Professional Documents
Culture Documents
Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at http://www.jstor.org/page/
info/about/policies/terms.jsp
JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range of content
in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new forms of scholarship.
For more information about JSTOR, please contact support@jstor.org.
Duke University Press and Philosophical Review are collaborating with JSTOR to digitize, preserve and extend access to The
Philosophical Review.
http://www.jstor.org
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
The PhilosophicalReview, Vol. XCVI, No. 4 (October 1987)
521
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
in any plausible psychological theory, and (b) these states are indi-
viduated by way of content: if two ascriptions differ in content,
then they ascribe different states.6 One who subscribes to this posi-
tion on psychological explanation, and at the same time wishes to
abide by the supervenience requirement, must suppose that these
attitudes are such as to satisfy this requirement-that physical
doppelgangers cannot differ in such states. The claim is not that
physical doppelgangers must agree in every psychological state of
the common-sense kind, but only in those states which figure in
psychological explanation. Thus such a theorist can allow that,
given appropriate differences in their respective worlds, one
doppelganger might know or remember that P while the other
fails to do so; one might have de re beliefs about a certain indi-
vidual, beliefs not shared by the other, and so on. The theorist can
allow for such differences without casting the supervenience re-
quirement into doubt by simply denying that such states need
figure as theoretical explanatory states. So far as knowledge goes
(and, likewise, memory, etc.), the explanatory theory need not ap-
peal to such a state as opposed to belief; the theorist can restrict
herself to belief, ignoring distinctions which might be made in the
light of truth, evidence, etc., distinctions which do not provide for
any genuine distinction in content.7 Likewise with de re beliefs. To
the extent that one's concern is with the explanation of behavior,
one need distinguish between psychological states only to the ex-
tent that they play different explanatory roles, and such states can
differ in explanatory role only if they differ in content (or in the
attitude of the agent to the content). The theorist can ignore de re
distinctions which are drawn in the light of differences in the sub-
jects' respective environments, but which do not provide for any
genuine difference in content. Genuine conflict arises only if it is
possible for psychological states of physical doppelgangers to
differ in content.
Though the notion of sameness of content is somewhat vague, I
shall employ the following criterion of difference: Where Cf and
Cg are sentences differing only in that the one contains the predi-
6For discussion and defense of these claims see, for example, Fodor (8)
and (9); and Stich (32).
7This general strategy is explicit in Fodor (9); Kim (17); and Stich (32).
524
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
J believes that Cf
and
J believes that Cg
525
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
"1Forsome discussion of such issues, see Kaplan (15); Perry (26); Burge
(2); Stalnaker (31); and Lewis (22).
527
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
'2Though Stich's paper (32) played a crucial role in setting the stage for
the resulting controversy about the implications (if any) of Twin-Earth
style examples for cognitive theory, it also generated some confusion in
that it treated the problems created by indexicals as being similar in kind
to those created by Twin-Earth style examples. Stich's assimilation of these
two kinds of problems was facilitated by his criterion of content differ-
ence, viz., contents (or thoughts) are different if they have different truth
conditions.
131t is true, of course, that Putnam interpreted natural kind predicates
as having an indexical component (Putnam (28)); but, as Burge has urged,
this claim seems to be unfounded (Burge (3)).
'4One finds both approaches in J. Fodor: The de re approach in (9) and
an appeal to an implicit indexical element in (10). The de re approach fails
for the simple reason that we can easily construct examples in which the de
re reading is inappropriate. The 'implicit indexical' story told by Fodor in
(10) might explain how the doppelgangers' beliefs differ in truth condi-
tions, but it does not explain why we are entitled to use the concept 'water'
in characterizing the belief of one and not the other. Burge, in (1), dis-
528
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
offer much promise, and has now largely been replaced by a very
different strategy; one which attempts, on the one hand, to grant
the point of the examples, while, on the other, attempts to mini-
mize their implications for cognitive theory and the supervenience
thesis. It is to this kind of response that we now turn.
II.
529
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
530
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
531
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
'7McGinn (24).
'8Field (6).
532
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
533
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
534
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
excluded from the outset. In specifying the relations that must ob-
tain between a given representation Ri and other representations
Rj, Rk, etc., if Ri is to have content P, these other representations
must be theoretically specifiable in some other way than as repre-
sentations having this and that content. The task, after all, is that
of providing an account of how any mental representation can be
said to have this kind of content in virtue of role; and this goal is
not realized by a theory which explains the content of a given rep-
resentation as resulting from its being appropriately related to
other representations where these others are theoretically specifi-
able only as representations having this or that content. To meet
the requirements of the account the various internal representa-
tions appealed to in the conceptual-role characterizations have to
be specifiable in some way other than as representations having
such and such specific content. This, of course, is a requirement
conceptual-role theorists willingly acknowledge, and indeed func-
tional definitions were introduced, in part, to meet this need, to
enable the theorist to characterize relations between arrays of
mental states SI, . . . S., without using any explicit mentalistic vo-
cabulary in the process (and so without explicitly characterizing
the states by way of content). Let us assume then that the internal
representations and their interrelations are characterized in some
standard abstract functionalist fashion, and turn our attention to
the language employed in input and output specifications, and, in
particular, to the descriptive language employed in output charac-
terizations. The output characterizations will have to employ a
substantial descriptive vocabulary if they are to play their assigned
role in the specification of psychological states; one can hardly
characterize both outputs and internal states in the same abstract
fashion, and hope to provide anything like appropriate conditions.
And in deciding upon this vocabulary the theorist is faced with the
familiar choice, to employ or eschew common-sense intentional ac-
tion terminology; both options seem doomed given the task she
has set herself-the provision of an account of mental content
compatible with the supervenience requirement.
Suppose that the theorist opts for the first alternative, and
taxonomizes outputs in an intentional fashion. Without losing any
generality we may, for the sake of argument, confine our attention
to the characterization of verbal behavior, where the contrast be-
tween the intentional and non-intentional is most sharply drawn.
535
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
536
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
537
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
538
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
tional systems and intentional states. No one, after all, wants to suppose
that ability of genetic engineers to fashion devices which warrant inten-
tional characterizationslends any support to the claim that intentional
states are best understood in terms of constructs drawn from genetic
theory.
26Intuitivelyit seems evident that such role-individuatedstates are too
fine-grained to do the work of psychologicalstates; and simply to assume
that there must be some way of generating the appropriate taxonomy
within this non-intentional framework smacks of the behaviorist'sinsis-
tence that there must be some way of characterizingbehavior which both
serves to express the requisite generalizationsand is 'observational'-not
requiring any 'interpretation'on the part of the theorist. Some fifty-odd
years of fruitless searching provide reason enough for thinking there
simply is no such vocabularyto be found.
539
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
540
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
dition for, say, believing that snow is white is that one be in the
appropriate role-individuated state. That is, to have this belief one
must, on this account, be in a state which bears certain specific
(causal) relations to other internal states (functionally specified),
and, in conjunction with these other states, to non-intentionallyspeci-
fied behaviors. It is this last element that is especially problematic.
There seems to be about as much reason to think that the attitudes
bear such necessary links to non-intentionally characterized be-
havior as there is to think that they admit of necessary neurophysi-
ological conditions; a claim that is now thoroughly discredited.27
However, to re-emphasize the point that such necessary conditions
are not forthcoming, it is helpful to turn once again to linguistic
behavior. And, for a particularly clearcut example, consider some
linguistic behavior which is taken to express a sophisticated belief,
for example, of a philosophical nature; one which need not find
expression in nonsymbolic behavior. Let P be such a belief. Then
whatever about the mental, we do know that it is not possible to
provide interesting necessarynon-intentional behavioralconditions(be
they grammatical, phonetic, or whatever) for saying that P, for ex-
ample, for saying that synonyms are not interchangeable in em-
bedded belief contexts. Given any set of allegedly necessary behav-
ioral conditions for saying this, one can always fashion a new way
of saying the same thing, one not in the original set. This belief can
be expressed in an indefinite variety of forms which bear no inter-
esting similarities to each other when characterized at the phonetic
level, at the grammatical level, or indeed at any currently available
non-intentional level. This difficulty is not avoided by appealing to
more abstract non-intentional characterizations. It won't do, for
example, to suppose that the necessary conditions are relational
conditions on the lexical items in the individual's repertoire-that
an expression could serve to encode the belief that P only if it bore
certain specific relations to other expressions in the individual's
language. Of course the expression must satisfy some substantial
constraints if it is really to express this belief; it must bear certain
relations to other expressions which encode conceptually related
beliefs. The difficulty however lies in the claim that there is some
27See Putnam (27) and Davidson (4), for arguments against there being
nomological necessary physiological conditions for mental states.
541
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
542
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
543
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
544
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
III.
545
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
29Fodor (9).
30See, for example, P. Kitcher (19).
546
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
The argument, I take it, goes like this: those who accept the super-
venience (or formality) thesis and those who reject it agree on the
principle that states with different contents can play different
causal roles in the genesis of behavior; difference in content can
provide for different "behavioral effects." The theorist who ac-
cepts the supervenience requirement can at least understand how
such difference in content can provide for different behaviors;
difference in content implies some physical (or formal) difference,
and it is this physical difference that makes for the possibility of
different behaviors. One who abandons the supervenience prin-
ciple, however, and attributes different psychological states to
physical doppelgangers, must allow that these different psycholog-
ical states could have different behavioral effects in the absence of
any physical difference. And this Fodor finds totally mysterious: if
the internal states of the doppelgangers are physically identical,
how could they generate different behaviors?
There is an air of mystery here, but it arises simply from the
same old ambiguity in the term 'behavior', from the fact that the
term is neutral vis-at-visthe intentional/non-intentional distinction.
This argument would tell against the opponent of supervenience if
that opponent were committed to the claim that the different con-
tents he ascribes to Alf and Alf* (in the light of relevant contextual
factors) were such as to provide for physically different behaviors
(for some non-intentional difference). If the different internal
psychological states he ascribes could cause different physical be-
547
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
haviors (in Alf and Alf*) then Alf and Alf* better differ in some
(internal) physical fashion. But the defender of Twin-Earth ex-
amples makes no such claim; indeed it is an essential part of the
story he tells that Alf and Alf* are exact counterparts when it
comes to physical behavior and dispositions to physical behavior.
Given that the different psychological states ascribed do not pro-
vide for the possibility of physically different behaviors, there is no
need to postulate some internal physical difference. In these ex-
amples the different psychological states do indeed provide for
different behavior, but only for behavior as intentionally charac-
terized. And such difference does not necessarily require some in-
ternal physical difference; the relevant contextual factors support
different intentional characterizations of both behavior and in-
ternal psychological state.
In general the opponent of supervenience can agree with
Fodor's principle: difference in content provides for different be-
haviors. But he accepts this principle in all its generality only when
it is construed as a principle to the effect that difference in content
provides for difference in intentionally characterized behavior;
and when construed in this weakened fashion it poses no threat to
his non-individualistic account of the mental.
One final argument for the supervenience principle is drawn
from S. Stich (who, remember, speaks of "psychological au-
tonomy" rather than "supervenience"):
I think the best defence of the autonomy principle begins with what
might be called the replacementargument. Suppose that someone were
to succeed in building an exact physicalreplica of me . .. the replica,
being an exact physical copy, would behavejust as I would in all cir-
cumstances.. . . But now, the argument continues, since psychologyis
the science which aspires to explain behavior, any states or processes
or properties which are not shared by Stich and his identically be-
having replica must surely be irrelevantto psychology.32
548
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
33Ibid., p. 167.
34Ibid.
549
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
35Ibid., p. 169.
550
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
STANDARD SPECIFICATIONS
Robot main body
Rotation ? 1500
Repeatability ? 0.2 mm
Weight 200 kg
551
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
552
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
IN DEFENSE OF A DIFFERENT DOPPELGANGER
REFERENCES
553
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions
JOSEPH OWENS
554
This content downloaded from 147.91.173.31 on Fri, 26 Jun 2015 16:18:42 UTC
All use subject to JSTOR Terms and Conditions