Вы находитесь на странице: 1из 12

Bertrand Russell

The Relation of Sense-data to Physics


(in his Mysticism and Logic (London: George Allen & Unwin Ltd.:
1917). Reprinted Totowa !ew "erse#: Barnes & !o$le Boo%s 19&1.
'ages 1()*1+1. 'agination here ,at-hes the latter.)
I. THE PROBLEM STATED
'./0120 is said to $e an e,piri-al s-ien-e $ased upon o$ser3ation
and e4peri,ent.
1t is supposed to $e 3eri5ia$le i.e. -apa$le o5 -al-ulating $e5orehand
results su$se6uentl# -on5ir,ed $# o$ser3ation and e4peri,ent.
7hat -an we learn $# o$ser3ation and e4peri,ent 8
!othing so 5ar as ph#si-s is -on-erned e4-ept i,,ediate data o5
sense: -ertain pat-hes o5 -olour sounds tastes s,ells et-. with -ertain
spatio*te,poral relations.
The supposed -ontents o5 the ph#si-al world are prima facie 3er#
di55erent 5ro, these: ,ole-ules ha3e no -olour ato,s ,a%e no noise
ele-trons ha3e no taste and -orpus-les do not e3en s,ell.
15 su-h o$9e-ts are to $e 3eri5ied it ,ust $e solel# through their
relation to sense*data: the# ,ust ha3e so,e %ind o5 -orrelation with
sense*data and ,ust $e 3eri5ia$le through their -orrelation alone.
But how is the -orrelation itsel5 as-ertained 8 A -orrelation -an onl#
$e as-ertained e,piri-all# $# the -orrelated o$9e-ts $eing -onstantl#
found together. But in our -ase onl# one ter, o5 the -orrelation
na,el# the sensi$le ter, is e3er found: the other ter, see,s
essentiall# in-apa$le o5 $eing 5ound. There5ore it would see, the
-orrelation with o$9e-ts o5 sense $# whi-h ph#si-s was to $e 3eri5ied
is itsel5 utterl# and 5or e3er un3eri5ia$le.
There are two wa#s o5 a3oiding this result.
(1) 7e ,a# sa# that we %now so,e prin-iple a priori, without the
need o5 e,piri-al 3eri5i-ation e.g. that our sense*data ha3e causes
other than the,sel3es and that so,ething -an $e %nown a$out these
-auses $# in5eren-e 5ro, their e55e-ts. This wa# has $een o5ten adopted
$# philosophers. 1t ,a# $e ne-essar# to adopt this wa# to so,e e4tent
$ut in so 5ar as it is adopted ph#si-s -eases to $e e,piri-al or $ased
upon e4peri,ent and o$ser3ation alone. There5ore this wa# is to $e
a3oided as ,u-h as possi$le.
(:) 7e ,a# su--eed in a-tuall# de5ining the o$9e-ts o5 ph#si-s as
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 1<9
5un-tions o5 sense*data. "ust in so 5ar as ph#si-s leads to e4pe-tations
this must $e possi$le sin-e we -an onl# expect what -an $e e4peri*
en-ed. And in so 5ar as the ph#si-al state o5 a55airs is in5erred 5ro,
sense*data it ,ust $e -apa$le o5 e4pression as a 5un-tion o5 sense*data.
The pro$le, o5 a--o,plishing this e4pression leads to ,u-h interesting
logi-o*,athe,ati-al wor%.
1n ph#si-s as -o,,onl# set 5orth sense*data appear as 5un-tions o5
ph#si-al o$9e-ts: when su-h*and*su-h wa3es i,pinge upon the e#e we
see su-h*and*su-h -olours and so on. But the wa3es are in 5a-t in5erred
5ro, the -olours not 3i-e 3ersa. 'h#si-s -annot $e regarded as 3alidl#
$ased upon e,piri-al data until the wa3es ha3e $een e4pressed as
5un-tions o5 the -olours and other sense*data.
Thus i5 ph#si-s is to $e 3eri5ia$le we are 5a-ed with the 5ollowing
pro$le,: 'h#si-s e4hi$its sense*data as 5un-tions o5 ph#si-al o$9e-ts
$ut 3eri5i-ation is onl# possi$le i5 ph#si-al o$9e-ts -an $e e4hi$ited as
5un-tions o5 sense*data. 7e ha3e there5ore to sol3e the e6uations
gi3ing sense*data in ter,s o5 ph#si-al o$9e-ts so as to ,a%e the,
instead gi3e ph#si-al o$9e-ts in ter,s o5 sense*data.
II. CHARACTERISTICS OF SESE-DATA
7hen 1 spea% o5 a ?sense*datu,? 1 do not ,ean the whole o5 what is
gi3en in sense at one ti,e. 1 ,ean rather su-h a part o5 the whole as
,ight $e singled out $# attention: parti-ular pat-hes o5 -olour
parti-ular noises and so on. There is so,e di55i-ult# in de-iding what
is to $e -onsidered one sense*datu,: o5ten attention -auses di3isions to
appear where so 5ar as -an $e dis-o3ered there were no di3isions
$e5ore. An o$ser3ed -o,ple4 5a-t su-h as that this pat-h o5 red is to
the le5t o5 that pat-h o5 $lue is also to $e regarded as a datu, 5ro, our
present point o5 3iew: episte,ologi-all# it does not di55er greatl# 5ro,
a si,ple sense*datu, as regards its 5un-tion in gi3ing %nowledge. 1ts
logical stru-ture is 3er# di55erent howe3er 5ro, that o5 sense: sense
gi3es a-6uaintan-e with parti-ulars and is thus a two*ter, relation in
whi-h the o$9e-t -an $e named $ut not asserted, and is inherentl#
in-apa$le o5 truth or 5alsehood whereas the o$ser3ation o5 a -o,ple4
5a-t whi-h ,a# $e suita$l# -alled per-eption is not a two*ter,
relation $ut in3ol3es the prepositional 5or, on the o$9e-t*side and
gi3es %nowledge o5 a truth not ,ere a-6uaintan-e with a parti-ular.
This logi-al di55eren-e i,portant as it is is not 3er# rele3ant to our
present pro$le,@ and it will $e -on3enient to regard data o5 per-eption
as in-luded a,ong sense*data 5or the purposes o5 this paper. 1t is to $e
o$ser3ed
11( A/0T1210A A!> L<G12
that the parti-ulars whi-h are -onstituents o5 a datu, o5 per-eption are
alwa#s sense*data in the stri-t sense.
2on-erning sense*data we %now that the# are there while the# are
data and this is the episte,ologi-al $asis o5 all our %nowledge o5
e4ternal parti-ulars. (The ,eaning o5 the word ?e4ternal? o5 -ourse
raises pro$le,s whi-h will -on-ern us later.) 7e do not %now e4-ept
$# ,eans o5 ,ore or less pre-arious in5eren-es whether the o$9e-ts
whi-h are at one ti,e sense*data -ontinue to e4ist at ti,es when the#
are not data. 0ense*data at the ti,es when the# are data are all that we
dire-tl# and pri,iti3el# %now o5 the e4ternal world@ hen-e in
episte,olog# the 5a-t that the# are data is all*i,portant. But the 5a-t
that the# are all we dire-tl# %now gi3es o5 -ourse no presu,ption that
the# are all that there is. 15 we -ould -onstru-t an i,personal
,etaph#si- independent o5 the a--idents o5 our %nowledge and
ignoran-e the pri3ileged position o5 the a-tual data would pro$a$l#
disappear and the# would pro$a$l# appear as a rather haphaBard
sele-tion 5ro, a ,ass o5 o$9e-ts ,ore or less li%e the,. 1n sa#ing this 1
assu,e onl# that it is pro$a$le that there are parti-ulars with whi-h we
are not a-6uainted. Thus the spe-ial i,portan-e o5 sense*data is in
relation to episte,olog# not to ,etaph#si-s. 1n this respe-t ph#si-s is
to $e re-%oned as ,etaph#si-s: it is i,personal and no,inall# pa#s no
spe-ial attention to sense*data. 1t is onl# when we as% how ph#si-s -an
$e known that the i,portan-e o5 sense*data re*e,erges.
III. SESIBILIA
1 shall gi3e the na,e sensibilia to those o$9e-ts whi-h ha3e the
sa,e ,etaph#si-al and ph#si-al status as sense*data without ne-es*
saril# $eing data to an# ,ind. Thus the relation o5 a sensibile to a
sense*datu, is li%e that o5 a ,an to a hus$and: a ,an $e-o,es a
hus$and $# entering into the relation o5 ,arriage and si,ilarl# a
sensibile $e-o,es a sense*datu, $# entering into the relation o5
a-6uaintan-e. 1t is i,portant to ha3e $oth ter,s@ 5or we wish to dis-uss
whether an o$9e-t whi-h is at one ti,e a sense*datu, -an still e4ist at a
ti,e when it is not a sense*datu,. 7e -annot as% ?2an sense*data e4ist
without $eing gi3en 8? 5or that is li%e as%ing ?2an hus$ands e4ist
without $eing ,arried8? 7e ,ust as% ?2an sensibilia e4ist without
$eing gi3en 8? and also ?2an a parti-ular sensibile $e at one ti,e a
sense*datu, and at another not 8? Unless we ha3e the word sensibile as
well as the word ?sense*datu,? su-h 6uestions are apt to entangle us in
tri3ial logi-al puBBles.
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 111
1t will $e seen that all sense*data are sensibilia. 1t is a ,etaph#si-al
6uestion whether all sensibilia are sense*data and an episte,ologi-al
6uestion whether there e4ist ,eans o5 in5erring sensibilia whi-h are not
data 5ro, those that are.
A 5ew preli,inar# re,ar%s to $e a,pli5ied as we pro-eed will
ser3e to elu-idate the use whi-h 1 propose to ,a%e o5 sensibilia.
1 regard sense*data as not ,ental and as $eing in 5a-t part o5 the
a-tual su$9e-t*,atter o5 ph#si-s. There are argu,ents shortl# to $e
e4a,ined 5or their su$9e-ti3it# $ut these argu,ents see, to ,e onl#
to pro3e physiological su$9e-ti3it# i.e. -ausal dependen-e on the sense*
organs ner3es and $rain. The appearan-e whi-h a thing presents to us
is -ausall# dependent upon these in e4a-tl# the sa,e wa# as it is
dependent upon inter3ening 5og or s,o%e or -oloured glass. Both
dependen-ies are -ontained in the state,ent that the appearan-e whi-h
a pie-e o5 ,atter presents when 3iewed 5ro, a gi3en pla-e is a 5un-tion
not onl# o5 the pie-e o5 ,atter $ut also o5 the inter3ening ,ediu,.
(The ter,s used in this state,entC?,atter? ?3iew 5ro, a gi3en pla-e?
?appearan-e? ?inter3ening ,ediu,?Cwill all $e de5ined in the -ourse o5
the present paper.) 7e ha3e not the ,eans o5 as-ertaining how things
appear 5ro, pla-es not surrounded $# $rain and ner3es and sense*
organs $e-ause we -annot lea3e the $od#@ $ut -ontinuit# ,a%es it not
unreasona$le to suppose that the# present some appearan-e at su-h
pla-es. An# su-h appearan-e would $e in-luded a,ong sensibilia. If
per impossibileCthere were a -o,plete hu,an $od# with no ,ind
inside it all those sensibilia would e4ist in relation to that $od# whi-h
would $e sense*data i5 there were a ,ind in the $od#. 7hat the ,ind
adds to sensibilia, in 5a-t is merely awareness: e3er#thing else is
ph#si-al or ph#siologi-al..
I!. SESE-DATA ARE PH"SICAL
Be5ore dis-ussing this 6uestion it will $e well to de5ine the sense in
whi-h the ter,s ?,ental? and ?ph#si-al? are to $e used. The word
?ph#si-al? in all preli,inar# dis-ussions is to $e understood as ,eaning
?what is dealt with $# ph#si-s?. 'h#si-s it is plain tells us so,ething
a$out so,e o5 the -onstituents o5 the a-tual world@ what these
-onstituents are ,a# $e dou$t5ul $ut it is the# that are to $e -alled
ph#si-al whate3er their nature ,a# pro3e to $e.
The de5inition o5 the ter, ?,ental? is ,ore di55i-ult and -an onl# $e
satis5a-toril# gi3en a5ter ,an# di55i-ult -ontro3ersies ha3e $een
dis-ussed and de-ided. =or present purposes there5ore 1 ,ust -on*
11: A/0T1210A A!> L<G12
tent ,#sel5 with assu,ing a dog,ati- answer to these -ontro3ersies. 1
shall -all a parti-ular ?,ental? when it is aware o5 so,ething and 1 shall
-all a 5a-t ?,ental? when it -ontains a ,ental parti-ular as a -onstituent.
1t will $e seen that the ,ental and the ph#si-al are not ne-essaril#
,utuall# e4-lusi3e although 1 %now o5 no reason to suppose that the#
o3erlap.
The dou$t as to the -orre-tness o5 our de5inition o5 the ?,ental? is o5
little i,portan-e in our present dis-ussion. =or what 1 a, -on-erned to
,aintain is that sense*data are ph#si-al and this $eing granted it is a
,atter o5 indi55eren-e in our present in6uir# whether or not the# are
also ,ental. Although 1 do not hold with Ala-h and "a,es and the
?new realists? that the di55eren-e $etween the ,ental and the ph#si-al is
merely one o5 arrange,ent #et what 1 ha3e to sa# in the present paper
is -o,pati$le with their do-trine and ,ight ha3e $een rea-hed 5ro,
their standpoint.
1n dis-ussions on sense*data two 6uestions are -o,,onl# -on*
5used na,el#:
(i) >o sensi$le o$9e-ts persist when we are not sensi$le o5 the, 8 in
other words do sensibilia whi-h are data at a -ertain ti,e so,eti,es
-ontinue to e4ist at ti,es when the# are not data 8 And (:) are sense*
data ,ental or ph#si-al 8
1 propose to assert that sense*data are ph#si-al while #et ,ain*
taining that the# pro$a$l# ne3er persist un-hanged a5ter -easing to $e
data. The 3iew that the# do not persist is o5ten thought 6uite
erroneousl# in ,# opinion to i,pl# that the# are ,ental@ and this has 1
$elie3e $een a potent sour-e o5 -on5usion in regard to our present
pro$le,. 15 there were as so,e ha3e held a logical impossibility in
sense*data persisting a5ter -easing to $e data that -ertainl# would tend
to show that the# were ,ental@ $ut i5 as 1 -ontend their non*
persisten-e is ,erel# a pro$a$le in5eren-e 5ro, e,piri-all# as-ertained
-ausal laws then it -arries no su-h i,pli-ation with it and we are 6uite
5ree to treat the, as part o5 the su$9e-t*,atter o5 ph#si-s.
Logi-all# a sense*datu, is an o$9e-t a parti-ular o5 whi-h the
su$9e-t is aware. 1t does not -ontain the su$9e-t as a part as 5or
e4a,ple $elie5s and 3olitions do. The e4isten-e o5 the sense*datu, is
there5ore not logi-all# dependent upon that o5 the su$9e-t@ 5or the onl#
wa# so 5ar as 1 %now in whi-h the e4isten-e o5 A -an $e logically
dependent upon the e4isten-e o5 is when is part o5 A. There is
there5ore no a priori reason wh# a parti-ular whi-h is a sense*datu,
should not persist a5ter it has -eased to $e a datu, nor
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 11+
wh# other si,ilar parti-ulars should not e4ist without e3er $eing data.
The 3iew that sense*data are ,ental is deri3ed no dou$t in part 5ro,
their ph#siologi-al su$9e-ti3it# $ut in part also 5ro, a 5ailure to
distinguish $etween sense*data and ?sensations?. B# a sensation 1 ,ean
the 5a-t .-onsisting in the su$9e-t?s awareness o5 the sense*datu,. Thus
a sensation is a -o,ple4 o5 whi-h the su$9e-t is a -onstituent and whi-h
there5ore is ,ental. The sense*datu, on the other hand stands o3er
against the su$9e-t as that e4ternal o$9e-t o5 whi-h in sensation the
su$9e-t is aware. 1t is true that the sense*datu, is in ,an# -ases in the
su$9e-t?s $od# $ut the su$9e-t?s $od# is as distin-t 5ro, the su$9e-t as
ta$les and -hairs are and is in 5a-t ,erel# a part o5 the ,aterial world.
0o soon there5ore as sense*data are -learl# distinguished 5ro,
sensations and as their su$9e-ti3it# is re-ogniBed to $e ph#siologi-al
not ph#si-al the -hie5 o$sta-les in the wa# o5 regarding the, as
ph#si-al are re,o3ed.
!. #Sensi$ilia# and #Thin%s#
But i5 ?sensi$ilia? are to $e re-ogniBed as the ulti,ate -onstituents o5
the ph#si-al world a long and di55i-ult 9ourne# is to $e per5or,ed
$e5ore we -an arri3e either at the ?thing? o5 -o,,on sense or at the
?,atter? o5 ph#si-s. The supposed i,possi$ilit# o5 -o,$ining the
di55erent sense*data whi-h are regarded as appearan-es o5 the sa,e
?thing? to di55erent people has ,ade it see, as though these ?sensi$ilia?
,ust $e regarded as ,ere su$9e-ti3e phantas,s. A gi3en ta$le will
present to one ,an a re-tangular appearan-e while to another it
appears to ha3e two a-ute angles and two o$tuse angles@ to one ,an it
appears $rown while to another towards who, it re5le-ts the light it
appears white and shin#. 1t is said not wholl# without plausi$ilit# that
these di55erent shapes and di55erent -olours -annot -oe4ist
si,ultaneousl# in the sa,e pla-e and -annot there5ore $oth $e
-onstituents o5 the ph#si-al world. This argu,ent 1 ,ust -on5ess
appeared to ,e until re-entl# to $e irre5uta$le. The -ontrar# opinion
has howe3er $een a$l# ,aintained $# >r T. '. !unn in an arti-le
entitled: ?Are 0e-ondar# Dualities 1ndependent o5 'er-eption Tl The
supposed i,possi$ilit# deri3es its apparent 5or-e 5ro, the phrase: ?in
the sa,e pla-e? and it is pre-isel# in this phrase that its E3ea%ness lies.
The -on-eption o5 spa-e is too o5ten treated in philosoph#Ce3en $#
those who on re5le-tion would not de5end su-h treat,entCas though it
were as gi3en si,ple and una,$iguous as Fant in his ps#-hologi-al
inno-en-e supposed.

Proc. Arist Soc. &19(9*191( pp. 191*:1).
11G A/0T1210A A!> L<G12
1t is the unper-ei3ed a,$iguit# o5 the word ?pla-e? whi-h as we
shall shortl# see has -aused the di55i-ulties to realists and gi3en an un*
deser3ed ad3antage to their opponents. Two ?pla-es? o5 di55erent %inds
are in3ol3ed in e3er# sense*datu, na,el# the pla-e at whi-h it
appears and the pla-e from whi-h it appears. These $elong to di55erent
spa-es although as we shall see it is possi$le with -ertain li,itations
to esta$lish a -orrelation $etween the,. 7hat we -all the di55erent
appearan-es o5 the sa,e thing to di55erent o$ser3ers are ea-h in a spa-e
pri3ate to the o$ser3er -on-erned. !o pla-e in the pri3ate world o5 one
o$ser3er is identi-al with a pla-e in the pri3ate world o5 another
o$ser3er. There is there5ore no 6uestion o5 -o,$ining the di55erent
appearan-es in the one pla-e@ and the 5a-t that the# -annot all e4ist in
one pla-e a55ords a--ordingl# no ground whate3er 5or 6uestioning their
ph#si-al realit#. The ?thing? o5 -o,,on sense ,a# in 5a-t $e identi5ied
with the whole -lass o5 its appearan-esCwhere howe3er we ,ust
in-lude a,ong appearan-es not onl# those whi-h are a-tual sense*data
$ut also those ?sensi$ilia? i5 an# whi-h on grounds o5 -ontinuit# and
rese,$lan-e are to $e regarded as $elonging to the sa,e s#ste, o5
appearan-es although there happen to $e no o$ser3ers to who, the#
are data.
An e4a,ple ,a# ,a%e this -learer. 0uppose there are a nu,$er o5
people in a roo, all seeing as the# sa# the sa,e ta$les and -hairs
walls and pi-tures. !o two o5 these people ha3e e4a-tl# the sa,e
sense*data #et there is su55i-ient si,ilarit# a,ong their data to ena$le
the, to group together -ertain o5 these data as appearan-es o5 one
?thing? to the se3eral spe-tators and others as appearan-es o5 another
?thing?. Besides the appearan-es whi-h a gi3en thing in the roo,
presents to the a-tual spe-tators there are we ,a# suppose other
appearan-es whi-h it would present to other possi$le spe-tators. 15 a
,an were to sit down $etween two others the appearan-e whi-h the
roo, would present to hi, would $e inter,ediate $etween the
appearan-es whi-h it presents to the two others: and although this
appearan-e would not e4ist as it is without the sense organs ner3es and
$rain o5 the newl# arri3ed spe-tator still it is not unnatural to suppose
that 5ro, the position whi-h he now o--upies some appearan-e o5 the
roo, e4isted $e5ore his arri3al. This supposition howe3er need ,erel#
$e noti-ed and not insisted upon.
0in-e the ?thing? -annot without inde5ensi$le partialit# $e identi5ied
with an# single one o5 its appearan-es it -a,e to $e thought o5 as
so,ething distin-t 5ro, all o5 the, and underl#ing the,. But $# the
prin-iple o5 <--a,?s raBor i5 the -lass o5 appearan-es will 5ul5il the
purposes 5or the sa%e o5 whi-h the thing was in3ented
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 11&
$# the prehistori- ,etaph#si-ians to who, -o,,on sense is dxtt,
e-ono,# de,ands that we should identi5# the thing with the -lass o5 its
appearan-es. 1t is not ne-essar# to deny a su$stan-e or su$stratu,
underl#ing these appearan-es@ it is ,erel# e4pedient to a$stain 5ro,
asserting this unne-essar# entit#. <ur pro-edure here is pre-isel#
analogous to that whi-h has swept awa# 5ro, the philosoph# o5
,athe,ati-s the useless ,enagerie o5 ,etaph#si-al ,onsters with
whi-h it used to $e in5ested.
!I. COSTR&CTIOS !ERS&S IFERECES
Be5ore pro-eeding to anal#se and e4plain the a,$iguities o5 the
word ?pla-e? a 5ew general re,ar%s on ,ethod are desira$le. The
supre,e ,a4i, in s-ienti5i- philosophiBing is this:
7here3er possi$le logi-al -onstru-tions are to $e su$stituted 5or
in5erred entities.
0o,e e4a,ples o5 the su$stitution o5 -onstru-tion 5or in5eren-e in
the real, o5 ,athe,ati-al philosoph# ,a# ser3e to elu-idate the uses
o5 this ,a4i,. Ta%e 5irst the -ase o5 irrationals. 1n old da#s irrationals
were in5erred as the supposed li,its o5 series o5 rationals whi-h had no
rational li,it@ $ut the o$9e-tion to this pro-edure was that it le5t the
e4isten-e o5 irrationals ,erel# optati3e and 5or this reason the stri-ter
,ethods o5 the present da# no longer tolerate su-h a de5inition. 7e
now de5ine an irrational nu,$er as a -ertain -lass o5 ratios thus
-onstru-ting it logi-all# $# ,eans o5 ratios instead o5 arri3ing at it $# a
dou$t5ul in5eren-e 5ro, the,. Ta%e again the -ase o5 -ardinal nu,$ers.
Two e6uall# nu,erous -olle-tions appear to ha3e so,ething in
-o,,on: this so,ething is supposed to $e their -ardinal nu,$er. But
so long as the -ardinal nu,$er is in5erred 5ro, the -olle-tions not
-onstru-ted in ter,s o5 the, its e4isten-e ,ust re,ain in dou$t unless
in 3irtue o5 a ,etaph#si-al postulate ad hoc. B# de5ining the -ardinal
nu,$er o5 a gi3en -olle-tion as the -lass o5 all e6uall# nu,erous
-olle-tions we a3oid the ne-essit# o5 this ,etaph#si-al postulate and
there$# re,o3e a needless ele,ent o5 dou$t 5ro, the philosoph# o5
arith,eti-. A si,ilar ,ethod as 1 ha3e shown elsewhere -an $e
applied to -lasses the,sel3es whi-h need not $e supposed to ha3e an#
,etaph#si-al realit# $ut -an $e regarded as s#,$oli-all# -onstru-ted
5i-tions.
The ,ethod $# whi-h the -onstru-tion pro-eeds is -losel# analo*
gous in these and all si,ilar -ases. Gi3en a set o5 propositions
no,inall# dealing with the supposed in5erred entities we o$ser3e
11H A/0T1210A A!> L<G12
the properties whi-h are re6uired o5 the supposed entities in order to
,a%e these propositions true. B# dint o5 a little logi-al ingenuit# we
then -onstru-t so,e logi-al 5un-tion o5 less h#potheti-al entities whi-h
has the re6uisite properties. This -onstru-ted 5un-tion we su$stitute 5or
the supposed in5erred entities and there$# o$tain a new and less
dou$t5ul interpretation o5 the $od# o5 propositions in 6uestion. This
,ethod so 5ruit5ul in the philosoph# o5 ,athe,ati-s will $e 5ound
e6uall# appli-a$le in the philosoph# o5 ph#si-s where 1 do not dou$t
it would ha3e $een applied long ago $ut 5or the 5a-t that all who ha3e
studied this su$9e-t hitherto ha3e $een -o,pletel# ignorant o5
,athe,ati-al logi-. 1 ,#sel5 -annot -lai, originalit# in the appli-ation
o5 this ,ethod to ph#si-s sin-e 1 owe the suggestion and the sti,ulus
5or its appli-ation entirel# to ,# 5riend and -olla$orator >r 7hitehead
who is engaged in appl#ing it to the ,ore ,athe,ati-al portions o5 the
region inter,ediate $etween sense*data and the points instants and
parti-les o5 ph#si-s. A -o,plete appli-ation o5 the ,ethod whi-h
su$stitutes -onstru-tions 5or in5eren-es would e4hi$it ,atter wholl# in
ter,s o5 sense*data and e3en we ,a# add o5 the sense*data o5 a single
person sin-e the sense*data o5 others -annot $e %nown without so,e
ele,ent o5 in5eren-e. This howe3er ,ust re,ain 5or the present an
ideal to $e approa-hed as nearl# as possi$le $ut to $e rea-hed i5 at all
onl# a5ter a long preli,inar# la$our o5 whi-h as #et we -an onl# see the
3er# $eginning. The in5eren-es whi-h are una3oida$le -an howe3er $e
su$9e-ted to -ertain guiding prin-iples. 1n the 5irst pla-e the# should
alwa#s $e ,ade per5e-tl# e4pli-it and should $e 5or,ulated in the
,ost general ,anner possi$le. 1n the se-ond pla-e the in5erred entities
should whene3er this -an $e done $e si,ilar to those whose e4isten-e
is gi3en rather than li%e the Fantian !ing an sich, so,ething wholl#
re,ote 5ro, the data whi-h no,inall# support the in5eren-e. The
in5erred entities whi-h 1 shall allow ,#sel5 are o5 two %inds: (a) the
sense*data o5 other people in 5a3our o5 whi-h there is the e3iden-e o5
testi,on# resting ulti,atel# upon the analogi-al argu,ent in 5a3our o5
,inds other than ,# own@ "b# the ?sensi$ilia? whi-h would appear 5ro,
pla-es where there happen to $e no ,inds and whi-h 1 suppose to $e
real although the# are no one?s data. <5 these two -lasses o5 in5erred
entities the 5irst will pro$a$l# $e allowed to pass un-hallenged. 1t
would gi3e ,e the greatest satis5a-tion to $e a$le to dispense with it
and thus esta$lish ph#si-s upon a solipsisti- $asis@ $ut thoseCand 1
5ear the# are the ,a9orit#Cin who, the hu,an a55e-tions are stronger
than the desire 5or logi-al e-ono,# will no dou$t not
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 117
share ,# desire to render solipsis, s-ienti5i-all# satis5a-tor#. The
se-ond -lass o5 in5erred entities raises ,u-h ,ore serious 6uestions. 1t
,a# $e thought ,onstrous to ,aintain that a thing -an present an#
appearan-e at all in a pla-e where no sense organs and ner3ous
stru-ture e4ist through whi-h it -ould appear. 1 do not ,#sel5 5eel the
,onstrosit#@ ne3ertheless 1 should regard these supposed appearan-es
onl# in the light o5 a h#potheti-al s-a55olding to $e used while the
edi5i-e o5 ph#si-s is $eing raised though possi$l# -apa$le o5 $eing
re,o3ed as soon as the edi5i-e is -o,pleted. These ?sensi$ilia? whi-h
are not data to an#one are there5ore to $e ta%en rather as an illustrati3e
h#pothesis and as an aid in preli,inar# state,ent than as a dog,ati-
part o5 the philosoph# o5 ph#si-s in its 5inal 5or,.
!II. PRI!ATE SPACE AD THE SPACE OF PERSPECTI!ES
7e ha3e now to e4plain the a,$iguit# in the word ?pla-e? and how
it -o,es that two pla-es o5 di55erent sorts are asso-iated with e3er#
sense*datu, na,el# the pla-e at whi-h it is and the pla-e from whi-h
it is per-ei3ed. The theor# to $e ad3o-ated is -losel# analogous to
Lei$niB?s ,onadolog# 5ro, whi-h it di55ers -hie5l# in $eing less
s,ooth and tid#.
The 5irst 5a-t to noti-e is that so 5ar as -an $e dis-o3ered no
sensi$ile is e3er a datu, to two people at on-e. The things seen $# two
di55erent people are o5ten -losel# si,ilar so si,ilar that the sa,e
words -an $e used to denote the, without whi-h -o,,uni-ation with
others -on-erning sensi$le o$9e-ts would $e i,possi$le. But in spite o5
this si,ilarit# it would see, that so,e di55eren-e alwa#s arises 5ro,
di55eren-e in the point o5 3iew. Thus ea-h person so 5ar as his sense*
data are -on-erned li3es in a pri3ate world. This pri3ate world -ontains
its own spa-e or rather spa-es 5or it would see, that onl# e4perien-e
tea-hes us to -orrelate the spa-e o5 sight with the spa-e o5 tou-h and
with the 3arious other spa-es o5 other senses. This ,ultipli-it# o5
pri3ate spa-es howe3er though interesting to the ps#-hologist is o5 no
great i,portan-e in regard to our present pro$le, sin-e a ,erel#
solipsisti- e4perien-e ena$les us to -orrelate the, into the one pri3ate
spa-e whi-h e,$ra-es all our own sense*data. The pla-e at whi-h a
sense*datu, is is a pla-e in pri3ate spa-e. This pla-e there5ore is
di55erent 5ro, an# pla-e in the pri3ate spa-e o5 another per-ipient. =or
i5 we assu,e as logi-al e-ono,# de,ands that all position is relati3e
a pla-e is onl# de5ina$le $# the things in or around it and there5ore the
sa,e pla-e -annot o--ur in
11) A/0T1210A A!> L<G12
two pri3ate worlds whi-h ha3e no -o,,on -onstituent. The 6uestion
there5ore o5 -o,$ining what we -all di55erent appearan-es o5 the sa,e
thing in the sa,e pla-e does not arise and the 5a-t that a gi3en o$9e-t
appears to di55erent spe-tators to ha3e di55erent shapes and -olours
a55ords no argu,ent against the ph#si-al realit# o5 all these shapes and
-olours.
1n addition to the pri3ate spa-es $elonging to the pri3ate worlds o5
di55erent per-ipients there is howe3er another spa-e in whi-h one
whole pri3ate world -ounts as a point or at least as a spatial unit. This
,ight $e des-ri$ed as the spa-e o5 points o5 3iew sin-e ea-h pri3ate
world ,a# $e regarded as the appearan-e whi-h the uni3erse presents
5ro, a -ertain point o5 3iew. 1 pre5er howe3er to spea% o5 it as the
spa-e o5 perspecti$es, in order to o$3iate the suggestion that a pri3ate
world is onl# real when so,eone 3iews it. And 5or the sa,e reason
when 1 wish to spea% o5 a pri3ate world without assu,ing a per-ipient
1 shall -all it a ?perspe-ti3e?.
7e ha3e now to e4plain how the di55erent perspe-ti3es are ordered
in one spa-e. This is e55e-ted $# ,eans o5 the -orrelated ?sensi$ilia?
whi-h are regarded as the appearan-es in di55erent perspe-ti3es o5 one
and the sa,e thing. B# ,o3ing and $# testi,on# we dis-o3er that two
di55erent perspe-ti3es though the# -annot $oth -ontain the sa,e
?sensi$ilia? ,a# ne3ertheless -ontain 3er# si,ilar ones@ and the spatial
order o5 a -ertain group o5 ?sensi$ilia? in a pri3ate spa-e o5 one
perspe-ti3e is 5ound to $e identi-al with or 3er# si,ilar to the spatial
order o5 the -orrelated ?sensi$ilia? in the pri3ate spa-e o5 another
perspe-ti3e. 1n this wa# one ?sensi$ile? in one perspe-ti3e is -orrelated
with one ?sensi$ile? in another. 0u-h -orrelated ?sensi$ilia? will $e -alled
?appearan-es o5 one thing?. 1n Lei$niB?s ,onadolog# sin-e ea-h ,onad
,irrored the whole uni3erse there was in ea-h perspe-ti3e a ?sensi$ile?
whi-h was an appearan-e o5 ea-h thing. 1n our s#ste, o5 perspe-ti3es
we ,a%e no su-h assu,ption o5 -o,pleteness. A gi3en thing will ha3e
appearan-es in so,e perspe-ti3es $ut presu,a$l# not in -ertain others.
The ?thing? $eing de5ined as the -lass o5 its appearan-es i5 % is the -lass
o5 perspe-ti3es in whi-h a -ertain thing & appears then ' is a ,e,$er
o5 the ,ultipli-ati3e -lass o5 I % $eing a -lass o5 ,utuall# e4-lusi3e
-lasses o5 ?sensi$ilia?. And si,ilarl# a perspe-ti3e is a ,e,$er o5 the
,ultipli-ati3e -lass o5 the things whi-h appear in it.
The arrange,ent o5 perspe-ti3es in a spa-e is e55e-ted $# ,eans o5
the di55eren-es $etween the appearan-es o5 a gi3en thing in the 3arious
perspe-ti3es. 0uppose sa# that a -ertain penn# appears in a nu,$er o5
di55erent perspe-ti3es@ in so,e it loo%s larger and in so,e
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 119
s,aller in so,e it loo%s -ir-ular in others it presents the appearan-e o5
an ellipse o5 3ar#ing e--entri-it#. 7e ,a# -olle-t together all those
perspe-ti3es in whi-h the appearan-e o5 the penn# is -ir-ular. These we
will pla-e on one straight line ordering the, in a series $# the
3ariations in the apparent siBe o5 the penn#. Those perspe-ti3es in
whi-h the penn# appears as a straight line o5 a -ertain thi-%ness will
si,ilarl# $e pla-ed upon a plane (though in this -ase there will $e ,an#
di55erent perspe-ti3es in whi-h the penn# is o5 the sa,e siBe@ when one
arrange,ent is -o,pleted these will 5or, a -ir-le -on-entri- with the
penn#) and ordered as $e5ore $# the apparent siBe o5 the penn#. B#
su-h ,eans all those perspe-ti3es in whi-h the penn# presents a 3isual
appearan-e -an $e arranged in a three*di,ensional spatial order.
;4perien-e shows that the sa,e spatial order o5 perspe-ti3es would
ha3e resulted i5 instead o5 the penn# we had -hosen an# other thing
whi-h appeared in all the perspe-ti3es in 6uestion or an# other ,ethod
o5 utiliBing the di55eren-es $etween the appearan-es o5 the sa,e things
in di55erent perspe-ti3es. 1t is this e,piri-al 5a-t whi-h has ,ade it
possi$le to -onstru-t the one all*e,$ra-ing spa-e o5 ph#si-s.
The spa-e whose -onstru-tion has 9ust $een e4plained and whose
ele,ents are whole perspe-ti3es will $e -alled ?perspe-ti3e*spa-e?.
!III. THE PLACI' OF #THI'S# AD #SESIBILIA# I
PERSPECTI!E SPACE
The world whi-h we ha3e so 5ar -onstru-ted is a world o5 si4
di,ensions sin-e it is a three*di,ensional series o5 perspe-ti3es ea-h
o5 whi-h is itsel5 three*di,ensional. 7e ha3e now to e4plain the
-orrelation $etween the perspe-ti3e spa-e and the 3arious pri3ate
spa-es -ontained within the 3arious perspe-ti3es se3erall#. 1t is $#
,eans o5 this -orrelation that the one three*di,ensional spa-e o5
ph#si-s is -onstru-ted@ and it is $e-ause o5 the un-ons-ious
per5or,an-e o5 this -orrelation that the distin-tion $etween perspe-ti3e
spa-e and the per-ipient?s pri3ate spa-e has $een $lurred with
disastrous results 5or the philosoph# o5 ph#si-s. Let us re3ert to our
penn#: the perspe-ti3es in whi-h the penn# appears larger are regarded
as $eing nearer to the penn# than those in whi-h it appears s,aller $ut
as 5ar as e4perien-e goes the apparent siBe o5 the penn# will not grow
$e#ond a -ertain li,it na,el# that where (as we sa#) the penn# is so
near the e#e that i5 it were an# nearer it -ould not $e seen. B# tou-h we
,a# prolong the series until the penn# tou-hes the e#e $ut no 5urther.
15 we ha3e $een tra3elling
1:< A/0T1210A A!> L<G12
along a line o5 perspe-ti3es in the pre3iousl# de5ined sense we ,a#
howe3er $# i,agining the penn# re,o3ed prolong the line o5
perspe-ti3es $# ,eans sa# o5 another penn#@ and the sa,e ,a# $e
done with an# other line o5 perspe-ti3es de5ined $# ,eans o5 the
penn# All these lines ,eet in a -ertain pla-e that is in a -ertain
perspe-ti3e. This perspe-ti3e will $e denned as ?the pla-e where the
penn# is?.
1t is now e3ident in what sense two pla-es in -onstru-ted ph#si-al
spa-e are asso-iated with a gi3en ?sensi$ile?. There is 5irst the pla-e
whi-h is the perspe-ti3e o5 whi-h the ?sensi$ile? is a ,e,$er. This is
the pla-e from whi-h the ?sensi$ile? appears. 0e-ondl# there is the pla-e
where the other thing is o5 whi-h the ?sensi$ile? is a ,e,$er in other
words an appearan-e@ this is the pla-e at whi-h the ?sensi$ile? appears.
The ?sensi$ile
&
whi-h is a ,e,$er o5 one perspe-ti3e is -orrelated with
another perspe-ti3e na,el# that whi-h is the pla-e where the thing is
o5 whi-h the ?sensi$ile? is an appearan-e. To the ps#-hologist the ?pla-e
5ro, whi-h? is the ,ore interesting and the ?sensi$ile? a--ordingl#
appears to hi, su$9e-ti3e and where the per-ipient is. To the ph#si-ist
the ?pla-e at whi-h? is the ,ore interesting and the ?sensi$ile?
a--ordingl# appears to hi, ph#si-al and e4ternal. The -auses li,its
and partial 9usti5i-ation o5 ea-h o5 these two apparentl# in-o,pati$le
3iews are e3ident 5ro, the a$o3e dupli-it# o5 pla-es asso-iated with a
gi3en ?sensi$ile?.
7e ha3e seen that we -an assign to a ph#si-al thing a pla-e in the
perspe-ti3e spa-e. 1n this wa# di55erent parts o5 our $od# a-6uire
positions in perspe-ti3e spa-e and there5ore there is a ,eaning
(whether true or 5alse need not ,u-h -on-ern us) in sa#ing that the
perspe-ti3e to whi-h our sense*data $elong is inside our head. 0in-e
our ,ind is -orrelated with the perspe-ti3e to whi-h our sense*data
$elong we ,a# regard this perspe-ti3e as $eing the position o5 our
,ind in perspe-ti3e spa-e. 15 there5ore this perspe-ti3e is in the a$o3e
denned sense inside our head there is a good ,eaning 5or the
state,ent that the ,ind is in the head. 7e -an now sa# o5 the 3arious
appearan-es o5 a gi3en thing that so,e o5 the, are nearer to the thing
than others@ those are nearer whi-h $elong to perspe-ti3es that are
nearer to ?the pla-e where the thing is?. 7e -an thus 5ind a ,eaning
true or 5alse 5or the state,ent that ,ore is to $e learnt a$out a thing $#
e4a,ining it -lose to than $# 3iewing it 5ro, a distan-e. 7e -an also
5ind a ,eaning 5or the phrase ?the things whi-h inter3ene $etween the
su$9e-t and a thing o5 whi-h an appearan-e is a datu, to hi,?. <ne
reason o5ten alleged 5or the su$9e-ti3it# o5 sense*data is that the
appearan-e o5 a thing
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 1:1
,a# -hange when we 5ind it hard to suppose that the thing itsel5 has
-hangedC5or e4a,ple when the -hange is due to our shutting our
e#es or to our s-rewing the, up so as to ,a%e the thing loo% dou$le. 15
the thing is de5ined as the -lass o5 its appearan-es (whi-h is the
de5inition adopted a$o3e) there is o5 -ourse ne-essaril# some -hange in
the thing whene3er an# one o5 its appearan-es -hanges. !e3ertheless
there is a 3er# i,portant distin-tion $etween two di55erent wa#s in
whi-h the appearan-es ,a# -hange. 15 a5ter loo%ing at a thing 1 shut ,#
e#es the appearan-e o5 ,# e#es -hanges in e3er# perspe-ti3e in whi-h
there is su-h an appearan-e whereas ,ost o5 the appearan-es o5 the
thing will re,ain un-hanged. 7e ,a# sa# as a ,atter o5 de5inition
that a thing -hanges when howe3er near to the thing an appearan-e o5
it ,a# $e there are -hanges in appearan-es as near as or still nearer to
the thing. <n the other hand we shall sa# that the -hange is in so,e
other thing i5 all appearan-es o5 the thing whi-h are at not ,ore than a
-ertain distan-e 5ro, the thing re,ain un-hanged while onl#
-o,parati3el# distant appearan-es o5 the thing are altered. =ro, this
-onsideration we are naturall# led to the -onsideration o5 matter, whi-h
,ust $e our ne4t topi-.
I(. THE DEFIITIO OF MATTER
7e de5ined the ?ph#si-al thing? as the -lass o5 its appearan-es $ut
this -an hardl# $e ta%en as a de5inition o5 ,atter. 7e want to $e a$le to
e4press the 5a-t that the appearan-e o5 a thing in a gi3en perspe-ti3e is
-ausall# a55e-ted $# the ,atter $etween the thing and the perspe-ti3e.
7e ha3e 5ound a ,eaning 5or ?$etween a thing and a perspe-ti3e?. But
we want ,atter to $e so,ething other than the whole -lass o5
appearan-es o5 a thing in order to state the in5luen-e o5 ,atter on
appearan-es.
7e -o,,onl# assu,e that the in5or,ation we get a$out a thing is
,ore a--urate when the thing is nearer. =ar o55 we see it is a ,an@ then
we see it is "ones@ then we see he is s,iling. 2o,plete a--ura-# would
onl# $e attaina$le as a li,it: i5 the appearan-es o5 "ones as we approa-h
hi, tend towards a li,it that li,it ,a# $e ta%en to $e what "ones
reall# is. 1t is o$3ious that 5ro, the point o5 3iew o5 ph#si-s the
appearan-es o5 a thing -lose to ?-ount? ,ore than the appearan-es 5ar
o55. 7e ,a# there5ore set up the 5ollowing tentati3e de5inition:
The matter o5 a gi3en thing is the li,it o5 its appearan-es as their
distan-e 5ro, the thing di,inishes.
1:: A/0T1210A A!> L<G12
1t see,s pro$a$le that there is so,ething in this de5inition $ut it is
not 6uite satis5a-tor# $e-ause e,piri-all# there is no su-h li,it to $e
o$tained 5ro, sense*data. The de5inition will ha3e to $e e%ed out $#
-onstru-tions and de5initions. But pro$a$l# it suggests the right
dire-tion in whi-h to loo%.
7e are now in a position to understand in outline the re3erse
9ourne# 5ro, ,atter to sense*data whi-h is per5or,ed $# ph#si-s. The
appearan-e o5 a thing in a gi3en perspe-ti3e is a 5un-tion o5 the ,atter
-o,posing the thing and o5 the inter3ening ,atter. The appearan-e o5 a
thing is altered $# inter3ening s,o%e or ,ist $# $lue spe-ta-les or $#
alterations in the sense*organs or ner3es o5 the per-ipient (whi-h also
,ust $e re-%oned as part o5 the inter3ening ,ediu,). The nearer we
approa-h to the thing the less its appearan-e is a55e-ted $# the
inter3ening ,atter. As we tra3el 5urther and 5urther 5ro, the thing its
appearan-es di3erge ,ore and ,ore 5ro, their initial -hara-ter@ and the
-ausal laws o5 their di3ergen-e are to $e stated in ter,s o5 the ,atter
whi-h lies $etween the, and the thing. 0in-e the appearan-es at 3er#
s,all distan-es are less a55e-ted $# -auses other than the thing itsel5
we -o,e to thin% that the li,it towards whi-h these appearan-es tend
as the distan-e di,inishes is what the thing ?reall# is? as opposed to
what it ,erel# see,s to $e. This together with its ne-essit# 5or the
state,ent o5 -ausal laws see,s to $e the sour-e o5 the entirel#
erroneous 5eeling that ,atter is ,ore ?real? than sense*data.
2onsider 5or e4a,ple the in5inite di3isi$ilit# o5 ,atter. 1n loo%ing at
a gi3en thing and approa-hing it one sense*datu, will $e-o,e se3eral
and ea-h o5 these will again di3ide. Thus one appearan-e ,a# represent
many things and to this pro-ess there see,s no end. .en-e in the li,it
when we approa-h inde5initel# near to the thing there will $e an
inde5inite nu,$er o5 units o5 ,atter -orresponding to what at a 5inite
distan-e is onl# one appearan-e. This is how in5inite di3isi$ilit# arises.
The whole -ausal e55i-a-# o5 a thing resides in its ,atter. This is in
so,e sense an e,piri-al 5a-t $ut it would $e hard to state it pre-isel#
$e-ause ?-ausal e55i-a-#? is di55i-ult to de5ine.
7hat -an $e %nown e,piri-all# a$out the ,atter o5 a thing is onl#
appro4i,ate $e-ause we -annot get to %now the appearan-es o5 the
thing 5ro, 3er# s,all distan-es and -annot a--uratel# in5er the li,it o5
these appearan-es. But it is in5erred approximately $# ,eans o5 the
appearan-es we -an o$ser3e. 1t then turns out that these appearan-es
-an $e e4hi$ited $# ph#si-s as a 5un-tion o5 the ,atter in our
i,,ediate neigh$ourhood@ e.g. the 3isual appearan-e o5 a
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 1:+
distant o$9e-t is a 5un-tion o5 the light*wa3es that rea-h the e#es. This
leads to -on5usions o5 thought $ut o55ers no real di55i-ult#.
<ne appearan-e o5 a 3isi$le o$9e-t 5or e4a,ple is not su55i-ient to
deter,ine its other si,ultaneous appearan-es although it goes a -ertain
distan-e towards deter,ining the,. The deter,ination o5 the hidden
stru-ture o5 a thing so 5ar as it is possi$le at all -an onl# $e e55e-ted $#
,eans o5 ela$orate d#na,i-al in5eren-es.
(. TIME
1
1t see,s that the one all*e,$ra-ing ti,e is a -onstru-tion li%e the
one all*e,$ra-ing spa-e. 'h#si-s itsel5 has $e-o,e -ons-ious o5 this
5a-t through the dis-ussions -onne-ted with relati3it#.
Between two perspe-ti3es whi-h $oth $elong to one person?s
e4perien-e there will $e a dire-t ti,e*relation o5 $e5ore and a5ter. This
suggests a wa# o5 di3iding histor# in the sa,e sort o5 wa# as it is
di3ided $# di55erent e4perien-es $ut without introdu-ing e4perien-e or
an#thing ,ental: we ,a# de5ine a ?$iograph#? as e3er#thing that is
(dire-tl#) earlier or later than or si,ultaneous with a gi3en ?sensi$ile?.
This will gi3e a series o5 perspe-ti3es whi-h might all 5or, parts o5
one person?s e4perien-e though it is not ne-essar# that all or an# o5
the, should a-tuall# do so. B# this ,eans the histor# o5 the world is
di3ided into a nu,$er o5 ,utuall# e4-lusi3e $iographies.
7e ha3e now to -orrelate the ti,es in the di55erent $iographies. The
natural thing would $e to sa# that the appearan-es o5 a gi3en
(,o,entar#) thing in two di55erent perspe-ti3es $elonging to di55erent
$iographies are to $e ta%en as si,ultaneous@ $ut this is not -on3enient.
0uppose A shouts to , and replies as soon as he hears A(s shout.
Then $etween A(s hearing o5 his own shout and his hearing o5 (s there
is an inter3al@ thus i5 we ,ade A(s and (s hearing o5 the sa,e shout
e4a-tl# si,ultaneous with ea-h other we should ha3e e3ents e4a-tl#
si,ultaneous with a gi3en e3ent $ut not with ea-h other. To o$3iate
this we assu,e a ?3elo-it# o5 sound?. That is we assu,e that the ti,e
when hears A(s shout is hal5*wa# $etween the ti,e when A hears his
own shout and the ti,e when he hears (s. 1n this wa# the -orrelation is
e55e-ted.
7hat has $een said a$out sound applies o5 -ourse e6uall# to light.
The general prin-iple is that the appearan-es in di55erent perspe-*
1
<n this su$9e-t -o,pare A )heory of )ime and *pace, $# Ar A.
A. Ro$$ (2a,$. Uni3. 'ress) whi-h 5irst suggested to ,e the 3iews
ad3o-ated here though 1 ha3e 5or present purposes o,itted what is
,ost interesting and no3el in his theor#. Ar Ro$$ has gi3en a s%et-h o5
his theor# in a pa,phlet with the sa,e title (.e55er and 0ons
2a,$ridge 191+).
1:G A/0T1210A A!> L<G12
ti3es whi-h are to $e grouped together as -onstituting what a -ertain
thing is at a -ertain ,o,ent are not to $e all regarded as $eing at that
,o,ent. <n the -ontrar# the# spread outward 5ro, the thing with
3arious 3elo-ities a--ording to the nature o5 the appearan-es. 0in-e no
direct ,eans e4ist o5 -orrelating the ti,e in one $iograph# with the
ti,e in another this te,poral grouping o5 the appearan-es $elonging to
a gi3en thing at a gi3en ,o,ent is in part -on3entional. 1ts ,oti3e is
partl# to se-ure the 3eri5i-ation o5 su-h ,a4i,s as that e3ents whi-h
are e4a-tl# si,ultaneous with the sa,e e3ent are e4a-tl# si,ultaneous
with one another partl# to se-ure -on3enien-e in the 5or,ulation o5
-ausal laws.
(I. THE PERSISTECE OF THI'S AD MATTER
Apart 5ro, an# o5 the 5lu-tuating h#potheses o5 ph#si-s three ,ain
pro$le,s arise in -onne-ting the world o5 ph#si-s with the world o5
sense na,el#:
1. the -onstru-tion o5 a single spa-e@
:. the -onstru-tion o5 a single ti,e@
+. the -onstru-tion o5 per,anent things or ,atter.
7e ha3e alread# -onsidered the 5irst and se-ond o5 these pro$le,s@
it re,ains to -onsider the third.
7e ha3e seen how -orrelated appearan-es in di55erent perspe-ti3es
are -o,$ined to 5or, one ?thing? at one ,o,ent in the all*e,$ra-ing
ti,e o5 ph#si-s. 7e ha3e now to -onsider how appearan-es at di55erent
ti,es are -o,$ined as $elonging to one ?thing? and how we arri3e at
the persistent ?,atter? o5 ph#si-s. The assu,ption o5 per,anent
su$stan-e whi-h te-hni-all# underlies the pro-edure o5 ph#si-s -annot
o5 -ourse $e regarded as ,etaph#si-all# legiti,ate: 9ust as the one
thing si,ultaneousl# seen $# ,an# people is a -onstru-tion so the one
thing seen at di55erent ti,es $# the sa,e or di55erent people ,ust $e a
-onstru-tion $eing in 5a-t nothing $ut a -ertain grouping o5 -ertain
?sensi$ilia?.
7e ha3e seen that the ,o,entar# state o5 a ?thing? is an asse,$lage
o5 ?sensi$ilia? in di55erent perspe-ti3es not all si,ultaneous in the one
-onstru-ted ti,e $ut spreading out 5ro, ?the pla-e where the thing is?
with 3elo-ities depending upon the nature o5 the ?sensi$ilia?. The ti,e
at whi-h the ?thing? is in this state is the lower li,it o5 the ti,es at
whi-h these appearan-es o--ur. 7e ha3e now to -onsider what leads us
to spea% o5 another set o5 appearan-es as $elonging to the sa,e ?thing?
at a di55erent ti,e.
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 1:&
=or this purpose we ,a# at least to $egin with -on5ine oursel3es
within a single $iograph#. 15 we -an alwa#s sa# when two ?sensi$ilia? in
a gi3en $iograph# are appearan-es o5 one thing then sin-e we ha3e
seen how to -onne-t ?sensi$ilia? in di55erent $iographies as appearan-es
o5 the sa,e ,o,entar# state o5 a thing we shall ha3e all that is
ne-essar# 5or the -o,plete -onstru-tion o5 the histor# o5 a thing.
1t is to $e o$ser3ed to $egin with that the identit# o5 a thing 5or
-o,,on sense is not alwa#s -orrelated with the identit# o5 ,atter 5or
ph#si-s. A hu,an $od# is one persisting thing 5or -o,,on sense $ut
5or ph#si-s its ,atter is -onstantl# -hanging. 7e ,a# sa# $roadl# that
the -o,,on*sense -on-eption is $ased upon -ontinuit# in appearan-es
at the ordinar# distan-es o5 sense*data while the ph#si-al -on-eption is
$ased upon the -ontinuit# o5 appearan-es at 3er# s,all distan-es 5ro,
the thing. 1t is pro$a$le that the -o,,on*sense -on-eption is not
-apa$le o5 -o,plete pre-ision. Let us there5ore -on-entrate our
attention upon the -on-eption o5 the persisten-e o5 ,atter in ph#si-s.
The 5irst -hara-teristi- o5 two appearan-es o5 the sa,e pie-e o5
,atter at di55erent ti,es is continuity. The two appearan-es ,ust $e
-onne-ted $# a series o5 inter,ediaries whi-h i5 ti,e and spa-e 5or,
-o,pa-t series ,ust the,sel3es 5or, a -o,pa-t series. The -olour o5
the lea3es is di55erent in autu,n 5ro, what it is in su,,er@ $ut we
$elie3e that the -hange o--urs graduall# and that i5 the -olours are
di55erent at two gi3en ti,es there are inter,ediate ti,es at whi-h the
-olours are inter,ediate $etween those at the gi3en ti,es.
But there are two -onsiderations that are i,portant as regards
-ontinuit#.
=irst it is largel# h#potheti-al. 7e do not o$ser3e an# one thing
-ontinuousl# and it is ,erel# a h#pothesis to assu,e that while we are
not o$ser3ing it it passes through -onditions inter,ediate $etween
those in whi-h it is per-ei3ed. >uring uninterrupted o$ser3ation it is
true -ontinuit# is nearl# 3eri5ied@ $ut e3en here when ,otions are 3er#
rapid as in the -ase o5 e4plosions the -ontinuit# is not a-tuall# -apa$le
o5 dire-t 3eri5i-ation. Thus we -an onl# sa# that the sense*data are
5ound to permit a h#potheti-al -o,ple,ent o5 ?sensi$ilia? su-h as will
preser3e -ontinuit# and that there5ore there may $e su-h a
-o,ple,ent. 0in-e howe3er we ha3e alread# ,ade su-h use o5
h#potheti-al ?sensi$ilia? we will let this point pass and ad,it su-h
?sensi$ilia? as are re6uired to preser3e -ontinuit#.
1:H A/0T1210A A!> L<G12
0e-ondl# -ontinuit# is not a su55i-ient -riterion o5 ,aterial identit#.
1t is true that in ,an# -ases su-h as ro-%s ,ountains ta$les -hairs
et-. where the appearan-es -hange slowl# -ontinuit# is su55i-ient $ut
in other -ases su-h as the parts o5 an appro4i,atel# ho,ogeneous
5luid it 5ails us utterl#. 7e -an tra3el $# sensi$l# -ontinuous
gradations 5ro, an# one drop o5 the sea at an# one ti,e to an# other
drop at an# other ti,e. 7e in5er the ,otions o5 sea*water 5ro, the
e55e-ts o5 the -urrent $ut the# -annot $e in5erred 5ro, dire-t sensi$le
o$ser3ation together with the assu,ption o5 -ontinuit#.
The -hara-teristi- re6uired in addition to -ontinuit# is -on5or,it#
with the laws o5 d#na,i-s. 0tarting 5ro, what -o,,on sense regards
as persistent things and ,a%ing onl# su-h ,odi5i-ations as 5ro, ti,e
to ti,e see, reasona$le we arri3e at asse,$lages o5 ?sensi$ilia? whi-h
are 5ound to o$e# -ertain si,ple laws na,el# those o5 d#na,i-s. B#
regarding ?sensi$ilia? at di55erent ti,es as $elonging to the sa,e pie-e
o5 ,atter we are a$le to de5ine motion, whi-h presupposes the
assu,ption or -onstru-tion o5 so,ething persisting throughout the ti,e
o5 ,otion. The ,otions whi-h are regarded as o--urring during a
period in whi-h all the ?sensi$ilia? and the ti,es o5 their appearan-e are
gi3en will $e di55erent a--ording to the ,anner in whi-h we -o,$ine
?sensi$ilia? at di55erent ti,es as $elonging to the sa,e pie-e o5 ,atter.
Thus e3en when the whole histor# o5 the world is gi3en in e3er#
parti-ular the 6uestion what ,otions ta%e pla-e is still to a -ertain
e4tent ar$itrar# e3en a5ter the assu,ption o5 -ontinuit#. ;4perien-e
shows that it is possi$le to deter,ine ,otions in su-h a wa# as to
satis5# the laws o5 d#na,i-s and that this deter,ination roughl# and
on the whole is 5airl# in agree,ent with the -o,,on*sense opinions
a$out persistent things. This deter,ination there5ore is adopted and
leads to a -riterion $# whi-h we -an deter,ine so,eti,es pra-ti-all#
so,eti,es onl# theoreti-all# whether two appearan-es at di55erent
ti,es are to $e regarded as $elonging to the sa,e pie-e o5 ,atter. The
persisten-e o5 all ,atter throughout all ti,e -an 1 i,agine $e se-ured
$# de5inition.
To re-o,,end this -on-lusion we ,ust -onsider what it is that is
pro3ed $# the e,piri-al su--ess o5 ph#si-s. 7hat is pro3ed is that its
h#potheses though un3eri5ia$le where the# go $e#ond sense*data are
at no point in -ontradi-tion with sense*data $ut on the -ontrar# are
ideall# su-h as to render all sense*data -al-ula$le when a su55i-ient
-olle-tion o5 ?sensi$ilia? is gi3en. !ow ph#si-s has 5ound it e,piri-all#
possi$le to -olle-t sense*data into series ea-h series $eing re*
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 1:7
garded as $elonging to one ?thing? and $eha3ing with regard to the the
laws o5 ph#si-s in a wa# in whi-h series not $elonging to one thing
would in general not $eha3e. 15 it is to $e una,$iguous whether two
appearan-es $elong to the sa,e thing or not there ,ust $e onl# one
wa# o5 grouping appearan-es so that the resulting things o$e# the laws
o5 ph#si-s. 1t would $e 3er# di55i-ult to pro3e that this is the -ase $ut
5or our present purposes we ,a# let this point pass and assu,e that
there is onl# one wa#. Thus we ,a# la# down the 5ollowing de5inition:
+hysical things are those series of appearances whose matter obeys the
laws of physics. That su-h series e4ist is an e,piri-al 5a-t whi-h
-onstitutes the 3eri5ia$ilit# o5 ph#si-s.
(II. ILL&SIOS) HALL&CIATIOS) AD DREAMS
1t re,ains to as% how in our s#ste, we are to 5ind a pla-e 5or
sense*data whi-h apparentl# 5ail to ha3e the usual -onne-tion with the
world o5 ph#si-s. 0u-h sense*data are o5 3arious %inds re6uiring
so,ewhat di55erent treat,ent. But all are o5 the sort that would $e
-alled ?unreal? and there5ore $e5ore e,$ar%ing upon the dis-ussion
-ertain logi-al re,ar%s ,ust $e ,ade upon the -on-eptions o5 realit#
and unrealit#.
Ar A. 7ol5
1
sa#s:
?The -on-eption o5 ,ind as a s#ste, o5 transparent a-ti3ities is 1
thin% also untena$le $e-ause o5 its 5ailure to a--ount 5or the 3er#
possi$ilit# o5 drea,s and hallu-inations. 1t see,s i,possi$le to realiBe
how a $are transparent a-ti3it# -an $e dire-ted to what is not there to
apprehend what is not gi3en.?
This state,ent is one whi-h pro$a$l# ,ost people would endorse.
But it is open to two o$9e-tions. =irst it is di55i-ult to see how an
a-ti3it# howe3er un* ?transparent? -an $e dire-ted towards a nothing: a
ter, o5 a relation -annot $e a ,ere nonentit#. 0e-ondl# no reason is
gi3en and 1 a, -on3in-ed that none -an $e gi3en 5or the assertion that
drea,*o$9e-ts are not ?there? and not ?gi3en?. Let us ta%e the se-ond
point 5irst.
(1) The $elie5 that drea,*o$9e-ts are not gi3en -o,es 1 thin% 5ro,
5ailure to distinguish as regards wa%ing li5e $etween the sense*datu,
and the -orresponding ?thing?. 1n drea,s there is no su-h -orresponding
?thing? as the drea,er supposes@ i5 there5ore
1
?!atural Realis, and 'resent Tenden-ies in 'hilosoph#? +roc.
Arist. *oc.
y
19()*19(9 p. 1H&.
1:) A/0T1210A A!> L<G12
the ?thing? were gi3en in wa%ing li5e as e.g. Aeinong ,aintains
1
then
there would $e a di55eren-e in respe-t o5 gi3enness $etween drea,s and
wa%ing li5e. But i5 as we ha3e ,aintained what is gi3en is ne3er the
thing $ut ,erel# one o5 the ?sensi$ilia? whi-h -o,pose the thing then
what we apprehend in a drea, is 9ust as ,u-h gi3en as what we
apprehend in wa%ing li5e.
;4a-tl# the sa,e argu,ent applies as to the drea,*o$9e-ts $eing
?there?. The# ha3e their position in the pri3ate spa-e o5 the perspe-ti3e
o5 the drea,er@ where the# 5ail is in their -orrelation with other pri3ate
spa-es and there5ore with perspe-ti3e spa-e. But in the onl# sense in
whi-h ?there? -an $e a datu, the# are ?there? 9ust as trul# as an# o5 the
sense*data o5 wa%ing li5e.
(:) The -on-eption o5 ?illusion? or ?unrealit#? and the -orrelati3e
-on-eption o5 ?realit#? are generall# used in a wa# whi-h e,$odies
pro5ound logi-al -on5usions. 7ords that go in pairs su-h as ?real? and
?unreal? ?e4istent? and ?non*e4istent? ?3alid? and ?in3alid? et-. are all
deri3ed 5ro, the one 5unda,ental pair ?true? and ?5alse?. !ow ?true? and
?5alse? are appli-a$le onl#Ce4-ept in deri3ati3e signi5i-ationsCto
propositions. Thus where3er the a$o3e pairs -an $e signi5i-antl#
applied we ,ust $e dealing either with propositions or with su-h
in-o,plete phrases as onl# a-6uire ,eaning when put into a -ontest
whi-h with the, 5or,s a proposition. Thus su-h pairs o5 words -an $e
applied to descriptions,
,
$ut not to proper na,es: in other words the#
ha3e no appli-ation whate3er to data $ut onl# to entities or non*entities
des-ri$ed in ter,s o5 data.
Let us illustrate $# the ter,s ?e4isten-e? and ?non*e4isten-e?. Gi3en
an# datu, x, it is ,eaningless either to assert or to den# that x ?e4ists?.
7e ,ight $e te,pted to sa#: ?<5 -ourse x e4ists 5or otherwise it -ould
not $e a datu,?. But su-h a state,ent is reall# ,eaningless although it
is signi5i-ant and true to sa# ?A# present sense*datu, e4ists? and it
,a# also $e true that (x is ,# present sense*datu,?. The in5eren-e 5ro,
these two propositions to (x e4ists? is one whi-h see,s irresisti$le to
people una--usto,ed to logi-@ #et the apparent proposition in5erred is
not ,erel# 5alse $ut stri-tl# ,eaningless. To sa# ?A# present sense*
datu, e4ists? is to sa# (roughl#): ?There is an o$9e-t o5 whi-h J,#
present sense*datu,J is a des-ription?. But we -annot sa#: ?There is an
o$9e-t o5 whi-h -x- is a des-ription? $e-ause K is (in the -ase we are
supposing) a na,e not a des-ription. >r 7hitehead and 1 ha3e
1
!ie .rfahrungsgrundlagen unseres /issens, *. :).
1
25. +rincipia Mathematica
y
Kol. 1L I 1G and 1ntrodu-tion 2hap.
111. =or the de5inition o5 existence, -5. I 1G. (:.
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 1:9
e4plained this point 5ull# elsewhere "0oe. cit.# with the help o5 s#,$ols
without whi-h it is hard to understand@ 1 shall not there5ore here repeat
the de,onstration o5 the a$o3e propositions $ut shall pro-eed with
their appli-ation to our present pro$le,.
The 5a-t that ?e4isten-e? is onl# appli-a$le to des-riptions is -on*
-ealed $# the use o5 what are gra,,ati-all# proper na,es in a wa#
whi-h reall# trans5or,s the, into des-riptions. 1t is 5or e4a,ple a
legiti,ate 6uestion whether .o,er e4isted@ $ut here ?.o,er? ,eans
?the author o5 the .o,eri- poe,s? and is a des-ription. 0i,ilarl# we
,a# as% whether God e4ists@ $ut then ?God? ,eans ?the 0upre,e Being?
or ?the ens realissimum( or whate3er other des-ription we ,a# pre5er. 15
?God? were a proper na,e God would ha3e to $e a datu,@ and then no
6uestion -ould arise as to .is e4isten-e. The distin-tion $etween
e4isten-e and other predi-ates whi-h Fant o$s-urel# 5elt is $rought to
light $# the theor# o5 des-riptions and is seen to re,o3e ?e4isten-e?
altogether 5ro, the 5unda,ental notions o5 ,etaph#si-s.
7hat has $een said a$out ?e4isten-e? applies e6uall# to ?realit#?
whi-h ,a# in 5a-t $e ta%en as s#non#,ous with ?e4isten-e?.
2on-erning the i,,ediate o$9e-ts in illusions hallu-inations and
drea,s it is ,eaningless to as% whether the# ?e4ist? or are ?real?. There
the# are and that ends the ,atter. But we ,a# legiti,atel# in6uire as to
the e4isten-e or realit# o5 ?things? or other ?sensi$ilia? in5erred 5ro, su-h
o$9e-ts. 1t is the unrealit# o5 these ?things? and other ?sensi$ilia? together
with a 5ailure to noti-e that the# are not data whi-h has led to the 3iew
that the o$9e-ts o5 drea,s are unreal.
7e ,a# now appl# these -onsiderations in detail to the sto-%
argu,ents against realis, though what is to $e said will $e ,ainl# a
repetition o5 what others ha3e said $e5ore.
(1) 7e ha3e 5irst the 3ariet# o5 nor,al appearan-es supposed to $e
in-o,pati$le. This is the -ase o5 the di55erent shapes and -olours whi-h
a gi3en thing presents to di55erent spe-tators. Lo-%e?s water whi-h
see,s $oth hot and -old $elongs to this -lass o5 -ases. <ur s#ste, o5
di55erent perspe-ti3es 5ull# a--ounts 5or these -ases and shows that
the# a55ord no argu,ent against realis,.
(:) 7e ha3e -ases where the -orrelation $etween di55erent senses is
unusual. The $ent sti-% in water $elongs here. 'eople sa# it loo%s $ent
$ut is straight: this onl# ,eans that it is straight to the tou-h though
$ent to sight. There is no ?illusion? $ut onl# a 5alse in5eren-e i5 we
thin% that the sti-% would 5eel $ent to the tou-h. The sti-% would loo%
9ust as $ent in a photograph and as Ar Gladstone used
1+( A/0T1210A A!> L<G12
to sa# ?the photograph -annot lie?.
1
The -ase o5 seeing dou$le also
$elongs here though in this -ase the -ause o5 the unusual -orrelation is
ph#siologi-al and would there5ore not operate in a photograph. 1t is a
,ista%e to as% whether the ?thing? is dupli-ated when we see it dou$le.
The ?thing? is a whole s#ste, o5 ?sensi$ilia? and it is onl# those 3isual
?sensi$ilia? whi-h are data to the per-ipient that are dupli-ated. The
pheno,enon has a purel# ph#siologi-al e4planation@ indeed in 3iew o5
our ha3ing two e#es it is in less need o5 e4planation than the single
3isual sense*datu, whi-h we nor,all# o$tain 5ro, the things on whi-h
we 5o-us.
(+) 7e -o,e now to -ases li%e drea,s whi-h ,a# at the ,o,ent
o5 drea,ing -ontain nothing to arouse suspi-ion $ut are -onde,ned
on the ground o5 their supposed in-o,pati$ilit# with earlier and later
data. <5 -ourse it o5ten happens that drea,*o$9e-ts 5ail to $eha3e in the
a--usto,ed ,anner: hea3# o$9e-ts 5l# solid o$9e-ts ,elt $a$ies turn
into pigs or undergo e3en greater -hanges. But none o5 these unusual
o--urren-es need happen in a drea, and it is not on a--ount o5 su-h
o--urren-es that drea,*o$9e-ts are -alled ?unreal?. 1t is their la-% o5
-ontinuit# with the drea,er?s past and 5uture that ,a%es hi, when he
wa%es -onde,n the,@ and it is their la-% o5 -orrelation with other
pri3ate worlds that ,a%es others -onde,n the,. <,itting the latter
ground our reason 5or -onde,ning the, is that the ?things? whi-h we
in5er 5ro, the, -annot $e -o,$ined a--ording to the laws o5 ph#si-s
with the ?things? in5erred 5ro, wa%ing sense*data. This ,ight $e used to
-onde,n the ?things? in5erred 5ro, the data o5 drea,s. >rea,*data are
no dou$t appearan-es o5 ?things? $ut not o5 su-h ?things? as the drea,er
supposes. 1 ha3e no wish to -o,$at ps#-hologi-al theories o5 drea,s
su-h as those o5 the ps#-ho*anal#sts. But there -ertainl# are -ases
where (whate3er ps#-hologi-al -auses ,a# -ontri$ute) the presen-e o5
ph#si-al -auses also is 3er# e3ident. =or instan-e a door $anging ,a#
produ-e a drea, o5 a na3al engage,ent with i,ages o5 $attleships and
sea and s,o%e. The whole drea, will $e an appearan-e o5 the door
$anging $ut owing to the pe-uliar -ondition o5 the $od# (espe-iall# the
$rain) during sleep this appearan-e is not that e4pe-ted to $e produ-ed
$# a door $anging and thus the drea,er is led to entertain 5alse $elie5s.
But his sense*data are still ph#si-al and are su-h as a -o,pleted
ph#si-s would in-lude and -al-ulate.
(G) The last -lass o5 illusions are those whi-h -annot $e dis-o3ered
1
25. ;dwin B. .olt )he +lace of Illusory .xperience in a
1ealistic /orld, The !ew Realis,? p. +(& $oth on this point and as
regards seeing double.
T.; R;LAT1<! <= 0;!0;*>ATA T< './0120 1+1
within one person?s e4perien-e e4-ept through the dis-o3er# o5
dis-repan-ies with the e4perien-es o5 others. >rea,s ,ight -on*
-ei3a$l# $elong to this -lass i5 the# were 9ointed su55i-ientl# neatl#
into wa%ing li5e@ $ut the -hie5 instan-es are re-urrent sensor#
hallu-inations o5 the %ind that lead to insanit#. 7hat ,a%es the patient
in su-h -ases $e-o,e what others -all insane is the 5a-t that within his
own e4perien-e there is nothing to show that the hallu-inator# sense*
data do not ha3e the usual %ind o5 -onne-tion with ?sensi$ilia? in other
perspe-ti3es. <5 -ourse he ,a# learn this through testi,on# $ut he
pro$a$l# 5inds it si,pler to suppose that the testi,on# is untrue and
that he is $eing wil5ull# de-ei3ed. There is so 5ar as 1 -an see no
theoreti-al -riterion $# whi-h the patient -an de-ide in su-h a -ase
$etween the two e6uall# satis5a-tor# h#potheses o5 his ,adness and o5
his 5riends? ,enda-it#.
=ro, the a$o3e instan-es it would appear that a$nor,al sense*data
o5 the %ind whi-h we regard as de-epti3e ha3e intrinsi-all# 9ust the
sa,e status as an# others $ut di55er as regards their -orrelations or
-ausal -onne-tions with other ?sensi$ilia? and with ?things?. 0in-e the
usual -orrelations and -onne-tions $e-o,e part o5 our unre5le-ti3e
e4pe-tations and e3en see, e4-ept to the ps#-hologist to 5or, part o5
our data it -o,es to $e thought ,ista%enl# that in su-h -ases the data
are unreal whereas the# are ,erel# the -auses o5 5alse in5eren-es. The
5a-t that -orrelations and -onne-tions o5 unusual %inds o--ur adds to
the di55i-ult# o5 in5erring things 5ro, sense and o5 e4pressing ph#si-s
in ter,s o5 sense*data. But the unusualness would see, to $e alwa#s
ph#si-all# or ph#siologi-all# e4pli-a$le and there5ore raises onl# a
-o,pli-ation not a philosophi-al o$9e-tion.
1 -on-lude there5ore that no 3alid o$9e-tion e4ists to the 3iew
whi-h regards sense*data as part o5 the a-tual su$stan-e o5 the ph#si-al
world and that on the other hand@ this 3iew is the onl# one whi-h
a--ounts 5or the e,piri-al 3eri5ia$ilit# o5 ph#si-s. 1n the present paper
1 ha3e gi3en onl# a rough preli,inar# s%et-h. 1n parti-ular the part
pla#ed $# time in the -onstru-tion o5 the ph#si-al world is 1 thin%
,ore 5unda,ental than would appear 5ro, the a$o3e a--ount. 1 should
hope that with 5urther ela$oration the part pla#ed $# unper-ei3ed
?sensi$ilia? -ould $e inde5initel# di,inished pro$a$l# $# in3o%ing the
histor# o5 a ?thing? to e%e out the in5eren-es deri3a$le 5ro, its
,o,entar# appearan-e.

Вам также может понравиться