Вы находитесь на странице: 1из 8

-- -- -- -- -- '

52

CH,9

EMPIIlIC ISM ANO SCIENTIF IC REALISM

l. Represelltig (!lld Intervellllg, Cambrid gc:


183.

University Pless,

of science confercn ce in DlIbro\'

1984 lt has bcncfiled trom Ihe cOlnmellls of Ihose


in part, by a granl frol1l S.s.H.R.C .

IAN HACK ING

Experimentatian and
Scientific Realisn1

Experim ental physics provides the strongesl evidenc e for scientific


realismo
Entities tha! in principie canllol be observed are regularly Illanipul
aled lo
produce new phenolll ena and lo nvcstigate other aspects of natme.
They
are tools, instrulllcnts nol for thinking but for
The philosop hcr's standard "theoret ical entity" is the clectron . I shall
llustra!e how clectmn s have becOllle experim ental entities, or experil1l
ent
er's entities. In the carly stages of our discovery of 3n entily, \vc
mav test
hypothescs abollt it. Then t is rncrely an hypothe tical entity. MlIch
if we come lo lIndcrstand SOl1le of its causal po\Vers and lo use it
to build
devices that achievc \Vell understo od effects in olher parts of na!ure,
then
il assul11es quite a different status.
DiscllSsions about scientific realism or 3nt-real iSI11 usually !alk about
theories, cxplanat ou and prcdictioll. Debates at that level are necessal
incondu sl'e. GlIly at the level of experim ental practice s seientifi
c real
sm unavoidable. Bul this realslll is not about theories amI tmth.
Thc

experimenlalst need on!v be a realist abont the entities used as tools,

A PIea for Experiments

No field in Ihe philosophy of ,dence is more s)'stematically neglecle


d Ihan
school teachers may have told LIS that scienlific
method is experim ental method, but histories of scicnce llave
become
histories of theory. Experim ents, the philosop hers say, are of value
whcn hey test theory. Experim ental work, Ihey ill1ply, has no life
of ils
own. So we lack even a lerminol ogy to describe Ihe many variecl
roles of
experim ent. Nor has this one-sidedness done theory any goocl, for
differen! types of theory are lIsed to think abont Ihe same "I""ir'.,1

FROM

Topics B (1982): 154-72.


1153

1154

CH. 9

EMPIRICI~M AND SCIENTlFIC REALISM

nomenon (e.g.. the magneto-optical effect). The philosophers of theory


have not noticed this ane! so misreport even theoretical inquiry.l
Different sciences at different times exhibit different relationships be
t",een "theory" ancl "experiment." Qne chief role of experiment is the
creation of phenomena. Experimenters bring into being phenomena that
do not natmally exist in apure state. These phenomcna are the touch
stones of physics, the keys to nature amI the source of much modern
lechnology. Manv are what ph)'sicists after the 1870s began to call "ef
fects": the photo-electric effect, the Compton effect, and so forth. \ rccent
high-energy extension of the creation of phenomena is tbe creation of
"evcnts," to use the jargon of the trade. Most of the phenomena, effects
and e\'ents creatcd by lhe experilllenter are like plutonium: they do not
exist in natme except possibly on vanishingly rare occasioI1S. 2
In this paper I leave asid' questions of methodologv, history, taxonomy
and the purpose of experiment in natural science. I turn to thc purely
philosophical issue of scientihc realismo Call it simply "realism" for short.
'There are two basic kinds: realism about entities ancl realism about the
ories. There is no agreement on the precise definition of either. Realism
about theories says we try to form true theories about the world, about the
inner constitution of lIlatter and about the outer reaches of space. This
realism gets its bite from optimism; we think we can do well in this project,
and ha\'e airead)' had partial success.
Realislll about entities-and I include processes, states, waves, cur
rents, interactioI1S, fields, black holes ancl the like among entities-asserts
the existence of at least some of the entities that are the stock in trade of
physics.'
The t\\o realisms ma)' seem identical. If you believe a theory, do you
not believc in thc existen ce of the entities it speaks about? If you believe
in some entities, must you not describe them in some theoretical wa)' that
you accepl? This seeming identity is illusory. Thc vast majority of exper
imental ph)'sicists are realists about entities without a commitment to re
alisI1l about theories. Thc experimentcr is con\'inced of the existence of
plent)' of "inferred" and "unobservable" entities. But no one in the lab
belie\'es in the literal truth of present theories about those entities. AI
though various properties are confidently ascribed to el ectroI1S, most of
these properties can be embedded in plenty of different inconsistent the
ories about which the experimenter is agnostic. Even people working on
adjacent parts of the same large experiment willuse different and mutually
incompatible accounts of what an electron is. That is beca use different
parts of the experiment will make different uses of electrons, and the mod
els that are useful for making calculations about one use mal' be com
pletel)' haywire for another use.
Do I describe a merel)' sociological fact about experimentalists? It is
flOt surprising, it wiII be said, that these goad practical people are realists.
They neeel that for their own self-esteem. But tbe self-vindicating realism

HACKING

EXPERIMENTATION AND SCIENTlFIC REALISM

1155

of experimenters shows nothing about what actually exists in the world.


In reply I repeat the distinction bet\veen realism about entities and realism
about theories and models. Anti-realism about models is perfectly coher
ent. Many research workers lIla)' in fact hope that their tbeories and even
their mathematical models "aim at the truth," but they seldom suppose
that any particular model is more tban adequate for a purpose. By and
large most experimenters seem to be instrumentalists about the 1II0dels
they use. The IIIodels are products of tbe intellect, tools for thinking and
calculating. They are essential for writing up grant proposals to obtain
further funding. Tbe)' are rules of thumb lISed to get things done. Some
experimenters are instrumentalists about theories and models, while sorne
are noto That is a sociological fact. But experimenters are realists about
the entities that they use in order to investigate other hypotheses or hy
pothetical entities. That is not a sociological fact. Their enterprise would
be incoherent without it. But their enterprise is not incoherejIt. It persis
tently crea tes new phenomena that become regular technology. M)' task
is to show that realism about entities is a necessary condition for the co
herence of most experimentation in natural science.

Our Debt to Hilary Putnam

It was once the accepted wisdorn that a word Iike "electron" gets its mean
ing from its place in a net\vork of sentences that state theoretical laws.
Hence arose the infamolIS problems of incommensurability and theory
change. For if a theory is modified, how could a word Iike "e1ectron"
retain its previous meaning? How could different theories about electrons
be compared, since the very word "electron" would differ in meaning from
theory to theory?
Putnam saves us from such questions by inventing a referential model
of meaning. He says that meaning is a vector, refreshingly like a dictionary
entry. First comes the syntactic marker (part of speech). Next the semantic
marker (general category of thing signified by the word). Then the stereo
type (clichs about the natural kind, standard examples of its use ane!
present day associations. The stereotype is subject to change as opinions
about the kind are modified). Finally there is the actual reference of the
word, the very stuff, or thing, it denotes if it denotes anything. (Evidently
dictionaries cannot inelude this in their entry, but pictorial dictionaries
do their best by iI1Serting illustrations whenever possible.)4
Putnam thought we can often guess at entities that we do not lite rally
point to. Qur initial guesses mal' be jejune or inept, and not every naming
of an invisible thing or stuff pans out. But when it cloes, and we frame
better and better ideas, tben Putnam says tbat although the stereotype
changes, we refer to the same kind of thing or stuff all along. We and

1158

ell..9

for realsm- is not that we inEer the


of eledruns frolll Ollr Sl1ccess. \Ve do nol makc lbe nstrlllllcnts
infer lhe realitv of lhe e1eclrulls, as when we test a hypolhesis,
passed the test.. That gets thc time-arder
relyiHg OH a moelest nLllnber of horne
SOllle other phenol11enon that we wish
lo
That may sounel as f we believe in the e1edrons because we preelkt
ho\\' our apparatus I\"il! hehave. That too s lllisleadillg. We have a llllmher
of general ieleas abont hOI\" lo prepare polarized e1ectrons, sayo \Ve spelld
a lot of tillle huilding protolypes that c1on't work. We
riel of irllllll11er
ahle bllgS. Often we have to give I1p a;d try another approach. Debllgging
is not ;1 matter of theoretically explaining or predicting what is going
wrong. It is partly a malter of getting rid of "noise" in the apparatus.
"Noise" often Ille<lns all the events that are 1l0t nnderstood by any tlIeory.
The instrul11enl must be able lo isolate, physically, tbe properties uf lhe
entibes that \Ve wish to use, amI dalllp down all he other effeets that
mght
in Ollf war. We are completel)' convillced or the realt)' ofelectrom

whm we regular/y set out to build- and orten eHough succeed in


-lIew kinds or devces that use various well !lllderstood causal propertes
electrolls to interfere in other more hypothetical parts or llature.
lt is not Dossible to grasp this wtbout an cxamplc. Familiar historical
bccome encrusted by false theory-oriented
So J shall take sOfllething new. This is a polarizing
electIOll gun whose acronym is PECCY 11. In 1978 it was usecl in a
fundamental experirnent that attracted attention even in The New York
Times. In lhe next sectoll I descrihe lhe point o[ rnaking PECCY Il. So
I llave lo tel! SOllle new Dhvsics. You can omit this and read only the
Yel it must be of interest lo kno\V the
O[ the rnain experimental results,
nalllcly, (l) parily is not con;~rvcd in seattering ""bT'7pcl eleetrons from
denterium, amI
more generally, Daritv is
rent interaetions.'

HACKING F.XPERIMENTATION ANI) SCIENTIFIC REALISM

EMPIRICISM AND SCIENTlFIC REALlSM

Methodological Remark
in the seetio!1 after

lt is the lalter that Illat


ters to my case, !1ot the formcr. Importantly, even if present
turns ont to need radical revision, the
PECGY !l, will stll work. J am concerned \Vith how it was macle lo

1159

and why. I shall sketch far more sheer engilleering than is seen in
papers. 1\1y reason is that the engineerillg is incoherent unless elee
trons are taken for granled. Quc canflot say this by merely reporting,
they macle an electro n glln for shooting polarized electrollS." An immense
knowledge of how to manipulate electrollS, of what sorts (lf
Ihey \ViII do reliably and how they tend lo misbehave-that is the kind
whieh grollnds lhe experimenter's realsm abolll e1eetrons. YOH
cannot grasp Ihis kind of knowledge in the abstract, for it is
kTlowledge. So I must Dainfullv inlroduce tite reader to sOllle IaDormO!
Luekily it is a

Parity and Weak Neutral Currents

Therc are four fundamental forces in nature, Ilolnecessarily distinct. Crav


ily ami electromaglletism are familiar. Then there are the strong alld weak
forees, the fulfilllllcnl of Newtou's program, in the Optics, whicll
that al! nature IVollld be uuderstood by the interaction of partdes wilh
variOllS forees that \Vere cffeetive in atlraetion or repulsion over various
different distances (i.e., wth different rates of extinetion).
Strong forces are 100 times stronger than eleetrolllagnetsIll bul <let
for a miniseule dstanee, al l1lost the diameter of a proton. Strong
forces aet 011 "hadrons," which include protons, lIeutrons, anclmore reeent
particles, but l10t eleetrons or any other members of the class of particlcs
ealled "leptons."
The \Veak forees are only 1f1O,i)UO times as strong as electromagnetism,
alld <Jet over a distancc l/iIIO times smaller than strong forees. But Ihey <Jet
on both hadrons amI Ieptons, including electrons. Tite most familiar ex
alllple of a weak force lllay be radioactvity,
The tlteory that motiva tes such speclIlation is quantlllll e1ectrodynam
es. lt is increclibly successful, yielding IIIally predctions belter than one
part in a milliol1, a miraele in experimental physics. It applies ayer dis
tances ranging frOln diallleters of the carth to l/lOO the diameter of the
proton. Tltis titear)' supposes t] all the forces are "carried" hy some sort
of partcle. Photons da the ob in electromagnetislll. We hvpotltesize
itOllS" for gravity.
In the case of interaetiollS involving weak forces, tltere are charged
currcnts. \Ve postulale that particles callecl bosolls carry these weak forees. H
For charged currents, the hosons lIIay be positive or negative, In Ihe 1970s
there arose the possibility thal there could be weak "neutral" currents in
whieh 110 charge is carried or exchanged. By sheer ana!ogy wth the vin
dicated parts of qllantulll elcctrodynarnics, neutral hosons \Vere postulaled
as the earricrs in weak interactiolls.
s the [ailure
energy
The most falllous discovery of recent

1160

CII. 9

EMPIRIClsM AND SCIENTlFIC REALlSM

of the cOllservation of parity. COlltrary to Ihe expectations of lllany


icists and philosophers, including Kant,9 nature makes an absolute
linction between righl-handedness and left-handedness. Apparently this
in weak inleractions. *
or left-handed in nature has an clemellt of
eleclrons have spin. Imagine your right hand
\Vith the lingers poillting in the direc
is said to point in the direction of the spin
are lraveling in a beam. consider the relatioll
_ vector alld the beam. If all the particles have their
vector in Ihe same direction as the beam, tbey have right-banded
polarization. while ir the spin \'ector is opposile to the beam direction,
they have lcft-handcd (linear) polarzation.

The original discovcry, of parity v;oJation showcd that une kind of


product of a particle deea}:, a so-called muan eutrino, cxisls only in left
hancled polarizatioll and never in right-handed polarization.
Parity violalons have beell found for weak clwrged inleractions, What
about weak neutral currenls? Thc remarkable Wcinberg-Salam model for
the four kinds of force was proposed independently by Steven Weinberg
in 1967 and Arbdusl Sala m in I9H, It illmlies a minute violatioll of party
model is sheer spcculation,
its suceess has been amazing, even awe uspiring. So it seellled worthwhile
to trv out the prcdcted failure of parl)' for weak neutral interactions. That
teaeh lIS more about hose weak forces that act over so Illinute a
distanee.
The prediction is: Slightly more left-handed polarzed
ting certain targets will seatter, than right-handed electrons. Slightly more!
lile diffcrenee in relative freqnency of be 1\\'0 killds of scattering is one
part in 10,000, comparable lo a differellce in probability between 0.50005
and 0.49995. Suppose oue llsed the standard eqllipment available al the
Stanford Linear Accclerator in the early 1970s, generating 120 pulses per
second, each pulse providing one e1ectron event. Then you \Vould have
lo nlll the entire SLAC lStanford Linear Aeeelerator Center] beam for 27
years in order to deteet so smal! a difference in relative freqllency. Con
sidering that olle uses Ihe same beam for lots of experiments simultane
by lettiug different experiments use differeut pulses, :lnd ronsder
Kant did not dcny that there is a real, intrinsk diffcrellce between
and left-handed ob'jects, Hacking is referring to Ihe belief, he Id universally prior
to !956, that the laws of nature are indifferent to the left-rght distinction: for
process that is physically p05sible, 50 too is ils mirror
It s t}is helief
parity-volatOll experments proved false, See Martin
rile New Ambidex
tmus Universe, 3d rev. ed. (New York: W, H. Freeman and Company, 1990),
ch. 22, and Jallles Van Cleve and Robert E, Fredcrick, eds" Tite Phlosop/), of
Riglzt ad Left: Incongruent Coullterparts and tile Natme of Space (Dorclrecht,

Netherlands: ..

HACKING EXPF.RIMF.NTATION AND SCTENTlFIC REALISM

1161

that no equipment remains stable for even a monlh, let aone 27 years,
slleh <In experiment is impossible, You need enorl11ously more electrons
off in ea eh pulse. We need between 1000 and 10,000 !llore dee
trons per pulse than was onee possible. The lir,t
mellt now called PEGGY l. lt had, in essence, a high-class version of
1- 1- Tholllson's hot eathode. Some lithiul11 was heated and electrons were
boiled off. PEGGY II uses quite different

PEGGY JI

The basic idea

artic1e in an

when C. y. Preseott
magazine about a

arsenidc, GaAs
a nUll1ber of clirious propertles
in laser
One of its quirks is that whell it is strllck by LllLlllauy
polarized light
the right frequeneies, it emils a lot of lineady
electrons. There is a good rough and ready quanlum
this
and why half the emitted eleetrons will be pobrized,
nolanzed in one drection ancI lA polarized in the other.
PEGGY II uses this fact, plus the faet Ihal GaAs emils lols of deetrons
dne to fealmes of its cryslal struclure. Then comes some engincering. It
lakes \York to liberate an e1ectron from a smface. We know thal paintillg
a surfaee wilh lhe rght sllbstanec helps. In this case, a thin layer of eesium
and oxygen is applied to the erystal. Moreover the less air pressme 3rouIICl
Ihe crystal, Ihe more eleclrons wiU escape for a given amount of work. So
the bOlllbardmellt takes place in a good vaCllUIll al the lelllDeratnre of
Iquid lIitrogen.
We neecl lhe right source of light. A laser with bursls of red Iighl
llgstroms) is trained on the cryslaL Tbe light first goes lhrough an
ordinary polarizer, a very old-fashioned prisl11 of caleite, or Iceland spar. 11l
This gives linearly polarized light. We want cireularly polarized ighl lo
hit he crystal. The polarized laser bemn [Jow goes throllgh a cnnning
lTlodern device, caBed a Pockels cell. It electrically tUfllS linearly poJarized
inlo circulady polarizeJ oncs. Being electric, it ad" as a very fasl
switch. The direction of circular polarizalion depends on the dreclion of
curren! in the cell. Hellce the direction of polarizalioll can be varecl
This is important, for we are trying lo delect a minute aSylll
metry between right- and Ieft-handed polarization. Ralldomizing helps us
againsl any systematic "drift" in the eql1ipmcnt. 11 lile randol1liza
tion is gencrated by a radioactivc deea)' device, and a computer rccords
lhe direchon of polarization for each
A circulady polarized pulse hits the GaAs crystal, resulting in a pulse
of linearly polarizcd electrons. A beam of such pulses is malleuvered by
magnets into lhe accelerator for the ncxt bit of the experimellt. It pass es

CI!, 9

116:1.

IlACKI1'<~G EXPERnlENTAT10N AND SCIE'ITlF'IC REALlnl

I\MPIRICIS\I AND SCIENTIFIC REALlSM

through device that check> 0[\ a proportiOll of polarization along the


wa\', The remaimler of the
requires othcr dcviees and e!eteclors
at PEGCY 11.
of comparablr im!elluit\'. but le! lIS

116 3

eleetrons IVere sealterecl fmm deuterilll11


eleetrons, This was the firsl
in a weak neutral cmrenl inter
aetioll, "

Bugs

Short deseriptiol1S make it all sOllne! too eas\', so lel llS pause to rcAect on
l\lal\\' of the bllgS are never underslood, TIJe) are c1iminated
allCl error. Let lIS illustrale three different kinds: (l) The essential
t~ehnical IinIitatons tlr< in the end have lo be faclored into the analysis
of error. (2) Simpler mechanical defects ron ne\er think of lIl1II they are
forced on \'OlI, C~) I-Iunches abollt whal ITIighl go \\Tong,
1, Laser beams are no! as COllstant as sccnce fiction teaches, and
there is always :m irremediable amOl1!lt of "itter" in the beam over any
stretch of time,
2, At a more humdrul1l le\'el lhe electrons from the GaAs erystal are
the saflle e hannel as the laser beam used
baek-scattcred and go baek
to hit the crystal. tvlost of them are then deAecled
reHeded frolll the Iaser apparatlls and get back into the system, So yOl!
to eliminate these !le\\' ambient eleclrons. This is done
erucle
mechanical llleans, making tbem foeus jllst off he erystal and so wander
~l\vay.

~,

Good
011 an
hit, them, ,lIld then stand
in lhe

the absurd,
011

llsing
antidnst sml\",

jusi i11 caseY

Results

SOllle IO IJ e\'ents "ere needed to obtain a reslIlt tha! eOldcl be


anc! statistical error. i\lthough lhe idea of s)"stem<ltle error
conceptllill prohlems, it seelllS to he nnkllo\\'ll to
There ,,'ere sl'stelllatic tlllcertainties in [he detection of right
there \las some ilter, and there \\ere other
ahout thc paralllcters of the two killds of bealll, These errors
ane! linead\' added lo lhe statistical error. '1'0 a student of
inference this is re,d seat-of-the-pants 111alysis with no rationale
whatsoever, Be that as it mal', thanks lo PECCY Il Ihe nlllllher of evenls
was bil! enOIlgh to give a r~sult that convinced the entirc Dhvsics COI11

Cornment

Tite making of PECCY JI \Vas

in ad\ anee the

the polarizatiol1 cffeet, il c10es


cr\'stal lIscd, No one has bcell
37 pcrccnt of lhe clectrollS,

lo polarize
in principIe 50
although \vc
and oxygen \Viii "produce
for c1eetro!ls to escape, IVe have !lO
increases effieiency to a ,eore of 37
Nor \Vas there aH\' guarantee
gether. To
IVork, briefly dcscribed later in th is
trons per pulse than PECCY 11
\Vas reported in The New York Times, a
the newspaper amI sal\' ",hal was
a crystal lattice for totallv unrelaled purposes, lt uses la\'crs of Ga/\s
a relatecl altllninull1 ('ompollncl, TIJe stmeture of this lattiee IC,lds (me lo
expect thal I'rtllallv alI the electrons emitted \\onld be polarized, So \\'e
might be able to clollble the efficiencv of PEGCY 11 Bnt at prcsent
1981) that nice idea has problems, The /le\l' attice sllOllld also be
in 1I'0rk-redm:ing paint. But the eesiulll-oxygen compolll1d is applied at
tClllperature, TIrcn Ihe alt!lllinlllll tenels to ooze lito the lleigbboring
of Ga/\s, alld tire pretty ,Htifieial latticc bceomes a bit llllevcn, lilll
ting its fine polarizecl-eleelron-emitting properties, So perhaps this \\ill
ne\'er \York,14 The group are sillllllhmeouslv rel'i\'ng a soupcd U]> !le\l
to get !llOre electrons, :YL1\'be PEGCY 11 \\'ould
thermionie cathode to
ha\'c shared the smne firte. llC\'er \\'orking, ami thermionic dcviees \\'Oulcl
have sto len the SllO\\',
Note,
, that Ihe Bell people d id not !leed lo ki 1011' a lot of

u64

CH. Q

weak neutral eurrenl lheory lo send


read The New York Times.

HACKING EXPERIM}:NTATION AND SClIlNTlI'IC REALlS~1

E~IPIRIGISM AND SCIENTlI'IC REALISM

Iheir

lattiee.

Moral

Once llpon a time it made good sense to doubl tbal !here are elcdrons.
clOllbt Illade
Even afler Millikan Iwd IlIcasurcd Ihe charge on the
sense. Perhaps Millibn was engaging in "inference to the best
lion." The charges on his carefully selec!ed oil drops were all
grallllultiples of a leasl charge. He inferred Ihal this is Ihe real leas!
in natme, and hence il is lhe eharge on Ihe
and hellce there are
of leasl. charge. In MilIikan's da}' mos!
ane ar more theories
Ihe eledroll. Hawever il is always adlllissible, al leasl for .
lo treal inferenecs lo Ihe besl explanaliall in a purely inslnlIllenlal
wilhoul aHy commilmcnt to the exislellce of entities used in Ihe
1, Bul il is IlO\\" sevellty rears after Millikan, ancl \\"c no longer
ha\'e lo i!lfer from explanalory success. Prescoll el al., clOl explain phe
nomena wilh elCclrollS. Thc)' kllow a greal clcal aboul how lo use Ihem.
Thc group of experimenters do no! knolY what electrons are, exactly,
Inevilably they think in lerms of particles. Therc is also a c10ud piclme
of <In electron ",hich he1ps liS thillk of complex wave flll1ctions of c1ectrons
ill a bonlld stale. The angular momenlull1 ami spin vector of a c1o11cl
make little sense outside a mathematical formalism, A beam of
c10nds is fantas)' so no experimenter uses that Illoclcl-nol b~callse of
doubting its lrulh, but beca me other 11l0clels help !llore with Ihe ealcu
lations, Nobody thinks that clcctrons "reallv" are lIsl little
abolll w]ich yOll could, with a small
find Ihe direction of 5Din alone: the
clse, e.g., weak neutral
eurrents amI nentral bosolls. \Ve kno\\" an enormous amOL1llt abonl the
hehavior of clectrolls. 'vVe also kno\\' wllat does not maller lo clcclrollS.
Tlllls we knolY lllal bcncling a polarized eleetron heam in magnetic eoils
cloes !lot atTecl polarization in ;my significant \Val'. \Ve have hU!lches. loo
strong to ignore although too trivial to test
dance Llllcler changes of directions of polarization. 1110se Illlllclles are
based on a hard-won sellSe of Ihc killds of thillgS clt:ctrons are. lt does not
maller al all to l!Jis hUllCh \vhelher e1edrons are cloucls or
The .
words relrieved from mecliacval sciellee bv
1l0Illelltl," On the cOlltrary, \Ve believe in <thelll because we use tbem to

116 5

create ne\\' phenoll1ena, such as the phC1l0111enOn of parity violalion in


IVeak neulral currenl illteractions.

When HypotheticaI Entites Become Real

contras! bet\\'eell eleclrons and neutral bosoll s.


can yet malllpulale a bllllCh of nentral
ir there are a11)'. En:ll
neulral currents are only
frolll lhe Illists of
1980 a sufficiellt range of
experiments had made thcl11 Ihe
of investi!!atiou. \Vhcn mi!!ht
lose their hypothelcal status and
clcclrons? \Vhen \l'e use thelll lo in

1 mcntioned the desirc lo make a betler gun Ihan PEGGY JI. Why?
Becallse we \lO\\' "know" thal parity is l'iolated in weak neutral inleractiol1S,
Perhaps by an even more grotesque statislical <lllal)!sis than that illl-olved
in the parit)! experiment, \\'e can isolale just the weak interacliolls. That
is, \Ve huye a lot of interadions, incllldng say
can censor these in variolls ways, but IVe cml also stallstJcally l1ick out a
class of weak interactiollS as precise1} those where
This wOllld possibly give llS a road lo
deep investigations of matler
3nd anti-Illatlcr. To do lhe slatistes O!le nceds even more eledrons per
If sueh a projecl \verc lo
than PEGGY 11 could hope lo
we should be begi\lning to use weak neulral currenls as a ma
1001 for looking al somelhing cIsc, Tlle llext step tOIV~rcls a re
alislll aboul such currents IVould have been made.
Thc message is general amI could be extmcled from almost any
branch of physics. Dudley Shapere has recenlly llscd "observation" of Ihe
sun's hot eore lo illllstratc !JOI\' physicists emplov Ihe cOllcept of observa
tion. They collect neutrinos fmm Ihe stln ill an cnOflllOllS disllScd lInder
grouncl mine Ihat has been fillcd ",ilh the old cbllliug fluid (i.e., earholl
letrachloride). \Ve \\'ould know a 101 about the inside of the sun if \\'e
kneIV bOl\" ll1anv solar ncutrilloS arrive Oll the emth. So Ihese are
in Ihe cleanillg fluid; a fe\\' \ViII form a ne\\"
llumber IImt do this can be counted.
nelltrinos lo investigate sornelhing
here we are plaillly
else. Yet llol manv years ago, llClllrinos IYere abollt as hypothetica! as an
eould
After 1946 il was rcalized Ihal whell mesOlls distintegwle,
Hlllong other things, high1y
electrons, Oile neecled an
particle to conserve lllomentt1l11 amI cucrgv. At that lime
"nculrino" W<lS thoroughly hypothelical, but uo\\' it is rou
Ilsed lo examine other Ihings.

1166

Cn. 9

EMPIRICISM AND SCIENTlFIC REALISM

Changing Times

Although realisl11s alld anti-realisms are part of Ihc


\Vell back iulo Greek prehistory, our prcsenl versions I110stly descend from
debales abollt alomism al he end of the nineleelllh century. Anli-realisrn
abont alollls \Vas partly a rnatter of physics: lhe cnergetcists 10llghl cnergy
\Vas at Ihe bottolll of everything, nol tiny bils of matter. It also \Vas con
necled with lhe positvism of Comle, Mach, Pearson nd even 1 S. MilI.
Mill's young associate Alexander Bain states the Doinl in a charaderslic
wa)', apt [or
Sume hypotheses consst of assumptiollS as to the mlmte slructme onu op
eraliollS of bodies. Fwm the natme of lhe case Ihesc assllmpliolls can ne"cr
be provcd by diree! mcans. Thcir lllerit is Iheir snitability to express
ella. Thev flre Representative Fietiolls. 16

"AI\ asscrlions as lo he ultilllale slruclure of lhe particles of matter," COIl


tinues BOln, "are and ever musl be hypolheticaL ..." The kinelic
of heat, he says, "serves an imporlant inlellectual functioll." But we cannol
hold il lo be a Ime descriplion of the world. It is a Rcpresentative Ficlon.
Bain was sllfely right a century ago. Asslllnptions about the minute
slructure of matter could nol be proved lhen. The only proof eould be
namcly that hypothcses seemed lo provide some explallation and
make good predictions. Sucb inferences !leed never produce eon
vielion in the philosopher inc!ined lo instrumenlalisll1 or SOl1le olher
brand of idealismo
Indeed the situation is quite silllilar lo sevenleenth-cenlury episle
IlIology. Al lhat time kno\Vledge was lhought of as corrcct representation.
But Ibcn one con Id never gel outside lhe represenlalions lo be sure hat
corresponded lo he world. Every test of a represenlalioll s jusi an
olher representation. "Nolhing is so ll1uch Iike an idea as iln idea," as
Bishop Berkeley had il. To altempl lo arguc for sclentific realisl11 al he
level of Iheory, lesting, explanatioll, prediclve Sllecess, convergence of
theories and so forth is lo be locked illto a world of rCDresentatOl1s. No
wOl}(lcr thal sccnlific anti-realisll1 is so
varianl 011 "Thc SDeclalor Theory of

HACKING EXPERIMENTATlON ANO SCIENTlFIC REALlSM

1167

as opposed lo pbilosophers, dd in general become realisls


1910. Michacl Garuner, in (me of lhe finest stlldies of
delails many of lhe factors that wenl inlo lhat
in clirnate of opinionP Despite the changing c1imate, some
of instrumentalism or fietionalism remained a
native in 1910 and in 1930. That is what lhe history
USo lis most recent lesson is Bas van Fraassen's The Scentfic
"construclive empiricism" is allolher theory-orienled anti-realislll.
son is: think aboul practice, not lheury.
Anti-realism about aloms was very sensible when Bain wrote a celllury
ago. An\-realism about any sub-microscopic entilies \Vas a soune! doclrine
in Ihose days. Things are different now. The "direct" proof of electrons and
the lke is our abilitv to l1laniplllate them IIsing wd! understood low-Ievel

properties. Ido 110t of course claitn that "reality" is constituted hy bu


man manipulability. We can, however, ealJ something real, in tite sense in
whieh il matters lo scientific realism, only when we understand quite welJ
\Vhal its causal properties are. The best evidence for Ihis kind uf understand
ing is that we can set out, fmm scrateh, lo build machlles thal will work
fairJy reliably, laking auvantage uf Ihis or that causal nexllS. Hcnce, engi
not theorizine. is Ihe proof of seientific realism about entites. lB

Notes

1. C. W. F. Everitt and an

"Which Comes First, Theory or Experi


.. of Jan Hackin2. Represe/lting

Z. Jan
IpVPU1/)Pl1rfJ:

der P];inome
(P. l)uerr,

3. Nallcv Carl-wright makes a similar dislinclion in a sequence of papers.


n (1982):
Leads lo Inference," [philo8ophical
She approaches realis11l from lhe top, dislinguishing theorclicaI la\\"s (which
do 110t stale the faels) from phenomenological laws (",hiel! do). She believes in
SOllle "lheorelical" entities and rejects lllllCh thcory on the basis of a slIbtle
of ll10deling in physics. 1 proceed in the opposite direclion. from
Both aooroaches share an interest in real-life phvsics as
. science. J\1y 0"'11 approach owes al! enOfmOllS amounl lo Carl
par'lllel developments, whch have often preceded m}" o\\"n. My use of
the t\Vo kinds of realislIl is a case in point.

4. Hilary Putnarn, "Ho\V Not To Talk About


ing'," and other papers in Ihe I\1nd. Language and
Vol. Z. Cambridge, 1975.
5. Francis Bacon, Tlle Creat Instauration, in '[711'
Bacon (J. M. Robertson, ed; Elli, amI Spedding,

,. "The Meaning of 'I'vlean

Philosophcal

Phi/nmi">'irr1!

1168

CII,9

EMPlIUClSM AND SCIF.NTlFIC REALlSM

6, an Hackillg, "Do We See Through a Micrmrnnp

Pacfic Philosophical Quar

62, (wintcr 1981), ~05-22

thank Melissa Franklill. of the Stallford Lnear Accelerator. for


lile lo PECCY 11 and lelling me how il tvorks, SlIe also ammged disCllssiolls
lllclllbers of Ihe PECCY 11 grollp, ,ome of whch are lucntioned below, The report
of eXDeriment E-I22 described here is
Non-conservatioll in Inelastic Elec
" C. y, Preseo\! el aL, l'h)'sics Let/ers, 1 have relied hcavilv 011 the
in-house oumal, Ihe SLAC fleam Line, Report No, 8, Odober.
ViolalOll in Polmized Eleclron Scattering," This \Vas prepared by tbe in-bollse
seienee \Vriter BiH Kirk, who is tlJe clcaresl, most readable DOIJlllarizer of dif/icult
Ile\\' exoerimcntal physics that I llave come across,

DAVID

B.

RESNIK

Hacking's Experimental Realism

8, The odd-SOlllldiug bosollS are !lamed after the ludian physicisl S, N, Bose
894-197+), also remembered in the name "Base-Einstein statistics" (which bo
SOIlS satis~'),

9, Bnt

Leibniz, who '''kncw'' there had to be some rcal, natural differ


and left-hundedncss,

Iceland spar is an e1egant cxample of hol\' experimental phellomena persist


cl'cn while theories abollt thcm nlldergo revollltons, Mariners brought calcite
from lteland lo Scandinavia. Erasllllls Barlholn experimented with it and ",role
about it in 1689, \Vhcll )'Oll look throllgh thcse beautiflll crystals yOll sce donble,
thanks lo the so-callcd ordiuaf\' and extraordinary rays, Calcite is a natural polar
which for 300 ycars \\'as lhe chief route lo
and tIJen eleclro
is ~I happ\' reminder of a great
11. It also tmm CaAs, a %-If
12, I (me these

hand

into a 50-50

lo conversation with Roger Miller of SLAC,

3, Tite cOllcept of a "comincing experiment" is fundalllental. Peter Calison has


done important \York on t!Jis idea, studying European ami American expcriments
011 weak IIcutml eurrcllts condllcted durillg the
14,

J o\\e Ihis informatioll to Charles Sinclair of SLAC,

15,
~
attilude to "inference lo the best
froll!
See, fOf exalllplc, her papcr I"When
ference," loe. cit,
l6, Alc\;lIlcler Bain,
p. ~62.

is one of lI1ally leamed


Leads lo In

Deduclive (lIld Inducfil'e, LOlldon ami New York, 1870,

Traditional debates about scientific realism tend lo focus on i,sues C011


presentation (broadly speaking) and de-emphasize is
sues cOllcernmg scicntific inlervention, Questions about the relation
belween Iheories amI the world, the nature of scientific inference, alld the
structure of scienlific explanations ha ve occupied a central place in the
realislll debate, while guestons abollt experimentation and
have not. lan
shifting the defense of realism awav from representation to intervCll
tion,1 Experimental realism, according to Hacking, cloes not reql1ire us to
believe that our theories are tflle (or approxim3tely tme), nar does ils
defense depend 011 inference to the best explanation, For Hacking, lhe
proof for realislll is that \Ve can manipl1late obiects: 'So [ar as
I'm concemed, if yOll can spray them, then Ihey are real'
In this papcr I arglle (1) that Hacking's argl1ment for
realisllI is, despite his slrong denials, another \'ersion of the 'slIceess of
science' argllmenl;
Ihat the experimental realist can onh- have knowl
edge aboul theoretical enhties if she assumes that Ihe
describe those enlities are at least approximately tme; and (~) tllat exper
imentation is not nearly as tbeory-free as Hacking l11aintains, The COllll11011
Ihrean in these three crilicisllls is that Hackill! does not slIceeed
the <lefense of realism away from questions
is a fonu of

17, l\lichacl Gardncr, "Rcalislll and Instrumentalislll in 19th-Cclllur)' Alomislll,"

Philosol.>",\' o{Scellce 46, (1979),1-34,


8, Added in proof. February, 1983.1, As inclicatcd in lhe texto this is a
Julr, 1981, ami henec s Ollt of date, For eX<lmple, neutral bosons are described
as purely hypothetical. Their status has changcd since CERN 311l1OUnced Oll Jan,
n, 198~, Ihat a group there had found W, the weak intermedian'
HlItiprotOIl dcca)' ,Jt 540 CeV, These
111\'
. book, Represelltillg <1lld
FROM Canadian !oumal o{ P'osoph)' 24 (1994): 395-412,

1169

Вам также может понравиться