother as merely a means. ll the woman aborts her letus, we
cannot condemn her without considering her as merely a means lor the ends ol
the letus. 1his violates the extended mere means principle. 8o we should at least
tolerate abortion, even il the letus is already sentient (and even il the letus would
have higher mental capacities such as rationality). 1his line ol reasoning is also
rellected in 1homson's argument ol the lamous violinist to delend abortion
(1homson, 191). lmagine that you are kidnapped by music lovers who connected
your body to an unconscious very lamous violinist. 1he violinist has a special
disease that takes nine months to cure, and you are the only person who can save


1
1here are exceptional cases, e.g. when the luture child will be seriously disabled, or when the
pregnancy involves serious health risks. 1hese have to be taken into account in the prioritarian
weighing ol well-being. Also, a letus has a very low psychological connectedness with hisJher luture
(see section 4.2.). 8ome impartial observers behind a veil ol ignorance might therelore have dillerent
estimates ol hisJher liletime well-being (value ol lile) or ascribe a low connectivity lunction between
the letus and hisJher luture momentaneous minds. 1his means that extending the lile ol the letus
should not get such a strong priority. (8ee also mcmahan, 2002.) Yet, abortion not only shortens the lile
ol the letus, but also prevents the existence ol all luture momentaneous minds and persons that the
letus could become. when these possible luture persons can no longer contribute to the wellare
lunction, the wellare lunction might decrease too strong. more mathematically: when a letus is aborted
at time t, its integrated well-being

is a little bit lower than without abortion (it is only a little bit,
due to low connectivity). As a result, this decrease ol integrated well-being due to abortion decreases
the wellare lunction a little. lut with abortion, something else happens that more strongly decreases
the wellare lunction. without abortion, the letus will become luture persons, and hence the wellare
lunction also includes an integral over all those luture integrated levels ol well-being

lor .
when abortion is perlormed, those luture integrated levels ol well-being no longer count, and this
results in a much stronger decrease ol the wellare lunction. ln other words: an abortion does not
strongly harm the letus (lowering its integrated well-being

a little bit), but it can still strongly


harm (decrease) the wellare lunction as a whole. 1his strong decrease ol the wellare lunction counts as
an impersonal harm (mcmahan, 2009).
1nc mcra| nanJs
281
him. ll you unplug yoursell lrom the violinist, he will die. most people have the
intuition that unplugging is permissible.
-1he ring linger: procreation is natural. One reply to 1homson's delense ol
abortion (the example ol the lamous violinist) is based on the natural-artilicial
distinction. According to larks (2006), pregnancy and procreation are natural, and
this is dillerent lrom the artilicial treatment to save the lamous violinist. As
procreation is natural, normal and necessary, one might say that the letus is
allowed to use the mother as merely a means.
-1he little linger: some level ol partiality is allowed. As prey are allowed to
delend themselves against being used by predators (even il that implies the death
ol the predator), we can state that also pregnant women are allowed to delend
themselves against being used by letuses. A doctor is allowed to be partial and to
choose lor the woman (i.e. help to perlorm abortion) as well as lor the letus (i.e.
reluse abortion).
1he result ol balancing the above lour principles (the lour lingers) is that early
abortions are permissible (and the doctor is allowed to reluse to help). when the
pregnancy is already very advanced, abortion might not be permitted. ll e.g. the
pregnancy takes only a lew more days, the extra use as merely a means during
those lew days will be low. lt will be the case that the violation ol the mere means
principle (the middle linger) becomes so low, that it will be canceled by the
naturalness principle (the ring linger) which permits the use as merely a means
lor the ends ol the letus. Alter this cancelation, what is lelt are the lorelinger and
little linger principles. 1he priority lor the liletime well-being ol the letus (the
lorelinger) will be very high, so it might become too partial to choose lor abortion
(the little linger principle might be too weak to justily abortion).
1!.1.2.! 1he lingers applied to the environmental problem
As a third example, we can look at what the moral hand says about the twin
environmental problem ol overconsumption and overpopulation. lmagine there is
a planet, the larth, that contains both moral agents and amoral sentient beings.
1he moral agents are the people who can rellect on their own behavior and can
have a strong inlluence on their own consumption and reproduction levels. On
this planet larth, the moral agents appear to have very high potential levels ol
liletime well-being, because they have a rich emotional lile, a long lilespan and a
high psychological connectedness. 1hose moral agents also typically use a lot ol
resources (not only the use ol material resources, but also the use ol the
ecosystem's absorption and processing capacities lor emitted substances).
we can write an equation lor the environmental impact generated by a group ol
moral agents. 1he impact (lm) is the product ol lour lactors: lm=l.A.c.1, where l
tnica| ccnsistcncy anJ Anima| ua|ity
282
equals the population size (the number ol moral agents in the population), A
equals the average allluence (or average liletime well-being ol those moral
agents), c equals the average consumption level ol resources per unit ol allluence
and 1 is a technology lactor which equals the average environmental impact per
unit ol resource consumption (this lactor is determined by the choice ol
technology).
2

lcosystems are very complex, but scientists have derived some rules ol thumb
to determine the ellects ol a high environmental impact generated by resource
consumption. 8ome uselul rules ol thumb are the lootprint indicators. lor
example the ecological lootprint measures the use ol bioproductive area (olN,
2010), and the carbon lootprint measures the emissions ol greenhouse gases.
1hose lootprint indicators each have a corresponding larth's carrying capacity l.
1hese carrying capacities, lor example the total available bioproductive area or
the larth's capacity to absorb greenhouse gases, are linite. ll the lootprint is
higher than the corresponding carrying capacity (i.e. il lm-l), then we can expect
that the current population ol moral agents has a negative inlluence on the
liletime well-being ol luture populations ol moral agents, as well as on current and
luture living amoral sentient beings.
-1he lorelinger: a lot ol current moral agents should decrease their
consumption and reproduction levels. looking at the wellare lunction, we see that
there is a current living population ol moral agents who have a high level ol
liletime well-being, but this population also generates an environmental impact
that is higher than the carrying capacity (lm-l, see olN, 2010) and hence
decreases the levels ol liletime well-being ol other current and luture individuals.
Also, the number ol sentient beings on larth is very high, so the wellare lunction
reduces to the average ol priority weighted levels ol liletime well-being (the
population lactor in the wellare lunction is close to 1, see appendix 2 1he
impartial observer behind the veil ol ignorance"). ll the environmental impact ol
the current moral agents decreases the liletime well-being ol other individuals,
this average decreases and hence the wellare lunction decreases.
1o reduce the environmental impact, the current population ol moral agents
has lour options, relerring to the lour lactors in the impact equation
!
: 1) decrease


2
1his equation is inspired by the lamous llA1 equation ol lhrlich and loldren (191).
!
Note that lowering the well-being ol luture people is a kind ol impersonal harm. 1he current moral
agents have two options: reduce or not reduce their environmental impact. ll they do not reduce the
impact, the luture will contain a population A that will have low levels ol well-being. ll on the other
hand the current moral agents reduce their impact, the luture will contain a Jijjcrcnt population l,
where the individuals have a higher liletime well-being compared to the individuals in population A.

1nc mcra| nanJs
28!
population size, 2) decrease liletime well-being, !) decrease consumption levels
and 4) decrease the impact ol resource consumption.
1he most ethical approach lor the lirst option consists in creating lair
opportunities lor a voluntary pregnancy restriction (by e.g. education lor women
and a good access to contraceptives and services lor sexual and reproductive
health). 1he second and third options reler to a lilestyle ol voluntary simplicity.
1hese two options imply that the moral agents should lirst cut on their resource
use that does not strongly contribute to their liletime well-being. uecreasing the
consumption lor luxury needs is a good starting point, because luxury needs (such
as resource intensive social status symbols) are delined by the lact that society can
create new circumstances where those needs no longer need to be satislied in
order to have an increase in well-being (see section 6.). 1he lourth option can be
done by technological innovations and scientilic research.
we also observe that on planet larth, the current population ol moral agents
has a lertility rate higher than the replacement level ol roughly 2,1 children per
lemale human. ll the lertility rate remains that high, the luture populations ol
moral agents will show an exponential growth. 1hat is again unsustainable. As a
consequence, luture populations ol moral agents and amoral sentient beings will
receive much lower levels ol liletime well-being. 1hat lowers the wellare
lunction.
4
ln the end, the lertility rate ol the population ol moral agents should
drop to the replacement level.
-1he middle linger: stop the consumption lor luxury needs. luxury needs not
only generate a high environmental impact, but they also result in a violation ol
the basic right (as was discussed in section 6.): we have to avoid the use ol non-
sentient living beings lor luxury needs.
-1he ring linger: a decrease in environmental impact decreases the loss ol
biodiversity. 1his is another important reason why current moral agents should
avoid overconsumption and overpopulation. 1he ring linger also says that even il


lopulation l is dillerent (the people in l are not the same people as in A), because the choices that the
current moral agents make inlluence who will be born in the luture. 8uppose that the liletime well-
being in population A is still higher than 0, i.e. the lives in this population are still worth living.
Nevertheless, population A sullers a harm, but this harm is impersonal because the alternative (when
the current moral agents reduce their impact) would be that the luture population A would not even
exist. 1he harm consists in the lowering ol the wellare lunction. 1he lorelinger uses the wellare
lunction, and hence looks at impersonal harms.
4
Note that - lor large populations - the population lactor in the wellare lunction is close to 1 (see
appendix 2 1he impartial observer behind the veil ol ignorance"). ll - lor smaller populations - this
lactor would be close to NJN
k
, an increase in the population size might increase the wellare lunction.
tnica| ccnsistcncy anJ Anima| ua|ity
284
the current planet larth is overpopulated by overconsuming moral agents,
everyone is allowed to procreate.
-1he little linger: moral agents are allowed to be a bit partial, but should relrain
lrom causing more harm by overconsuming resources. 1he little linger cannot
justily the current levels ol resource consumption.
-1he thumb: every moral agent should give the good example by lollowing a
universalizable rule. lor example having more than two children should be
avoided as long as the lertility rate ol the population ol moral agents is higher
than the replacement level. A moral agent cannot lind a universalizable rule that
allows lor having more than two children in the current situation where the
lertility rate is too high.

1he moral agent is allowed to lollow a rule like lveryone


may have as many children as one likes, as long as the lertility rate is not above
replacement level." 1his is comparable to the universalizable rule lveryone may
take a train that one prelers, as long as there is some place available on the train."
1!.1.! lntermezzo: maps ol the moral landscape
lour ol the live lingers can be expressed in a mathematical equation ol the moral
weight (see appendix 2): V-w-k-8. 1he lirst term w is the wellare lunction ol the
lorelinger. 1he second term k represents the violations ol the basic right ol the
middle linger, as well as the tolerated partiality principle ol the little linger. 1he
third term 8 corresponds with the value ol biodiversity. 1his moral weight
combines dillerent moral lorces into a quantity that should be maximized, just as
the standard model ol physics combines dillerent physical lorces into a quantity
that is extremized (a lagrangian, see weinberg, 1996).
1he thumb principle should be applied to this moral weight: derive those
universalized guiding rules that result in a best situation under universal
compliance. we should not try to maximize the moral weight directly by our
actions. lnstead, we should lollow those rules that, when those rules are lollowed
by everyone who is capable ol lollowing them, maximize the moral weight.
1he moral weight is a multidimensional lunction ol controllable variables that
can be inlluenced by moral agents when those moral agents select rules. 1he
controllable variables are e.g. distributable goods and liberties. 1hese controllable

1o decrease the environmental impact, l suggest that lor a period ol time the lertility rate ol the
population ol moral agents should drop below the lertility rate, such that the population size can
decrease to a sullicienty low level. Alter that, the lertility rate should increase to the replacement level
to reach a sustainable steady state.
1nc mcra| nanJs
28
variables and universalized rules are embedded in the moral weight as lollows.
1he moral weight can be written as V=V(x(n(v(r)))), i.e. the moral weight is a
lunction ol values x which represent e.g. the levels ol liletime well-being, the
strength ol basic rights violations and the amount ol biodiversity. 1he values x are
lunctions ol the world histories n (see appendix 2). 1hese world histories are
dependent on the controllable variables v. linally, these controllable variables are
lunctions ol the universalized rules r.
lt is the rules that we (moral agents) have to select. 8o we lirst start with
selecting a rule. 1hen we derive what distribution ol controllable variables we
would get il the rule is universalized. lence, these universalized rules determine
the distribution ol the controllable variables. Next, a chosen distribution ol
controllable variables generates a number ol possible world histories (there can be
more than one world history lor a unique choice ol controllable variables, because
the world can contain probabilistic uncertainties). 1aking expectation values over
those possible world histories gives us the values liletime well-being, basic rights
violations and biodiversity.
1his moral weight lunction can be represented as a moral landscape, with peaks
and valleys
6
. 1he peaks correspond with the best situations (e.g. the best
distribution ol goods and liberties).
As an example, take the controllable variable v that corresponds with a
behavior that is natural, normal and necessary. 1his variable might represent e.g.
the level ol predation or the level ol procreation that moral agents can control. ll
predation is universally prohibited, il the selected universalized rule r is stop
predation always and everywhere", then we move to one end in the moral
landscape as the variable v goes to zero. 1hen the 8-term becomes very low
because a lot ol biodiversity gets lost. 1his decrease ol the 8-term outweighs the
increase ol the w- and k-terms, and hence the moral weight decreases under
universal compliance ol a predation prohibition rule. A duty to stop predation
moves us to a valley on the moral landscape.
Also, the 8-term might have another very special property to llatten the moral
weight lunction in some areas, such that e.g. V(v)=V() lor two levels ol predation
v and within an interval v. 1his means that we do not have a duty to (but we
are allowed to) decrease the level ol predation lrom to v. 1he more natural,
necessary and normal a behavior is, the wider the range might be ol the interval


6
1he concept ol moral landscape" introduced by larris (2010) might correspond with a moral weight
that only contains the wellare lunction w in a simplilied lorm that represents sum-utilitarianism
instead ol quasi-maximin prioritarianism.
tnica| ccnsistcncy anJ Anima| ua|ity
286
v that corresponds with the level ol that behavior. 1his means that these types ol
behavior have a wide range ol permissibility: we are allowed to increase or
decrease those levels within that range.
1he moral landscape allows us to visualize what kinds ol actions are obligatory,
permissible and impermissible. 8uppose that il you do not do something (e.g. you
do not help or harm anyone), we are at a specilic point in the moral landscape, on
a mountainside. ll you harm others, you push location downwards to the valley on
the moral landscape. ll you help others, the location moves upwards and climbs
the mountain. my intuition says that no-one should move the location downwards.
moving downwards is impermissible. my intuition also says that in moving
upwards, we are allowed to choose in what direction the location can move
upwards. 1his is the little linger principle ol tolerated partiality (section .1). Any
upward moving direction is permissible.
lut sometimes some upward moving direction might be obligatory instead ol
merely permissible. 8ometimes we should take a certain prelerred path to climb
up the mountain. 8ometimes we have a duty to help others in a prelerable way.
when are you obligated to help in a specilic way, even when you do not want to
help in that way7 1his question can be answered by the lollowing procedure. ll
someone lorces you to help in that way, sJhe uses you as merely a means (you
have to do something that you do not want, and your presence is required to help
others), and that means that the k-term adds a negative amount to the total moral
weight. All else equal, this would result in a downward movement on the moral
landscape. lut by helping others, the w-term adds a positive amount, which
means an increase ol the total moral weight and an upward movement. ll this w-
term outcompetes the k-term (il the increase in the w-term is larger than the
absolute value ol the change in the k-term), it means that the increase in wellare
trumps the violation ol your basic right not to be used as merely a means. ll that
would be the case when you are lorced to help, it implies that you have an
obligation to help in that way.
we also have to consider the problem that the moral landscape is not unique
and objective (impersonal): each moral agent might have hisJher own prelerence
lor the parameters in the moral weight. 1hese parameters correspond with e.g. the
level ol risk aversion or the relative strength ol the k-term. Not everyone needs to
have the same levels ol risk aversion, need lor elliciency, estimates ol well-being,
or intuitive balancing choices between the strengths ol dillerent principles. Also
the choice ol wellare lunction can be dillerent among moral agents. 1herelore,
each moral agent might look at hisJher own map ol the moral landscape. 1hose
dillerent maps might indicate dillerent locations ol the peaks and valleys. 1he
highest peak on moral agent a's map might also be much higher than the highest
1nc mcra| nanJs
28
peak on u's map. No map is the unique and objective one. No moral agent can be a
dictator about e.g. the level ol risk aversion that one should have.
1here is a procedure that allows lor democratic assessments between the
prelerences ol all moral agents (appendix 2, uemocratic impartial prelerences ol
moral agents"). lirst, all moral agents calculate their own moral weights and
generate their own maps ol the moral landscape. 8econd, lor each moral agent a
weighted moral weight is calculated: the wellare lunction

ol moral agent a is
divided by its maximum value

. ll we now look at the moral landscapes


representing the weighted moral weights according to moral agents a and u, we
see that the highest peak on a's map is as high as the highest peak on u's map. ln
other words, the optimal situation according to moral agent a is as valuable as the
optimal situation according to u. ln the third and linal step, a democratic average
ol all weighted wellare lunctions ol all moral agents is taken:


All prelerences ol all moral agents (i.e. all persons who are able to do the
exercise to derive what should be done) can and should be taken into account
equally (democratically), making the theory more objective (impersonal).
1his democratic procedure removes an important arbitrariness ol the theory:
one could object that my selection ol parameters (e.g. the parameters that
measures my risk aversion or my estimate ol the strength ol the basic right) is
arbitrary. why not take another level ol risk aversion7 1he democratic procedure
implies that everyone's prelerences should be taken into account. 1his also means
that e.g. the prelerences ol a utilitarian (i.e. no risk, loss and uncertainty aversion
behind a veil ol ignorance, no strengths lor the mere means principle and the
tolerated choice principle, and no value ol biodiversity) can be included. lence, it
is possible to include the moral choices ol someone who wants to delete some
lingers (e.g. delete the deontological principle ol the middle linger) and simplily
the remaining lingers (e.g. simplily the lorelinger principle).


constructing a huge number ol maps (generated by weighted moral weights ol
all moral agents), each map representing a vast, multidimensional moral
landscape, and taking the democratic average ol those maps, will be very
complicated in real lile. 1herelore, in daily lile it is better to work with simpler

ueleting the thumb principle without opening the door lor huge levels ol arbitrariness, might be
impossible. At least some version ol a universalization principle should be preserved.
tnica| ccnsistcncy anJ Anima| ua|ity
288
rules ol thumb to roughly approximate the location ol mountainous areas.
8
And
we should set priorities to act against the greatest lorms ol injustice, such as
poverty and animal abuse. lt is clear that those lorms ol injustice are really lar
away lrom any peak. lven il we can't determine the location ol the highest peak
on the moral landscape, we do know that we better move to a mountainous area
instead ol remaining in a low area.
1!.2 A second moral hand ol meta-ethics
Alter removing the arbitrariness ol the choice ol parameters, only one important
arbitrariness remains: why not take other moral lingers, why not include more
principles, why not add more terms to the moral weight7
lerhaps such inclusions are possible and are allowed, under one important
condition: they should be based on a coherent set ol strong, shared moral
intuitions which are translated in clear, universalized ethical principles. And that
is not easy. lncluding a sixth principle to the moral hand might be incoherent with
background theories (e.g. some scientilic lacts), moving the new theory away lrom
a wide rellective equilibrium and turning this sixth principle into a moral illusion.
An ethical system consists ol ethical principles that impose conditions on our
behavior. 1he question is which ethical principles lorm a good, coherent ethical
system. Are there rules to determine what ethical systems are good7 1his is a
meta-ethical question, because it is about rules about rules: meta-ethical rules that
determine which moral rules ol conduct are good. 1he meta-ethical hand is a
metaphor lor live meta-ethical ground rules lor constructing a good ethical
system. Numerous ethical systems can be constructed. 1o avoid an anything goes
ethical relativism as much as possible the meta-ethical hand imposes strong
conditions.
An example ol a concrete ethical system which was constructed with the meta-
ethical hand, is the abovementioned system ol the normative moral hand. 1he
principle ol rule universalism (the thumb principle ol the moral hand) can be
extended to a meta-ethical level: it does not only apply to the choice ol actions or


8
Again, we can compare this with physics. 8olving the complete lagrangian ol the universe will be very
dillicult. 1his impractical dilliculty does not imply that the standard model is wrong. lnstead, physicists
use approximations (e.g. mean lield theory) to simplily things.
1nc mcra| nanJs
289
behavioral rules (ll l am allowed to do something or lollow a rule, then so are
you."), but also to the choice ol ethical systems (ll l am allowed to construct and
lollow an ethical system, then so are you."). 1he other lour lingers ol the
normative moral hand strongly restrict what kind ol action or rule we are allowed
to lollow. 8imilarly, there are lour extra requirements that place strong
constraints on the kind ol ethical system we are allowed to construct. 1hese lour
meta-ethical principles generate the other lour lingers ol the meta-ethical hand.
lence, there are two moral hands: the above mentioned moral hand with live
moral principles ol ncrmativc ctnics, and a second moral hand ol mcta-ctnics. Note
the similarities between the live meta-ethical principles given below and the live
normative principle ol the moral hand given above.
1he thumb: the principle ol rule universalism. when constructing your ethical
system, you must lollow those rules that everyone must lollow in similar ways
when constructing their ethical systems. when constructing your ethical system,
you may lollow those rules that anyone may lollow in similar ways when
constructing their ethical systems. An ethical system is a set ol universalized
ethical principles, applicable to all (real and hypothetical) situations.
lor example: il you can rely on your intuitions, then everyone can rely on their
own moral intuitions. ll no one may introduce ad hoc principles or larletched
rules at will, then neither do you.
1he thumb principle is a very abstract principle that not yet decides what rules
one must lollow when constructing an ethical system. }ust like we have to place
the thumb against the other lingers in order to grasp a construction tool, we have
to apply the meta-ethical thumb ol rule universalism to the other meta-ethical
lingers in order to construct an ethical system.
1he lorelinger: compatibility and agreement with basic inlormation. lasic
moral judgments lorm the basic inlormation in the construction ol an ethical
system. lasic judgments are lor example moral intuitions that olten
spontaneously emerge in concrete situations or thought experiments. 1he basic
inlormation also includes background theories such as reliable scientilic
knowledge (lor example about biological species and evolution
9
, well-being and
consciousness
10
,.). Non-scientilic (e.g. pseudoscientilic or religious) ideas should
be excluded lrom the set ol basic inlormation. with this condition, the ethical
system will be in line with science.


9
1he ethical system should respect the uarwinian lact that species do not have essences.
10
1he ethical system should respect the scientilic lacts about personal identity, degrees ol
consciousness,.
tnica| ccnsistcncy anJ Anima| ua|ity
290
1he lorelinger basically says that ethical principles must reler to the basic
moral judgments as good as possible and that we should give a strong priority to
the strongest moral judgments and most reliable empirical beliels. 1he strength ol
a basic judgment is determined by our willingness to give up the judgment when it
conllicts with other judgments: il we do not lind it so bad that the basic moral
judgment does not lit into the constructed ethical system, then it is a weak basic
judgment.
1he middle linger: completeness and internal consistency. lach situation
should generate one and only one linal moral verdict. A linal moral verdict is
generated by the ethical principles when everything is taken into account.
consistency means not (p and not p)". lor example, a behavior in a specilic
situation cannot both be allowed and prohibited at the same time. 1here should be
no 'true' or 'hard' dilemmas in an ethical system: no judgment that something is
both obligatory and not obligatory, all things considered. ll p" is equal to not
(not p)", then lrom consistency lollows completeness: p or not p". 8o in any
situation an act is either allowed or not. 1he ethical system should be able to
generate a unique answer to the question which actions are permitted, prohibited
and obligatory. 1his goes lor each possible action in each possible situation.
1he middle linger is the longest linger, so consistency is the most important
condition in the construction ol an ethical system. lnconsistent systems are not
valid.
1he ring linger: clarity. 1he ethical principles in the ethical system should be
clearly lormulated, so that they can be understood by everyone (who has the
capacity ol understanding) and they can always be applied without ambiguities.
1he meaning or interpretation ol moral terms should therelore be clear.
1he little linger: parsimony and simplicity. }ust as the little linger can deviate a
little bit lrom the other lingers, one may add additional, deviating ethical
principles in an ethical system to a limited degree. One has to avoid as much as
possible any artilicial ad hoc adjustments (lor example exceptions to exceptions to
rules, or rules that are restricted to a specilic situation). One may therelore
introduce only a little bit ol complexity or artiliciality, provided one is willing to
tolerate everyone else adding artiliciality to the same degree in the construction
ol their ethical systems (everyone, because one has to place the thumb against the
little linger). 1his parsimony principle is an example ol the philosophical principle
ol Occam's razor.
1hese lingers are held together by the palm, which says that one must show
goodwill in constructing an ethical system, without arbitrariness and cognitive
bias.

1nc mcra| nanJs
291
Note that the lingers ol the second moral hand correspond with epistemic
virtues ol scientilic research (luhn, 190). lence, constructing an ethical system is
similar to the way one ought to do science: deriving clear and mutually consistent
principles (e.g. natural laws) lrom basic inlormation (experimental data), thereby
minimizing ad hoc constructions to the theory. A scientilic theory should be as
parsimonious as possible (the little linger), and should consist ol clearly delined
laws (the ring linger) that are consistent with each other (the middle linger) and
correspond as close as possible to the most reliable experimental data (the
lorelinger).
1!.2.1 An analogy with crossword puzzles
constructing a coherent ethical system is like solving a crossword puzzle. A white
box ol a crossword symbolizes a particular situation or a moral point ol view. A
letter corresponds with a linal moral verdict: an answer to the question what we
may or should ultimately do in that particular situation, or what - all things
considered - is valuable lrom the moral point ol view. A word correspond with a
universalized ethical principle.
1he thumb: equivalent solutions ol a crossword puzzle are equally correct,
provided that they respect certain rules ol the game to solve the puzzle.
1he lorelinger: the completed words must reler to the given descriptions (the
basic inlormation). 1his also implies that a large crossword puzzle can be more
coherent than a small puzzle. lence, extending a 'moral crossword puzzle' by
introducing new moral thought experiments and testing moral intuitions in exotic
cases might increase coherence il these tests are successlul.
1he middle linger: in a white box you must lill in one and only one letter.
consistency means not both a letter and a dillerent letter. completeness means
either a letter or a dillerent letter (so no empty white box).
1he ring linger: the words must lorm existing, clear words.
1he little linger: one has to avoid new words, larletched words, incorrectly
written words or ad hoc adjustments to words as much as possible, and give a
prelerence to the most common words.
1he palm: one should not arbitrarily lill in some letters in adjacent white boxes,
one should not arbitrarily change some given descriptions.
1!.2.2 live principles ol anti-arbitrariness
Anti-arbitrariness (or regularity) is an overarching theme in the meta-ethical
hand: it is present in all live lingers. }ust as the normative moral hand creates live
tnica| ccnsistcncy anJ Anima| ua|ity
292
kinds ol equality (anti-discrimination), so does the meta-ethical hand create ol live
kinds ol anti-arbitrariness. lence, meta-ethical anti-arbitrariness is analogous to
moral equality (anti-discrimination).
1he thumb: democracy ol ethical systems. All equally coherent ethical systems
are equivalent lrom a meta-ethical point ol view. ll dillerent people adhere to
dillerent but equally coherent ethical systems, those people should seek to achieve
an acceptable compromise through a democratic decision procedure, because no
one can argue that their own ethical system (based on their own moral intuitions)
are more important than those ol others.
1he lorelinger: one should not arbitrarily give weaker moral intuitions stronger
priority, one should not arbitrarily change or exclude basic moral judgments. ln
the crossword puzzle analogy, one should not arbitrarily say that this word has to
match this given description, whereas that word does not have to match that
description.
1he middle linger: one should not arbitrarily allow inconsistencies and gaps in
the ethical system. ln the crossword puzzle analogy, one should not arbitrarily say
that this white box can contain two letters, that one can contain none, and the
others just one.
1he ring linger: one should not arbitrarily introduce a vague ethical principle
that one can interpret and apply arbitrarily in concrete situations. ln the
crossword puzzle analogy, one should not arbitrarily say that in this lelt corner
you can write a newly invented word whereas there in the right corner you
cannot.
1he little linger: one should not arbitrarily add artilicial, complex, ad hoc
constructions to the ethical system. ln the crossword puzzle analogy, one should
not arbitrarily say that it is permissible to change the spelling ol this word a little
bit.
1!.2.! Applications ol the meta-ethical hand
let us look at some examples how to apply the live lingers ol meta-ethics. As a
lirst example, we can apply the thumb to the little linger: ll you are allowed to
deline discrimination in a way that it relers to an arbitrary group (e.g. humans),
then l may also pick an arbitrary group lor my ethical rules, principles and
delinitions. my prelerred group might exclude you, which you cannot want.
1herelore, relerring to arbitrary groups in rules, principles and delinitions is not
allowed.
8imilarly, introducing a principle that prohibits gay sex and marriage violates
the little linger, because we can all too easy ask the question: why gay sex7 One
1nc mcra| nanJs
29!
could try to justily a prohibition on gay sex and marriage by relerring to a
principle ol (sexual) purity. lut then we get into trouble with the ring linger,
because this concept ol purity needs clarilication (as l clarilied the concept ol e.g.
naturalness). And then this clarilied principle needs to be universally tested to see
whether it is compatible with other important principles and intuitions. l doubt
whether the notion ol purity can be clarilied and universalized to make it as clear
and coherent as the live principles that l derived. 1he new anti-gay principle
might easily get into conllict with scientilic lacts about gay people. 1herelore it is
unlikely that someone could simply include a principle (a sixth moral linger) that
prohibits gay sex and marriage, because inconsistencies or strong levels ol
arbitrariness might appear alter critical rellection and attempts to universalize
this new principle.
1he little linger (together with the lorelinger) also has as a result that scientilic
lacts can inlluence the ethical principles. 8uppose lor example that an ethical
principle states that there is a sharp dillerence between individuals with moral
status and those beings lacking moral status. 1his binary property ol moral status
cannot be matched non-arbitrarily to a natural property that comes in degrees.
lor example biological science can come to the conclusion that being human is a
matter ol degree (look at human ancestors and the possibility ol hybrids, chimeras
and genetically modilied humans). As a result, matching an all-or-nothing
interpretation ol moral status with the natural property ol being human cannot be
done, except when one introduces an artilicial or arbitrary cut-oll point lor being
human. ln the ethical system that l constructed in this dissertation, l always
avoided matching a discrete (e.g. binary) property to a gradual property. ln that
sense, scientilic lacts (e.g. about gradual evolution, personal identity or levels ol
complexity and mental capacities) had some inlluence in the construction ol the
ethical system.
As another example, we can apply the thumb to the lorelinger, lrom which we
can derive that an ethical system cannot be based on e.g. the 1en commandments.
ll you are allowed to base your ethical system on ideas (e.g. the existence ol a
christian ood) that lack evidence, l am allowed to do so as well. 8o l am allowed to
invent things that equally lack evidence. lor example l can introduce another ood
with other commandments that will harm you, and you cannot want that. 1he
beliel in e.g. a christian ood is too arbitrary (it violates the little linger): the
evidence lor the existence ol lrishna, Apollo or 2eus is as high as the evidence lor
the existence ol ood. ln this sense, all theistic believers are in lact inconsistent
atheists: they do not believe in all the other gods, they are not willing to have
blind laith in another god, they are not open-minded towards another god. 8uch
levels ol arbitrariness and inconsistency are not tolerable in ethics. 1herelore,
tnica| ccnsistcncy anJ Anima| ua|ity
294
there is no room lor religious laith-based ethical principles in the construction ol
an ethical system.
1he lirst moral hand ol normative ethics required some intuitive balancing. lor
example the mere means principle ol the long middle linger is stronger than the
tolerated partiality principle ol the small little linger. 8imilarly, the moral hand ol
meta-ethics requires an intuitive balancing between the dillerent criteria lor a
good ethical theory. lor example the consistency requirement ol the middle linger
is stronger than the simplicity requirement ol the little linger.
1he most important point is that the meta-ethical moral hand should set very
high standards lor a good ethical system. ligh standards are required to restrict
the number ol possible ethical systems, to avoid an 'anything goes' attitude
towards ethics and to limit moral relativism. lor example applying the meta-
ethical thumb principle to the lorelinger results in a completeness requirement:
we should use analogies or thought experiments involving all possible situations.
A good ethical theory has to be able to deal with all realistic as well as hypothetical
situations. lven very hypothetical situations count: the better an ethical system is
able to deal with all kinds ol hypothetical situations, the better the system is.
8etting high standards is what l did in my derivation ol a system ol animal
equality. l dealt with very hypothetical situations: what il insects or plants were
sentient7 what il a predator (a lion) acquired lull blown moral agency7 what il we
bred and used mentally disabled humans as slaves7 what il we could teleport and
make (inexact) copies ol persons7 what il the side track in the trolley dilemma
loops back to the main track7 what il human-animal hybrids or chimeras would
exist7
Lsing the ring linger ol the meta-ethical hand, l tried to clarily concepts such as
well-being, biodiversity, naturalness, the body and use as merely a means. ln these
clarilications, l tried to avoid arbitrariness and artiliciality in delinitions. Also the
use ol mathematical expressions should be understood as an attempt to clarily
principles.
ly restricting the complete normative ethic to live principles, l also tried to
make the theory as simple and parsimonious as possible, respecting the little
linger ol the meta-ethical hand. Ltilitarians might complain that my theory
remains too complex. 1heir theory is more economical, but it goes at the cost ol
violating the meta-ethical lorelinger (it has a lower match with moral intuitions
that many people share). 8o here we end up with an important trade-oll between
the lorelinger and the little linger. As the lorelinger is a little bit longer, my
constructed ethical system tends to lean a bit more towards compatibility (with
intuitions) than simplicity.
tnica| ccnsistcncy anJ Anima| ua|ity
296
principles, simply erase one ol the three corners. 1hen we end up with a bar with
two endpoints (two principles). 1his bar can exist in reality. Another possibility is
to redraw the picture, turning one ol the corners in the other direction.
1he question l want to ask to the meat eater is: which corner do you want to
turn7 uo you want to allow cannibalism7 uo you want to allow racism, sexism or
other lorms ol discrimination7 Or do you want to prohibit eating animals7 l am
convinced that most meat eaters would preler the third option, simply because the
other two are much less attractive. lirst, the intuition that we are allowed to eat
meat is much less strong than the intuition that cannibalism or discrimination are
not allowed. Adopting the third principle ol lreedom ol choice is not so dillicult.
we could simply state that we should respect lreedom ol choice, except when the
basic right is violated. 8o we are still allowed to eat and choose what we like, but
we have to restrict our choices a little bit.
8econd, lrom behind a veil ol ignorance, we would not like to live in a world
where cannibalism or discrimination is allowed. ll the meat eater is rational and
cares lor hisJher well-being, sJhe would preler a world where sJhe could not
enjoy the taste ol meat above a world where sJhe might be the victim ol
cannibalism or discrimination.
8o we have two coherent arguments that indicate that it is best to turn the
lower right corner ol the impossible triangle. Or stated in another way: il we (l and
a meat eater) put all our intuitions on the table, we restrict ourselves to those
intuitions that we both share, we give those intuitions a ranking according to their
strength, we translate those intuitions into universalized ethical principles, we
construct a consistent ethical system with those principles whereby we
systematically give priority to the strongest principles, then we end up with a
system ol animal equality. 1hat system implies veganism.
Veganism is a very simple rule ol thumb: do not use animal products. lrom a
political perspective, the technical implementation ol veganism might be rather
easy. Already today, a lot ol animals are subject to wellare laws. 1hese are the
animals that are sentient, according to scientists. 8o we already deal with
sentience in the law. we do not have to change the set ol beings that are subject to
wellare laws, we only need to change the content ol the rights that those beings
have. lveryone who is now already subject to wellare laws (everyone who already
has some wellare rights), should get the same rights as mentally disabed humans
already have. Lse your wildest imagination: what il we treat mentally disabled
humans in the same way that we treat some animals in e.g. medical experiments
or lactory larms7 ll we should not treat those humans in those ways, then we
should also abstain lrom treating in those ways everyone else who is subject to
current wellare laws.


29
where to go lrom here7 Questions lor luture
research
Ol course not all problems related to ethical consistency and animal equality are
solved. low can we reline the argument about speciesism7 low can we strengthen
the method to detect moral illusions7 low can we lurther clarily the basic ethical
principles7 1hese are the rather obvious questions lor luture research. 8o let us
have a briel look at the less obvious but more intriguing questions.
ln have constructed a pluralist, principlist ethical system ol the moral hand,
which contains live ethical principles.
1) why this set ol principles instead ol another pluralist principlist system such
the lour principles ol leauchamp & childress (2001), or the seven principles ol
koss (19!0)7 l think that the live principles ol the moral hand are in lact some
reshullling ol the principles ol leauchamp, childress and koss, where the
reshullling is done such that the resulting live principles can be expressed in a
more compact and yet clear way. l will not lurther elaborate on this issue here.
2) ls live too much or not enough7 consequentialist wellare ethicists would
lavor a simplilied system with only one linger: the lorelinger. 8um-utilitarians
would lurthermore simplily the lorelinger, deleting the priority lor the worst-oll.
kule consequentialists might preler two lingers: the thumb (which relers to the
universalized rules) and the lorelinger. On the other hand (no pun intended), some
deontological libertarians might restrict their ethical system to the basic right
principle ol the middle linger. lut perhaps moral particularists might argue
live is not enough7 lerhaps moral particularists are right: maybe moral judgments
are like esthetical judgments ol music. when you hear a piece ol music, you
automatically judge it to be good or bad, but can you capture all ol your musical
taste prelerences and judgments in a small set ol principles7 uoing that would
seem to be a miracle, discovering the magic potion ol music. lerhaps ethics, like
music, is lar too rich to be expressed in live principles. lerhaps not even hundred
principles will do.
!) what is the strength ol the dillerent principles7 ln other words: how long
and strong are the live lingers7 A sum-utilitarian is a pluralist who gives absolute
strength to only one principle: the lorelinger. 1he other lingers have zero length,
they have zero value. A deontological libertarian gives absolute strength to the
middle linger principle. lut most ol us are real pluralists: dillerent principles have
some non-zero strengths and lingers have some non-zero lengths. low do we
tnica| ccnsistcncy anJ Anima| ua|ity
298
balance the strengths ol the dillerent principles7 ln previous sections l have
suggested a kind ol democratic procedure to solve this problem: every moral agent
has an equal vote to put lorward hisJher intuitive judgments about the relative
strengths ol the dillerent principles. ll most ol us put greater weight to the mere
means principle ol the middle linger, we have to accept this outcome, we have to
accept that the average middle linger is longer than the average lorelinger. 8o,
everyone has his or her moral hand, and we measure the average lengths ol
everyone's thumb, lorelinger and so on to derive the 'average' or 'platonic' moral
hand.
lut there are two problems with this procedure.
lirst, how could we lollow the democratic procedure in practice7 1aking the
intuitive judgments ol all moral agents into account becomes as complicated as
constructing a combustion engine using the standard model ol elementary particle
physics. lut at least we can try to lind some rough moral rules ol thumb to guide
us, just as we use the laws ol thermodynamics.
A second, more intriguing problem is that studies in moral psychology clearly
demonstrated that the intuitive strength ol a moral principle depends on external
circumstances. lor example induced leelings ol disgust (8chnall et al., 2008) and
happiness (Valdesolo & ue8teno, 2006) can inlluence moral intuitions in the
trolley dilemmas, making us more or less reluctant to sacrilice someone in order
to save others. 8o the lengths ol the lingers ol the moral hand depend on whether
or not we are disgusted, happy, tired, snilled some oxytocine, saw a good movie,.
lt is like lingers can grow and change lengths at dillerent speeds. lmagine that we
could 'nudge' someone's moral intuitions, such that we can turn a utilitarian into a
libertarian by tweaking hisJher intuitions. what would this imply lor our
procedure to derive the strengths ol the principles7 ls there a 'neutral state' where
our moral intuitions are not inlluenced by leelings ol disgust or happiness7
4) linally, let me pose the most intriguing question: where did it come lrom7 l
have constructed an ethical system, but what exactly did l do7 uid l discover it or
invented it7 lt is like the question what mathematicians do: are they discovering
mathematical lacts in a platonic world ol mathematics, or are they rather
inventing and constructing theories like engineers do7 Are the moral lingers just
some clever conlabulations and rationalizations ol a bunch ol mysterious
intuitions7 where do these intuitions come lrom in the lirst place7 uo they have
an evolutionary psychological explanation7 uid evolution really have some
inlluence in how l delined the mere means principle or the notions ol well-being
and biodiversity7 1he more l think about this, the more mysterious it all seems.
And yet, l hope that l have made some progress.

299
Appendix 1: a review and systematization ol the
trolley problem
Abstract
1

1he trolley problem, lirst described by 1homson (196) and loot (198), is one ol
the most lamous and inlluential thought experiments in deontological ethics. 1he
general story is that a runaway trolley is threatening the lives ol live people. uoing
nothing will result in the death ol those persons, but acting in order to save those
persons would unavoidably result in the death ol another, sixth person. lt appears
that, depending on the situation, we have dillerent moral judgments about the
permissibility ol action. we will review and systematize all the proposals in the
literature ol the past ! years that have attempted to grasp our moral intuitions in
a simple deontological principle. ln particular, seventeen proposals will be
classilied: six algorithmic, seven psychological, and lour other invalid accounts.
1his review and classilication sheds light on some subtle dillerences and clarily a
lew issues.
lntroduction
1he trolley problem consists ol a series ol moral dilemmas involving a runaway
trolley threatening the lives ol a certain number ol people.
2
1he basic structure ol


1
1his appendix is based on lruers & lraeckman (201!).
tnica| ccnsistcncy anJ Anima| ua|ity
!00
all the dilemmas is the same: il you do not act, live people will die, il you act, one
other person will be killed and the live will be saved. kesearch into the way people
deal ethically with the trolley dilemmas has shown that most people's intuitions
do not correspond either with pure (extreme) deontology or with utilitarianism
(oreene et al., 2001, waldmann & uieterich, 200, lauser et al., 2008). ly 'pure'
deontology we mean here, lor simplicity's sake, that people should comply with
the lollowing rule: never act il the act results in harming people who were not
threatened il you had not acted. ly 'pure' utilitarianism we mean that people
should comply with the rule: always choose the action that maximizes the number
ol lives saved (i.e., least total harm).
uillerent trolley dilemmas have the same consequential structure but yet,
conlronted with those dilemmas, people hardly ever say that one should never act,
or that one should always act. when presented with dillerent dilemmas, most
people say that we must act in one trolley situation, but in another dilemma we
are not allowed to act, it is as il people make inconsistent choices. Only pure
utilitarian consequentialism states that we should always act in all the trolley
dilemmas. 8o most people's moral intuitions deviate lrom these consequentialist
ethics, and therelore the trolley problem is an interesting thought experiment lor
studying deontological ethics. 1he basic question is the lollowing: what is the
morally relevant dillerence between uilemma A and uilemma l, such that it is
morally allowable to act in A, but not to act in l7 Also, in this article we state that
a consistent moral solution ol the trolley problem should contain a clear
description ol a rule or principle that best lits, justilies, and explains the diversity
ol people's moral intuitions in the diverse cases. ln other words, the best solution
to the trolley problem is a clear algorithm to decide whether one should act or not,
and the answers that this algorithm generates should be in line with intuitions.
many people have tried to solve the abovementioned basic question. ln this
article, we present an overview ol the many proposals that ethicists have come up
with during the last ! years, and we discuss their dillerences, mutual relations,
strengths, and weaknesses. ln addition to covering the most relevant versions ol
trolley dilemmas and the solutions proposed in the literature, we also present
some new hypothetical solutions. lut probably the most important contribution ol
this article to the existing literature is a systematic classilication ol all those
solutions. And looking at new trolley dilemmas, we clarily the dillerences between
the proposed solutions (principles).


2
lor a highly readable overview and historic background ol 'trolleyology', see ldmonds (201!).
AppcnJix !: rcvicw trc||cy prcu|cm
!01
1here are several reasons why this new classilication is important. lirst, it
certainly helps to avoid conlusion between dillerent solutions (we will mention
some conlusions in the literature). 8econd, the classilication ol the dillerent
principles gives us insights into which ol those principles and underlying moral
intuitions could be something like 'moral illusions' (e.g., Lnger, 1996). 1hird,
people adhering to deontological ethics might be able to see which proposed
principle they would most preler, that is, which ol the proposals is most
compatible with their own moral intuitions. lourth, our lindings will have
implications lor lurther empirical studies in moral psychology (e.g., oreene, 2002,
cushman et al. 2006, mikhail, 200, oreene, 2008). 1his systematization opens up
some new questions. uo people preler one ol the proposed solutions7 low many
people would agree with which solution7 would they change their judgments in
some dilemmas in order to make them lit with their prelerred solution7 And il
there are dillerent proposed solutions related to dillerent (psychological or
algorithmic) mechanisms, does that mean that there would be more 'moral
modules' in our brains (e.g., the brain research on trolley dilemmas done by
oreene et al., 2001)7
we start this review with a number ol trolley dilemmas that cover all the
important issues and elements that are discussed in the literature. 1hen, we select
six possible solutions to the trolley problem that are described in the literature
(the sixth is in lact a new solution), lrom which we will suggest that these can be
grouped in pairs, so that there are in lact only three groups ol principles with
more or less strong support in the literature. 1hese accounts have an algorithmic
character, with a clearer and more objective decision procedure than the other,
psychological accounts.
Ol course, other hypothetical principles are possible, but they have lew or no
supporters or they remain dubious and are still debated. 8o alter describing the
three groups ol 'algorithmic' accounts, we give an overview ol seven other
proposals that do not distinguish permissibility lrom non-permissibility so clearly.
8ome ol these other proposals might be relevant, as they are more 'psychological'
in nature and psychology strongly inlluences our moral judgments. lowever,
these psychological explanations are not always clear or do not always make a
distinction uctwccn the dillerent 'agent-neutral' trolley dilemmas: they make
distinctions witnin one dilemma. An agent-neutral trolley dilemma is a description
tnica| ccnsistcncy anJ Anima| ua|ity
!02
ol a situation that excludes agent-related inlormation.
!
1he inclusion ol agent-
related inlormation such as a person's position (e.g., distance lrom the victim) or
mental state (e.g., knowledge, risk attitude, intention to harm) give rise to lurther
distinctions within the same agent-neutral trolley dilemma.
Alter the six algorithmic and seven psychological accounts, we linally brielly
highlight lour other proposals encountered in the literature that do not solve the
trolley problem, because they result in pure deontology. ln summary, seventeen
proposals in the literature are classilied as lollows: six solutions that make
objective (algorithmic) distinctions between dillerent dilemmas, seven solutions
that make distinctions within one and the same dilemma, depending on some
psychological state, and lour invalid proposals that always result in pure
deontology in all dilemmas.
8o let's start the trolley's engine.
1he trolley dilemmas
ln this section we will brielly present and systematize the most commonly
discussed versions ol the trolley dilemma (lor lurther details, see 1homson, 198,
lamm, 1989, 1998, Otsuka, 2008, lischer & kavizza, 1992b).
ui|cmma !: 1nc switcn. A trolley is moving towards live people on the main track.
You are standing at a switch. ll you turn the switch, the trolley will be diverted to
a side track, but there is one person on this side track. 1urning the switch will
result in that person's death, and the live people on the main track will be saved.
8hould you turn the switch7 most people (roughly 90/ according to lauser et al.,
2008) say you are allowed to do so.
ui|cmma .: 1nc uriJqc. A lat man is standing on a bridge above the track. You can
save the live people on the track below by pushing the lat man lrom the bridge in
lront ol the trolley, so that the trolley will be stopped by his heavy weight. 1he lat
man will die, and the live people will be saved. Only a lew people (roughly 10/
according to lauser et al., 2008) say that we are allowed to push the lat man. most
people either reluse to push the lat man or condemn pushing the lat man.


!
1he only (trivial) agent related inlormation in all situations is that il the agent acts, it is supposed that
sJhe acts with the intention or plan to save the people on the main track. l.e. malicious intentions (e.g.
to kill a hated person) are excluded.
AppcnJix !: rcvicw trc||cy prcu|cm
!0
dies and is no longer controlling the barrier, can the trolley be blocked and the
live saved.
4

8ix algorithmic accounts
lthicists have looked lor morally relevant dillerences between the above
dilemmas. 1hey want to lind a moral rule that generates answers that are
consistent with the answers (intuitive moral judgments) ol the majority ol people.
As we have seen, the majority ol people are very clear about the lirst two
dilemmas, the switch and the bridge. 1herelore, we think that a good solution
should consist ol a most precise lormulation as possible ol a moral criterion to
distinguish the dilemmas, such that action is at least permissible in the switch
dilemma but not in the bridge dilemma. concerning the other dilemmas, there is
less consensus about people's moral intuitions, so the solutions might diller in
these cases. we preler a moral rule that works like a kind ol algorithm, that is, a
clear procedure applicable to all dilemmas, which provides an unambiguous
answer as to whether action is allowed or not, and without relerence to luzzy or
ambiguous concepts.
1here are live algorithmic accounts proposed in the literature. we will
introduce a sixth. 1hey can be grouped together in pairs, hence, we have
structured them into three groups ol accounts. 1he lirst proposal in each group is
a rather vague account, vulnerable to misinterpretations or borderline cases. 1he
second account in each group is more precise, leaving less room lor interpretation.
ln other words, the second accounts correspond to more accurate interpretations
ol the lirst ones in each group. we will apply these algorithmic accounts to the
above dilemmas and show that these three explanations are dillerent lrom each
other.



4
we note that lippert-kasmussen (1996) gave another dilemma that has a resemblance to uilemma 8,
according to the accounts mentioned in the next section.
tnica| ccnsistcncy anJ Anima| ua|ity
!06
oroup A: the 'mere means' accounts
A!: 0sc as mcrc|y a mcans tc an cnJ. 1he lantian inspired right not to be treated
solely as a means is based on the unalienable dignity ol persons. 1his right trumps
the right to lile ol other persons: it is never allowable to kill and use a person as
merely a means - even il this means that by this act the lives ol others could be
spared. lt is considered disrespectlul to treat someone merely as means. 1his mere
means account is mentioned in 1homson (198), and as we will see in a later
section, it is related to an interpretation ol the doctrine ol double ellect (Quinn,
1989b).
looking at the dilemmas, we can say that the lat men on the bridge (uilemma 2)
and on the loop track (dilemma loop) are used as 'trolley blockers' or 'human
shields.' 8o only in these dilemmas is action not allowed (they violate the dignity
ol the victim). ln the other dilemmas action is permitted.
1his account can sometimes be a bit vague, as it is not always easy to
understand what use as merely a means really is. 8ometimes one might have to use
an element ol lantasy to reler to an analogous means or instrument. we believe
the lollowing account is equivalent to this mere means account, but it provides a
more algorithmic way (a clear test) to decide whether the lantian right is violated
or not.

A.: 1nc ccuntcrjactua| acccunt aucut tnc rcuircJ prcscncc cj tnc victim (mentioned
in 1homson, 198, larlit, 2011). ll the presence ol the potential victim is (causally)
required in order to save the live people on the main track (i.e., il it would be
impossible to save the live people without the victim's body), then you should not
act. lere we can easily decide whether the lantian right is violated, by asking
ourselves what would happen il the one person in the trolley dilemmas was not
present. ll nobody is on the bridge, it is impossible to push someone in lront ol the
trolley to stop it. lut in considering uilemma 1 (8witch), saving the live would still
be possible il the one person on the side track had not been present. ln uilemma 4
(loop and stone) one can still turn the switch and let the trolley be stopped by the
stone. ln uilemma (1ruck) one can still move the truck, even il there is nobody in
it. 1he platlorm (uilemma ) can still be shilted when there is no one standing
next to it. ln uilemma 8 (loop with avalanche) one could still start an avalanche
when the victim on the side track is not present.
According to waldmann and uieterich (200), using a person as a means is not a
criterion that people use, because a lot ol people say that we are allowed to turn
the switch in uilemma ! (loop). lowever, in their study, the person on the side
track was sitting in a bus, and people might think it was the bus that is blocking
the trolley. 8o the bus is used as the means, not the person's body. ll the person
AppcnJix !: rcvicw trc||cy prcu|cm
!0
was not sitting on the bus, saving the live by turning the switch still works.
lurthermore, lauser et al. (2008) noted a slight dillerence in people's responses
between uilemmas ! (loop) and 4 (loop and stone). 1his dillerence can only be
explained by the mere means accounts.
1he mere means account also has another property: il the victim's body needs
to be present (il there is no other heavy object that could replace the person as a
trolley blocker), it also implies that there is logically no possibility ol saving the
one person alter the live people on the main track are saved. Alter turning the
switch in uilemma 4 (loop with stone), you could still try to save the person on the
side track. lut in uilemma ! (loop), saving this person is impossible: even il you
manage to run to the person and pull him away lrom the tracks, you cannot do
this without endangering the live people again. 1his property might point at an
evolutionary explanation ol the moral intuition: in rescuing members lrom your
group, it is advantageous to choose the option that allows you to try to save all ol
them. 8aving everyone is not logically impossible in uilemma 1 (8witch) and
uilemma 4 (loop with stone). You lirst save the live people and then run to the
side track. ll you do not run last enough, you still have saved the live, and il you
can run last enough, you can save everyone. 1he latter is logically impossible in
uilemmas 2 and ! (lridge and loop).
compared with the next lour accounts, the mere means accounts are the most
reliable: they are more accurate, have less boundary cases and generate less
judgments that are strongly counter-intuitive in some dilemmas (the only
counter-intuitive judgment occurs in the loop dilemma).
oroup l: the 'same threat' accounts
1he two same threat accounts that we are about to discuss have something in
common, they both claim that it is only permissible to act il two conditions are
satislied: (1) no new threat is introduced, but a pre-existing threat is redirected or
redistributed lrom the larger to the smaller group (this is also relerred to as the
lermissible uiversion lypothesis in lostow, 1989), and (2) another condition is
satislied - about this latter condition, we will discuss two candidates (related to
rights or interventions), but we expect that they are equivalent.
1he lirst condition is not satislied in uilemma 6 (kockslide) and uilemma
(llatlorm), because electrocution and rockslides are new threats. 8o in these
dilemmas, action is already not permitted. lowever, a problem ol this condition is
its lack ol clarity: it is not always clear when a new threat is introduced. 1here are
some borderline cases ol redirected threats that more resemble new threats (e.g.,
situations where trolleys change alter being sent to a side track). 1hese borderline
tnica| ccnsistcncy anJ Anima| ua|ity
!08
cases might undermine the objective, algorithmic nature ol the same threat
account. leaving this issue aside, let's look at the second condition. 1here are two
versions ol the second condition, leading to two accounts.

8!: vic|aticn cj riqnts. 1homson (196, 198) had the idea that action in the case
ol uilemma 2 (lridge) is not allowable, because pushing the lat man is an
inlringement ol an important right. On the other hand, in uilemma 1 (8witch)
turning the switch does not violate a similar right ol the person on the side track
because you do not do anything to him. ln particular, you do not push him, so his
right not to be pushed is not violated. ln other words in uilemma 2 (lridge), you
do something to a person (which is a violation ol rights), whereas in the switch
you act on the threat (which is not a rights violation). Also in all the loop dilemmas
(!, 4, 8) nobody is pushed, so no right not to be pushed is violated. 1urning the
switch is itsell morally neutral and not a violation ol rights. 1herelore, 1homson
claimed that it is permitted to turn the switch in the loop dilemmas.
1homson's idea has been criticized as being too vague and lor contradicting
moral intuitions (lamm, 1989, lostow, 1989). what rights are we talking about -
the right not to be killed by trolleys or the right not to be pushed7 1he lollowing
describes another candidate condition, which in lact might be equivalent to what
1homson had in mind, but stated a bit more clearly.

8.: 5cnJinq a victim tc tnc trc||cy. 8ome ethicists claim that there is a morally
relevant dillerence between throwing a bomb at a person and throwing a person
at a bomb, or in trolley language, sending a trolley to a person versus sending a
person to a trolley. 1his is relerred to as intervention myopia (waldmann &
uieterich, 200) and locuses at the locus ol intervention: do you in the lirst
instance intervene in the path ol the aggressor (the trolley, the bomb) or the path
ol the victim7 1his criterion has some supporters (loorse, 1994, larris, 2000,
waldmann & uieterich, 200) and some critics (lischer, 1992, lischer & kavizza,
1994). montmarquet (1982) also ollered a same threat principle, but this approach
was criticized by lamm (1989).
Only in the bridge and truck dilemmas does one sends the victim (the lat man
or the passenger) to the trolley. ln these cases, action is not allowed, even il the
threat is the same. we also note that in uilemma (llatlorm), the victim is sent to
a new threat (electrocution), so both conditions ol the same threat account are not
satislied. Yet, we expect that most people's intuitions would allow action in this
dilemma. 1his gives a strong counter example to the same threat account.
lurthermore, there are some boundary cases between sending victims and
threats to each other. As an example, consider a loop dilemma whereby turning
the switch also shilts a platlorm, positioning the victim exactly on the side track to
AppcnJix !: rcvicw trc||cy prcu|cm
!09
block the trolley. Also, according to Lnger (1996, p. 101), the dillerence between
sending the victim to the trolley versus sending the trolley to the victim is an
illusion, based on what he called 'protophysics.'

ln his book, Lnger (1996) also


gives other similar irrelevant protophysical dillerences that inlluence our moral
judgments. lor example: in some dilemmas it is worse to save some people and
harm someone else by increasing the speed ol a trolley than by decreasing it.
oroup c: the 'causal chain' accounts
1he next two principles look at the causal chain that is the result ol action or
inaction. we note that these principles should be taken with a grain ol salt,
because a clear and consistent interpretation ol them might just be impossible il
we think about them more critically. Nevertheless, we present them here.
c!: rincip|c cj (m)crmissiu|c uarm (/u). we cite lamm (1989, in uarwall, 200!,
p. 16), who introduced this hypothesis: lt is permissible to cause harm to some in
the course ol achieving the greater good ol saving a greater number ol others lrom
comparable harm, il events which produce the greater good are not more
intimately causally related to the production ol harm than they are to the
production ol the greater good." 1his is a complicated lormulation that needs
more explication, so let us look at the dilemmas to see what is meant by
intimately causally related."
looking at the switch dilemma, the action is turning the switch, and this action
has two consequences that appear at the same instant in the causal chain: the live
are saved, and the one is threatened by the trolley. 1he production ol the harm
(the threat to the one on the side track) is causally related to the action ol turning
the switch, and also the saving ol the live is causally related to the turning ol the
switch. loth are in this dilemma equally intimately causally related to the turning
ol the switch, because both are the direct consequences ol turning the switch. 1he
condition ol the llJll is satislied, so it is allowable to turn the switch.
ln the bridge dilemma, however, the action is pushing the lat man. As a lirst
consequence, the lat man is threatened, a second consequence is that the lat man

1he loop dilemma is olten used by some philosophers (e.g., 8inger, 200, 8canlon, 2008) to
demonstrate the invalidity ol the deontological mere means account, the abovementioned mere means
principle, by claiming that a lot ol people have the intuition that it is permissible to act in the loop
dilemma, even when the victim is used as a trolley blocker. lowever, il this protophysical explanation
is correct, the judgment in the loop dilemma (the permissibility to act), might be a moral illusion. we
will demonstrate in more detail in another study.
tnica| ccnsistcncy anJ Anima| ua|ity
!10
blocks the trolley and saves the live. lut looking at the causal chain, we see that
the 'causal distance' between the action (the pushing) and the harm (or the threat)
to the lat man is smaller than the causal distance between the action and the
saving ol the live. 1he saving ol the live happens lurther up in the causal chain.
1herelore, llJll says that action is not permitted.
As it is not always clear how to calculate intimate causal relatedness, there are
some borderline cases. ln uilemmas ! (loop), the situation is similar to uilemma 1
(8witch), according to lamm (1989).
6
lowever, we might disagree with this, as can
be seen in 8ituations 2 and ! (ligure 1). ln both the bridge and the loop dilemmas,
the lat man is simply placed in the path ol the trolley, either by changing the path
ol the trolley (8ituation !, loop) or changing the position ol the lat man (8ituation
2, lridge). causally speaking, both are equivalent. 8o we should be a bit skeptical
about this account.
1he sixth possible explanation is not mentioned in the literature (as lar as we
are aware), and perhaps it is identical to an interpretation ol lamm's llJll
hypothesis above. lt is also vulnerable to borderline cases, such as the loop
dilemma.

c.: 1nc incrcascJ tnrcat acccunt. 1his rule says: you are allowed to act il the victim
ol your action dies (is harmed) alter the others are saved. ln the causal chain, we
can represent it as lollows. lach person has a value: 1 equals a|ivc anJ savcJ, 0
means JcaJ (cr narmcJ), X means actua||y tnrcatcncJ (uut sti|| a|ivc), and Y means
pctcntia||y tnrcatcncJ (meaning it is possible to turn a threat towards that person).
1he six persons in the dilemmas all have a value at each step, so we can represent
the starting situation as X,X,X,X,X,Y, that is, live persons are actually threatened
and one person is potentially threatened (we are able to act so that his position
would become a threatened one). 1urning the switch in uilemma 1 changes the
situation to Y,Y,Y,Y,Y,X, which means that one person is really threatened and the
live people are potentially threatened: we can turn the switch back to change the
situation back to the initial situation. 8o the live people are not yet absolutely sure
about their survival. At a particular point, when the trolley passes the bilurcation
in the track, the live people are actually saved and the situation turns into
1,1,1,1,1,X. And alter a lew moments, the person on the side track dies, resulting in
1,1,1,1,1,0. 8o the causal chain in the switch dilemma looks like:


6
ln her later work, lamm (200) introduced new relinements (e.g. causal versus non-causel llip sides,
directly versus indirectly causing a lesser evil, producing versus sustaining a greater good and
substituting versus subordinating persons). lut this was criticized by Norcross (2008) as being heavily
ad hoc and unclear.
AppcnJix !: rcvicw trc||cy prcu|cm
!11
X,X,X,X,X,YY,Y,Y,Y,Y,X1,1,1,1,1,X1,1,1,1,1,0.
Action is allowable il the causal chain looks like that above. lowever, in
uilemma 2 (lridge), matters are more complicated: it all depends on whether the
lat man is heavy enough to block the trolley (i.e. whether the live are delinitely
saved once you pushed the lat man). lt might be the case that the trolley is too last
and is able to kill all six people, because all six people are placed in the trajectory
ol the trolley. ln other words, it is not clear that the live people are absolutely
saved a|rcaJy at tnc mcmcnt wncn cnc pusncs tnc jat man. 1he causal chain now can
look like:
X,X,X,X,X,Y X,X,X,X,X,X1,1,1,1,1,0.
1he possibility ol the X,X,X,X,X,X situation (everyone is in danger),
distinguishes the bridge lrom the switch dilemma. 1he number ol threatened
people is increased. ln uilemma 8 (loop with avalanche) we clearly see a moment
where everyone is in danger: when turning the switch and the trolley passes the
bilurcation, the one person on the side track is threatened, but the live other
people are also still threatened, because the one person is not heavy enough to
block the trolley. Only alter the one person is killed does it become possible to
relieve the threat to the live, by initiating an avalanche that blocks the trolley. 1he
causal chain in uilemma 8 looks like:
X,X,X,X,X,YX,X,X,X,X,XX,X,X,X,X,01,1,1,1,1,0
1he causal chain account is only able to distinguish uilemma 1 (8witch) lrom
uilemma 2 (lridge) il the causal chain can have a point where everyone is in
danger, lor example il we suppose that the train might kill all six people in lridge.
1he existence ol this increased threat situation in the causal chain disallows
action. lut then we have to suppose a similar possibility in uilemma ! (loop),
which disallows action. lowever, uilemma 4 (loop and stone) becomes creates a
boundary case: il the stone was a real mountain, the possibility ol situation
X,X,X,X,X,X is as unlikely as it is in uilemma 1 (8witch). 8o the permissibility
depends on whether the stone is really heavy enough to block the trolley. lut the
same can be said about the lat man, who might be heavy enough to block the
trolley. ll we know the lat man is heavy enough, there is no distinction between
8witch and lridge: alter pushing the lat man, the path ol the trolley changes (it
stops), just as the path ol the trolley changes in the switch dilemma. lushing the
lat man and turning the switch automatically guarantee the immediate salety ol
the live people. we can introduce a distinction by claiming that the agent cannot
be sure whether the lat man is heavy enough, but this turns the algorithmic
account into a psychological account (see the section in risk aversion below).
1he increased threat account reveals a kind ol 'causal myopia' (similar to the
term 'intervention myopia' related to the same threat account, waldmann and
uieterich, 200). ll in the series ol consequences ol your action you do not threaten
tnica| ccnsistcncy anJ Anima| ua|ity
!12
someone belore or at the moment when others are really saved, then you are
allowed to act. lt is as il you were blind to the lurther consequences in the causal
chain.
1able 1 presents the results ol the trolley dilemmas according to the above
three principles: the mere means, same threat, and causal chain accounts. A plus
means that the action is allowed, a minus means that it is not. As discussed, the
causal chain accounts have some question marks.


uilemma mere means
account
8ame threat
account
causal chain
account
1. 8witch - - -
2. lridge
!. loop -
4. loop and stone - -
. 1ruck -
6. kockslide - -
. llatlorm - -
8. loop and avalanche - -
1able 1: answers to the trolley dilemmas, according to the three accounts.

Note that the switch and bridge cases get all plus and minus signs respectively,
so lor these all three accounts can be considered as a solution to the trolley
problem. lut the answers diller when looking at other dilemmas.
8even psychological accounts
1he accounts that we presented above are all objective, in the sense that they did
not reler to mental states, but to events, counterlactual requirements, number ol
threats, points ol intervention, directions or causal consequences. ln this section,
we give an overview ol some other proposals discussed in the literature. 1hese
proposals olten involve some psychological inlluences, such as intentions, risk
aversion, personal versus non-personal conllict, and so on. 1hese psychological
accounts have some llaws: sometimes they are not able to derive a clear judgment
in a certain dilemma (especially the loop dilemma generates problems ol
interpretation), or they do not always make a clear distinction uctwccn dilemmas
AppcnJix !: rcvicw trc||cy prcu|cm
!1!
such as the switch and the bridge. 1hey make distinctions even witnin one
dilemma. 8o depending on the situation (related to the psychological states) it
might be possible that it is not permitted to act in the switch dilemma, or that it is
permitted to act in the bridge dilemma.

!. 1nc ucctrinc cj ucuu|c jjcct (uu). 1his doctrine is mentioned in quite a lew
discussions about the trolley problem (loyle, 1980, uavis, 1984, lischer & kavizza,
1992a, keibetanz, 1998, mclntyre, 2001, 8haw, 2006, ldmonds, 201!). 1he doctrine
says that there is a moral dillerence between the intentional harm as a means and
the loreseen harm as a side-ellect (Quinn, 1989b). lt has been criticized by, lor
example, mclntyre (2001).
1he uul is an agent-centered, psychological account, as it makes a dillerence
between what the agent intends or loresees.

we could try to interpret the uul in


a more agent-neutral way, that is, without too much relerence to the mental states
ol agents. keinterpreting the uul as an agent-neutral principle moves it close to
the mere means account discussed above, because the uul relers to 'harm as a
means.' lowever, we have to be carelul not to conluse the use ol a person's body
as a means versus the use ol, lor instance, a switch as a means or a plan as a means.
1he dillerence between the uul and the mere means accounts can be most
clearly seen in uilemma 8 (loop and avalanche). 1he person on the side track is
not used as a means, because the presence ol his body is not necessary to save the
live (on the contrary, his presence has prevented the initiation ol the necessary
avalanche). lut the agent intends the killing (removal) ol the person on the side
track, because this removal is necessary in order to initiate the avalanche. 1he uul
says that action is not allowed, because it involves an intentional harm.
lence, the uul is not simply equivalent to the mere means account, a lact that
might result in misinterpretations in the literature. lor example costa (1986), in
his application ol the uul to the trolley dilemma, combined (or conlused7) the
mere means account with a version ol the causal chain account. And to make it
even more extraordinary (or conlusing), in a later article costa (198) also
included a version ol 1homson's 'same threat' principle in the description ol the
uul, as il the uul is a conlusing mixture ol all three groups ol accounts discussed
in the previous section.

Ol course, the objective accounts also include a trivial mental state ol the agent: il the agent acts, sJhe
is supposed to have an intention or plan to save the people on the main track. lowever, the uul relers
to a non-trivial mental state: the intention to harm (distinguished lrom loreseeing the harm).
tnica| ccnsistcncy anJ Anima| ua|ity
!14
1he major problem with the uul is the loop dilemma: is the death ol the person
on the side track intended or merely loreseen7 when lamm (2000) tried to apply
the uul to the loop trolley dilemma, she promoted a new doctrine ol triple ellect
(u1l).

.. 1nc ucctrinc cj 1rip|c jjcct (u1). lollowing lamm's doctrine, turning the
switch in the loop case is permissible according to triple ellect. 1hat is because
apart lrom intentional harm (doing something in crJcr tc bring about an evil) and
merely loreseeing a side ellect (doing something in spitc cj bringing about an evil),
lamm claims that there is a third option, in which one does something uccausc it
brings about an evil (which should be distinguished lrom 'in order to bring about
an evil'). 1his u1l approach was lurther delended by 8haw (2006) but criticized by
larris (2000) and more recently by Otsuka (2008) and liao (2009) using the loop
dilemma: triple ellect does not solve the loop case either. liao argued that the
because olJin order to distinction does not apply to the loop case, and lurthermore
questions whether this distinction has a normative signilicance.
Otsuka (2008) gives an example ol a trolley dilemma where this triple ellect
becomes clearer: suppose you are at a switch, and on the side track there is one
person in lront ol six other people. ll you turn the switch, the live on the main
track are saved, the lirst person on the side track will block the trolley, and the six
people behind him are saved. lere, we can say that we would turn the switch, not
in crJcr tc kill the one on the side track, but rather uccausc he will be killed and stop
the trolley. Nothing new is added however, we think, because action in this
dilemma is also allowed according to our abovementioned three accounts.
8


1. lcc|inqs cj tnc victim. 1homson (199!) invited us to locus on what the potential
victim would leel about what the agent does. ll you were thrown lrom a bridge you
might leel dillerently about the agent, than il a trolley were directed towards you.
And it is this dillerence that plays a role. lowever, this claim also involves some
complex knowledge ol psychology, this time not ol the agent, but ol the victim. lt
does not yet solve the trolley problem, because one can imagine switch and bridge
situations where the victim leels the same.



8
lor some lurther subtlety, however, we can say that the one person on the side track is a means to
save the six behind him, but he is not used as means to save the live. ll the person was not present, the
plan to turn the switch and save the live would still work (but six other people would be threatened).
AppcnJix !: rcvicw trc||cy prcu|cm
!1
!. rcjcctivc qrcupinq. leter Lnger speculated about another psychological
mechanism behind our moral judgments: projective grouping and projective
separating (Lnger, 1996, p. 9). [when certain people are in a situation that is
taken to be their problem], we tend to think it is badly wrong to spare them the
serious losses that might stem lrom their problem by imposing serious loss on
other people, who don't have that problem." ln the lirst trolley dilemma (8witch),
all six people on the tracks are considered to be in a similar position in that they
have something in common: they are all on a track and could be run over by a
trolley. 8o the live on the main track and the one on the side track are grouped
together as having the same problem, and the one on the side track can therelore
be considered as 'lair game' to be sacriliced. lowever, in the bridge dilemma, the
lat man is in a dillerent position: he is not on a track, but on a bridge. 8o the lat
man is psychologically separated lrom the live people on the track, which makes
us decide not to sacrilice the lat man. A lot ol people, when responding to the
trolley dilemmas, give spontaneous answers that rellect this projective separation
(people say something like, lut the lat man had nothing to do with it, he was just
passing by"). Also lanna (1992) proposed a lrinciple ol moral lnertia, which is
basically the same as the projective separating. A distinction is made between
participants (such as the person on the side track in the switch dilemma) who are
part ol an ongoing causal process, and bystanders (such as the lat man on the
bridge) who are not part ol the ongoing process. lut this explanation is not lully
satislactory, however, because as Lnger himsell argued, it can be twisted. And it is
at the least very vague: there are no clear criteria to separate people into groups.
1here is no consensus about what the relevant dillerences should be. lnowing
whether someone is a participant or a bystander is not straightlorward. And what
about 8ituation 2 in ligure 1, where the lat man was on a side track on a bridge7

. pistcmic acccunts: risk avcrsicn. kisk aversion is a psychological attitude that
might give an interesting explanation lor the moral intuitions in the trolley
problems. can we know whether our plan to save the live would really work7 ll the
lat man is not heavy enough and the trolley were to keep on moving, then all six
will die, which is an even worse outcome. 1here is the risk ol a worse outcome. ll
the trolley could have stopped in time, even without the lat man blocking it, then
the lat man would have died unnecessarily. ln the switch dilemma, however, we
can be pretty sure that the live are saved and nobody dies in vain.
According to this epistemic account, action would be impermissible il there is a
possibility that the rescue plan will lail and all six people will die. ln particular,
action might not be allowed in uilemmas 2, !, and (lridge, loop and 1ruck).
1he problem with this hypothesis is that certainty is a matter ol degree. 1ake
uilemma 4 (loop and stone): what il the stone was really heavy so that you could
tnica| ccnsistcncy anJ Anima| ua|ity
!16
be sure that it would stop the train7 8urely a mountain ol stones would be
convincing. And even in the switch dilemma, suppose that the side track bends
behind a hill. You cannot be sure that there are no people on the side track behind
the hill. lerhaps there are ten people on the side track, but you cannot see them.
8o the epistemic account in lact makes distinctions even within one dilemma,
instead ol between dilemmas. Nevertheless, there might be some interesting truth
in this approach. lt is related to the amount ol risk aversion that the agent has.
8uppose in the bridge dilemma there is a 10/ probability that the plan ol pushing
the lat man lails and all six people die instead ol one, an 80/ probability that the
plan will work and one person will die instead ol live, and a 10/ probability that
the trolley could have stopped anyway without the lat man, so that one person
dies instead ol nobody. A person with a high level ol risk aversion would choose
not to act. A person with maximum risk aversion would never act, even il the
probability ol lailure were 0.0001/. ln this context, we note that most people have
a high but not maximum level ol risk aversion.
ooing back to the switch dilemma, risk aversion would imply not turning the
switch il there is a possibility that there are ten more people down the side track.
lut be aware that the same could apply to the main track: it might cua||y be
possible that there are ten people behind the live, and you did not see them. Not
turning the switch would result in lilteen deaths. Notice the word 'equally'. 1here
is a kind ol symmetry in the switch dilemma, whereas in most bridge dilemma
situations that we imagine we do not see such a symmetry, and risk aversion has a
stronger inlluence in those dilemmas.

e. pistcmic acccunts: unccrtainty avcrsicn. Next to risk aversion there is
uncertainty (or ambiguity) aversion, whereby the probabilities ol success are not
even known. 1he probability that the plan involving pushing the lat man will work
is not 10/: it is usually not known. 8o we have to choose between two games ol
chance. 8uppose that you are one ol the six people in the bridge trolley dilemma,
but you do not know which one. ll the lat man is not pushed, you know that the
trolley will continue moving and kill live people. 8o you have a survival
probability ol one sixth, because you have a one sixth probability ol being the lat
man who survives. 1his is the lirst game ol chance. ln the second game, the lat
man is pushed, and there is still a possibility that the trolley continues on and kills
one or more ol the live people on the track. lerhaps all might die. which game ol
chance would you preler to play7 1he situation is very similar to lllsberg's
paradox (lllsberg, 1961). 8uppose we have an urn and you know three things: it
contains six balls, has six (or lewer) dillerent colors, and there is one green ball.
1he choice is between two games ol chance. ln the lirst, you win when you draw
the green ball. Your probability ol winning is one sixth. ln the second, you win
AppcnJix !: rcvicw trc||cy prcu|cm
!1
when you draw a blue ball. Your probability ol winning is now unknown
(somewhere between zero and live sixths), because you do not know how many
blue balls there are. 8ome (or most) people preler the lirst game, because they
have uncertainty aversion. 1he similarity with the trolley game is obvious.

.crscna| vcrsus impcrscna| Ji|cmmas. oreene (2008), linally, points - using
psychological and brain research - to an important aspect in the trolley dilemmas:
the distinction between personal versus impersonal dilemmas, related to the
relative position ol the agent towards the victim. lushing the lat man is an action,
which is close up and personal, whereas turning the switch is a more detached
action. 1his is certainly something that inlluences people's choices, but it is not
sullicient to solve the trolley problem, because it is easy to invent scenarios such
as the bridge dilemma to make the action more detached (e.g., you are standing lar
away lrom the bridge and the lat man, but you can push a button, overturning the
bridge). 8o this criterion would also make a distinction within the bridge dilemma.
when most people imagine the bridge dilemma as a close up and personal
situation, some emotion reaction in their brains will be triggered and tell them not
to push the lat man.
lnterestingly, in their research, oreene et al. (2001) classilied personal dilemmas
using some criteria, one ol them reads: where this harm is not the result ol
dellecting an existing threat onto a dillerent party" (oreene, 2002, p168). 1his
relers to the same threat account.
lour invalid accounts
ln this section, we summarize some proposals that in lact would all result in 'pure'
deontology, so they do not solve the trolley problem.

!. 1nc ucctrinc cj ucinq vcrsus A||cwinq (uuA). 1his principle ol uuA is that there is
a moral dillerence between killing and letting die. Quinn (1989a) lor example,
relerred to the uuA to distinguish between trolley dilemmas. lut as lischer and
kavizza (1992a) argued, matters get very complicated in applying the uuA to
trolley dilemmas, because one needs to include unsatislactory relerences to
concepts such as 'transler ol intentions,' 'causal isolation,' and so on.
ln line with the uuA, loot (198) made a distinction between positive duties
(aid) versus negative duties (non-interlerence), and applied this to the trolley
tnica| ccnsistcncy anJ Anima| ua|ity
!18
dilemmas. lut this approach was criticized by 1homson (2008), who showed that
loot's idea basically results in pure deontology, whereby no action is permitted in
all the trolley dilemmas.
lnterestingly, 1homson (2008) also ends up with pure deontology. lowever, we
think that she is mistaken at some point. 1homson (2008, p.!6) used a wrong
argument (wrong analogy) to demonstrate that action is never allowed. let us
digress on this a little, because it is a recent discussion. 1homson starts with the
'three options' dilemma: you are at a switch - il you do nothing the trolley will kill
live people on the main track. You can also turn the switch to the right hand track,
where one person will be killed, or turn it to the lelt hand track, where you will be
killed. 1he argument goes that nobody is willing to sacrilice himsellJhersell (apart
lrom real altruists or depressed people), and it is really unlair, 1homson claims, to
turn the switch to the one victim on the right hand track. 1o show that this is
unlair, 1homson uses another example: You are asked to give money to a charity,
in order to save people. You are able to send your own money, but you instead leel
like stealing the money ol someone else and sending that money to the charity.
we claim that the analogy does not apply, because in the charity dilemma, you are
usinq something ol someone else. lt is comparable to the 'transplant dilemma'
(1homson, 198), whereby a surgeon can save live patients by sacrilicing an
innocent person and use that person's organs lor transplantation. 1he transplant
dilemma is similar to the bridge trolley dilemma, whereby you also use something
ol the victim, namely his body, without his consent. 8o 1homson's analogy can be
used to argue that pushing the lat man in the bridge dilemma is not allowed. lut
lrom the charity dilemma analogy, it does not yet lollow that turning the switch is
not allowed. litzlatrick (2009) and 8haver (2011) also commented on 1homson's
new turn towards pure deontology (1homson, 2008).

.. ||iqitimatc p|ans. kussell (199) relerred to 'illegitimate plans' to argue against
the permission ol, lor example, pushing the lat man lrom the bridge. lut this idea
was criticized by lamm (1989), and is in lact equal to pure deontology, as all the
actions are shown to be illegitimate plans.

1. 1nrcatcncJ pcrscns. montmarquet (1982) stated that only when a person was
not threatened is action impermissible. lut he claimed that the person on the side
track is already threatened. 1his claim, however, is lalse, as was argued by oorr
(1990). montmarquet's approach would result in pure deontology, just like the
uuA (oorr himsell, by the way, relers to a 'same threat' account).

!. karc situaticns. oert (199!) claimed that - in contrast with impermissible
actions - the permissible actions occur in very rare situations. 8itting in a truck
AppcnJix !: rcvicw trc||cy prcu|cm
!19
next to a railway (uilemma ) or standing on a bridge above a railway (uilemma 2)
are more typical situations than standing on the railway with no escape possible,
standing next to a moving platlorm, or standing in the pathway ol a rock, and so
on. 1he problem with this approach is that it is dillicult to quantily the rarity ol a
situation and derive lrom this the permissibility ol actions.
conclusion and lurther research
1he basic question we now have to ask is whether our pattern ol answers
(lollowing our moral intuitions) is given by one or more ol the three accounts
presented above: the mere means, same threat, or causal chain accounts. ll not,
there must be other principles, or we must have moral illusions (comparable to
optical illusions and cognitive biases). we know that there are some very peculiar
examples ol irrationalities in people's answers to the trolley dilemmas. Lnger
(1996) demonstrated that people's responses to the trolley dilemmas are olten
more (inconsistent) psychology than ethics, by pointing out that judgments about
the permissibility ol an option (e.g. the choice to push the lat man) depend on the
availability ol other options that people consider as being irrelevant (see also
Norcross, 2008). 1here is the well known ellect ol wording and lraming
(letrinovich & O'Neill, 1996, 8innott-Armstrong, 2008, lanteri et al., 2008, kay &
lolyoak, 2010). lspecially the order in which dillerent trolley dilemmas are
presented, has some inlluence (letrinovich & O'Neill, 1996, liao et al. 2011,
8chwitzgebel & cushman, 2012, ui Nucci, 2012). And also induced leelings ol
disgust (8chnall et al., 2008) and happiness (Valdesolo & ue8teno, 2006) can
inlluence moral intuitions in the trolley dilemmas. 1he turn that the trolley
problem has made towards empirical studies in moral psychology (oreene, 2008,
cushman et al. 2006, mikhail, 200) is very lruitlul, especially in discovering moral
illusions. 8till, the abovementioned studies in experimental philosophy do not
indicate that the gap between the two paradigmatic cases 8witch and lridge can
be closed.
oiven the classilication ol dillerent accounts above, we can now ask the
lollowing (empirical) questions: low many people can agree with one or more ol
the three principles7 which account will have the most lollowers7 what happens
il respondents learn about these accounts7 uo people leel satislied with these
accounts, and would they pick a prelerred one7 will this inlluence their judgments
in some dilemmas, and how7 ll, lor example, a person is permissive towards the
action in the loop dilemma, but learned that the mere means account is perlectly
tnica| ccnsistcncy anJ Anima| ua|ity
!20
compatible with all ol hisJher intuitions in all dilemmas, except lor the loop
dilemma, would that change the judgment in the loop dilemma7 can such
rellections easily override intuitions7 would that eventually inlluence the
intuition in that dilemma7 can those accounts be used in the method ol rellective
equilibrium (kawls, 191)7 1hese can be questions lor luture research.

!21
Appendix 2: aversions behind the veil ol
ignorance (a mathematical description lor a
theory ol justice)
why a mathematical model7
ln this section l want to unily dillerent theories ol justice and equality, by placing
them in a coherent lramework. ln order to do this, l will try to use mathematical
modeling as much as possible. lconomists and natural scientists are lamiliar with
the use ol mathematical models. ln moral philosophy however, only a lew theories
ol justice (e.g. utilitarianism) have some more or less explicit relerence to
quantitative objects (e.g. utility).
Lsing a mathematical lramework will help us to see dillerent theories ol justice
and their mutual relationships in more clarity. mathematical modeling ollers an
ellicient toolbox that helps us to work towards a more unilied theory ol justice.
1he mathematical equations in this section should therelore not to be taken too
literally, but they should be used as ways to simplily expressions ol complex ideas.
what l will attempt to do, is combine dillerent theories (utilitarianism, maximin,
prioritarianism and egalitarianism) into a mathematical expression that contains
some parameters. 1hese parameters can take dillerent values, and lor specilic
values we get a specilic theory ol justice.
As mentioned in a previous section, there are two arguments lor quasi-maximin
(Qmm) prioritarianism: one is based on a kawlsian argument ol impartiality (the
veil ol ignorance), whereby we assume that the person in the original position has
a high but not absolute need lor salety (high but not maximum risk aversion), and
one based on empathy lor the worst-oll individuals, combined with a low but non-
zero need lor elliciency in terms ol well-being. lence, elliciency is inversely
related to risk aversion. ln this section l derive a mathematical lormulation ol the
quasi-maximin prioritarian principle, using the veil ol ignorance as starting point.
tnica| ccnsistcncy anJ Anima| ua|ity
!22
1he mathematics ol consequentialist wellare ethics
lrioritarianism is a consequentialist theory that looks at the outcomes ol actions
in terms ol well-being. lt was made popular by larlit (1991, 199) and states that
we should maximize everyone's well-being, giving priority to the worst-oll
individuals. As a consequentialist theory, it lends itsell to mathematical modeling
using e.g. utility lunctions. ln particular a priority weighted utility lunction is used
to describe prioritarianism (see e.g. lroome 1991, lrown, 200, loltug 2006,
kabinowicz 2002, mccarthy 200!, 2008). 1hese utility lunctions are the elements ol
a wellare lunction, a quantity that represents the consequentialist betterness
relations between dillerent choices (dillerent situations or world histories).
A consequentialist wellare ethic such as prioritarianism laces serious problems
when it comes to choices involving variable and luture populations. 1hese
problems are relevant in animal ethics, because animals are consciously bred and
brought into existence by our choices. lopulation ethics (Arrhenius, 2000,
llackorby et al., 200) is the branch ol ethics that deals with variable populations.
lopulation ethics is perhaps the branch ol ethics that is mostly plagued with
impossibility theorems: using mathematics, some ethicists proved that we cannot
lind a theory or wellare lunction lor variable populations that meets certain basic
moral intuitions. Always some moral intuition has to lall (lor an overview ol such
impossibility theorems, see e.g. Arrhenius, 2000, llackorby et al. 200!). 1he goal is
therelore reduced to linding a wellare lunction that still satislies the strongest
moral intuitions regarding variable populations, such that only the weakest
intuitions are violated.
ln this section l derive a general wellare lunction lrom a 'veil ol ignorance'
thought experiment (larsanyi, 19!, kawls, 191), borrowing some concepts ol
prospect theory (lahneman & 1versky, 199). ln particular, l suppose that the
impartial observer (decision maker) behind the veil can have dillerent decision
aversions: risk aversion (Arrow, 196), loss aversion (lahneman & 1versky, 1984)
and uncertainty aversion (lpstein, 1999). kawls (191) took only the latter
uncertainty aversion to arrive at his maximin principle behind a veil ol ignorance,
but an impartial observer might have or use other aversions as well. 1hose
aversions set the parameters in the wellare lunction. Applying the veil ol
ignorance to population ethical situations (problems with variable populations
and potential beings), is tricky. lut as l will demonstrate, the wellare lunction that
corresponds with those three aversions also corresponds with some moral
intuitions in population ethics.
kisk avcrsicn deals with the problem that the impartial observer behind the veil
does not know whose lile sJhe will live once the veil is lilted. 1he observer has a
AppcnJix .: matncmatica| Jcscripticn
!2!
probability to become any ol the individuals born in the real world, with known,
unilorm probability distribution: sJhe has a probability 1JN to become any ol the
N individuals. laving risk aversion results in a prioritarian ethic: it corresponds
with the moral intuition that some priority lor the worst-oll is important. 1his
prioritarian intuition rellects a trade-oll between elliciency and equality.
combining prioritarianism with a liletime perspective, where the liletime well-
being levels count as the utility variables in the wellare lunction, also solves the
replaceability and non-identity problems (larlit, 1984). lence, risk aversion
behind a veil ol ignorance results in a wellare lunction that is consistent with
strong moral intuitions about elliciency, equality and replaceability.
css avcrsicn deals with an asymmetry between prelerences lor gains and losses.
leople have a tendency to preler avoiding losses to acquiring gains (lahneman &
1versky, 1984). An impartial observer with loss aversion can lix the parameters ol
the wellare lunction such that it includes number-dampening population lactors.
1hese population lactors allow avoiding some counter-intuitive conclusions in
population ethics: the repugnant conclusion (larlit, 1984, Arrhenius et al., 2010)
and the reverse repugnant conclusion lor positive levels ol well-being and the
strong sadistic conclusion lor negative levels ol well-being (Arrhenius, 2000).
lence, loss aversion behind a veil ol ignorance results in a wellare lunction that is
consistent with strong moral intuitions about variable populations.
0nccrtainty avcrsicn occurs when the veil is thickened in a way that the impartial
observer no longer knows the probability to become any ol the individuals. lt
rellects a prelerence lor known risks over unknown risks: when the possible
outcomes and the probability to become an individual are known, the risks are
known. when the veil is thickened, the risks are not known. 1his lack ol
knowledge ol risks, which is stronger than the lack ol knowledge ol outcomes,
inlluences the prelerences ol the impartial observer. l will argue below that this
uncertainty aversion generates a second kind ol prioritarian theory called
moderate egalitarianism (}ensen, 200!). lt dillers lrom prioritarianism in the sense
that the level ol priority lor the worst-oll does not depend on the absolute values
ol liletime well-being ol the worst-oll, as in prioritarianism, but depends on the
relative positions ol the worst-oll, relative to the better-oll. moderate
egalitarianism has a generalized oini wellare lunction (weymark, 1981). 1his
moderate egalitarianism solves the intransitivity problem (1emkin, 198) and the
problem ol the misery lor the ultra rich (uorsey, 2009), at the serious cost ol losing
independence (or strong separability, see mccarthy, 2008). 1he problem ol
independence is related to Allais paradox (Allais, 19!).
l will demonstrate that combining the three aversions, together with the
rellection ellect ol prospect theory (lahneman & 1versky, 199), results in a
wellare lunction as a sum ol two terms: a positive, number-dampened, weighted
tnica| ccnsistcncy anJ Anima| ua|ity
!24
power mean prioritarianism, and a negative, weighted total utilitarianism.
lncluding the number-dampening lactor in the lirst term generates a trade-oll
between quantity (the population size), quality (elliciency in terms ol maximally
increasing everyone's well-being) and equality (equalizing well-being).
1he weighted power mean ol the lirst term contains lree parameters. 1he
power p ol the power mean can vary lrom minus inlinity, which results in a
maximin theory, to 1, which results in a weighted average version ol moderate
egalitarianism. A negative value lor this power corresponds with a quasi-maximin
(Qmm) prioritarian theory. 1he weight lactors in the power mean can also take
dillerent values, ranging lrom an absolute weight lor the worst-oll individual,
which corresponds with maximin, to a unilorm distribution ol weights, resulting
in unweighted power mean prioritarianism. when the power is 1 and the weights
are unilorm, we get average sum-utilitarianism. lence, there are two ways to
move lrom sum-utilitarianism to maximin: using a power mean and using a
weighted averaging. 1hese two ways are based on respectively risk aversion and
uncertainty aversion.
lowever, some mathematically proven impossibility theorems in the literature
(see e.g. Arrhenius, 2000, llackorby et al. 200!) indicate that the proposed wellare
lunction violates some moral intuitions in population ethics. l discuss the three
most important counter-intuitive implications ol the number-dampened
prioritarian theory. 1hese moral intuitions might be moral illusions, and l brielly
present solutions or ways to deal with those two problems.
Another challenge lor prioritarianism, apart lrom the problems related to
variable populations, are lotteries. A lottery represents a policy choice which has
dillerent possible world histories as outcomes. A probability distribution over
world histories introduces some new complications.
linally, l will re-examine the problem ol interpersonal comparability ol liletime
well-being. lach moral agent can perlorm the thought experiment ol the veil ol
ignorance, using his own risk attitude and his own evaluations ol liletime well-
being. 1o make the theory as objective as possible, we look at distributions ol
measurable distributable goods (i.e. resources and liberties distributed among all
sentient beings). lach moral agent can maximize his own prelerred wellare
lunction, respecting the constraints on those distributable goods. lence, each
moral agent can derive his own optimal distribution ol goods, and based on a
democratic equality ol all moral agents, we can take an average ol all those
distributions as the impartial, optimal distribution ol goods.
AppcnJix .: matncmatica| Jcscripticn
!2
1he impartial observer behind the veil ol ignorance
lmagine there are N
8
potential beings behind a veil ol ignorance. You are one ol
them. when the veil is lilted, you will live the lile ol a sentient being in the real
world, but behind the veil you do not know yet who you will be. You are an
impartial decision maker (an impartial observer) behind the veil, and you can
decide between dillerent world histories. ln order to study world histories, you
can lirst look at a linite time interval t. ln this time interval, the number ol
sentient beings born in lront ol the veil (i.e. born in the real world) is linite. You
can be born and live the lile ol one ol these sentient beings. Alter deriving the
optimal world history that you preler lor this linite time interval, you can take a
longer time interval and perlorm the same process. ll the time interval gets
longer, more luture beings are taken into account, and you might derive a slightly
dillerent optimal world history lor that longer time interval. ln theory, this
process should be repeated to the limit ol an inlinite time interval, encompassing
the complete luture containing a potential inlinite number ol beings. lut in
practice, it will be enough to stop at a sulliciently long time interval.
8o consider a world history n limited to the time interval t. ln this world
history, a number N
l
(n) ol individuals are born in jrcnt ol the veil during that time
interval. 1he number ol beings behind the veil, N
8
, is equal to or larger than N
l
.
1he dillerence between N
8
and N
l
is the number N
0
ol unborn beings, the potential
beings who are not born in the relevant time interval ol the world history. 1he
population size N
l
can be split in three parts: N
-
is the number ol individuals with a
positive liletime well-being, N

the number with a zero well-being and N
-
the
number with a negative well-being.
lach individual i in world history n has a liletime well-being x
i
(n).
1
lor all

we can deline


1
we have to assume some properties lor the liletime well-being levels. lirst, the values

and

lor
the same person are ordinal numbers, which means they can be ordered in a complete well-ordered
set. ln other words, it is meaninglul to say that e.g.

, even though these values cannot be


quantilied. 1he order relation is complete il lor all

and

we have either

or

.
1his assumption is not a strong assumption: in nearly all our choices we can compare our dillerent
needs and leelings allecting our value ol lile. we might preler visiting a lriend over reading a book, we
might preler short term satislaction ol one need over long term satislaction ol another need,. 8o we
are able to compare the values ol lile ol dillerent choices.
A much more dillicult assumption is the lollowing step: there is an ordinality relationship between
dillerent individuals. l.e.: we should be able to compare

with

. 1his is the central most dillicult (or


vulnerable) point in our theory ol justice: how to compare the value ol lile ol dillerent individuals7 ls

tnica| ccnsistcncy anJ Anima| ua|ity
!26
x

+

x

il x


il x



i.e. the well-being vector x
+
x

=1
N

is a projection ol all well-being levels on

+
, the positive and zero values. 8imilarly, x
-
is a projection on all negative and
zero values.
1he wellare lunction
As an impartial observer, you can construct a wellare lunction w, i.e. a lunction
that evaluates world histories. 1he world history which has the highest value
should be prelerred: il w(n)-w(q), then world history n is better than world history
q. we can start with the lollowing general expression lor the wellare lunction:
wxb P
sN

N
+
x
+

up
P
tN

N
-
x
-

bq

with
P
sN

N
+

N
+
sN
+
N
R

a population lactor (a number-damping lunction ol N
-
containing parameter
s and a constant relerence population size N
k
), and
x
+

up


N
P
o

p
N

=1

1
p

the weighted power mean (llder mean
2
) ol x
-
with power p between and
weights a
i
such that o

=1
N
P
. 1he same goes lor the negative well-being levels
with parameters t, b

=1
N

and .


my satislaction ol visiting a lriend higher that your satislaction ol reading a book7 1here is no clear
method to solve these kind ol questions. All we have are two heuristic methods: empathy and the
kawlsian thought experiment ol the veil ol ignorance (kawls, 191, 2001).
A third assumption we have to make seems a big leap into superliciality: the values x

and y
]
are
assumed to be cardinal numbers, i.e. quantitative numbers that can be multiplied, added, subtracted,.
Although this step might seem superlicial, it is in lact only lor didactical purposes that l assume
cardinality, because now we can use clear mathematical expressions. 1herelore, l will speak ol a
quantitative model" lor a theory ol justice.
2
1he power mean in the wellare lunction can be lurther generalized to a weighted generalized l-mean
or weighted lolmogorov mean (lolmogorov, 19!0) with an invertible lunction j:

AppcnJix .: matncmatica| Jcscripticn
!2
lnvariances
1he above general expression ol the wellare lunction satislies some important
invariances.
A lirst property ol the wellare lunction is time scale invariance. 1he number ol
individuals with positive well-being N
-
born in time interval t is proportional to
the time interval t, and the same goes lor N
-
. we can write the relerence
population size N
k
as 8
k
t, with 8
k
a constant relerence birth rate. Now all
numerators and denominators in the wellare lunction are linear in t, which
allows us to take the limit t without expanding or shrinking a term in the
wellare lunction.
A second invariance is the scale invariance ol the well-being (klc-invariance in
lrown, 200). kescaling all x, with positive, results in w. 1his means
that the ordering between histories n and q remains the same alter rescaling. 1his
invariance is important when the unit ol well-being is not lixed. lt is like
comparing the lengths ol two sticks in terms ol meters or centimeters.
when the weights a
i
are unilorm (equal to 1), the wellare lunction has a third
important property: permutation symmetry ol the well-being levels ol the
dillerent individuals. keordering the individuals gives the same wellare lunction.
1his means that all individuals are treated impartially. when the weights are not
unilorm, we can relormulate the theory to keep it impartial. write the liletime
well-being vector as

, i.e. in ascending order (

). ln this interpretation,

does not reler to the well-being ol individual |, but


to the |-th level ol well-being.
ueriving the wellare lunction behind the veil ol ignorance
1he parameters in the above proposed wellare lunction will be determined behind
the veil ol ignorance, by borrowing some elements lrom prospect theory
(lahneman & 1versky, 199).
A lirst important element in prospect theory is the relerence point relative to
which losses and gains are measured. 1he liletime well-being is the value you as
impartial observer would ascribe to the complete lile ol a sentient being, i.e. your
prelerence lor living that lile. A liletime well-being higher (lower) than 0 is a lile
that is (not) worth living. You preler not being born to being born as someone


tnica| ccnsistcncy anJ Anima| ua|ity
!28
with a negative liletime-well-being. 8o we can assume that living a lile ol value 0 is
equally prelerable to not being born at all
!
, to living a completely unconscious lile
without any experiences, or to living a conscious lile without any positive and
negative leelings or prelerences (these three options are equal in terms ol well-
being, although only the latter, conscious lile is the lile ol a sentient being and is
included in the wellare lunction). 1he zero value is the relerence point, and
positive (negative) liletime well-being levels are considered as gains (losses)
behind the veil.
As the impartial observer does not know the identity ol the individual that sJhe
will be in the real world, sJhe does not know what liletime well-being sJhe will
get. lor the impartial observer, it becomes a game ol chance. ln the next three
sections, l apply three elements lrom prospect theory, to our game ol chance: risk
aversion towards gains, the rellection ellect (no risk aversion towards losses), and
loss aversion. Next, l also discuss uncertainty aversion.
kisk aversion lor positive well-being levels
8uppose the probability distribution is unilorm: you have an equal probability ol
being born as any sentient being. ll there are N sentient beings in time interval t,
then your probability ol being individual i is 1JN. uue to impartiality, this unilorm
probability distribution implies unilorm weights a
i
in the wellare lunction.
when the possible outcomes are gains, people olten tend to be risk averse
(lahneman & 1versky, 199). 8uppose there are two histories, both having two
individuals. ln history

the individuals have well-being

. lehind


!
1his assumes that a non-existing lile has a zero liletime well-being. lt requires that liletime well-being
is an 'extensive' instead ol an 'intensive' quantity. ln physics, examples ol extensive quantities are
lengths, masses, energies and electric charges, whereas densities, pressures, temperatures and
chemical potentials are examples ol intensive quantities. Also averages are intensive quantities.
consider the example ol me holding a non-existing cup ol tea in my hand. 1he question how much tea l
hold in my hand is well delined, even il the cup does not exist: 0 liter. ln contrast, the question what the
average amount ol tea is per cup in my hand, is not well delined: 0 liter divided by 0 cups is
mathematically not delined. 8uch an average is an intensive quantity, and intensive quantities cannot
properly deal with non-existence. l believe that liletime well-being is an extensive quantity, which
means that it is delined in cases ol non-existence. lut the wellare lunction is an intensive quantity, as it
is an average over sentient beings. 1his implies that we can compare a lile that has a well-being with
the absence ol that lile. lut we cannot compare the value ol a world that contains sentient beings with
a world without sentient beings. As the wellare lunction is an average, a non-sentient or non-existing
being adds 0 liletime well-being to the numerator and 0 to the population size in the denominator. On
the other hand, an existing, sentient being that has 0 liletime well-being adds 0 to the numerator but 1
to the denominator. lence, in the denominator we can see the dillerence between non-existing, non-
sentient beings and existing, sentient beings.
AppcnJix .: matncmatica| Jcscripticn
!29
the veil ol ignorance, you don't know whether you will get the lower or the higher
level ol well-being, so it becomes a game ol chance. ln history

there is no
gambling, which means that

.
ln the theory ol risk aversion (Arrow, 196), there is the notion ol a utility
lunction u(x) which allows comparison ol both histories. 1his utility lunction can
be derived by using the condition that when the two histories are equivalent, the
average utilities in both situations should be the same. lence, il the wellare
lunction represents your prelerence lor a certain history, then we lirst have to ask
what is the relation between the levels ol well-being in both histories such that
you would have an equal prelerence lor both histories. ln other words: when
would the certain outcome in history 2 be equal to the game ol chance ol history
17 lquality ol the wellare lunction

solves this question.


1his gives


Lsing the condition that when the two histories are equivalent, the average
utilities in both situations should be the same, we get as a condition:


we see that this equality is solved when the utility lunction is identical to the
power lunction: u(x)=x
p
.
1he Arrow-lratt measure ol relative risk aversion (Arrow, 196, lratt, 1964)
with respect to the well-being is delined as:


1his relative risk aversion is constant, which means that the utility lunction is
so-called iso-elastic. when p=1, the relative risk aversion is 0 and we get average
utilitarianism. when p-, the risk aversion goes to inlinity, and we get a
maximin wellare lunction. ln between these two extremes, a non-zero and non-
inlinite risk aversion behind a veil ol ignorance results in prioritarianism.
4



4
Note that when , then -x
-
-
p
=0 as soon as there is at least one x
i
less or equal to 0. As a result, once
there is at least one being with zero or negative liletime well-being, the positive part ol the wellare
lunction either becomes trivially zero, or the risk aversion parameter p should be higher than 0. 1he
latter restriction on the risk aversion implies that we cannot move close to a maximin theory. One
could try to take a power p that depends on the levels ol well-being (i.e. p-p(x)), such that p-0 when
there are individuals having a well-being at or below 0. Another, perhaps more elegant option to avoid

tnica| ccnsistcncy anJ Anima| ua|ity
!!0
lriority lor the worst-oll
lor positive well-being levels, the lirst term ol the wellare lunction corresponds
with the priority view (8chelller, 1982, weirich, 198!, larlit, 1991, larlit, 199,
loltug, 2006): in distributing benelits to individuals and maximizing liletime well-
being, the worst-oll individuals should get a priority. mathematically, this can be
expressed as the ligou-ualton principle: translerring a quantity x lrom the
better-oll (liletime well-being x
1
) to the worse-oll (liletime well-being x
2
x
1
),
without reversing the order (i.e. x
2
-xx
1
-x) increases the wellare lunction.
Note that the wellare lunction lor positive well-being levels has maximin and
sum-utilitarianism as limits. when the power p1, the wellare lunction becomes a
sum ol well-being levels, which means that no-one has a priority, as in sum-
utilitarianism. when the power p goes to minus inlinity, the theory becomes
maximin, where the worst-oll individual gets absolute priority lor its well-being.
1he latter is not true lor traditional expressions ol concave prioritarianism, where
the wellare lunction w=j(x) with j a concave lunction (as in lroome, 1991, lrown,
200, loltug, 2006, lumer, 2006, kabinowicz, 2002, mccarthy, 2008). 1he power 1Jp
in our wellare lunction allows us to take this limit p -. when p is very negative,
we get a quasi-maximin prioritarianism.
ln the previous section (4.4), l mentioned a second justilication lor
prioritarianism, next to impartiality (the veil ol ignorance) with a high but not
maximal risk aversion: empathy with a low, but not zero need lor elliciency. 1he


this problem is to take an exponential lunction ol the liletime well-being, instead ol a power root
lunction (see lumer, 2006). 1ake the exponential liletime well-being

u




-x

instead ol
p

p
, with the parameter . ll this parameter a equals 1, we get (number-
dampened) sum-utilitarianism because
1
. ln the limit ol the parameter going to inlinity, the
wellare lunction becomes maximin. 1he positive part ol the wellare lunction now becomes a (number-
dampened) lolmogorov mean (lolmogorov, 19!0) with exponential lunctions:

+


+

+

R

u
-1

=1

with

u
-1



the inverse ol the exponential lunction. 1his new wellare lunction has a priority lor the worst-oll and
rellects a constant absolute risk aversion, delined as


AppcnJix .: matncmatica| Jcscripticn
!!1
elliciency can be measured by looking at extended ligou-ualton translers
(Vallentyne, 2009, p.18). 8uch a transler is given by the inequality conditions:

. 1his means that switching lrom history


to

, the well-being ol the lowest level increases and the well-being ol the highest
level decreases (but the order ol the levels doesn't change). 1he extended ligou-
ualton transler elliciency is determined by the ratio ol the benelit lor the lowest
level to the cost lor the highest level, where benelits and costs mean increases and
decreases in well-being:


ll we take small, neutral extended ligou-ualton translers, i.e. inlinitesimal
translers with w(n
1
)=w(n
2
), then using the derivatives ol the power lunction
evaluated in

and

, the elliciency becomes approximately



when p=1, the elliciency is always 1, i.e. maximal. when relative risk aversion 1-
p increases, the elliciency decreases. lnlinite risk aversion always corresponds
with zero elliciency, which results in maximin prioritarianism.
Avoiding the replaceability problem
An objection to a total utilitarian theory (taking a sum ol the well-being ol all
individuals) is that sentient beings are treated as nothing but receptacles ol well-
being. ln total utilitarianism it is not a moral problem il a person is simply
replaced by another person with the same level ol well-being. lor example, you
are allowed to kill someone as long as you let another person be born, who will
have the same expected well-being as the murdered person would have il sJhe was
not killed. 1his is counter-intuitive.
One can counter this replaceability problem by simply stating that persons have a
unique intrinsic value, which simply means that these persons cannot be replaced
without violating something deemed important. lowever, an advantage ol using
the liletime perspective is that we can avoid this replaceability problem without a
need to introduce such an intrinsic irreplaceability value. uue to the liletime
perspective, the number N
l
ol individuals over time is well delined. lence, we can
construct a wellare lunction that uses this number ol individuals. we can use lor
example an average instead ol a total ol (priority weighted) well-being (i.e. we can
include a division by N
l
) or include a number-dampening lactor in the wellare
lunction.
tnica| ccnsistcncy anJ Anima| ua|ity
!!2
As an example, compare situation n=(100), i.e. a situation where one person has
a liletime well-being equal to 100, with situation q=(0,0), i.e. a situation where
one person is killed somewhere in the middle ol his lile (so he has liletime well-
being 0), and is replaced by a second person who will get the remaining liletime
well-being ol 0. According to total utilitarianism, both situations are equally
good. lut applying the prioritarian theory, we see that w(q)w(n). lence, the
person should not be killed and replaced. uue to the liletime perspective we can
avoid the replaceability problem.


Avoiding the non-identity problem
1he non-identity problem (larlit, 1984) asks questions like: can we harm someone
il the other option we had would be that this sentient being would not have
existed7 Or can we harm a luture being by bringing that being into existence (e.g.
breeding an animal in the livestock industry)7 1he problem is that both yes and no
answers somehow violate our moral intuitions. lt is dillicult to imagine how we
can harm someone il we'd say to that person: lut il you didn't want to be harmed,
it means you'd preler not to exist." On the other hand, it also seems wrong to let
someone be born, knowing that this person will suller tremendously.
1he reason why the non-identity problem is avoided in our Qmm-prioritarian
theory is because the Qmm-theory is actually not about harming someone, but it's
about just distributions ol liletime well-being. with the above wellare lunction, we
can simply avoid the tricky questions raised by larlit, because l didn't say
anything about the identity ol the dillerent persons. lt doesn't matter il person 1
in situation X equals person 1 in situation Y.
6


Average utilitarianism combined with the liletime perspective also avoids the replaceability problem,
because in average utilitarianism the wellare lunction is divided by the number ol individuals, and we
get 100J1 -(0-0)J2.
6
1he Qmm-theory uses a notion ol 'impersonal harm' (larlit, 1984, p.!8) or 'wide person allecting
harm' (Visak, 2011). One could argue that every notion ol harm requires an impersonal or wide person
allecting view. consider diachronic harm: A harms l at time t il l gets a lower momentaneous well-
being alter t than belore t. lut as discussed in section 4.2, the notion ol a personal identity over time is
tricky and vague. ln lact, l might become a (slightly) dillerent person alter time t, so A harms a
dillerent person. 1he same applies to a counterlactual notion ol harm: A harms l at time t il the well-
being ol l alter t in a world where A does act is lower than the well-being that l would have got alter t
in a parallel, counterlactual world where A did not act. As with the larlitian thought experiments ol
copying and splitting persons, we can say that person l in the real world is a (slightly) dillerent person
than person l' who lives in the counterlactual world. An example: A builds a house lor l. 1he house is
quickly built, which means that l is able to move in the house a month earlier. As a consequence, that
lirst month in his new town, l happens to meet a girl who will later become his wile and who will

AppcnJix .: matncmatica| Jcscripticn
!!!
1he rellection ellect and risk neutrality lor negative well-being
levels
According to the rellection ellect ol prospect theory (lahneman & 1versky, 199),
the risk attitude towards losses is dillerent than the risk attitude towards gains.
lor negative well-being levels, people behind the veil ol ignorance are no longer
risk averse, they become more risk seeking. As the parameter p determined the
risk attitude lor positive well-being levels, the wellare lunction lor negative well-
being levels is dependent on a similar parameter . when 1, we get risk seeking
behavior lor losses. 1his corresponds with the proposal ol triage as discussed in
lrown (200). lowever, triage can be counter-intuitive, as it gives priority to the
better-oll ol the negative well-being levels. 1herelore, and lor simplicity, l take
risk neutrality lor losses. 1his corresponds with =1, i.e. total utilitarianism lor
negative well-being levels.
loss aversion
According to prospect theory, people have a tendency to preler avoiding losses to
acquiring gains (lahneman & 1versky, 1984). loss aversion can be introduced in
the wellare lunction, by taking the parameters s=1 and t=0 in the number-damping
lunctions.
consider a history n with a population divided in two equal subpopulations
N
-
=N
-
and lor each individual i in the positive population there is a corresponding
individual j in the negative population with:

. ln other words: the well-


being levels are distributed symmetrically around the relerence value 0.
looking at the wellare lunction lor this situation, noting that


and

, we get w(n)0. 1his indicates a loss aversion: one


would rather have a well-being 0 with certainty, than taking a gamble with a
symmetric distribution ol well-being levels around zero, because the possible
losses count heavier than the gains.


change his lile (and personality) proloundly. lut as the house was quickly built, it has a weaker
construction, and at one day collapses and kills l. ln the parallel world, l' would have to wait another
month belore he can move in the house. As a consequence, l' does not meet the girl and he will live a
completely dillerent luture lile. lut the house does not collapse and l' is not harmed. 1he question is:
at the moment ol collapse, is l in the lirst world the same person as l' in the counterlactual world7 ll
not, then the non-identity problem already occurs in counterlactual notions ol harm, and the narrow
person allecting view (Visak, 2011) would run into counterintuitive troubles.
tnica| ccnsistcncy anJ Anima| ua|ity
!!4
1his loss aversion is consistent with some moral intuitions in population ethics.
uue to loss aversion, and in particular the choice ol the parameters s=1, we can
avoid the repugnant and reverse repugnant conclusions. 1he choice ol t=0 allows
us to avoid the strong sadistic conclusion. 1his will be explained in the next
sections.
Avoiding the repugnant conclusion
1he kepugnant conclusion (larlit, 1984) is one ol the most challenging arguments
in population ethics. 1he argument goes as lollows. 8tart with a population ol very
happy people who have well-being 100. 8o, situation A=(100) 1he total-
utilitarianist would say that a situation 8=(100,98) is better than A: the total
number ol happy persons increased, and the total happiness increased by adding
people who are only slightly less happy than the existing population. 1his addition
ol happy people is the lirst step. 1he second step consists ol equalizing the levels
ol well-being: situation c=(99,99) is considered to be even better than 8, because
now there is more equality. One can repeat step 1, by introducing a third
population with well-being levels equal to 9. Alter repeating step 1 and step 2, we
move to the optimal situation 7 which contains an almost inlinite number ol
people with an almost zero (but still positive) well-being. lach ol those individuals
still has a lile that is worth living, because their liletime well-being remains
positive, though very low. 8o 7 is better than A, or in other words, we should boost
population growth, even il all well-being levels become very low. lut this
conclusion is repugnant.
1he lirst term in the proposed wellare lunction shows how the repugnant
conclusion can be avoided: lor large populations N
-
, the population lactor

becomes constant and we end up with the power mean ol well-being


levels. Adding more and more persons with lower and lower well-being, lowers
this power mean, and hence lowers the wellare lunction.
As a side remark, the repugnant conclusion might also be avoided in total
utilitarianism, i.e. when p=1, when three conditions are met.

lirst, an individual
liletime well-being is dependent on a resource with decreasing marginal utility.
lence, the liletime well-being can be written as a concave lunction j(r) ol a

8hiell (200) gave a more general prool that total utilitarianism avoids the repugnant conclusion
when there are live restrictions which rellect universal properties ol physics, biology and prelerences:
essentiality ol material consumption (the lact that x depends on r), positive subsistence consumption (a
positive critical consumption level c), upper bounds on resources (k) and non-material goods, and the
law ol conservation ol matter.
AppcnJix .: matncmatica| Jcscripticn
!!
resource r available to the individual. 8econd, there is a critical consumption level
c ol the resource required lor a positive well-being. And third, the total amount ol
the resource k is limited. ll all individuals have an equal access r=k/N to the
resource, then we can write well-being lor each individual as


1he wellare lunction

is maximal lor a linite, optimal population

. ln this optimum, the liletime well-being ol an individual becomes


. Lnder these realistic conditions, the population size does not explode as in the
repugnant conclusion. Nevertheless, a worry remains: what il the average well-
being is very low7 1his optimum might be repugnant enough in total
utilitarianism. 1o play it sale, l preler the power mean prioritarianism to avoid the
repugnant conclusion.
Avoiding the reverse repugnant conclusion
ll using a power mean ol well-being levels solves the repugnant conclusion, we
have to be aware ol a reverse repugnant conclusion. maximizing a power mean
implies that it is not good to give birth to beings who will get a lower well-being
than the power mean ol well-being ol the existing individuals. 8o il we
systematically exclude births ol potential beings who will have lower well-being
levels, in the end, only the person with the highest well-being should be born.
lnstead ol overpopulating the world, an average well-being wellare lunction says
that we should underpopulate the world.
uue to the choice lor s=1, the population lactor in the positive term ol the
wellare lunction is a concave lunction ol the population size N
-
.
8
1hat means that
lor low populations (lower than the relerence value N
k
) the wellare lunction
increases linearly in N
-
. 1his pulls us away lrom an under-populated world,
because it is good to increase the population size.
9



8
Note that this adapted moral weight expression has a similar structure as the value ol lile expressed
in section 4.2.4, generating a trade-oll between quantity (number ol individuals, or length ol a liletime
ol a single individual) and quality (values ol lile ol dillerent individuals, or experienced well-being ol a
single individual).
9
Another way to avoid the reverse repugnant conclusion, is by introducing a deontological permission
that is related to the !-N-principle to be discussed in section 1.1(10.4. 8uppose the wellare lunction
simply contains a generalized l-mean (lolmogorov, 19!0)

tnica| ccnsistcncy anJ Anima| ua|ity
!!6
1rade-oll between quantity, quality and equality
As the priority view lies between maximin and utilitarianism, it represents a
trade-oll between equality and elliciency. According to maximin, inequality is
only allowed il it benelits the worst-oll. According to utilitarianism inequality is
always allowed as long as the distribution ol well-being is ellicient, i.e. as long as
total well-being is maximized.
Adding the population lactor P
1N

N
+
, we get a trade-oll between three
elements: quantity (population size), quality (elliciency) and equality. 1hese three
elements correspond with three lactors in the wellare lunction
10
:
w
+
xb
N
+
N
+
N
R
Ax
+
I
A
x
+

with Ax
+
x
+

1
the average well-being and
I
A
x
+

x
+

p
x
+

1

the Atkinson inequality index (Atkinson, 190, 8en, 1982). when p1, this
inequality index is zero only when all x
i
are equal.
Avoiding the strong sadistic conclusion
1he negative mere addition principle (Arrhenius, 2000, p. 66)
11
says that adding
individuals with a negative liletime well-being always lowers the wellare lunction.
lowever, consider a negative average utilitarianism, with average well-being -100.


w
]

xb
u
-1


N
P

u
x

=1

with the weight lunction lor example the exponential liletime well-being (lumer, 2006):

u
x
o
o
o
-x

1his wellare lunction does not include a population lactor. lt is simply an average, so the repugnant
conclusion is avoided, but the reverse repugnant conclusion is not avoided. lowever, we can add a
deontological permission which says that everyone is allowed to procreate, even il the new individuals
would get a liletime well-being lower than the generalized l-mean ol all other individuals. As we will
see in section 10.6, procreation is always allowed, because procreation is natural, normal and necessary
lor e.g. biodiversity.
10
1his expression ol the wellare lunction gives 2-dimensional indillerence surlaces in a !-dimensional
space, as discussed in carter (1999).
11
1his corresponds with the lell 1hree" thought experiment in larlit (1984, p. 422).
AppcnJix .: matncmatica| Jcscripticn
!!
ln such a theory, adding someone with liletime well-being -1 would increase this
average, and hence would be an improvement. 1his is the strong sadistic
conclusion ol negative average utilitarianism: adding someone whose lile is not
worth living might increase the wellare lunction.
Our wellare lunction contained a negative term
tN

q
. ll the
parameter t=0, then the strong sadistic conclusion is avoided, because in this case
the negative term does not have a
-
in the denominator. ll the negative term
would have a
-
in the denominator ol the population lactor, the term can
increase (become less negative) when
-
increases, i.e. when people are added
whose lives are not worth living.
12

lriority lor negative levels ol well-being
1he wellare lunction is discontinuous around 0: when a positive well-being level
decreases till it reaches the zero value, the population lactor suddenly drops lrom

1N

+
to
1N

+
, because the number positive well-being levels
decreases. 8imilarly, the population lactor
0N

-
suddenly increases il the
well-being lurther drops below zero.
1he discontinuity ol the population lactors might seem counter-intuitive to
some people, but a nice leature is that the wellare lunction now contains a
sullicientarian (critical threshold) element: lilting a negative well-being up to a
positive well-being becomes very important.
Note that avoiding the discontinuity by taking a population lactor
1N

P

instead ol
1N

+
generates stronger counter-intuitive problems, such as a
strong sadistic conclusion. ll the population lactor contains the total number ol
beings N
l
, and il N
l
is much lower than N
k
, then the wellare lunction might
increase when a new being with a small negative well-being is introduced.


12
Another way to avoid the strong sadistic conclusion, is by introducing a deontological constraint that
is related to the mere means principle to be discussed in section 6.2. 8uppose the wellare lunction
simply contains a generalized l-mean (lolmogorov, 19!0)


u
-1

=1

1his wellare lunction includes the negative levels ol well-being and does not include a population
lactor. lence it is vulnerable to the strong sadistic conclusion when the generalized l-mean is negative:
the wellare lunction increases when a new individual is added who has a negative liletime well-being
which is higher than this generalized l-mean. lut this new individual will be used as merely a means to
increase the wellare lunction: the presence ol the individual is required and the individual does not
want to live because sJhe has a negative liletime well-being.
tnica| ccnsistcncy anJ Anima| ua|ity
!!8
Note also that the power mean

contains the total number ol beings N


l

instead ol only

. ll the power mean was restricted to the positive well-being


levels, another very strong sadistic conclusion might occur: decreasing someone's
well-being till it reaches the zero value might suddenly increase the wellare
lunction.
lrelerence lor (not) being born
1he choice ol the population lactors

and

generates loss
aversion relative to the zero well-being relerence. 1hese lactors can also have a
very dillerent interpretation, not related to loss aversion. we can lormulate the
veil ol ignorance in such a way that the population lactors correspond with
conditional probabilities ol being born as an individual with a positive or negative
liletime well-being. 1his conditional probability is related to the question: how
many potential beings are sitting behind the veil, and how many ol them will be
born in lront ol the veil7
8uppose you are an impartial observer behind the veil ol ignorance. 1he
number N
l
ol beings actually born in jrcnt ol the veil in time interval t, depends
on your choice ol history n. lor the number ol beings behind the veil, there are
three options.
Option 1: you are as impartial observer alone behind the veil, and alter your
choice ol world history, N
l
-1 extra beings are created, so N
8
= N
l
. ln this case, you
are certain to be born.
Option 2: the number N
8
ol potential beings behind the veil is already
determined belore you, as impartial observer, choose a certain history. N
8
can be
written as 7t, with 7 the maximum possible birth rate. Ol those N
8
potential
beings, only N
l
will be born. Your probability to be born might be very low,
because N
8
might be very high.
Option !: something in between the previous two option: there is an inlinite
pool ol all possible beings. Once a world history n is chosen, a number

is
drawn lrom the pool. N
l
ol them will actually be born. Your probability to be born
is now N
l
JN
8
, which might be low, but not as low as in option 2.
8uppose lirst that all well-being levels are positive or zero. According to the
lirst lormulation, you are sure to be born, so you can as well maximize the (power)
average well-being:


1his expression laces the reverse repugnant conclusion.
According to the second lormulation, you can increase your probability ol being
born by increasing the actual population size N
l
. 1he wellare lunction reads
AppcnJix .: matncmatica| Jcscripticn
!!9


with the denominator a constant (independent lrom N
l
). 1his expression
laces the repugnant conclusion.
1he third option avoids both repugnant conclusions. Not knowing whether you
have to maximize total or average weighted well-being, we could take a
combination, such as:

B

with N
8
= N
l
- N
k
. 1his equals number-dampened prioritarianism.
when some well-being levels are negative, the third option can be
reinterpreted. You lirst get a probability N
-
J(N
-
-N
k
) ol being born with a positive
well-being. You want to maximize this probability, but there is a trade-oll with the
quality, i.e. the power average well-being. ll you are not selected to have a positive
well-being, you are lelt with a probability ol being born with a negative well-being
equal to N
-
JN
k
. 1his equals the population lactor lor negative well-being levels.
Note that this is a conditional probability (conditional on not being born with a
positive well-being), so it no longer contains a term N
-
in the denominator. 1he
number ol unborn beings equals N
0
- N
k
N
-
N

. 1his number is variable, such that


N
k
can be treated as a constant (independent lrom N
-
), to avoid the strong sadistic
conclusion. Note that il N
-
is big, N
k
should also be big. 1hat means that when there
are a lot ol individuals with negative well-being levels, the population lactor lor
positive well-being levels
1N

+
increases almost linearly in N
-
. ln other words:
the more individuals whose lives are not worth living are born, the more
individuals should be born whose lives are worth living.
we see that the population lactors
1N

+
and
0N

-
have three
dillerent justilications: 1) they correspond with moral intuitions to avoid the
repugnant, reverse repugnant and strong sadistic conclusions, 2) they correspond
with a loss aversion behind a veil ol ignorance and !) they correspond with
conditional probabilities in a dillerently constructed veil ol ignorance thought
experiment. 1hese dillerent perspectives generate a coherent picture.
8ummary: positive number-dampened power mean prioritarianism and
negative total utilitarianism
lor unilorm probability distributions (a
i
= u
j
= 1), the rellection ellect combined
with risk and loss aversion allows taking the parameters , =1, s=1 and
t=0. 1hen the wellare lunction reads:
tnica| ccnsistcncy anJ Anima| ua|ity
!40


+

+

R

p
N

=1

1
p

]
-
N

]=1

1his expression has two parameters (p and N
k
) that can be tuned to correspond
with our intuitions.
1he lirst term ol the wellare lunction is a numucr-JampcncJ pcwcr mcan
pricritarianism. lere, prioritarianism relers to a wellare lunction with a sum ol
concave utility lunctions such as u(x)=x
p
, with (lroome, 1991, lrown,
200, loltug, 2006, lumer, 2006, kabinowicz, 2002, mccarthy, 2008). 1he power
mean prioritarianism is a generalization to negative powers (this generalization is
made possible due to the overall reverse power 1Jp in the wellare lunction). 1he
number-dampened property relers to the extra lactor that is linear in N
-
lor very
small populations and nearly constant lor very large populations (see number-
dampened utilitarianism: lurka, 198!, llackorby et al. 2002, Ng, 1986).
1he second term ol the wellare lunction corresponds with a kind ol ncqativc
tcta| uti|itarianism, a theory where only negative liletime well-being levels count,
and where the only objective is to decrease this total amount ol negativity.
1his wellare lunction corresponds with some moral intuitions in population
ethics. lt gives a priority lor the worst-oll (and a special priority lor the beings
with a negative well-being), generates a trade-oll between quantity, quality and
equality, and avoids the replaceability problem, the repugnant conclusion, the
reverse repugnant conclusion and the strong sadistic conclusion.
lowever, in population ethics, a lot ol impossibility theorems are proven (see
e.g. Arrhenius, 2000, llackorby et al. 200!, lrown, 200). 1hat means that our
wellare lunction cannot escape some problems. ln lact, it laces three troublesome,
counter-intuitive conclusions.
lroblematic properties ol number-dampened prioritarianism
lndependence and the mere addition paradox
koughly speaking, independence or strong separability (see e.g. mccarthy, 2008)
says that the choice that maximizes the wellare lunction should not depend on
individuals whose presence and well-being levels cannot be inlluenced. 1his
means that the moral judgment ol a change that only allects a subpopulation does
not depend on the rest ol the population. more accurately: suppose we have to
decide between two histories n and q, and in n there is an unallected
subpopulation ol N
un
people who have the same levels ol liletime well-being as the
AppcnJix .: matncmatica| Jcscripticn
!41
corresponding N
un
people in history q. 1his subpopulation consists ol the
unallected people, because in both choices their well-being is the same. lmagine
this subpopulation on a lar away island, outside the inlluence ol our choices. Now
we translorm n and q to and respectively. lor people nct in the subpopulation,
i.e. lor people in the allected population ol size N
al
=N
tot
-N
un
, individuals in have
the same well-being as individuals in n, and the same goes lor and q. lut lor the
people in the unallected subpopulation, the well-being levels are changed in the
same way lor both histories: lor the subpopulation, the translormation lrom n to
is the same as the translormation lrom q to . 8o, lor example the well-being ol all
people in the unallected subpopulation is raised (e.g. the people on the lar away
island discovered new resources). ln that case, independence says that w(n)-w(q)
il and only ol w()-w(), i.e. the order ol prelerence should not change.
1he principle ol independence (strong separability) is valid in our prioritarian
theory in situations where the numbers ol people with positive and negative levels
ol well-being remain constant. lowever, the principle is violated in two cases: in
mixed populations (where choices can inlluence the numbers ol people with
positive and negative well-being) and in variable populations (where choices can
inlluence the total number ol people).
consider mixed populations lirst: suppose that in history n the allected
subpopulation has some people with positive and some with negative liletime
well-being levels, whereas in q the allected population has dillerent numbers ol
people with positive and negative levels. lor example some people with positive
well-being in n get a negative well-being in q. 8uppose that the unallected people
in n have positive levels ol well-being, whereas they have negative levels in . ln
that case, it is easy to demonstrate that w(n)-w(q) does not necessarily result in
w()-w(), because the unallected population inlluences the population lactors.
1!

ll most ol our policy choices do not llip the sign ol someone's well-being (i.e. do
not change a positive into a negative well-being or vice versa), this violation ol
independence is not a serious problem. 1he violation only implies that the amount
ol priority lor negative levels ol well-being (how important it is to raise someone's
negative well-being instead ol raising someone else's positive well-being) can
depend on unallected populations.


1!
Note that a generalized l-mean (lolmogorov, 19!0)


that includes the negative levels ol well-being and does not include a population lactor, has
independence lor mixed populations (but not lor variable populations).
tnica| ccnsistcncy anJ Anima| ua|ity
!42
As most policy choices can inlluence the population size, there is a second,
perhaps more serious kind ol violation ol independence in situations with variable
populations. 1ake lor example the power p=1J2, N
k
very low and consider two
times two histories, each with populations having well-being levels n
1
=(1), n
2
=(1,9),

1
=(1,81) and
2
=(1,9,81). we see that history
1
is related to history n
1
, just as
history
2
is related to n
2
: l simply added a population (having a well-being 81 per
individual) to both histories n
1
and n
2
in order to get histories
1
and
2
. we can
imagine that this happy population is added on a lar away island, so we expect
that this population does not inlluence our choices here. 1he choice we lace is
between a situation with a population having well-being 1, versus a situation with
a doubled population, whose individuals have well-being levels 1 and 9. lowever,
w(n
1
)w(n
2
) but w(
1
)-w(
2
). ln other words: the existence ol a population on a lar
away island might reverse the order ol prelerence here.
1hese violations ol independence are counter-intuitive. Nevertheless, restoring
independence would result in the violations ol two other moral intuitions: the
prelerences that an impartial observer has behind the veil ol ignorance and the
intuition that we should avoid the repugnant conclusion. 1he latter two moral
intuitions might be stronger (have more coherence) than the single moral
intuition ol independence. ll that is the case, then constructing the most coherent
rellective equilibrium would imply that it is better to violate independence.
kelated to this violation ol independence lor variable populations is the mere
addition paradox. we encountered this paradox in the lirst step towards the
repugnant conclusion: how can it be bad to add very happy people7 ln order to
avoid the repugnant conclusion, this badness came down to the observation that
the average well-being decreases when people are added who are (slightly) less
happy than the existing group. ly moving lrom A=(100) to 8=(100,98), no-one is
harmed, the total well-being almost doubles, and all lives are worth living. lor
small populations, such additions are good according to our wellare lunction. Yet,
such additions are not good when N
-
becomes larger than the relerence N
k
. lor
large populations, quality dominates over quantity.
lt seems paradoxical that merely adding happy people could lower the wellare
lunction. 1he paradox gets worse when we imagine lar away planets with trillions
ol super happy extraterrestrial beings. Our human happiness cannot compete with
theirs, so we only lower the average well-being il we keep on procreating. lt would
be better il humans stopped procreating. 1he same goes lor a world history time
interval t that includes a past era ol dinosaurs: what il scientists discovered that
they were happier than humans7 low is it possible that the ethics ol human
procreation would depend on such discoveries7
1hree remarks are in order. combining those three remarks will sulliciently
weaken the problem ol the mere addition paradox.
AppcnJix .: matncmatica| Jcscripticn
!4!
lirst, a weak remark: this paradox only occurs in situations when populations
are already large. 1he mere addition paradox is avoided in small populations, due
to the population lactor (N) that becomes linear in N. lowever, lor large
populations (or il populations include all lar away planets), this is not much ol a
consolation.
8econd, and more importantly, we can solten the contra-intuitive mere
addition paradox by writing the wellare lunction as a sum ol a changeable and an
additional part: w= w
cn
-w
aJJ
. 1he changeable part looks like the expression
presented above and contains individuals whose liletime well-being can be
changed amongst each other. 1he additional part is simply the sum ol liletime
well-being ol additional beings. 1he point is that this additional part contains only
those beings whose liletime well-being cannot be exchanged with beings lrom the
changeable group. ln particular it is impossible to transler well-being lrom the
changeable to the additional group. Adding individuals who have a lower liletime
well-being than the critical level determined by w
cn
(e.g. lower than the population
weighted, power mean ol the levels ol well-being ol the changeable group) will
increase the total wellare lunction il those individuals belong to the additional
group. writing the wellare lunction like this avoids the repugnant conclusion,
because the argument towards the repugnant conclusion breaks down at the
second step: a move lrom A=(100) to 8=(100,98) is allowed (satislying mere
addition), but the second move lrom 8=(100,98) to c=(99,99) is impossible because
the second person belongs to the additional group.
1hird, and most importantly, even though adding people who belong to the
changeable group and who have a too low liletime well-being is not good
according to the wellare lunction w
cn
, we can include some deontic permissions.
1hese are permissions that are always allowed, even when they violate the
prioritarian wellare ethic.
most people have moral intuitions about three such deontic permissions. 1hey
become particularly visible when we include non-human animals as sentient
beings in the wellare ethic. 1he lirst deontic permission says that predation is
allowed (we do not have a duty to protect prey lrom predators), even when
predators violate the wellare ethic by killing a lot ol prey. 8econd, animals are
allowed to move around, even il small insects would have a well-being that is
lowered when they are in huge numbers trampled by the large animals. And third,
procreation is allowed, even il an animal species does not contribute enough to the
wellare lunction. compare a lrog with a low liletime well-being (a poor emotional
lile over a short, eight year lilespan, with a low psychological connectivity) with a
normal human with a high liletime well-being (a rich emotional lile over a longer,
eighty years lilespan). 1he lrog still has a lile worth living: his liletime well-being
is positive.
tnica| ccnsistcncy anJ Anima| ua|ity
!44
1wo explanations might justily such deontic permissions. lirst, biodiversity
might have a moral value. ll lrogs (and all other lile lorms with lower liletime
well-being) were not allowed to procreate when happier humans exist (or il
humans are not allowed to procreate when happier l1's or dinosaurs existed),
then biodiversity would drastically decrease. 1his decrease in biodiversity trumps
the prioritarian wellare ethic. 1he same goes lor the extinction ol all predators
when they are not allowed to prey on animals and lor large animals when they are
not allowed to move and kill insects by accident. A second solution is a kind ol
behavioral lairness: il humans are allowed to procreate, then so are lrogs. ll zebras
are allowed to eat lor survival, then so are lions. ll insects are allowed to move,
then so are elephants.
loth arguments ol biodiversity and lairness can be combined into the lollowing
deontic permission ol procreation: il person X is allowed to do something that
necessarily contributes to biodiversity (such as procreation), then so is person Y,
even il the wellare lunction decreases (i.e. even il Y's child has a positive liletime
well-being below the power mean). 1hese explanations ol deontic permissions can
be lurther relined, but l will leave that to a later chapter on the predation problem
(chapter 10). lere, it is sullicient to note that there might exist such deontic
permissions that are coherent with moral intuitions about e.g. predation, motion
and procreation.
ln summary, even il some behavior such as procreation would lower the wellare
lunction, it is always allowed (but not obligatory)
14
. wellare lunctions are
applicable to dillerent, incomparable regimes, corresponding to dillerent
populations. Once new beings with lower well-being are added, we enter a new
regime. 8uch a shilt lrom regime is always permissible. ln this new regime, we
have to move on maximizing the wellare lunction. lilling the added beings would
not be an improvement: it is impossible to return to a previous regime.
1he weak sadistic conclusion
Arrhenius (2000, p6) pointed out that a theory such as number-dampened
prioritarianism is vulnerable to a weak version ol the sadistic conclusion. As
mentioned, the addition ol people with low positive well-being levels lowers the
average. Adding a lot ol persons with a well-being slightly above 0 might result in
a stronger decrease ol the wellare lunction compared to adding one individual


14
8ome restrictions might be included: perhaps it is never permitted to add an individual with a
negative liletime well-being. lence, species who can only have negative well-being levels are not
allowed to procreate.
AppcnJix .: matncmatica| Jcscripticn
!4
with a small ncqativc liletime well-being. ln other words, adding one person whose
lile is not worth living might be better than adding thousands ol people whose
lives are (barely) worth living.
1his seems counter-intuitive, but l believe we are dealing with a moral illusion
here. As with the mere addition paradox, three remarks are in order to sulliciently
weaken the weak sadistic conclusion.
lirst ol all, both additions decrease the wellare lunction, and it is possible to
avoid such decreases by simply not adding any ol those people. 8o it is not a
dilemma between adding the one sullerer versus adding the barely happy people,
but a trilemma between those two options and a third option: no addition. 1he
latter option is always prelerable.
8econd, the weak sadistic conclusion only occurs when the mentioned well-
being levels ol the added people are as high as they can possibly get. ln more
realistic cases, the well-being levels ol the added people are not maximal. lor
example, in the above scenarios moving to the repugnant conclusion, we assumed
that we could redistribute well-being lrom 8=(100,98) to c=(99,99). 8o the well-
being 98 ol the added person is not its maximal possible level. ll the well-being was
already maximal, then the repugnant conclusion is already avoided because a
move to situation c would not be possible. 8imilarly, il the well-being levels ol the
added people in the problem ol the sadistic conclusion are not lixed or maximal, a
redistribution ol well-being is possible. And the sacrilice lor the better-oll people
is low when they have to redistribute their well-being with the one person at a
negative level. ln contrast, when the better-oll people have to redistribute their
well-being with the thousands ol people at a low positive level ol well-being, their
well-being might drop drastically towards a very low averaged level.
kelated to this is the abovementioned suggestion to write the wellare lunction
as a sum ol two parts w= w
cn
-w
aJJ
. ll the added person with a negative well-being
belongs to the changeable group, i.e. il other people can transler their well-being
to this miserable person, the weak sadistic conclusion becomes weak, because
those other people would preler a small redistribution ol their well-being towards
the one miserable person over a huge redistribution ol their well-being towards a
huge number ol added people who have small positive levels ol well-being. lut il a
transler ol well-being towards the added person is impossible, the added person
belongs to w
aJJ
, which contains a sum ol well-being. Adding a miserable person
with negative well-being would lower w
aJJ
.
1hird, as mentioned above, we can say that procreation is allowed as a deontic
permission, as long as the new lives are worth living.
tnica| ccnsistcncy anJ Anima| ua|ity
!46
No replication invariance
1he above wellare lunction contains a positive and a negative part. 1hose two
parts are weighted with two dillerent population lactors

and

.
1his implies that the theory is not replication invariant when a situation X is
replicated into a situation =X-X, doubling the population size. 1he wellare
lunction ol this doubled situation is not simply proportional to the wellare
lunction ol the old situation: w() does not equal lor example 2w(X). 1his means
that reversals might occur alter replication: il w(X)-w(Y) it can happen that
w()w(). keplication invariance is delined as the impossibility ol such
reversals.
lt is easy to demonstrate that our wellare lunction has no replication
invariance. 1here are in lact two reasons why there is no replication invariance.
lirst, suppose that there is no negative part in the wellare lunction. ln that case,
the theory is replication invariant only when the parameter s=0. when the
parameter s-0, we would still have a violation ol replication invariance (unless we
would also duplicate N
k
into N
k
'=2N
k
, but that is cheating). lut this is a rather weak
violation ol replication invariance, because it only occurs lor intermediate
population sizes. when N is very low or very high, the theory becomes replication
invariant.
A second, more serious violation ol replication invariance occurs lor mixed
populations when the two population lactors are dillerent.
1
we could restore
replication invariance lor such mixed populations by adapting the wellare
lunction into lor example:


1he population lactors are now the same lor the positive and negative parts,
and they use the total number ol beings N.
1his wellare lunction has replication invariance lor mixed populations, but
unlortunately it implies the strong sadistic conclusion: in some situations the
wellare lunction might increase by adding a person with a (small) negative
liletime well-being. Yet, this strong sadistic conclusion can be avoided by
introducing another deontological constraint that prohibits the addition ol a
miserable person to increase the wellare lunction. 1his deontological constraint
relers to the mere means principle that will be discussed in section 6.2. 1he
miserable person would be used as merely a means to increase the wellare


1
1his kind ol violation ol replication invariance is equivalent to larlit's 'absurd conclusion' (larlit,
1984, p. 410).
AppcnJix .: matncmatica| Jcscripticn
!4
lunction, because the person has to undergo and do something that sJhe does not
want: sJhe has to be born and live, whereas sJhe would rather not live at all. 8uch
deontological mere means principle helps to avoid the strong sadistic conclusion.
Nevertheless, when s-0, we are still stuck with a violation ol replication
invariance. 1here does not seem to be a solution, an adaptation ol the wellare
lunction, that respects replication invariance without violating another important
moral intuition. 1he question is: which intuition is the weakest7 l would answer
that the violation ol replication invariance is less bad than a violation ol e.g. the
(reverse) repugnant conclusion or the strong sadistic conclusion.
16


ln summary, the mere addition paradox, the weak sadistic conclusion and the
lack ol replication invariance are three weak counter-intuitive implications ol the
theory. we can simply split the wellare lunction into changeable and additional
parts and we can lurthermore add a deontic permission: procreation is always
allowed (as long as the maximum attainable level ol well-being ol the added
person is positive). 1his is sullicient to deal with those problems.
lntermezzo: a more complex lormulation to solve the replaceability
problem
1he above wellare lunction uses the levels ol liletime well-being as input
parameters. lence, an impartial observer behind the veil ol ignorance is required
to group momentaneous minds together into mutually exclusive subsets that
represent strictly separated persons. lowever, as we have seen in the section on
personal identity and psychological continuity, personal identity between
momentaneous minds is not always an all-or-nothing issue.
ln this intermezzo l want to propose a more accurate wellare lunction that
deals with a more complex account on personal identity. All we have is a set ol
momentaneous minds (indexed by , i.e. momentaneous person at time )
who experience a momentaneous well-being
t
. 1he impartial observer behind


16
Note that a generalized l-mean (lolmogorov, 19!0)


u
-1

P

u

=1

that includes the negative levels ol well-being and does not include a population lactor, has replication
invariance. lut this theory requires a deontological constraint (the mere means principle discussed in
section 6.2) and a deontological permission (the !-N-principle discussed in section 10.4) to avoid
respectively the strong sadistic conclusion and the reverse repugnant conclusion.
tnica| ccnsistcncy anJ Anima| ua|ity
!48
the veil does not have to group these momentaneous minds into an individual's
liletime well-being

. lence, dilliculties in grouping momentaneous minds into


subsets that represent all-or-nothing personal identities can be avoided. lnstead,
the impartial observer can work with a connectivity lunction

. 1his
connectivity can have two dillerent interpretations that are coherent with each
other.
lirst, behind the veil ol ignorance, the connectivity lunction might be
proportional to the conditional probability: il the impartial observer would
experience momentaneous well-being

, then hisJher probability to


experience momentaneous well-being

(ol another momentaneous person

at another time

) will be proportional to the connectivity. 1he impartial


observer, once incarnated in lront ol the veil, might travel around between
momentaneous minds and have multiple momentaneous experiences belonging to
multiple minds. lence, the idea is that the impartial observer behind the veil lirst
calculates hisJher probability to experience momentaneous well-being

and
next ascribes conditional probabilities to experience other momentaneous minds
given that sJhe already experienced (or will experience)

somewhere during
hisJher stay in lront ol the veil.
As a second interpretation, the connectivity lunction represents how strong the
two momentaneous minds and

belong to the same person over time.


1he lunction represents the psychological and physical connectivity between two
momentaneous minds. 1he luturistic larlitian thought experiments (e.g.
teleportation, mind copying, mind swapping, splitting minds or changing
personalities) imply that momentaneous minds can be mutually related in
degrees: it is not an all-or-nothing question whether or not two momentaneous
minds belong to the same personal identity over time.
ll lor example the personal identity splits like the splitting ol a rope into two
branches (i.e. the momentaneous minds generate a Y-shape in space-time), we can
look at three momentaneous minds at times t, -t and '-t: and

are
connected through

, and also and

are connected
through

, but

, and hence

and

do
not belong to the same person although and

do and and


do.
Now we have to derive the wellare lunction over a considered time-interval .
lirst, we calculate the time-average number ol momentaneous minds


AppcnJix .: matncmatica| Jcscripticn
!49
with N
P
t the number ol momentaneous minds in lront ol the veil at time t. ll
the impartial observer will experience p
t
, then hisJher integrated well-being
can be written as
p
t
c
t

Jt
N

=1
t
0

As with the values ol lile x

+
and x

-
, these integrated well-being levels now
have to be projected to the positive and negative values p
t
+
and p
t
-
. write v
+

and v
-
as the time-average number ol positive and negative values ol integrated
well-being, and v
R
as a relerence number. 1he wellare lunction can then be
written as a summation (integral) over the momentaneous minds nt
wc p

v
+
v
+
v
R


v
P
t
c
t

Jt
N

=1
t
0

+

p
Jt
N

t
t=1
At
0

1
p

v
-
v
R

v
P
t
c
t

Jt
N

=1
t
0

-
Jt
N

t
t=1
At
0


1his expression does not require a grouping in subsets that represent dillerent
beings with a unique personal identity over time. 1he impartial observer merely
has to ascribe values to the levels ol connectivity between each two
momentaneous minds. 1his is the connectivity lunction. 1he higher the
connectivity, the more the impartial observer believes that two momentaneous
minds belong to the same person over time.
1he connectivity lunction is new in a consequentialist wellare ethic. lt indicates
that not only momentaneous well-being matters, but also connections between
momentaneous minds matter. 1he connectivity lunction can be represented as a
web where the momentaneous minds are the nodes and the threads connect the
dillerent momentaneous minds. 1he thicker the thread, the more the two minds
are psychologically and physically connected, the more they can be said to belong
to the same person over time, and the higher the connectivity lunction will be.
Now the problem ol replaceability can be understood in a simple way: il you kill
a person (who has a strong identity over time) and replace him by another person
whose momentaneous minds are not connected to the minds ol the killed person,
it is as il you cut some threads in the connectivity web. 8ome connectivities are set
to zero. lence, the wellare lunction decreases. ln other words: not only the levels
ol well-being at the nodes ol the web (the momentaneous minds) have moral
(intrinsic) value, also the threads between the nodes have moral value.
tnica| ccnsistcncy anJ Anima| ua|ity
!0
let's study the consequences ol dillerent connectivity lunctions. lirst, the most
trivial connectivity lunction is c

, i.e. the connectivity is


inlinite il nt n

and 0 otherwise.
1
1here is no connectivity between
dillerent momentaneous minds, there is no personal identity over time. ln this
case, the dillerent momentaneous minds are treated as completely dillerent
persons. lt is as il all persons brielly pop up into existence. All persons
immediately die and are replaced by other persons. ln this case, the problem ol
replaceability will not be solved: momentaneous minds are lully replaceable,
because they are in lact replaced all the time.
1he other extreme is to take c

, i.e. all momentaneous minds at all


times are connected, as il there was only one superperson who experiences
everything. ln this case, we end up with sum-utilitarianism, simply adding up the
momentaneous well-being ol all minds. ll the connectivity lunction remains equal
to 1 even when persons are replaced, the problem ol replaceability will not get
solved: replacing a person will not inlluence the wellare lunction.
ln between the above two extreme options lor the connectivity lunction, there
is an interesting one. 8uppose the momentaneous minds can be easily grouped
into sets that correspond with persons having a clear personal identity over time.
we can write the connectivity as
c

IJ

nt n

NI


with IJ

nt n

il the two momentaneous minds nt and n


belong to the same person i, and it is 0 otherwise. N(1) is a normalization lactor
that can be included in order to avoid a kind ol double counting (the wellare
lunction has a double integral, so the well-being ol some momentaneous minds
will be counted multiple times). lor simplicity we can taken N(1)=1.
18
1he lactor
R

might correspond with the impartial observer's probability to


experience momentaneous mind n

given the experience ol nt ln another


interpretation, the parameter R

is a relerence time-length ol individual i and I

is
the lilespan ol that individual. we brielly encountered this expression in the
section on the liletime perspective (section 4.2.4). As was mentioned there, the
relerence time-length was related to the psychological connectivity. 1he above
expression ol the connectivity lunction clearly demonstrates this relation. Alter


1
mathematically speaking, this is the uirac delta lunction.
18
uue to this normalization lactor, the connectivity lunction approaches the uirac delta lunction when
I .
AppcnJix .: matncmatica| Jcscripticn
!1
plugging the connectivity lunction in the wellare lunction and setting N(1)= 1, we
get a simplilied expression:


+

+

R


p
N

=1

1
p

=1

with

+
the time-average momentaneous well-being ol individual i, averaged over
the lilespan

, and projected on the positive values. we encountered this wellare


lunction belore, when the value ol lile equals


As mentioned belore, this expression avoids the problem ol replaceability. lt
also presents a new solution to the excessiveness (demandingness) problem ol
prioritarianism raised by loltug (200): the animals with the shortest lives are the
worst-oll and should get the highest priority. As humans have a long lilespan,
humans should sacrilice a lot in order to increase the liletime well-being ol
animals such as lrogs.
1his prioritarianism is very demanding lor humans. 1o weaken the
demandingness objection, loltug (200) proposed a time-slice prioritarianism: at
each separate moment ol time, momentaneous well-being should be distributed in
a prioritarian way. 1his is less demanding lor humans than a liletime perspective
prioritarianism, because as we have seen above, the gap in momentaneous well-
being between a human and a lrog is lower than the gap in liletime well-being (the
dillerence in lilespan and the dillerence in psychological connectivity between
humans and lrogs add to the gap in liletime-well-being).
lut loltug also mentioned some counter-intuitive problems ol his proposal. lis
time-slice prioritarianism is insensitive to inter-temporal compensations that can
take place in a lile. 8uppose individual A has a high liletime well-being, but at time
t she has a very low momentaneous well-being. ln contrast, individual l has a high
momentaneous well-being at time t but a low liletime well-being. 1hen time-slice
prioritarianism says that at time t well-being should be redistributed lrom l to A.
lut il this is the only moment when a redistribution between A and l is possible, A
ends up with an even higher liletime well-being.
ln contrast, the above presented connectivist prioritarianism not only has a
weaker demandingness objection just as loltug's proposal, but it avoids the
counter-intuitive conclusion ol time-slice prioritarianism as well. with the above
expression, il the animal has a low connectivity and hence a low value k
i
, a unit
increase in average well-being or in lilespan will give a relatively low increase in
liletime well-being, compared with a unit increase in average well-being or in
lilespan lor a person having a strong connectivity (a high value k
j
). 1his will
tnica| ccnsistcncy anJ Anima| ua|ity
!2
compensate the priority lor the shortest lilespan, resulting in a less demanding
ethic lor humans.
we can also look at the issue ol abortion. As mcmahan (2002) argued, killing an
adult human is worse than killing an early human embryo, because the human has
a stronger connectedness with his luture sell than the embryo has with hers. lor
the embryo, death means the loss ol a higher amount ol luture well-being, because
the embryo has a longer potential luture ahead ol her compared to the adult. lut
this amount ol luture well-being should be multiplied by the psychological
connectedness that the embryo has with her luture sell at the moment she is
killed. As this psychological connectedness is very low, the result is a lower loss ol
weighted luture well-being lor the embryo.
1he above can be mathematically expressed with the integrated well-being

, which includes the connectivity lunction. 1his means that the dillerence
between

(the integrated well-being ol the non-killed embryo at time t) and

(the integrated well-being ol the embryo when she is killed at time t) is lower
than the dillerence between

(the integrated well-being ol the non-killed


human adult at time ) and

(the integrated well-being ol the adult when


he is killed at time ).
lven il this dillerence ol the integrated well-being is lower lor the embryo, it
does not yet imply that the dillerence ol the wellare lunction between killing and
not killing the embryo is lower compared to killing and not killing the adult. 1hat
is because killing an embryo not only results in a slightly lower integrated well-
being

, but it also results in the loss ol all luture integrated well-being states.
1he latter loss can easily be bigger than the loss ol all luture integrated well-being
states ol the adult.

ln this intermezzo l argued that not only the momentaneous experienced well-
being is important in a consequentialist wellare ethic. 1he momentaneous well-
being is experienced by momentaneous minds, and these minds lorm the nodes ol
a vast connectivity web. letween two nodes can be a link: the connectivity that
measures how strongly the two momentaneous minds are psychologically
connected. Not only has each node a moral value, the momentaneous well-being,
but also each link has a moral value, the connectivity. lowering the value ol a
node lowers the wellare lunction, but also lowering the connectivity between
nodes lowers the wellare lunction.
with this connectivist wellare ethic, we get a new, elegant relraming ol the
replaceability problem: killing and replacing persons that have the same
momentaneous well-being means cutting links between nodes, setting some
AppcnJix .: matncmatica| Jcscripticn
!!
connectivities to zero. 1he number and values ol the nodes remain the same, but
the connectivity web is less connected. lence, the wellare lunction decreases even
il the momentaneous well-being remains the same.
1he connectivity web also allows lor intrapersonal, intertemporal shilts in well-
being: well-being can be shilted lrom node to node, lollowing a connected path
between the nodes. 1he connectivity allows lor those shilts, because connectivity
means that the two nodes belong to the same person. lowever, il the connectivity
between two nodes is very weak, not much well-being is allowed to be shilted lrom
one node to the other. ll there is a weak connectivity between my current sell and
my luture sell over !0 years, my luture sell should be treated as an almost
dillerent person. As a consequence, smoking today will not only be imprudent but
also to some degree immoral: it harms (without permission) an almost dillerent
luture person who will get cancer.
linally, the connectivity is not necessarily a binary lunction that takes only two
values: one (personal identity between two momentaneous minds) and zero (no
identity). 1his means that it is not an all-or-nothing issue whether two
momentaneous minds belong to the same person: it can be a matter ol degree.
lersonal identity over time can be more lluid, especially il luturistic thought
experiments such as teleportation, mind copying, mind splitting and mind
swapping would one day become reality. we now already have an ethical theory
that is lit to deal with those tricky situations.

wnat aucut jracticna| ccnscicusncss
my intuition says that consciousness is an all-or-nothing issue: either it is
switched on or it is switched oll, either the neural program runs correctly or it
doesn't. 8till one might object that scientists might discover that consciousness
comes in matters ol degree. we should distinguish quality versus quantity ol an
experience. A lractional consciousness deals with the quantity ol an experience,
not the quality. 1he quality can have e.g. an intensity, and intensity ol leelings can
come in degrees. lor example pain is stronger than another. 1his dillerence in
intensity is a dillerence in the quality ol an experience: a weak pain leels dillerent
lrom a strong pain. On the other hand, lractional consciousness deals with the
quantity ol an experience and asks the question: what il there is not an integer,
but a lractional number ol momentaneous minds7 what il consciousness itsell is a
matter ol degree instead ol an onJoll switch7 lostrom (2006) ollers the most
challenging thought experiments to argue that at least in theory, lractional minds
are conceivable. lostrom's thought experiments extend those ol larlit (1984) and
deal with brain-duplication to argue that consciousness comes in degrees, that
there might be lor example 1, instead ol two conscious people.
tnica| ccnsistcncy anJ Anima| ua|ity
!4
1he above wellare lunction can be easily adapted to deal with lractional
consciousness. lirst, we replace the integer number ol momentaneous minds at
time t by:


with

a lractional number lor a momentaneous mind

. 1his lractional
number can be e.g. 0, or 1,!2 or whatever. lt corresponds with the lractional
number ol minds that have the same mental state, i.e. the mental state ol the
momentaneous mind

. ln the numerator ol the wellare lunction, we replace

. ll the lractional number is 1, we get the previous wellare


lunction. ll a momentaneous mind comes with a very low level consciousness, the
lractional number is so low that the momentaneous mind does not strongly
inlluence the wellare lunction.
Lncertainty aversion
ln the previous sections, we assumed a unilorm probability distribution behind the
veil ol ignorance: the impartial observer knows that sJhe will be born as individual
i with probability 1JN. 1his assumption can be justilied: it is the only probability
distribution with maximal inlormation entropy, i.e. least inlormation content
(cover & 1homas, 1991, chapter 11), at least il you have no lurther inlormation
about, lor example, what the average well-being or the standard deviation will be.
lowever, we can also assume that behind the veil, you do not know the
probability distribution, and you have an uncertainty aversion.
19
1his is what
kawls originally intended in his theory ol justice (kawls, 191). A person who has a
maximum (unrestricted) uncertainty aversion prelers playing any gamble with
known probabilities above a gamble with unknown probabilities. Any gamble also
means: the gamble with the worst probability. 1he maximum uncertainty averse
person would preler playing the gamble where sJhe will become the worst-oll
person lor sure, instead ol playing the gamble where sJhe does not know the
probability to become the worst-oll person. As mentioned above, the weights a
i

reler to the probability to get a liletime well-being level

. lence, the worst


probability distribution is the one with a
1
=1, i.e. the highest probability to become
the worst-oll individual. As with the situation ol maximum risk aversion,


19
kisk aversion assumed a knowledge about the probability distribution, and resulted in a concave
utility lunction j=x
p
. Lncertainty aversion assumes a lack ol knowledge ol the probability distribution.
AppcnJix .: matncmatica| Jcscripticn
!
maximum (i.e. unrestricted) uncertainty aversion also results in maximin (see
oilboa & 8chmeidler, 1989).
oajdos & landil (2008) gave a mathematical prool lor a wellare lunction that is
a linear combination ol maximin and average utilitarianism
20
:
w
uK
0x
1
0x
1

where x
1
is the worst-oll and is a parameter between 0 and 1. ln other words:
a
1
= 1J(1-(N1)(1 )) and a
i-1
=(1 )J(1-(N1)(1 )). 1he prool is based on a
restricted kind ol uncertainty aversion (restricted mixture neutrality). No
uncertainty aversion corresponds with = 0, unrestricted (maximum) uncertainty
aversion corresponds with = 1.
1he restricted mixture neutrality used in the prool, is only one way to restrict
uncertainty aversion. Although l do not prool it here, l postulate that there are
other restrictions ol uncertainty aversion that will result in other weights a
i
, and
perhaps one kind ol restriction results in a moderate egalitarianism (}ensen, 200!)
discussed below.
moderate egalitarianism
like oajdos & landil (2008), l choose the weights in such a way that we get average
utilitarianism and maximin as limits. 1ake lor example:
o
I
N

I-1

]-1 N
]=1

with . ll q=0, we get maximin, il q=1 (and the power p=1, i.e. no risk
aversion), we get average utilitarianism. A wellare lunction with the above
weights is called a generalized oini wellare lunction (weymark, 1981). when
distributing benelits, the worst-oll should get a priority because they have a
higher weight lactor a
|
. ln contrast with (power mean) prioritarianism, the priority
in this moderate egalitarianism depends on the position ol an individual rc|ativc tc
tnc wc||-ucinq |cvc|s cj tnc ctncrs. lrioritarianism uses a concave utility lunction u(x)
that does not depend on the relative position (relative to the well-being ol others).
1his generalized oini wellare lunction can be derived as lollows. lirst, start
with a population ol two individuals and two situations: X x
1
x
2
, and
y
1
y
2
. we can write three consequentialist theories ol justice in a simple
set ol mathematical inequalities:
8trict egalitarianism: X is better than Y il and only il


20
oajdos & landil assumed a risk neutral but uncertainty averse impartial observer.
tnica| ccnsistcncy anJ Anima| ua|ity
!6

2

1

2

1

i.e. the dillerence between the values ol lile should be minimized.
maximin: X is better than Y il and only il

1

1

i.e. the value ol lile ol the person in the worst position should be
maximized.
8um-utilitarianism: X is better than Y il and only il

1

2

1

2

i.e. the total value ol lile (total utility) should be maximized.
1hese expressions can be unilied in one inequality

2

1

2

where the parameter q takes the values:
: strict egalitarianism,
: maximin,
: utilitarianism.
we can write
1

2
as the wellare lunction ol situation X.
oeneralizing to situations with N number ol individuals can be done using a
recursive relation. 1he wellare lunction ol situation X then reads:

1

N

1

3

-1

N
=1

ln order to avoid the repugnant conclusion, we have to normalize this wellare
lunction with the sum
]-1 N
]=1
. 1he result is the oini wellare lunction

un



-1

N
=1

]-1 N
]=1


keplication invariance
1he weight can be adapted to make the theory replication invariant. ll we start
with a world with two individuals and


1

2
, then replication invariance
means that making a copy ol this world, i.e.


1

1

2

2
, does not
change the wellare lunction. Lsing the above weights, the generalized oini wellare
lunction does not remain invariant under replication (see e.g. lleurbaey et al.
2009). more generally, start with the N-individual situation


N
and replicate this k times


KN

,
with
1

1

,
2

K+1


2K

and so on lor the


other

. Applying the wellare lunction gives


AppcnJix .: matncmatica| Jcscripticn
!
w
un
x

=

]-1
x
]
N
]=1

k-1 N
k=1
=
-
K
-
KN

KI-1
x
I
=

KI-1
x
I
N
I=1

Kk-1 N
k=1
N
I=1

1his expression is not the same as the above w
un
x

, but there is a similarity.


ll we rescale
K
, we get the same expression. 1hat means that invariance
under replication is restored il the parameter q depends on the number ol
individuals. 8o we can start with a lixed parameter in the two-individuals world:

2
. 1he parameter in a world with N individuals then becomes
N
=
2
2N
. 1his is
in line with our intuition: the more individuals, the higher the parameter
N

should become, because otherwise the higher well-being levels will rapidly get
extremely low weight lactors. 8o the replication invariant expression lor the
wellare lunction ol oini-moderate egalitarianism becomes:
w
un
x

b =

2
2
N
I-1
x
I
b
N
I=1

2
2
N
k-1
N
k=1

1his shows that a geometric generalized oini wellare lunction can be made
replication invariant, il the weights properly depend on the population size N.
21

Avoidance ol intransitivity and misery lor the ultra rich
1he moderate egalitarian wellare lunction solves 1emkin's intransitivity problem
(1emkin, 198). 1his intransitivity problem comes down to a line ol reasoning
where choice A is better than l, and l better than c, but c is better than A (as in
the rock-paper-scissors hand game). ln his argument, 1emkin relers to a lirst
8tandard View (l8V) and a 8econd 8tandard View (88V).


21
As with power mean prioritarianism, we can adapt this expression to deal with non-trivial
psychological connectivities. write p
nt
= c
ntn

p
n

(t

)
Jt
N
F
(t

)
n

=1
t
0

in increasing order,
and nt = v
P
t with N
P
tJt
At
0
= v
P
t. 1hen we can write w =


v
F
At
v
F
At
0

d:

v
F
At
v
F
At
0
d:

when
c
ntn

=
R
i
1
i
+R
i
IJ

nt n

, the lunction p
nt
becomes an increasing step lunction between 0
and v
P
t where each step corresponds with the lile ol an individual (and the width ol a step, 1
i
,
corresponds with the lilespan 1
i
such that I

= v
P
t
N
F
=1
). 1he wellare lunction then reduces to
w =

T
] ]<i
v
F
At
1-
T
i
v
F
At
x
i
N
F
i=1
1-
with x

=
1
i
R
i

i
1
i
+R
i
. 1his is a slight generalization ol the oini wellare
lunction: only when all 1
i
-1
j
, i.e. all lilespans are equal, we get w
un
x

=

1
N
F
i-1
x
i
N
F
i=1

1
N
F
k-1
N
F
k=1
.
tnica| ccnsistcncy anJ Anima| ua|ity
!8
1he l8V states that a situation in which one person sullers a lot is better than a
situation where two people suller a lot, but their sullering is a little bit less than
the sullering ol the one person in the lirst situation. ln other words: two almost
extreme sullerers is worse than one extreme sullerer. And three almost almost
extreme sullerers is worse than two almost extreme sullerers. And lour almost
almost almost extreme sullerers is worse still. continuing in this way, we end up
with a situation where a very high number ol people suller only a little bit (lrom a
light headache). ll transitivity applies, then the l8V means that this linal situation
is the worst ol all.
lut that conclusion violates another intuition, the 88V, which says that the lirst
situation, where everyone is really happy except one extreme sullerer, is much
worse than the linal situation where a very high number ol people are almost
really happy and no-one really sullers.
1his 88V corresponds with the problem ol the 'misery lor the Lltra-kich'. One
critique ol sum-utilitarianism and (power mean) prioritarianism is that according
to these theories, there are situations where it is good to sacrilice the well-being ol
a worst-oll person il this sacrilice results in a huge benelit that can be distributed
amongst a huge amount ol best-oll people (the ultra rich), each getting a tiny
share that increases well-being. As long as the total number ol ultra-rich
beneliciaries is high enough, any amount ol cost lor the worst-oll will be ollset by
the sum ol small benelits lor the many ultra rich. 1he ultra rich can get richer
while the poorest person gets poorer. 1his seems counterintuitive according to
many people (lroome, 2004 p.8, loltug, 2006 p.1!4, uorsey, 2009, p.4).
let's apply the intransitivity problem to the equations ol moderate
egalitarianism with the oini wellare lunction. we lirst note that in contrast with
the repugnant conclusion, in the intransitivity problem the total number ol people

z
is constant and very large. let's start with history
1
where
1
(much smaller
than
z
) ol extreme sullerers have a well-being w below the happy state u, and all
the other
z

1
people are in the happy state u. 8o we get:

un

1


-1

]-1

]=N

+1
N

=1

k-1
N

k=1

with

N


2
2N


ln history
2
there are
2

1
people with well-being
1
(with
1

small but positive), and the others still have well-being u. lor history
3
there are

3

2
people with well-being
1

2
. moving on, we get history
p

where
p
people have well-being
p
, with
k
p-1
k=1

p
.
AppcnJix .: matncmatica| Jcscripticn
!9
let's suppose that the l8V is always valid. 1hen lrom the inequality

un

1

un

p
and using

-1
N
=1
=
-
N
-

we get
-
-
N

-
+
-
N

-
-
-
N

-
( - +
p
)
-
N

-

when
p
becomes very large, and q is lower than 1, the lactor
N

goes to zero.
1his simplilies the above inequality to:
p

N

< .
we can conclude that il the l8V is always satislied, the total increment
p
is
always strict|y lower than w. 8o we will never be able to get the well-being
arbitrarily close to u. ll we want to proceed and move close to u, we at one point
have to violate the l8V. 8o we have an history
]
where the inequality llips:

1
>
2
> > (
]
) < (
]+1
) < Only when q = 1 (sum-
utilitarianism) we have the constraint
p
that allows us to move close to u.
And il q = 0, then
p
and l8V is always violated. moderate egalitarianism
solves 1emkin's intransitivity paradox, in the sense that it has a point in the series
where the l8V no longer becomes valid and the 88V takes over.
1he problem ol independence and Allais paradox
As we have seen, unweighted power mean prioritarianism has a wellare lunction
that has the property ol independence or strong separability (see e.g. mccarthy,
2008), except lor mixed populations and variable populations. 1he situation with
moderate egalitarianism is worse: strong separability is violated, even when
populations are not mixed and population size is constant.
8uppose we have two times two histories, each involving three individuals.
listory n
1
=(1,1,1), n
2
=(1,0,),
1
=(0,1,1) and
2
=(0,0,). we see that history
1
is
related to history n
1
, just as history
2
is related to n
2
. 8uppose the impartial
observer behind the veil can choose between either n
1
and n
2
or between
1
and

2
. lndependence now says that
1
is better than
2
il and only il n
1
is better than
n
2
. 1he presence ol the lirst person should not matter (whether sJhe has well-
being 1 in the lirst two histories, or well-being 0 in the last two histories). we can
apply the wellare lunction
un
. when q 1J2, we get
un

1
= + + >

un

2
= + . lut il q - 1J4, we see that
un

1
= + <

un

2
= ln other words, when q is in the range
1
4
< <
1
2
, it is possible
that situation
2
is better than situation
1
, although situation n
2
was worse than
situation n
1
. lndependence is restored when we exclude the lirst person.
tnica| ccnsistcncy anJ Anima| ua|ity
!60
ln decision theory, Allais paradox (Allais, 19!, lahneman and 1versky, 199) is
similar to the problem ol independence. An example ol the paradox goes as
lollows. 8uppose we have two experiments, each consisting ol two gambles. ln the
lirst experiment, you can choose between gamble 1, where you have probability
100/ to gain $100, and gamble 2, where you have probability 89/ to gain $100, 1/
to gain nothing and 10/ to gain $00. most people would choose to play gamble 1,
especially when they have risk aversion, because in gamble 2 they risk to gain
nothing. ln gamble 1 they are always certain to receive some benelit.
ln a second experiment, people can choose between gambles 1' and 2'. oamble
1' has the lollowing probabilities: 89/ to gain nothing and 11/ to gain $100.
oamble 2' has 90/ to gain nothing and 10/ to gain $00. ln this second
experiment, people would preler to play gamble 2', because they gain $00 at an
almost equal probability as gaining $100 in gamble 1'. 1he curious thing is that
both experiments are in lact quite similar, as can be seen by writing it as in the
lollowing table.
Gamble 1 Gamble 2 Gamble 1` Gamble 2`
89% 100$ 89% 100$ 89% 0$ 89% 0$
1% 100$ 1% 0$ 1% 100$ 1% 0$
10% 100$ 10% 500$ 10% 100$ 10% 500$

ln the lirst experiment, we can decouple a 89/ probability to gain $100. 8o we
can write lor gamble 1 that you have 89/-1/-10/ to gain $100. lut il we would
now set the gains lor those 89/ in gambles 1 and 2 equal to zero, we get gambles 1'
and 2'. 8o it is strange why people preler gamble 1 over 2, but 2' over 1', because
gamble 2' is quite similar to gamble 2, and 1' is similar to 1. oambles 1 and 2 are
both changed in the very same way (setting a gain to zero), and yet the order ol
prelerence ol the gambles suddenly changes.
1he analogy with this Allais paradox and the above veil ol ignorance example is
clear.
22
ll the impartial observer behind the veil would have prelerences as in
Allais paradox, sJhe could end up with a wellare lunction ol moderate
egalitarianism.


22
ln the veil ol ignorance example, the impartial observer might ascribe a probability 1J! lor each
individual, because we have a population ol ! persons. ll we would like to use the probabilities 89/, 1/
and 10/ as in the lormer example, we can take a population ol 100 individuals, whereby 89 individuals
have well-being 1 in the lirst history, and so lorth.
AppcnJix .: matncmatica| Jcscripticn
!61
8ummary
ln summary, we basically have two theories that have maximin as a limit:
moderate egalitarianism with a generalized oini wellare lunction (having maximin
in the limit ) and prioritarianism with a power mean wellare lunction
(having maximin in the limit when the power ). moderate egalitarianism
solves 1emkin's intransitivity problem, but this theory is not strongly separable,
not even when population size is lixed. lower mean prioritarianism does not solve
1emkin's intransitivity paradox, because it always respects the l8V and hence it
laces the problem ol the misery lor the Lltra-kich. ll the latter problem is
considered less bad than the problem ol independence (strong separability) lor
unmixed, lixed populations (i.e. lor situations with constant numbers ol people
with positive and negative levels ol well-being), then prioritarianism is the better
theory. loth moderate egalitarianism and power mean prioritarianism can be
derived lrom a veil ol ignorance with respectively uncertainty aversion and risk
aversion. loth theories can be unilied in one expression: a weighted power mean.
lrioritarian theories lor lotteries
ln the above descriptions, the impartial observer had to choose between dillerent
histories lrom behind the veil ol ignorance. lowever, it is possible that outcomes
are not certain. ln this case ol probabilistic outcomes, we have to apply the theory
to lotteries (kabinowicz, 2002, mccarthy 200!, 2006, 2008, Otsuka & Voorhoeve,
2009).
A lottery can be written as a set ol m histories, where each history has a
probability and all probabilities sum to one:
1

1

2

2

m

m

with


m
=1
and

the r-th possible history. lrom behind the veil ol


ignorance, we now have to choose between dillerent lotteries (instead ol dillerent
histories). 8o there are now two elements ol risk: lirst you don't know who you
will be, and second you don't know which history will be actualized.
we can write the expected well-being ol individual i in lottery over all
histories as:

m
=1

1he wellare lunction
+
can take three dillerent lorms: an ex-ante, ex-inter
and ex-post (lor simplicity l take again a unilorm distribution a
i
=1).

+
LA

p
N

=1

1
p

m
=1

p
N

=1

1
p

tnica| ccnsistcncy anJ Anima| ua|ity
!62

+
LI

p
N

=1

1
p

m
=1

p
N

=1

1
p

+
LP

p
N

=1

1
p

m
=1

p
N

=1

1
p


ln the ex-ante version, we lirst calculate the expectation values ol the well-
being levels, and alterwards use them in the wellare lunction, whereas in ex-post
we apply the expectation value at the end, alter calculating the wellare lunctions
lor the dillerent histories. lx-inter is a so-called weighted power mean, weighted
by the probabilities ol the lotteries.
let's apply our three theories ol prioritarianism to the lollowing example.
8uppose there are lour lotteries, each with two persons (A and l) and two histories
(n
1
and n
2
). 1he values ol well-being are summarized in the lollowing table.
L
1
L
2
L
3
L
4
h
1
h
2
h
1
h
2
h
1
h
2
h
1
h
2
A 2 2 2 0 2 0 1 1
B 0 0 0 2 2 0 1 1

1
is the worst lottery, because we know lor sure that only A will win (l will get
0), ending up with an inequality. lottery
2
is a bit better, because now both A and
l at least get an equal chance to win.
!
is better still, because we will always end
up with a situation ol equality, and both A and l get an equal probability to win.
linally, lrom a risk averse point ol view, lottery
4
is the best, because now each
person will at least win something, end both will end up equal.
1he lollowing table summarizes the wellare lunctions lor all lour lotteries.
L
1
L
2
L
3
L
4
W
+
EA

1-
1
p
1 1 1
W
+
EI

1-
1
p

1-
1
p

1-
1
p
1
W
+
EP

1-
1
p

1-
1
p
1 1

we see that each ol the three wellare lunctions satislies one ol our judgments.
2!

lor example according to the ex-ante w
A
, the lirst lottery is the worst, the other


2!
As with populations ethics (problems related to variable populations), situations ol risk also has
unavoidable counter-intuitive implications: it is not possible (or very dillicult) to respect all ol our

AppcnJix .: matncmatica| Jcscripticn
!6!
three lotteries are equal. lowever, the ex-ante wellare lunction encounters a
problem when it comes to variable populations, i.e. when the population size N
dillers between histories. ll person l does not exist in history n
2
, we cannot
calculate w
A
.
24

low to decide between the ex-inter and ex-post versions ol the wellare
lunction7 we have seen that lrom behind a veil ol ignorance, not knowing who
you will be, there is some risk involved, and prioritarian theories show a non-zero
risk aversion. when the choice is between lotteries, a second kind ol risk is
introduced. ln the above example, there is the risk ol becoming person A instead
ol l (a risk related to impartiality), and there is the risk ol having one history
instead ol the other (a risk related to lotteries). 1he ex-inter wellare lunction
treats these two risks in a same way: w

has coherent risk aversion towards both


kinds ol risks (this can be seen in the mathematical expression: the two
probabilities p
r
and 1JN
l
are treated in the same way). 1he ex-post wellare
lunction, however, treats these two kinds ol risks dillerently.
1his apparent incoherence ol ex-post prioritarianism is not a real threat to the
theory, because one might argue that the two kinds ol risk are not comparable and
should not be treated in the same way. 1he impartiality risk can be related to the
notion ol separability ol persons, which is - morally speaking - something other
than a separability ol outcomes ol a lottery. ll the separability ol persons is
morally more relevant than the separability ol outcomes, the two kinds ol risks
can be treated dillerently.
lurthermore, Otsuka and Voorhoeve (2009) expressed the intuition that a one-
person decision under the risk ol a lottery should be treated dillerently than a
multi-person decision (without the risk ol a lottery). 1he one-person decision
under risk should lollow the expected utility theory to rellect the 'unity ol the
individual' (see also lorter, 2012). when there is only one individual, the ex-post
wellare lunction indeed becomes the expected well-being:

.
1he multi-person decision should lollow the prioritarian approach to rellect the
separability ol persons. 8o, the incoherence ol ex-post prioritarianism might not
be a weakness alter all, as it might rellect the intuitions shared by e.g. Otsuka and
Voorhoeve.


seemingly sell-evident moral intuitions. 8ee e.g. uougherty (201!) who argued that we have intuitions
both lor and against the ex-ante view.
24
Assuming that person l has well-being 0 in history n
2
would not work, because why should one
include this potential person in the calculation and not include all other potential persons as well7
lncluding all potential persons, setting their non-existent well-being levels equal to zero, will result in a
division by an inlinite population size N.
tnica| ccnsistcncy anJ Anima| ua|ity
!64
1he veil ol ignorance can also be relramed in order to make it compatible with
this apparent incoherence ol the dillerence in risk attitude between situations
with multiple persons versus multiple outcomes. ll there are N individuals in lront
ol the veil, you are going to live the lile ol just one ol those individuals. You will be
person i with a probability 1JN, and it is not irrational to be risk averse with
respect to the liletime well-being ol this one person. On the other hand, il you
would 'reincarnate' N times and live the lives ol all N persons, you will experience
everything and there is no reason to be risk averse towards an individual well-
being behind the veil ol ignorance.
1he point is that behind the veil ol ignorance, lotteries can be understood as
reincarnations. ll in a lottery person i has a probability p-5/1 to have an outcome O
and a probability -(1-5)/1 to have outcome O' (with 1 and 5 natural numbers), it is
as il you will live the lile ol person i a number ol 1 times, 5 ol which you will get
outcome O, and 1-5 ol which you will get outcome O'. lt is as il you will be
reincarnated 1 times into parallel worlds, leading again the lile ol person i. 1here
is no reason to be risk averse il you will experience everything ol the 1 copies ol
person i.
1he ex-post approach does however have a problem ol independence. compare
the above lotteries
2
and
!
. ln both lotteries, the outcomes ol individual A are the
same (outcome 2 ol history n
1
is realized, outcome 0 il the second history is
realized). ll there is independence between the two persons, we can change the
outcomes ol person A in both lotteries in the same way. 1he lollowing table
represents such a translormation into new lotteries
2
and
!
.

2

3
h
1
h
2
h
1
h
2
A 0 2 0 2
B 0 2 2 0

we saw that w

(
!
) - w

(
2
), because
!
has always equal outcomes lor both
individuals (either both 2 or both 0). lut now we have w

(
!
) w

(
2
), so
independence is no longer valid in an ex-post theory (see also lleurbaey and
2uber, 2012).

AppcnJix .: matncmatica| Jcscripticn
!6
combining the prioritarian theory with the basic right and
biodiversity principles
can we incorporate the basic right principle (see chapter 6) and the value ol
biodiversity (see section 10.) in our mathematical lormulation7
l suggest the lollowing possibility: apart lrom their values ol lile, all individuals
have a 'basic right parameter'

, which is zero il the basic right ol individual in


situation (world history) X is not violated, and very large il her basic right is
violated. 1he quantity

can also take dillerent values, depending on what the


'ends' are and how seriously someone's will is violated. with these basic right
parameters, we add a new term to the wellare lunction: the basic right lunction


when the basic right ol person is violated, the value

should be very large,


but not inlinite: there might be a threshold value, above which one prelers the
consequentialist outcome.
2
lor example: when there are, say, a billion people on
the main track ol the trolley dilemma (all threatened), one might be tempted to
push the heavy man lrom the bridge in order to save those billion people. Another
reason why

should not be inlinite, is because it is dillicult to count with


inlinities, as inlinity plus inlinity equals inlinity. 8o when we have to choose
between situation X where one basic right is violated and situation Y where the
basic rights ol two persons are violated, we should be able to conclude that the
situation X is prelerred over Y.
1here is a basic right equality il

when the use as a means ol subject in


situation X is similar to the use ol subject in situation Y (i.e. lor the same kind ol
ends).
1he value ol biodiversity can be incorporated with a term 8 that is proportional
to a measure ol momentaneous biodiversity at time , as il this biodiversity is
a momentaneous well-being ol an ecosystem.


2
1he strength ol someone's basic right (the value r) can depend on both the number ol lives saved and
the number ol lives at risk. ll lor example a trolley is about to kill N number ol people (N lives at risk)
and pushing a heavy man in lront ol the trolley saves m number ol people, the permissibility ol
pushing the man can depend on m as well as on the ratio mJN. An experimental study lound that the
higher N, the higher m needs to be to make the act permissible (kai & lolyoak, 2010). 1his dependency
on the ratio mJN passed a coherence test.
tnica| ccnsistcncy anJ Anima| ua|ity
!66
laving said this, we can now write the new mathematical lormulation ol the
Qmm-principle extended with the deontological rule ol the basic right and the
value ol biodiversity. ln the previous intermezzo, l presented a wellare lunction
that depends on the momentaneous well-being and a connectivity lunction.
8imilarly, we should write the basic right term as r
t
, which represents the level
ol basic rights violations ol the momentaneous mind n at time t. l will also include
the ex-inter approach to lotteries, i.e. using expectation values, written by the
brackets -, that calculate the probability weighted average.
Now we can write a moral weight, which is composed ol three terms: the
wellare lunction ol consequentialist ethics, a basic right lunction ol deontological
ethics and a biodiversity lunction ol environmental ethics. 1his moral weight
lunction reads:
Hp c r b = w
MM
p c +Rr + Bb
=
v
+
v
+
+ v
R


v
P
t
_ __ c
t

(t

)
Jt
N

(t

=1
t
0

+
_
p
Jt
N

t
t=1
At
0

1
p
+
v
-
v
R


v
P
t
_ _ c
t

(t

)
Jt
N

(t

=1
t
0

-
Jt
N

t
t=1
At
0

-

v
P
t
_ r
t
Jt
N

t
t=1
At
0
+

t
_ btJt
At
0


1his moral weight lunction can be considered as the standard model ol ethics,
just as a lagrangian represents the lundamental quantity in the standard model ol
physics (weinberg, 1996). 1he moral weight combines the consequentialist wellare
ethic with the deontological mere means principle and the biodiversity principle.
we should not try to maximize the moral weight directly. lnstead, we should use
the method ol rule universalism: derive those rules that, under universal
compliance, would maximize the above moral weight. 1hose rules should be
lollowed by all moral agents who are capable ol lollowing them.
ln section 6.6, we saw an extended mere means principle, which generates the
tolerated partiality principle. lence, the basic right lunction Rr in the above
expression can reler to the extended mere means principle, which says that we
should not usc nor ccnsiJcr someone as merely a means. Not considering someone
AppcnJix .: matncmatica| Jcscripticn
!6
as merely a means implies that we should allow lor some level ol partiality.
26
ll we
consider a person as merely a means, i.e. il we do not allow that person to be
partial whereas we should allow such degree ol partiality according to the
extended mere means principle, then a negative r-term is added to the moral
weight. Or in other words: when you want to be partial to a degree that should be
universally permissible, and when l prohibit you to be partial in that way, then a
negative term is added to the equation.
2

ln this way, all live principles ol the normative moral hand (section 1!.1) are
included in the above mathematical expression: rule universalism, Qmm-
prioritarianism, the mere means principle, the biodiversity principle and tolerated
partiality. with this expression ol the moral weight, including the tolerated
partiality principle in the basic right lunction, we can get a very rich ethical
system that is compatible with a lot ol moral intuitions that a lot ol people have.
uemocratic impartial prelerences ol moral agents
1he prioritarian theory uses undetermined values: the risk parameter p
28
, the
relerence population size N
k
, the choice between ex-ante, ex-inter and ex-post
decisions, and the values ol lile (liletime well-being levels) x
i
. As mentioned in a
previous section (4.2), the values ol lile are lunctions ol experienced well-being ol
individuals, but these experienced well-being levels are not interpersonally
comparable, just like l cannot compare my perception ol a red color with your
perception ol red. 1herelore, the values ol lile are the values attributed by an
impartial observer behind a veil ol ignorance. 1his impartial observer uses
empathy to guess how situation x lor person i would compare to situation y lor
person j.


26
1he extended mere means principle does not allow lor levels ol partiality where someone's basic
right is violated. 8aving your child by killing another child and using its organs lor transplantation, is
not allowed. 8uch behavior would be too partial.
2
ll you do not want to be impartial, l cannot prohibit you to be partial. 1he only thing l can do is try to
convince you to behave more impartially. 8uppose l convinced you such that you want to be impartial.
1hen l do not consider you as merely a means when l want you to be impartial. ln that case, no basic
right is violated.
28
Or ln() in case ol the exponential liletime well-being in the lolmogorov mean.
tnica| ccnsistcncy anJ Anima| ua|ity
!68
ln reality, we do not have an ideal impartial observer. what we do have, are
moral agents. 1hese are all persons who are able to perlorm the thought
experiment, using empathy or imagination. lach moral agent will imagine
himJhersell as an impartial observer behind the veil.
ln a lirst step, a moral agent a classilies all momentaneous minds in subsets that
represent the N
l
a
dillerent real world persons: the number ol persons in lront ol
the veil, according to moral agent a (see the section on personal identity and
continuity). 1his classilication rellects the importance ol personal identity and
non-replaceability according to moral agent a (see the section on the
replaceability problem above). ll the moral agent has no problem with
replaceability ol persons, sJhe can treat all momentaneous minds independently
as dillerent persons. Alternatively, the moral agent can ascribe a connectivity
lunction between all the momentaneous minds to represent the level ol
irreplaceability ol persons (see the intermezzo).
ln a second step, the moral agent a perlorms the thought experiment, choosing
hisJher prelerred wellare lunction (e.g. a positive number-dampened power mean
prioritarianism with negative total utilitarianism, or an exponential liletime well-
being generalized l-mean), deriving hisJher own prelerred risk aversion
parameter p
a
, population relerence N
k
a
, lottery decision rule and estimates ol
liletime well-being x
i
a
(or momentaneous well-being

and connectivity lunction


c
a
). 1hese parameters and values can lie in a certain range: the broader the range
ol e.g. the risk aversion parameter, the more llexible is moral agent a's attitude
towards risk aversion.
Next, the moral agent calculates hisJher own wellare lunction w
a
(p
a
, N
k
a
, x
i
a
)
with respect to the range ol parameters. Ol course, dillerent moral agents will end
up with dillerent wellare lunctions and dillerent maximizations, even when they
do the thought experiment as sincerely as possible. who is right7 who has the best
risk attitude7 who has the best empathy7 we have to respect the principle ol
universalization: il you are allowed to do the exercise (the thought experiment ol
the veil ol ignorance) using your prelerences (e.g. your risk attitude), then
everyone who is capable (i.e. every moral agent) is allowed to do the exercise using
one's own prelerences.
lurthermore, each moral agent has hisJher own wellare lunction w
a
, but the
wellare lunctions ol dillerent moral agents are not mutually well-calibrated.
uillerent moral agents might use dillerent scales or units lor e.g. the levels ol well-
being. As estimates ol well-being are not interpersonally comparable, a level ol
well-being equal to 100 according to moral agent a might mean a level equal to 10
according to moral agent u. As a result, the wellare lunction used by moral agent a
might have much higher values than the wellare lunction used by moral agent u.
low do we compare those wellare lunctions when they are not properly gauged7
AppcnJix .: matncmatica| Jcscripticn
!69
1hese problems can be solved as lollows. consider the set 6 ol all objective (i.e.
quantiliable) distributable goods and burdens. 1hese are goods and burdens in the
sense that they can positively or negatively inlluence someone's well-being, they
are distributable in the sense that we are able to distribute those goods between
sentient beings (the goods are in our direct control, whereas someone's well-being
is not in our direct control), and they are objective in the sense that they should be
measurable (quantiliable). 1he latter property is important, because we want to
avoid subjective estimates that diller between moral agents. lxamples ol such
objective distributable goods are resources (economic wealth, income, energy,
materials) and liberties (e.g. primary goods according to kawls (191), capabilities
according to Nussbaum (2000)). 1he goods are subject to some constraints, such as
maximum available resource levels and logically possible distributions ol liberties.
moral agents can now do the thought experiment as sincerely as possible,
distributing the goods and burdens to maximize their own prelerred wellare
lunctions, respecting the constraints on the goods. lach ol the momentaneous
minds that compose person i gets a part ol the distributed goods compatible with
the liletime well-being

u
where q
a
is the distribution ol goods (lor example a
person i gets a certain income at a certain time). lach moral agent a can now
calculate the optimal distribution ol goods
u
opt
that maximizes moral agent a's
wellare lunction at
mux
u
. As 6 is a compact (bounded) set, the maximal wellare
lunction is linite (it cannot grow to inlinity).
we can now maximize the average ol weighted wellare lunctions

mux
u
N

u=1

with
u
the number ol moral agents. 1his averaging means that the subjective
estimates ol all moral agents count equally.
29
Lsing the relative wellare lunctions


29
1here are three subtleties. 1) uo all moral agents (who exist and will exist here and everywhere) have
a vote, or do only those moral agents who are able to inlluence the distribution ol the respective goods
have a vote7 ldeally, l would say all moral agents should have an equal vote, but this is impossible in
practice, so l go lor the second option: only those moral agents who exist at the time ol decision
making and who can inlluence the decision, count. 2) what il the prelerences (e.g. the level ol risk
aversion) ol a moral agent changes over time7 lither we take the average ol the prelerences ol the
moral agent over hisJher lile, or only the prelerences at the moment ol decision making. l would leave
this choice up to the moral agents themselves. most importantly, we have to be aware that the moral
agents do not have cognitive biases at the moment ol decision making. lllects ol e.g. lraming and
priming or the inlluence ol e.g. smells (triggering disgust) should be avoided. lxperimental moral
psychology can help us to determine all inlluences that generate such biases. !) what il all except one
ol the moral agents do not leel an emotional problem with the resulting average distribution7 lmagine
that you would leel very unhappy when no-one except you has risk aversion behind the veil, i.e. when

tnica| ccnsistcncy anJ Anima| ua|ity
!0
(w/w
max
) implies that the optimal distribution ol goods according to moral agent a
is as valuable as the optimal distribution ol goods according to moral agent u. 1his
solves the problem ol gauging the wellare lunctions ol dillerent moral agents.
lven il moral agents might use dillerent scales or units lor e.g. the levels ol well-
being, taking the relative wellare lunction gives a good calibration lor the wellare
lunctions.
1his is the democracy in impartial prelerences ol moral agents (a slightly
dillerent approach was proposed in moreno-1ernero & koemer, 200). lt turns a
collection ol subjective estimates into an objective, impartial solution.
ln reality, this optimal impartial distribution ol goods will be very dillicult to
determine, because each one ol all the moral agents have to imagine being each
one ol all the current and luture living sentient beings, experiencing each
momentaneous well-being. lt is like using the standard model ol elementary
particle physics to solve problems with many particles (e.g. to study the elliciency
ol a combustion engine). low do physicists deal with the complexity7 1wo
considerations can make the principle easier to apply in daily lile situations.
lirst, we can restrict the set ol moral agent and sentient beings: in a lot ol daily
lile situations, only a limited number ol sentient beings are measurably
(relevantly) allected by the policy ol only a limited number ol moral agents. 1hese
moral agents can hence apply the democratic approach, limiting the thought
experiment ol the veil ol ignorance to only those sentient beings who are
measurably allected. 1his is similar to what physicists do when they study
interactions between a lew number ol particles (or a lew number ol planets and
stars), assuming interactions with a background lield ol particles lar away are
negligible.
8econd, il we know that moral agents do not have strongly divergent risk
attitudes (that they can easily lind a consensus on the issue ol priority lor the
worst-oll), and il the liletime well-being estimates according to those moral agents
are not so divergent when considering important, extreme cases (e.g. cases ol


everyone chooses sum-utilitarianism whereas you really would want that we give some priority to the
worst-oll. lor you, prioritarianism is very important. 8o imagine, that your well-being in a chosen sum-
utilitarian ethic will be much lower than everyone's well-being in a chosen prioritarian ethic. ll
everyone has one vote, the result is close to a sum-utilitarian ethic, which has lower well-being due to
your unhappiness. 1his loss ol well-being should be taken into account, so a weighted democratic
voting might be required. 1his is comparable to situations in physics, where a gravitational lield
created by heavy objects inlluences the path ol another object, but the presence ol the mass ol this
other object itsell also inlluences the gravitational lield to which it is subjected. 1his results in
complex, non-linear interactions and ellects. 1he (physicalJmoral) lorce (the gravitational lieldJthe
wellare lunction) is inlluenced by the objects (the massesJthe decisions ol moral agents).
AppcnJix .: matncmatica| Jcscripticn
!1
extreme poverty or lactory larms), we can easily derive rules ol thumb to quickly
decide the best estimate lor the optimal distribution q
cpt
. compare it with
thermodynamics in physics: when a lot ol particles interact, physicists use
approximations (rules ol thumb) to study e.g. the elliciency ol combustion engines
(l used the same analogy between branches ol physics and approaches to ethics in
section !.). 1hese approximated rules are e.g. the laws ol thermodynamics and
lluid mechanics. ln the end, they can be derived lrom the standard model ol
elementary particles, using a lot ol statistical mechanics. lut as engineers do not
use the standard model ol elementary particles to study combustion engines,
ethicists and politicians do not have to use the complex wellare lunction all the
time. 1hey can derive equations or laws that are easier to apply in those daily lile
situations. And these new, derived moral rules can look completely dillerent than
the underlying Qmm prioritarian principle, just as the laws ol thermodynamics
look completely dillerent than the standard model in particle physics.
Ol course, even when there was only one moral agent, sJhe should appeal to
approximate, derived moral rules, because it will be practically impossible to
estimate past and predict luture levels ol liletime well-being and to calculate the
wellare lunction over a vast period ol time. 8cience is necessary to tell whether
the derived moral rules match the target principle ol Qmm prioritarianism.


!2
libliography
AuA (2009). losition ol the American uietetic Association: Vegetarian uiets, jcurna| cj tnc
Amcrican uictctic Asscciaticn, 109():1266-82.
Adams, c. (199). 1nc 5cxua| c|itics cj Vcat, continuum, New York.
Adams, c. (199b). Ncitncr Van ncr 8cast: lcminism anJ tnc ucjcnsc cj Anima|s, continuum,
New York.
Adams c. & uonovan }. (1996). 8cycnJ Anima| kiqnts: A lcminist carinq tnic jcr tnc 1rcatmcnt
cj Anima|s, continuum lntl lub oroup.
Ahluwalia, A. (198). An intra-cultural investigation ol susceptibility to 'perspective' and
'non-perspective' spatial illusions. 8r. j. sycnc|., 69(2): 2!!-241.
Allais, m. (19!). le comportement de l'homme rationnel devant le risque: critique des
postulats et axiomes de l'cole Amricaine, ccncmctrica, 21: 0!-46.
Allen, m.w., wilon m., Ng, 8.l. & uunne m. (2000). Values and beliels ol vegetarians and
omnivores. jcurna| cj 5ccia| sycnc|cqy, 140 (4): 40-422.
Allen, m. w., oupta k. & monnier A. (2008). 1he lnteractive lllect ol cultural 8ymbols and
luman Values on 1aste lvaluation. jcurna| cj ccnsumcr kcscarcn, !: 294-!08.
Arneson, k.}. (1989). lquality and lqual Opportunity lor wellare. ni|cscpnica| 5tuJics, 6:
-9!.
Arneson, k.}. (2008). lquality ol Opportunity. 5tanjcrJ ncyc|cpcJia cj ni|cscpny. lall 2008
ldition.
Arrhenius, o. (2000). luturc 6cncraticns: A cna||cnqc jcr Vcra| 1nccry, lhu dissertation,
Lppsala Lniversity.
Arrhenius, o., kyberg, }. & 1nnsj, 1. (2010). 1he kepugnant conclusion, 1nc 5tanjcrJ
ncyc|cpcJia cj ni|cscpny (lall 2010 ldition), ldward N. 2alta (ed.).
Arrow, l.}. (196!). 5ccia| cncicc anJ nJiviJua| va|ucs, 2
nd
ed., New laven and london: Yale
Lniversity lress.
Arrow, l.}. (196). 1nc 1nccry cj kisk Avcrsicn, in Aspccts cj tnc 1nccry cj kisk 8carinq, by Y.}.
8aatio, lelsinki.
Atkinson, A.l. (190). On the measurement ol lconomic lnequality. jcurna| cj ccncmic
1nccry, 2 (!):244-26!.
larilan, Y.m. (200). 8peciesism as a precondition to justice. c|itics anJ tnc ijc 5cicnccs,
2!(1):22-!!.
leauchamp, 1.l. & childress, }.l. (2001). rincip|cs cj 8icmcJica| tnics, th ed. New York:
Oxlord Lniversity lress.
lekoll, m. (200). 1nc mcticna| ivcs cj Anima|s: A caJinq 5cicntist xp|crcs Anima| jcy,
5crrcw, anJ mpatny anJ wny 1ncy Vattcr. New world library.
lekoll, m. & lierce, }. (2009). wi|J justicc: 1nc Vcra| ivcs cj Anima|s. Lniversity ol chicago
lress
lenson, }. (2001). nvircnmcnta| tnics: An ntrcJucticn witn kcaJinqs. london: koutledge.
lernstein, m. (2004). Neo-speciesism. jcurna| cj 5ccia| ni|cscpny, ! (!):!80-!90.
llackorby, c., lossert w., & uonaldson, u. (2002). lopulation lrinciples with Number-
uependent critical levels, jcurna| cj uu|ic ccncmic 1nccry, 4:!4-68.
llackorby, c., lossert w., & uonaldson, u. (200!). 1he Axiomatic Approach to lopulation
lthics. c|itics ni|cscpny ccncmics, 2(!): !42-!81.
llackorby, c., lossert w., & uonaldson, u. (200). cpu|aticn ssucs in 5ccia| cncicc 1nccry,
wc|jarc ccncmics, anJ tnics. cambridge Lniversity lress.
lloom, l. (2010). why we like what we like. ouscrvcr, 2! (8):!.
8iu|icqrapny
!!
lognar, o. & lerstein, 8.}. (2010). 8aving lives and kespecting Others. jcurna| cj tnics
5ccia| ni|cscpny, (2):1-20.
loorse, c. (1994). uucking 1rolleys, jcurna| cj 5ccia| ni|cscpny, 2(!):146-12.
lostrom, N. (2006). Quantity ol experience: brain-duplication and degrees ol
consciousness. VinJ Vacn 16:18-200.
lostrom, N. & Yudkowski, l. (2011). 1he lthics ol Artilicial lntelligence. ln kamsey w. and
lrankish, l. (eds.) camuriJqc uanJucck cj Artijicia| ntc||iqcncc, cambridge
Lniversity lress.
loyle, }.m. (1980). 1oward Lnderstanding the lrinciple ol uouble lllect. tnics 90(4): 2-
!8.
lroome, }. (1991). wciqninq 6ccJs, Oxlord, lasil llackwell.
lroome, }. (2004). wciqninq ivcs. Oxlord: Oxlord.
lrophy, m. (2009). Vcra| ntuiticns in kcj|cctivc ui|iurium: App|yinq 5cicntijic VctncJc|cqy tc
tnics. uissertation, Lniversity ol minnesota.
lrown, c. (200). lrioritarianism lor Variable lopulations. ni|cscpnica| 5tuJics, 1!4:!2-
!61.
lruers, 8. (201!a). 8peciesism as a moral leuristic. ni|cscpnia, 41(2): 489-01.
lruers, 8. (201!b). uieren, dilemma's en discriminatie: naar een samenhangende ethiek
van dierengelijkheid. tnick Vaatscnappij, 14 (4):!!-9.
lruers, 8. en lraeckman, }. (201!) A review and systematization ol the trolley problem.
ni|cscpnia, uOl: 10.100Js11406-01!-90-.
lyrne, A. (2010). lnverted Qualia, 1nc 5tanjcrJ ncyc|cpcJia cj ni|cscpny, 8pring 2010
ldition.
carruthers, l. (1992). 1nc Anima|s ssuc: Vcra| 1nccry in racticc, cambridge Lniversity
lress, cambridge.
carter, A. (1999). moral 1heory and olobal lopulation, rccccJinqs cj tnc Aristctc|ian 5ccicty,
99: 289-!1!.
cll (201!). our Nutricnt wcr|J. center lor lcology and lydrology, olobal lartnership on
Nutrient management.
changizi m.A., lsieh A., Nijhawan k., lanai k. & 8himojo 8. (2008). lerceiving the lresent
and a 8ystematization ol lllusions. ccqnitivc 5cicncc, !2(!):49-0!.
chappell, 1. (2011). On the Very ldea ol criteria lor lersonhood. 5cutncrn jcurna| cj
ni|cscpny, 49 (1):1-2.
chomsky, N. (1986). kncw|cJqc cj |anquaqc: ts naturc, criqin, anJ usc. New York, NY:lraeger.
clark, 8. (19). 1nc Vcra| 5tatus cj Anima|s, OLl.
cohen, c. (1986). 1he case lor the use ol animals in biomedical research, 1nc Ncw nq|anJ
jcurna| cj VcJicinc !1(14):866.
cohen, c. (199). uo Animals lave kights7 tnics anJ 8cnavicr, larvard Lniversity, (2):91-
102.
cohen, c & kegan, 1. (2001). 1nc Anima| kiqnts ucuatc. lanham, mu: kowman & littlelield.
cohen, o. A. (1989). On the currency ol lgalitarian }ustice. tnics, 99:906-944.
cohnitz, u., lggqvist, 8. (2009). 1he kole ol lntuitions in lhilosophy. 5tuJia ni|cscpnica
stcnica 2:1-14.
cosmides, l., 1ooby, }. & lurzban, k. (200!). lerceptions ol kace. 1rcnJs in ccqnitivc
5cicnccs, (4), 1!-19.
costa, m.}. (1986). 1he 1rolley lroblem kevisited. 1nc 5cutncrn jcurna| cj ni|cscpny,
24(4):4!-149.
costa, m.}. (198). Another 1rip on the 1rolley. 1nc 5cutncrn jcurna| cj ni|cscpny, 2(4):461-
166.
cover 1.m. & 1homas, }. A. (1991) |cmcnts cj njcrmaticn 1nccry. wiley-lnterscience.
cowen, 1. (200!). lolicing Nature. nvircnmcnta| tnics, 2 (2):169-182.
crisp, k. (2008). well-being. 5tanjcrJ ncyc|cpcJia cj ni|cscpny. ldward N. 2alta (ed.).
tnica| ccnsistcncy anJ Anima| ua|ity
!4
cushman, l., Young, l., & lauser, m. (2006) 1he kole ol conscious keasoning and lntuition
in moral }udgment: 1esting 1hree lrinciples ol larm. sycnc|cqica| 5cicncc,
1(12):10821089.
uaniels, N. (199). wide kellective lquilibrium and 1heory Acceptance in lthics. jcurna| cj
ni|cscpny, 6():2682.
uarwall, 8. (ed.) (200!). uccntc|cqy. Oxlord: llackwell lublishing.
uavis, N. (1984). 1he uoctrine ol uouble lllect: lroblems ol lnterpretation. acijic
ni|cscpnica| quartcr|y 6: 10-12!.
uavis, 8. (200!). least larm. jcurna| cj Aqricu|tura| anJ nvircnmcnta| tnics, 16(4).
uawkins, k. (2004). 1nc Anccstcr's 1a|c, loston: loughton milllin.
ueutsch, u. (1992). laradoxes ol musical litch. 5cicntijic Amcrican, 26 (2): 889.
uevine, l.o. (2001). lmplicit lrejudice and 8tereotyping: low Automatic Are 1hey7 jcurna|
cj crscna|ity anJ 5ccia| sycnc|cqy 81(): 9.
uiamond, c. (198). atinq Vcat anJ atinq ccp|c. ni|cscpny, !(206):46-49.
uick, }. (19). low to }ustily a uistribution ol larnings. ni|cscpny anJ uu|ic Ajjairs, 4: 248-
2.
ui Nucci, l. (2012). 8ell-8acrilice and the 1rolley lroblem. ni|cscpnica| sycnc|cqy,
lorthcoming.
uombrowski, u. (199). 8auics anJ 8casts. 1nc Arqumcnt jrcm Varqina| cascs. Lniversity ol
lllinois lress, chicago.
uonaldon 8. And lymlicka w. (2011). 7ccpc|is. A c|itica| 1nccry cj Anima| kiqnts. Oxlord
Lniversity lress, L8A
uorsey, u. (2009). leadaches, lives, and Value. 0ti|ita, 21:4-6.
uougherty, 1. (201!). Aggregation, lenelicence and chance. jcurna| cj tnics 5ccia|
ni|cscpny, (2):1-19.
uworkin, k. (1981). what is lquality7 lart 1: lquality ol kesources. ni|cscpny anJ uu|ic
Ajjairs, 10: 18-246.
lbert, k. & machan, 1. (2012). lnnocent 1hreats and the moral lroblem ol carnivorous
Animals. jcurna| cj App|icJ ni|cscpny, Vol. 29, No. 2, pp:146-19.
ldmonds, u. (201!). wcu|J Ycu ki|| tnc lat Van 1nc 1rc||cy rcu|cm anJ wnat Ycur Answcr
1c||s 0s aucut kiqnt anJ wrcnq. lrinceton Lniversity lress.
ll8A (200). Opinion ol the 8cientilic lanel on Animal lealth and wellare (AlAw) on a
request lrom the commission related to the aspects ol the biology and wellare ol
animals used lor experimental and other scientilic purposes. l5A jcurna|.
doi:10.290!Jj.elsa.200.292
ll8A (2009). 6cncra| apprcacn tc jisn wc|jarc anJ tc tnc ccnccpt cj scnticncc in jisn. 8cientilic
Opinion ol the lanel on Animal lealth and wellare, lL lood 8alety Authority.
lhrlich, l. and loldren, }. (191). lmpact ol lopulation orowth, 5cicncc, 11:1212-121.
llster, }. and koemer, }. (1991). ntcrpcrscna| ccmpariscns cj wc||-8cinq. cambridge
Lniversity lress, cambridge.
lllsberg, u. (1961). kisk, Ambiguity, and the 8avage Axioms. quartcr|y jcurna| cj ccncmics,
(4): 64!669.
lpstein, l.o. (1999). A uelinition ol Lncertainty Aversion. 1nc kcvicw cj ccncmic 5tuJics, 66
(!): 9.
lverett, }. (2001). lnvironmental ethics, animal wellarism and the problem ol predation, A
lambi lovers respect lor nature. tnics tnc cnvircnmcnt, 6(1):42-6.
lairlie, 8. (200). can lritain leed ltsell7 1nc anJ, 4.
lAO (2006), ivcstcck`s cnq 5naJcw. Lnited Nations lood and Agriculture Organisation,
kome.
link, c.l. (200). 1he lredation lroblem. 8ctwccn tnc spccics, lssue V:1-16.
linnis, }. (199). A lhilosophical case Against luthanasia. ln: leown }. (ed.) utnanasia
xamincJ. cambridge Lniversity lress.
8iu|icqrapny
!
lischer, }.m. (1992). 1houghts on the 1rolley lroblem. ln: lthics: rcu|cms anJ rincip|cs,
lischer, }. m. & kavizza, m., New York: lolt, kinehart & winston.
lischer, }.m. & kavizza, m. (1992a). Quinn on uoing and Allowing, 1nc ni|cscpnica| kcvicw,
Vol. 101, No. 2.
lischer, }. m. & kavizza, m. (1992b). tnics: rcu|cms anJ rincip|cs. New York: lolt, kinehart
& winston.
lischer, }. m. & kavizza, m. (1994). uucking harm and sacrilicing others, jcurna| cj 5ccia|
ni|cscpny, 2:!.
litzlatrick, w.}. (2009). 1homson's 1urnabout on the 1rolley. Ana|ysis 69(4):6!6-64!.
lleurbaey, m. (2009). Asscssinq risky sccia| situaticns. l8l choice oroup working paper
series, vol. , no. 9.
lleurbaey, m., 1ungodden, l. & Vallentyne, l. (2009). On the possibility ol nonaggregative
priority lor the worst oll, 5ccia| ni|cscpny anJ c|icy, 26:28-28.
lleurbaey, m. & 2uber, 8. (2012). lnequality aversion and separability in social risk
evaluation. ccncmic 1nccry. uOl 10.100Js00199-012-0!0-2.
loot, l. (198). 1he lroblem ol Abortion and the uoctrine ol uouble lllect. ln: virtucs anJ
viccs. Oxlord: lasil llackwell, pp.19-!2 (originally appeared in the oxjcrJ kcvicw
, 196.)
lox, m.A. (1999). uccp vcqctarianism. 1emple Lniversity lress.
lkA (2010). 1nc kiqnt tc c|itica| articipaticn cj crscns witn Vcnta| uca|tn rcu|cms anJ
crscns witn ntc||cctua| uisaui|itics. luropean Lnion Agency lor lundamental
kights, Vienna, Austria.
lrancione, o. (2000), ntrcJucticn tc Anima| kiqnts: Ycur cni|J cr tnc ucq lhiladelphia:
1emple Lniversity lress.
lrey, k. (1980). ntcrcsts anJ kiqnts: 1nc casc Aqainst Anima|s. Oxlord Lniversity lress.
oajdos, 1. & landil, l. (2008). 1he lgnorant Observer. 5ccia| cncicc anJ wc|jarc, !1(2):19!-
2!2.
oelman, 8. (200!). 1nc sscntia| cni|J: oriqins cj sscntia|ism in vcryJay 1ncuqnt. New York:
Oxlord Lniversity lress.
oert, l. (199!). 1ransplants and 1rolleys. ni|cscpny anJ ncncmcnc|cqica| kcscarcn,
llll(1):1!-19.
olN, (2010). National lootprint Accounts, ldition 2010, data year 200. olobal lootprint
Network, Oakland, L8A.
oilboa, l. & 8chmeidler, u. (1989). maxmin lxpected Ltility with Non-Lnique lrior. jcurna|
cj Vatncmatica| ccncmics, 18(2):141-1!.
oilligan, c. (1982). n A uijjcrcnt vcicc. cambridge mA: larvard Lniversity lress.
oil-white, l.}. (2001). Are ethnic groups biological species" to the human brain7
lssentialism in our cognition ol some social categories. currcnt Antnrcpc|cqy,
42(4): 1-.
odel, l. (19!1). 8ome basic theorems on the loundations ol mathematics and their
implications. ln 8. lelerman, ed., 199. kurt 6cJc| cc||cctcJ wcrks, vc|. . Oxlord
Lniversity lress.
oodlovitch, k., oodlovitch 8. & larris }. (eds.) (191). Anima|s, Vcn anJ Vcra|s: An nuiry
intc tnc Va|trcatmcnt cj Ncn-uumans, orove lress, New York.
ooldman, m. (2001). A 1ranscendental uelense ol 8peciesism. 1nc jcurna| cj va|uc nuiry
!:9-69.
ooodin, k. (1992). 6rccn c|itica| 1nccry, lolity lress, london.
oorr, m. (1990). 1homson and the 1rolley lroblem. ni|cscpnica| 5tuJics, 9(1):91-100.
orau, c. (2010). moral 8tatus, 8peciesism and liao's oenetic Account. jcurna| cj Vcra|
ni|cscpny, (!).
oreene, }. u. (2002). 1nc 1crriu|c, ucrriu|c, Nc 6ccJ, vcry 8aJ 1rutn Aucut Vcra|ity anJ wnat 1c
uc Aucut t. uissertation. uepartment ol lhilosophy, lrinceton Lniversity.
tnica| ccnsistcncy anJ Anima| ua|ity
!6
oreene, }. u. (2008). 1he 8ecret }oke ol lant's 8oul. ln: 8innott-Armstrong, w. (ed.) Vcra|
sycnc|cqy, vc|. 1: 1nc Ncurcscicncc cj Vcra|ity. cambridge, mA: ml1 lress.
oreene, }., Nystrom l., lngell A., uarley }. & cohen }. (2004). 1he neural bases ol cognitive
conllict and control in moral judgment. Ncurcn, 44:!89-400.
oreene, }., 8ommerville k., Nystrom l., uarley }., & cohen }. (2001). An lmkl investigation
ol emotional engagement in moral judgment. 5cicncc, 29!:210-2108.
oreenwald, A.o., & lanaji, m.k. (199). lmplicit social cognition: Attitudes, sell-esteem,
and stereotypes. sycnc|cqica| kcvicw, 102:4-2.
oreenwald, A.o., mcohee, u.l., & 8chwartz, }.l.l. (1998). measuring individual dillerences
in implicit cognition: 1he lmplicit Association 1est. jcurna| cj crscna|ity anJ 5ccia|
sycnc|cqy, 4:1464-1480.
orillin, u. (2001). Anima| VinJs: 8cycnJ ccqniticn tc ccnsicusncss, Lniversity ol chicago
lress.
ounnarsson l. (2008). 1he oreat Apes and the 8everely uisabled: moral 8tatus and 1hick
lvaluative concepts. tnica| 1nccry anJ Vcra| racticc 11(!):!0-!26.
laack, 8. (199!). viJcncc anJ nuiry, Oxlord, Ll: llackwell.
laidt, }. (2001). 1he lmotional uog and lts kational 1ail: A 8ocial lntuitionist Approach to
moral }udgment. sycnc|cqica| kcvicw, 108(4): 814-8!4.
laidt, }. (2012). 1nc kiqntccus VinJ. lantheon.
lammond, l. (196). why ethical measures ol inequality need interpersonal comparisons.
1nccry anJ uccisicn, :26!-24.
lanna, k. (1992). morality ue ke: kellections on the 1rolley lroblem. ln: lischer, }. m. &
kavizza, m., (eds.) tnics: rcu|cms anJ rincip|cs. New York: lolt, kinehart &
winston.
lare, k.m. (1991). 1nc anquaqc cj Vcra|s, clarendon.
largreaves-leap, 8.l., Varoulakis, Y. (2004). 6amc 1nccry: a critica| 1cxt. london and New
York: koutledge.
larris, }. (2000). 1he moral dillerence between throwing a trolley at a person and
throwing a person at a trolley: a reply to lamm. rccccJinqs cj tnc Aristctc|ian
5ccicty, 5upp|cmcntary vc|umcs, Vol. 4 pp.41-.
larris, 8. (2004). 1nc nJ cj laitn. w.w.Norton.
larris, 8. (2010). 1nc Vcra| anJscapc: ucw 5cicncc can uctcrminc uuman va|ucs. lree lress.
larsanyi, }.c. (19!): cardinal Ltility in wellare lconomics and in the 1heory ol kisk-
1aking. jcurna| cj c|itica| ccncmy, 61:4!4-4!.
larsanyi, }.c. (19). cardinal wellare, individualistic ethics, and the interpersonal
comparison ol utility. jcurna| cj c|itica| ccncmy, 6!:!09-!21.
lauser, m., Young l., & cushman l. (2008). 'keviving kawls' linguistic Analogy: Operative
lrinciples and the causal 8tructure ol moral Actions. ln: 8innott-Armstrong w.
(ed.) (2008), Vcra| psycnc|cqy anJ uic|cqy. Oxlord Lniversity lress, New York.
lerzog, l. (2010), 5cmc wc cvc, 5cmc wc uatc, 5cmc wc at: wny t`s 5c uarJ tc 1nink 5traiqnt
Aucut Anima|s, larper.
loekstra, A.Y. (2010). 1he water lootprint ol animal products, ln: u'8ilva, }. and webster, }.
(eds.) 1nc mcat crisis: ucvc|cpinq mcrc sustainau|c prcJucticn anJ ccnsumpticn,
larthscan, london, Ll, pp.22-!!.
loltug, N. (2006). lrioritarianism. ln: qa|itarianism, ncw cssays cn tnc naturc anJ va|uc cj
cua|ity, loltug & lippert-kasmussen (eds.), clarendon lress.
loltug, N. (200). lquality lor Animals. ln }. kyberg, 1.8. letersen & c. woll (eds.), Ncw
wavcs in App|icJ tnics. lalgrave macmillan.
looker, l. (2011). kule consequentialism, 1nc 5tanjcrJ ncyc|cpcJia cj ni|cscpny (8pring
2011 ldition), ldward N. 2alta (ed.).
lorta, O. (2010). 1he lthics ol the lcology ol lear against the Nonspeciesist laradigm: A
8hilt in the Aims ol lntervention in Nature. 8ctwccn tnc spccics, lssue X:16!-18.
8iu|icqrapny
!
lorta, O. (2010b). what is 8peciesism7 1nc jcurna| cj Aqricu|tura| anJ nvircnmcnta| tnics,
2!:24!-266.
lorta, O. (2010c). uebunking the idyllic view ol natural processes: population dynamics
and sullering in the wild. 1c|cs, 1:!-88.
loward-8nyder l. (2011). uoing vs. Allowing larm. 1nc 5tanjcrJ ncyc|cpcJia cj ni|cscpny
(wintcr .!! Jiticn), ldward N. 2alta (ed.).
lull, u. (1986). On luman Nature. 5A: rccccJinqs cj tnc uicnnua| mcctinq cj tnc pni|cscpny cj
scicncc asscciaticn, 2: !-1!.
lurka, 1. (198!). Value and lopulation 8ize. tnics, 9!: 496-0.
lursthouse, k. (2000). tnics, uumans anJ otncr Anima|s, koutledge, london.
luther, c. (200). can 5pccicsism uc ucjcnJcJ A uiscussicn cj tnc 1raJiticna| Apprcacn tc tnc
Vcra| 5tatus cj Anima|s. munich: ludwig-maximilians-Lniversitt mnchen.
}amieson, u. (1990). kights, }ustice and the uuty to lrovide Assistance: A critique ol
kegan's 1heory ol kights. tnics, 100(2): !49-62.
}ensen, l.l. (200!). what is the uillerence letween (moderate) lgalitarianism and
lrioritarianism7 ccncmics anJ ni|cscpny, 19 (1), pp.89-109.
}oy, m. (2002). sycnic Numuinq anJ Vcat ccnsumpticn: 1nc sycnc|cqy cj carnism, lh.u.
dissertation, 8aybrook oraduate 8chool.
}oy, m. (2009). wny wc |cvc Jcqs, cat piqs anJ wcar ccws, An intrcJucticn tc carnism, conari
lress.
lagan, 8. (1999). ua|ity anJ ucscrt. ln: l.l. lojman and O. mcleod (eds.), wnat uc wc
ucscrvc A kcaJcr cn justicc anJ ucscrt, Oxlord and New York: Oxlord Lniversity
lress, pp. 298-!14.
lahneman, u. (200!). maps ol lounded kationality: lsychology lor lehavioral lconomics.
Amcrican ccncmic kcvicw, 9! (): 1449-14.
lahneman, u. (2011). 1ninkinq, last anJ 5|cw, macmillan.
lahneman, u. & 8hane, l. (2002). kepresentativeness kevisited: Attribute 8ubstitution in
lntuitive }udgment. ln 1homas oilovich, uale orillin, uaniel lahneman. ucuristics
anJ 8iascs: 1nc sycnc|cqy cj ntuitivc juJqmcnt. cambridge: cambridge Lniversity
lress. pp. 49-81.
lahneman, u. & 1versky, A. (199). lrospect 1heory: An Analysis ol uecision under kisk.
ccncmctrica, XlVll, 26!-291.
lahneman, u. & 1versky, A. (1984). choices, Values, and lrames. Amcrican sycnc|cqist, !9
(4): !41-!0.
lahneman, u., 1versky, A. & 8lovic, l., eds. (1982). juJqmcnt unJcr 0nccrtainty: ucuristics
8iascs. cambridge, Ll, cambridge Lniversity lress.
lamm, l.m. (1989). larming 8ome to 8ave Others. ni|cscpnica| 5tuJics, :22-60.
lamm, l.m. (1998). Vcra|ity, Vcrta|ity: ucatn anJ wncm tc 5avc jrcm t. New York: Oxlord
Lniversity lress.
lamm, l.m. (2000). 1he uoctrine ol 1riple lllect and why a kational Agent Need Not
lntend the means to lis lnd. rccccJinqs cj tnc Aristctc|ian 5ccicty, 8upplementary
Volumes, 4:21-!9.
lamm, l.m. (200). ntricatc tnics. kiqnts, kcspcnsiui|itics, anJ crmissiu|c uarm, Oxlord
Lniversity lress. Oxlord.
lant, l. (18), translated by }.w. lllington (199!). 6rcunJinq jcr tnc Vctapnysics cj Vcra|s
1rJ cJ. lackett.
larban, k. & l. 8hiojiri (2009). 8ell-recognition allects plant communication and delense.
cc|cqy cttcrs, 12.
laulman, l. (1998). 8peciesism and the Argument lrom mislortune, jcurna| cj App|icJ
ni|cscpny, 1 (2): 1-16!.
lerstein 8. (2009). 1reating Others merely as means. 0ti|itas, 21(2):16!-180.
tnica| ccnsistcncy anJ Anima| ua|ity
!8
lolmogorov, A. (19!0). On the Notion ol mean. ln: Vatncmatics anJ Vccnanics. lluwer,
pp. 144-146.
lorsgaard, c. (1996). 1he kight to lie: lant on uealing with lvil. ln crcatinq tnc kinqJcm cj
nJs. cambridge: cambridge Lniversity lress.
luhn, 1. (1962). 1nc 5tructurc cj 5cicntijic kcvc|uticns. 1he Lniversity ol chicago lress.
luhn, 1. (190). Objectivity, Value }udgment, and 1heory choice. ln: 5cicncc, kcascn, anJ
kca|ity. New York: larcourt lrace, 1998.
lumar, k. (2008). lermissible lilling and the lrrelevance ol leing luman. 1nc jcurna| cj
tnics 12:-80
lurzban, k., 1ooby, }. & cosmides, l. (2001). can race be erased7 coalitional computation
and social categorization. NA5, 98(26):1!8-1!9!.
lamey, A. (200). lood light! uavis versus kegan on the ethics ol eating beel. jcurna| cj
5ccia| ni|cscpny, Vol. !8 No. 2,
lamont, }. (1994). 1he concept ol uesert in uistributive }ustice. 1nc ni|cscpnica| quartcr|y,
44: 4-64.
lanteri, A., chelini, c., & kizzello 8. (2008). An lxperimental lnvestigation ol lmotions and
keasoning in the 1rolley lroblem. jcurna| cj 8usincss tnics, 8!:89-804.
lee, l. (2004). 1he lro-lile Argument lrom 8ubstantial ldentity: A uelence. 8icctnics
18(!):249-26!.
lee, l. & oeorge, k. (2008). 1he Nature and lasis ol luman uignity. katic juris, 21(2):1!-
9!.
levy N. (2004). cohen and linds: A kesponse to Nathan Nobis. jcurna| cj App|icJ ni|cscpny,
21(2):21!-21.
liao, m. (2009). 1he loop case and lamm's uoctrine ol 1riple lllect. ni|cscpnica| 5tuJics,
146(2):22!-2!1.
liao, m. (2010). 1he lasis ol luman moral 8tatus. jcurna| cj Vcra| ni|cscpny, (2): 19-19.
liao, m., wiegmann, A., Alexander, }. & Vong, o. (2011). lutting the 1rolley in Order:
lxperimental lhilosophy and the loop case. ni|cscpnica| sycnc|cqy, 2():1-11.
lippert-kasmussen, l. (1996). moral 8tatus and the lmpermissibility ol minimizing
Violations. ni|cscpny uu|ic Ajjairs, 2(4), pp. !!!-!1.
loomis, l.8. (1968). 1nc ytnaqcrcan rcpcsiticn (2nd ed.). 1he National council ol 1eachers
ol mathematics.
lumer, c. (2006). lrioritarian wellare lunctions. An llaboration and }ustilication. ln:
uaniel 8choch (ed.): ucmccracy anJ wc|jarc. laderborn: mentis.
machan, 1. (2004). uttinq uumans lirst: wny wc Arc Naturc`s lavcritc. kowman & littlelield.
maclean, u. (2010). ls luman leing" a moral concept7 ni|cscpny uu|ic c|icy quartcr|y
!0(!J4):16-20.
margolis, l. (198). attcrns, 1ninkinq anJ ccqniticn: A 1nccry cj juJqcmcnt. chicago:
Lniversity ol chicago lress.
maslow, A., (194!). A 1heory ol luman motivation. sycnc|cqica| kcvicw, Vol. 0, nr.4,
pp. !0-!96.
masson, }.m. (199). wncn |cpnants wccp: 1nc mcticna| ijc cj Anima|s, }ellrey m. masson,
uelta.
matheny, o. (200!). least harm: A delense ol vegetarianism lrom 8teven uavis's
omnivorous proposal. jcurna| cj Aqricu|tura| anJ nvircnmcnta| tnics, 16: 0-11.
matheny, o. & chan, l.m.A. (200). luman uiets and Animal wellare: 1he lllogic ol the
larder. jcurna| cj Aqricu|tura| anJ nvircnmcnta| tnics, 18: 9-94.
mccarthy, u. (200!). rcspccts jcr ricritarianism. Lnpublished
mccarthy, u. (2006). Ltilitarianism and lrioritarianism l. ccncmics anJ ni|cscpny, 22:!!-
6!.
mccarthy, u. (2008). Ltilitarianism and lrioritarianism ll. ccncmics anJ ni|cscpny, 24:1-
!!.
8iu|icqrapny
!9
mccloskey, l. }. (196). A Non-Ltilitarian Approach to lunishment. nuiry, 8:2!9-2.
mcuowell, }. (1984). Values and 8econdary Qualities. ln londerich, 1. (ed.), Vcra|ity anJ
oujcctivity. koutledge.
mclntyre, A. (2001). uoing Away with uouble lllect, tnics, Vol. 111, No. 2, pp. 219-2.
mclntyre A. (2011). uoctrine ol uouble lllect. 1nc 5tanjcrJ ncyc|cpcJia cj ni|cscpny (la||
.!! Jiticn), ldward N. 2alta (ed.)
mcmahan, }. (2002). 1nc tnics cj ki||inq: rcu|cms at tnc Varqins cj ijc. Oxlord: Oxlord
Lniversity lress.
mcmahan, }. (200). Our lellow creatures. 1nc jcurna| cj tnics, 9: !!-!80.
mcmahan, }. (2008). lating Animals the Nice way. uacJa|us, winter 2008.
mcmahan, }. (2009). Asymmetries in the morality ol causing leople to lxist. ln koberts m.
& wasserman, u. (eds)., uarminq luturc crscns. lnternational library ol lthics,
law, and the New medicine, Volume !.
melden, A.l. (1980). uo lnlants lave moral kights. ln Aiken, w. and lalollette, l. (eds)
wncsc cni|J, 1otowa, New }ersey: littlelied, Adams & co., pp. 210-211.
mikhail, }. (2000). kaw|s` |inquistic ana|cqy: A 5tuJy cj tnc 'qcncrativc qrammar` mcJc| cj mcra|
tnccry JcscriucJ uy jcnn kaw|s in 'A tnccry cj justicc.` lhu dissertation, cornell
Lniversity, lthaca, NY.
mikhail, }. (200). Lniversal moral orammar: 1heory, lvidence and the luture. 1rcnJs in
ccqnitivc 5cicnccs, 11(4): 14!-12.
miller, u. (196). 5ccia| justicc, Oxlord: clarendon lress.
milne, l. (1986). uesert, ellort and equality. jcurna| cj App|icJ ni|cscpny, !: 2!-24!.
montmarquet, }. (1982). uoing oood: 1he kight and the wrong way. jcurna| cj ni|cscpny,
lXXlX:4!9-4.
moreno-1ernero, }.u. & koemer, }.l. (200). mpartia|ity anJ ricrity. art !: tnc vci| cj
qncrancc. working paper.
mulgan, 1. (2006). luturc ccp|c. A VcJcratc ccnscucntia|ist Acccunt cj cur ou|iqaticns tc
luturc 6cncraticns, Oxlord: clarendon lress.
mller-lyer, l.c. (1889). Optische Lrteilstuschungen. Arcniv jur nysic|cqic 5upp|., 26!-20.
Narveson, }. (196). Ltilitarianism and New oenerations. VinJ, 6: 62-2.
Narveson, }. (19). Animal kights. canaJian jcurna| cj ni|cscpny, :161-8.
Narveson, }. (198). On a case lor Animal kights. 1nc Vcnist, 0(1):!1-49.
Nelkin, u.l. (201!), moral luck, 1nc 5tanjcrJ ncyc|cpcJia cj ni|cscpny (8ummer 201!
ldition), ldward N. 2alta (ed.).
Ng, Y-l. (1986). 8ocial criteria lor lvaluating lopulation change: an Alternative to the
llackorby-uonaldson criterion. jcurna| cj uu|ic ccncmics, 29:!-!81.
Nobis, N. (2004). carl cohen's 'lind' Arguments lcr Animal kights and Aqainst luman
kights. jcurna| cj App|icJ ni|cscpny, 21 (1):4!-9.
Noddings, N. (2002). 5tartinq at ucmc: carinq anJ 5ccia| c|icy, lerkeley, cA.: Lniversity ol
calilornia lress.
Nolt, }. (201!). comparing 8ullering Across 8pecies. 8ctwccn tnc 5pccics, 16(1):86-104.
Norcross, A. (2008). Oll ler 1rolley7 lrances lamm and the metaphysics ol morality.
0ti|itas, 20(1): 6-80.
Nozick, k. (194). Anarcny, 5tatc, anJ 0tcpia. lasic looks, New York.
Nussbaum, m. (1992). luman lunctioning and 8ocial }ustice: ln uelense ol Aristotelian
lssentialism. c|itica| 1nccry, 20:202-246.
Nussbaum, m. (2000). wcmcn anJ uuman ucvc|cpmcnt: 1nc capaui|itics Apprcacn, cambridge:
cambridge Lniversity lress.
Nussbaum, m. (2006) lrcnticrs cj justicc. uisaui|ity, Naticna|ity, 5pccics Vcmucrsnip, 1he
lelknap lress ol larvard Lniversity, cambridge, mass.
Olewski, }. (2010). calculating our Nitrogen lootprint. llliciency ol nitrogen use in
dillerent larming systems on a global scale, 6rcwinq 6rccn ntcrnaticna|, Nr.26.
tnica| ccnsistcncy anJ Anima| ua|ity
!80
O'Neill, l., & letrinovich, l. (1998). A lreliminary cross-cultural 8tudy ol moral lntuitions.
vc|uticn anJ uuman 8cnavicr, 19(6): !49-!6.
Oswald, }. (191). 1nc cry cj Naturc, cr, An Appca| tc Vcrcy anJ tc justicc, cn 8cna|j cj tnc
crsccutcJ Anima|s.
Otsuka, m. (2008). uouble lllect, 1riple lllect and the 1rolley lroblem: 8quaring the circle
in looping cases. 0ti|itas, 20(1):92-110.
Otsuka, m. & Voorhoeve, A. (2009). why lt matters 1hat 8ome Are worse Oll 1han Others:
An Argument against the lriority View. ni|cscpny uu|ic Ajjairs, !(2):11-99.
larlit, u. (1984). kcascns anJ crscns, Oxlord: clarendon lress.
larlit, u. (1991). ua|ity cr ricrity, 1nc inJ|cv ccturc, lawrence: Lniversity ol lansas.
larlit, u. (199). lquality and lriority. katic, 10:202-221.
larlit, u. (2011). on wnat Vattcrs. Oxlord Lniversity lress.
larks, l. u. (2006). 1he Natural-Artilicial uistinction and conjoined 1wins: A kesponse 1o
}udith 1homson's Argument lor Abortion kights. Naticna| catnc|ic 8icctnics
quartcr|y, 6(4):61-680.
letrinovich, l., & O'Neill, l. (1996). lnlluence ol wording and lraming lllects on moral
lntuitions. tnc|cqy anJ 5ccicuic|cqy 1:14-11.
lhillips c., wagers w. and lau, l.l. (2010). orammatical lllusions and 8elective lallibility
in keal-1ime language comprehension. ln }. kunner (ed.), xpcrimcnts at tnc
ntcrjaccs, 8yntax & 8emantics, vol. !. lmerald lublications.
livato, m. (2009). 5ccia| cncicc witn Apprcximatc ntcrpcrscna| ccmpariscns cj wc||-ucinq.
mlkA laper No. 1222.
llous, 8. (200!). ls there such a thing as prejudice towards animals7 ln 0nJcrstanJinq
prcjuJicc anJ Jiscriminaticn, llous (ed.), mcoraw-lill, New York.
llumwood, V. (199!). lcminism anJ tnc Vastcry cj Naturc, koutledge, london.
lohl, k. (2004). ccqnitivc ||usicns: A uanJucck cn la||acics anJ 8iascs in 1ninkinq, juJqcmcnt
anJ Vcmcry. lsychology lress.
lorter, 1. (2012). ln uelense ol the lriority View. 0ti|itas, 24:!49-!6.
lostow, l.c. (1989). 1homson and the 1rolley lroblem. 5cutncrn jcurna| cj ni|cscpny,
2(4):29-!.
lratt, }. w. (1964). kisk aversion in the small and in the large. lconometrica !2:122-1!6.
lurves, u. & lotto, l. (2002). wny wc 5cc wnat wc uc: An mpirica| 1nccry cj visicn,
8underland, mA: 8inauer Associates.
lylyshyn, 2. (1999). ls vision continuous with cognition7 1he case lor cognitive
impenetrability ol visual perception. 8cnav 8rain 5ci.,22:!41-6.
Quattrone, o.A. & }ones, l.l. (1980). 1he perception ol variability within in-groups and
out-groups: lmplications lor the law ol small numbers. jcurna| cj crscna|ity anJ
5ccia| sycnc|cqy, !8(1):141-12.
Quinn, w. (1989a). Actions, lntentions, and consequences: 1he uoctrine ol uoing and
Allowing. ni|cscpnica| kcvicw 98:28-!12.
Quinn, w. (1989b). Actions, lntentions, and consequences: 1he uoctrine ol uouble lllect.
ni|cscpny anJ uu|ic Ajjairs, 18(4):!!4-!1.
kabinowicz, w. (2002). lrioritarianism lor lrospects. 0ti|itas, 14:2-21.
kachels, }. (1990). crcatcJ lrcm Anima|s. 1nc Vcra| mp|icaticns cj uarwinism. Oxlord
Lniversity lress.
kai, 1. & lolyoak, l. (2010). moral lrinciples or consumer lrelerences7 Alternative
lramings ol the 1rolley lroblem. ccqnitivc 5cicncc !4:!11-!21.
kamachandran, V.8. and lubbard, l.m. (2001). 8ynaesthesia a window into perception,
thought and language. jcurna| cj ccnscicusncss 5tuJics, 8: !-!4.
kawls, }. (191). A 1nccry cj justicc. cambridge: larvard Lniversity lress.
kawls, }. (2001). justicc as lairncss: A kcstatcmcnt, ed. by l. lelly, cambridge: larvard
Lniversity lress.
8iu|icqrapny
!81
keibetanz, 8. (1998). A lroblem lor the uoctrine ol uouble lllect. rccccJinqs cj tnc
Aristctc|ian 5ccicty, New 8eries, Vol. 98, pp. 21-22!
kegan, 1. (198!). 1nc casc jcr Anima| kiqnts, lerkeley: Lniversity ol calilornia lress.
kiley, }. (1989). }ustice Lnder capitalism. ln: Varkcts anJ justicc, ed. }.w. chapman, New
York: New York Lniversity lress, 122-162.
kitson, }. (1802). An ssay cn Austincncc jrcm Anima| lccJ as a Vcra| uuty.
koemer, }.l. (1996). 1nccrics cj uistriuutivc justicc, cambridge: larvard Lniversity lress.
kolston lll, l. (1988). nvircnmcnta| tnics: uutics tc anJ va|ucs in tnc Natura| wcr|J.
lhiladelphia: 1emple Lniversity lress.
kosenberg, m. (200!). Ncnvic|cnt ccmmunicaticn: A anquaqc cj ijc. 2
nd
ldition. lncinitas,
cA: luddleuancer lress.
koss, w. u. (19!0, 2002). 1nc kiqnt anJ tnc 6ccJ. (with an introduction and bibliography by
lhilip 8tratton-lake). Oxlord Lniversity lress.
kowlands, m. (199). contractarianism and Animal kights. jcurna| cj App|icJ ni|cscpny,
14(!):2!-24.
kowlands, m. (1998). Anima| kiqnts: A ni|cscpnica| ucjcncc, macmillanJ8t martin's lress.
kubin, m. & ladea, c. (2012). 1hey're all the same!...but lor several dillerent reasons: A
review ol the multicausal nature ol perceived group variability. currcnt uirccticns
in sycnc|cqica| 5cicncc, 21: !6-!2.
kussell, l. (199). 1he kelative 8trictness ol lositive and Negative uuties. ln: 8teinbock, l.
(ed.) ki||inq anJ cttinq uic. NY: lrentice-lall.
kyder, k.u. (19). victims cj 5cicncc: 1nc 0sc cj Anima|s in kcscarcn, uavis-loynter.
kyder, k.u. (2001). ainism: A VcJcrn Vcra|ity, Opengate lress.
8achdeva, 8. e.a. (2011). culture and the quest lor universal principles in moral reasoning.
ntcrnaticna| jcurna| cj sycnc|cqy, 46(!).
8adurski, w. (198). 6ivinq ucscrt ts uuc, uordrecht, lolland: u. keidel.
8alt, l. (1892). Anima|s' kiqnts: ccnsiJcrcJ in kc|aticn tc 5ccia| rcqrcss.
8apontzis, 8. (1984). lredation. tnics anJ anima|s, (2):2-!8.
8canlon, 1.m. (1998). wnat wc owc tc acn otncr. cambridge, mass., lelknap lress.
8canlon, 1.m. (2008). Vcra| uimcnsicns. larvard Lniversity lress.
8chelller, 8. (1982). 1nc kcjccticn cj ccnscucntia|ism, Oxlord and New York: Oxlord
Lniversity lress.
8chnall, 8., laidt, }., clore, o.l., }ordan, A.l. (2008). uisgust as lmbodied moral }udgment.
58, !4(8): 1096-1109.
8chwitzgebel, l. & cushman l. (2012). lxpertise in moral keasoning7 Order lllects on
moral }udgment in lrolessional lhilosophers and Non-lhilosophers. VinJ
anquaqc 2:1!-1!.
8cruton, k. (1998). Anima| kiqnts anJ wrcnqs, uemos.
8cruton, k. (2004). 1he conscientious carnivore. ln 8. 8apontzis (ed.), lccJ lcr 1ncuqnt: 1nc
ucuatc cvcr atinq Vcat. lrometheus, Amherst, NY.
8cruton, k. (2006). lating Our lriends. kiqnt kcascn (may 26, 2006).
8egall, m.l., campbell, u.1. & lerskovits, m.}. (196!). cultural uillerences in the
lerception ol oeometric lllusions. 5cicncc, New 8eries, Vol. 1!9, No. !6, pp. 69-
1.
8en, A. (1982). cncicc, wc|jarc anJ Vcasurcmcnt, ml1 lress.
8en, A. (1992). ncua|ity kccxamincJ, cambridge: larvard Lniversity lress.
8erpell, }. (1996). n tnc ccmpany cj Anima|s, a stuJy cj numan-anima| rc|aticnsnips, cambridge
Lniversity lress.
8haver, k. (2011). 1homson's 1rolley 8witch, jcurna| cj tnics 5ccia| ni|cscpny, discussion
note.
8haw, w.l. (1999). ccntcmpcrary tnics: 1akinq Acccunt cj 0ti|itarianism. Oxlord: llackwell.
8haw, }. (2006). lntentions and 1rolleys, ni|cscpnica| quartcr|y, 6 (222):6! - 8!.
tnica| ccnsistcncy anJ Anima| ua|ity
!82
8hermer, m. (2004). 1nc 5cicncc cj 6ccJ anJ vi|. New York: 1imes looks.
8hiell, l. (200). 1nc kcpuqnant ccnc|usicn cn kca|istic cncicc 5cts, Lniversity ol Ottawa.
8immons, A. (2009). Animals, predators, the right to lile and the duty to save lives. tnics
tnc cnvircnmcnt, 14(1):1-2.
8inger, l. (19!). lccJ jcr 1ncuqnt [keply to uavid kosinger]. New York keview ol looks
20.10 19!.
8inger, l. (19, 1990). Anima| iucraticn, a ncw ctnics jcr cur trcatmcnt cj anima|s. 2nd ed.,
New York keview ol looks.
8inger, l. (199!). ractica| ctnics. cambridge Lniversity lress.
8inger, l. (200). lthics and lntuitions. 1nc jcurna| cj tnics 9:!!1!2.
8inger, l. & cavalieri, l. (eds.) (199!). 1nc 6rcat Apc rcjcct: ua|ity 8cycnJ uumanity.
lourth lstate publishing, london, lngland.
8innott-Armstrong, w. (2008). lraming moral lntuitions. ln w. 8innott Armstrong (ld.)
Vcra| sycnc|cqy, vc|umc .: 1nc ccqnitivc 5cicncc cj Vcra|ity, (pp. 4-6). cambridge,
mA: ml1 lress.
8innott-Armstrong, w., Young, l., & cushman, l. A. (2010). moral lntuitions as leuristics.
ln }. uoris et al. (lds.), 1nc oxjcrJ uanJucck cj Vcra| sycnc|cqy. Oxlord Lniversity
lress.
8lote, m. (2001). Vcra|s jrcm Vctivcs. Oxlord: Oxlord Lniversity lress.
8mart, }.c.c. & williams, l. (19!). 0ti|itarianism: lcr anJ Aqainst. cambridge Lniversity
lress.
8neddon, l.L., lraithwaite, V.A. and oentle, m.}. (200!a). uo lishes have nociceptors7
lvidence lor the evolution ol a vertebrate sensory system. rccccJinqs oj 1nc
kcya| 5ccicty oj cnJcn 5crics 8 8ic|cqica| 5cicnccs, 20 (120): 111-1121.
8neddon, l.L., lraithwaite, V.A. and oentle, m.}., (200!b). Novel object test: lxamining
nociceptionand lear in the rainbow trout. jcurna| oj ain, 4 (8): 4!1-440.
8obel, u. (200). 1he lmpotence ol the uemandingness Objection. ni|cscpncrs` mprint,
(8):1-1.
8tehlest, l., louwman, l., van Vuuren, u., den llzen, m., lickhout, l., labat, l. (2009).
climate lenelits ol changing uiet. c|imatic cnanqc 9:8!102.
8unstein, c. (200). moral leuristics. 8cnavicra| anJ 8rain 5cicnccs, 28:!1-!.
1ajlel, l. (1981). uuman 6rcups anJ 5ccia| catcqcrics. cambridge Lniversity lress,
cambridge.
1anner, }. (2006). marginal humans, the argument lrom kinds and the similarity
argument. lacta 0nivcrsitas, (1):4-6!.
1anner, }. (2009). 1he Argument lrom marginal cases and the 8lippery 8lope Objection.
nvircnmcnta| va|ucs, 18:1-66.
1aylor, l. (1986). kcspcct jcr Naturc: A 1nccry cj nvircnmcnta| tnics. lrinceton Lniversity
lress.
1emkin, l. (198). lntransitivity and the mere Addition laradox. ni|cscpny anJ uu|ic
Ajjairs, 16 (2): 1!818.
1homas, l. (2010). Animals and Animals. 8ctwccn tnc 5pccics X:108-20!.
1homson, }.}. (191). A uelense ol Abortion. ni|cscpny anJ uu|ic Ajjairs, 1(1):4-66.
1homson, }.}. (196). lilling, letting uie, and the 1rolley lroblem. 1nc Vcnist, 9:20421.
1homson, }.}. (198). 1he 1rolley lroblem. 1nc Ya|c aw jcurna|, 94:1!941.
1homson, }.}. (199!). keply to commentators. ni|cscpny anJ ncncmcnc|cqica| kcscarcn,
!(1):18194.
1homson, }.}. (2008). 1urning the 1rolley. ni|cscpny anJ uu|ic Ajjairs, !6(4):!9!4.
Lnger, l. (1996). ivinq uiqn anJ cttinq uic, Oxlord: Oxlord Lniversity lress.
Valdesolo, l. & ue8teno, u. (2006). manipulations ol lmotional context 8hape moral
}udgment. sycnc|cqica| 5cicncc, 1:46-.
8iu|icqrapny
!8!
Vallentyne, l. (2006). Ol mice and men: lquality lor Animals. ln: qa|itarianism, ncw cssays
cn tnc naturc anJ va|uc cj cua|ity, loltug & lippert-kasmussen (eds.), clarendon
lress.
Vallentyne, l. (2012). libertarianism. 1nc 5tanjcrJ ncyc|cpcJia cj ni|cscpny (5prinq .!.
Jiticn), ldward N. 2alta (ed.)
Van den lerg, l. (2011). uarminq otncrs: 0nivcrsa| 5uujcctivism anJ tnc xpanJinq Vcra| circ|c.
lhu dissertation, Lniversiteit leiden.
Van ue Veer, u. (199). Ol leasts, lerson, and the Original losition, 1nc Vcnist, 62:!68-!.
Velleman, u. (1991). well-being and 1ime. acijic ni|cscpnica| quartcr|y, 2:0.
Visak, 1. (2011). ki||inq uappy Anima|s. xp|craticns in 0ti|itarian tnics. uissertation, Ltrecht
Lniversity.
waldmann m. & uieterich }. (200). 1hrowing a bomb on a person versus throwing a
person on a bomb. lntervention myopia in moral intuitions, sycnc|cqica| 5cicncc,
18(!):24-2!.
weinberg, 8. (1996). 1nc quantum 1nccry cj lic|Js. cambridge Lniversity lress.
weirich, l. (198!). Ltility 1empered with lquality. Ncus, 1: 42!-!9.
weymark, }.A. (1981). oeneralized oini lnequality lndices. Vatncmatica| 5ccia| 5cicnccs,
1:4094!0.
whitley, l.l., & lite, m.l. (2010). 1nc sycnc|cqy cj rcjuJicc anJ uiscriminaticn. lelmont, cA:
wadsworth.
williams, l. (190). 1he 8ell and the luture, 1nc ni|cscpnica| kcvicw, 9(2):161-180.
williams, l. (2006). 1he luman lrejudice. ni|cscpny as a uumanistic uiscip|inc. lrinceton.
wilson, 8. (2001). carruthers and the Argument lrom marginal cases. jcurna| cj App|icJ
ni|cscpny, 18 (2):1!-14.
wreen, m. (1984). ln uelense ol 8peciesism. tnics anJ Anima|s (!):4-60.
Young, 1. (1984). 1he morality ol lilling Animals: lour Arguments. tnics anJ Anima|s
(4):88-101.


Buy your books fast and straightforward online - at one of worlds
fastest growing online book stores! Environmentally sound due to
Print-on-Demand technologies.
Buy your books online at
www.get-morebooks.com
Kaufen Sie Ihre Bcher schnell und unkompliziert online auf einer
der am schnellsten wachsenden Buchhandelsplattformen weltweit!
Dank Print-On-Demand umwelt- und ressourcenschonend produzi-
ert.
Bcher schneller online kaufen
www.morebooks.de
VDM Verlagsservicegesellschaft mbH
Heinrich-Bcking-Str. 6-8 Telefon: +49 681 3720 174 info@vdm-vsg.de
D - 66121 Saarbrcken Telefax: +49 681 3720 1749 www.vdm-vsg.de

Вам также может понравиться