I. Self-knowledge, Externalism, and an Incompatibility Worry
It has been disputed whether an externalist conception of the individuation of
intentional states, such as beliefs and desires, is compatible with self-knowledge, that
is, the claim that one's judgments about one's intentional states, are
non-evidential, non-inferential, and authoritative.(1) The
worry has been raised that if externalism is true, then for a subject to know, say, that
she believes that p, the subject would need to know, on the basis of some evidence, the
external conditions which determine the belief's content. And thus externalism would
be incompatible with self-knowledge. But many philosophers have accepted an objection
suggesting that this worry is mistaken, because even if one's belief content is
externally determined, one need not know the external conditions determining that content
in order to have the belief. And, thus, the subject's reflexive judgment about the
belief would not need to rest on evidence about those external conditions.(2)
But this objection rests, in turn, on a crucial assumption according to which mental
content is reflexively transparent in the sense that a subject could not judge that
she has an intentional state and be mistaken about the content of her state, even if
content is externally determined.(3)
My main purpose is to question this crucial assumption. Now the claim that mental
content is reflexively transparent is extremely compelling and, if it is correct
while externalism is true, then this would indeed support the compatibility of externalism
and self-knowledge. But, I want to argue that mental content is not reflexively
transparent on the assumption of externalism. If my argument is correct, the upshot is
that self-knowledge and externalism are incompatible.
To set the general background, let me first say a bit more about externalism,
self-knowledge, and the worry that these theses may be incompatible. Though externalism is
not universally accepted, it is very well-known and, recently, it has become almost the
received view in philosophy of mind and philosophy of language. Externalism is a
metaphysical thesis concerning the individuation conditions on intentional states, such as
beliefs, desires, and the like. According to this thesis, intentional states are
individuated by their content, and content depends essentially on some relations between
the subject and her environment.(4)
As regards self-knowledge, it has its historical roots in Descartes's ingenious
intuition about the cogito. Now, on both sides of the compatibility issue, it is
agreed that self-knowledge should be understood as a weaker claim than what we may call
the "unrestricted cartesian conception". On this unrestricted understanding,
self-knowledge would be the mark of the mental in the sense that (i) a subject who has a
mental state could not fail to know that she has it, and (ii) all reflexive judgments
about one's own mental states could not fail to be true and, hence, no evidence (and
no inference) could show them to be mistaken. Of course, this unrestricted cartesian
conception seems much too strong in so far as it rules out both the possibility that there
exist unconscious mental states and that some judgments about one's mental states may
be cases of self-deception.(5) However, it seems hard to
deny that typically when a subject judges that she has a certain intentional state,
then this judgment is warranted even if it rests on no empirical evidence and no
inference. And it is in this weaker sense, vague though it may be, that self-knowledge is
generally understood in the incompatibility issue.
Now, some have voiced the worry that externalism might be incompatible with
self-knowledge. The worry is simply that if an intentional state, say, my belief that
water is a liquid, is individuated in accord with externalism, then it depends on some
conditions that hold in my environment. But, goes the worry, how could I know on the basis
of no evidence that I have that belief since my knowledge of these external conditions
would certainly need to be based on some evidence? But, some philosophers have insisted
that this worry rests on the following mistaken assumption: for a reflexive judgment about
one's intentional state to be warranted, one would need to know the external
conditions which are essential to its content. For instance, to know that I believe that
water is a liquid, I would need to know that I inhabit an environment where there is
water, and to know that I would need some empirical evidence. But, goes the
objection, to have a belief with a certain externally determined content, it is not
necessary to know the external conditions which determine the belief's content. The
latter may be what it is simply because the subject happens to be appropriately
"hooked" to her environment. And if that's correct, then why should the
subject need to know that those external conditions obtain in order to know that she has
that belief?
However, it's important to see that this argument, in support of the compatibility
of externalism and self-knowledge, rests on the crucial assumption that I have already
noted, namely, that mental content is reflexively transparent in the sense that a
subject could not judge that she has an intentional state with a certain mental content
and be mistaken about the content of her state.
II. Reflexive Transparency
First, let me try to make a bit clearer the notion of reflexive transparency.
This notion is best understood with some examples which illustrate Descartes's
paradigm of the cogito. Consider the following:
(1) I am now thinking.
(2) I refer to myself.
(3) I exist.
Now, these judgments are reflexively transparent simply in the following sense:
the property (or the relation) which is self-ascribed is instantiated by the very
instantiation of the judgment. For instance, by judging that I refer to myself, I actually
stand in the referring relation to myself. Now, the sense in which judgments like these
are non-inferential, non-evidential, and authoritative and, thus, constitute
self-knowledge, is transparent enough. It seems that anyone who considers these examples,
and who actually makes such reflexive judgments would be struck by their reflexive
transparency.
However, it should be noted that in such paradigmatic cases no externally determined
content seems too involved. Thus, prima facie, even if these judgments are indeed
reflexively transparent, this does not entail that judgments about one's externally
determined mental contents are also reflexively transparent. In other words, though the
claim that mental content is reflexively transparent, may seem prima facie
quite compelling in its own right, still it is doubtful that the reflexive transparency of
judgments like (1)-(3) entails that judgments about one's externally determined
intentional states are also reflexively transparent.
Now, this should help understand the crucial assumption involved in the objection to
the incompatibility worry, that is, the claim that the content of one's
intentional state is reflexively transparent even if it is externally determined. Indeed,
it is only in so far reflexive judgments are transparent in that sense, that they would
not need to be supported by empirical evidence about the relevant external conditions.
Thus, according to this crucial assumption, I could not judge that I have a certain
intentional state, while the content of that reflexive judgment might fail to involve the
very externally determined content of that state. So, reflexive transparency would have to
be true not only of judgments like (1)-(3), which correspond quite strictly to the
intuition of the cogito, but it would also have to be true of judgments about the
content of one's thoughts. In other words, reflexive judgments like (4), would also
have to be reflexively transparent:
(4) I think that water is a liquid.
But, in so far as (4) is supposed to illustrate the reflexive transparency of mental content,
we must insist that "I think", in (4), is used in a generic sense. On this
reading, the judgment expressed by (4) is only about a certain propositional content the
subject believes she entertains. Thus, (4) could be taken as expressing the same thing as
(4') I grasp the proposition that water is a liquid.(6)
Now, it seems hard to deny that such judgments are reflexively transparent in the same
sense that (1)-(3) are. But I want to suggest otherwise.
III. Could a subject be mistaken about her own mental content?
If the claim that mental content is reflexively transparent is to support the
compatibility of externalism and self-knowledge, then this claim must also presuppose,
quite obviously, the externalist character of mental content. So this general assumption
of the compatibilist objection can be stated as follows. For convenience, I call it
"Reflexive Transparency of Externalist Content".
(RTEC) For any subject S, necessarily if S makes a reflexive judgment of the form
"I think that p" (where "I think" is used in the sense of grasping),
then this judgment is co-instantiated with the subject's very grasp of the thought
that p, and no evidence (and no inference) could show the judgment to be false, even if
the mental content, that p, is externally determined.
This claim seems quite plausible. After all, for one to judge that one is thinking that
p, it does seem that one must think that p. But could it not be possible that one judges
that she thinks that p, when in fact she does not, because she thinks that q instead? Of
course such a possibility may seem a bit paradoxical, but I now want to suggest an example
supporting the possibility of such an error through misidentification. For convenience, I
call this possibility "Error through misidentification of content" (or
"EMC"):
(EMC) One can judge that one thinks that p and be mistaken, because one does not think
that p, but one thinks that q.
Consider the following example.(7)
Suppose that a subject
is in a building located on the docks. The subject (call her "S") is facing the
wall of a room where there are two windows facing the harbor. The windows are at a certain
distance from each other. From those windows, ships getting in an out of the harbor can be
observed. S's angle of gaze is such that through the left window she sees the part of
a ship, say the front deck, and through the right window she sees another part of the very
same ship, say the pilot cabin at the rear. Now we can suppose that S does not realize
that it is one and the same ship she is observing. For instance, perhaps the paint of the
ship is being redone, and perhaps there is a discrepancy between the look of the front
part, and that of the rear part. Moreover, suppose that S has recently observed two
different ships in the harbor: one having some striking resemblance with the front part of
the ship she now observes, and the other having features similar to those of the rear
part. Furthermore, the angle of S's gaze is such that it would be physically possible
that what she sees through the two windows be two different ships. We can even suppose
that on several occasions S has indeed observed pairs of different ships from just that
perspective, and that she has never before observed only one ship through both windows as
it now happens.
Now, suppose that in that situation S comes to believe that the ship she sees through
the left window is different from the ship she sees through the right window. Hence, her
belief's content could be expressed thus:
(5) This ship (mentally gazing in the direction of the left window) is different from
that ship (mentally gazing in the direction of the right window).
Moreover, S could come to make a reflexive judgment about the content of her belief:
she may judge that what she believes is that this ship is different from that ship, which
reflexive judgment can be expressed as follows:
(6) I think that this ship ¹ that ship.
(For brevity's sake, I drop the qualification about the indexical mental gaze,
though it is still assumed here and in what follows.) But now if RTEC is true, then (6) is
reflexively transparent and, thus, it could not possibly be mistaken. But here the
following question arises: when S makes judgment (6) is it necessary that the that-clause
express the very proposition that occurs in her world-directed belief, namely the one
expressed by (5)? I think it is not. In other words, I want to argue that there is ground
to resist the claim that (6) could not possibly be mistaken and, thus, ground to deny
RTEC.
First, if the content of S's world-directed belief, that is, the content expressed
by (5), is indeed determined by the external conditions holding in S's environment,
namely by the fact that there is actually only one ship referred to, then this very mental
content entails that the ship is not self-identical which violates the general principle
that all objects are self-identical (or "PSI"). But we can certainly assume that
S accepts that principle. Moreover, it is in no way incoherent that in making reflexive
judgment (6), S uses the expression "I think" with the intention to refer, at
once, both to the content of the world-directed belief that just occurred to her, the
content expressed by (5), and to the content expressed by the that-clause in (6), meaning
that both thoughts are identical. But, by "that this ship ¹
that ship" in (6), S may well express a content different from the content of the
belief that just occurred to her. This is why it does not seem incoherent that in (6) S
may make an error through misidentification of her mental content. For instance, by the
that-clause in (6) the subject may express a proposition which, contrary to (5), does not
entail that the ship is not self-identical. Indeed, we may suppose that by the that-clause
in (6) S expresses the following proposition or a similar one:
(5*) This ship part is the part of a ship which is identical to the ship of which that
ship part is a part.(8)
Now, we may ask why it is (5*), rather than (5), that gives the content of the
that-clause in (6). In reply we must note, first, that I need not claim that the
that-clause in (6) must necessarily be interpreted as expressing (5*), or a similar
proposition. But to support the coherence of EMC, all I need to claim is that it is not
incoherent that the that-clause in (6) may express (5*), while by "I think", in
(6), the subject may also intend to refer to the different world-directed thought that
occurred to her, namely the one which violates PSI. The reason why this is not
implausible, let alone incoherent, can be put in a nutshell: Charity begins at home. It is
one thing for someone to believe a proposition which violates a principle she accepts,
namely PSI; but it is quite a different thing to self-ascribe a belief the content of
which does violate that accepted principle. It seems extremely plausible that when we
ascribe beliefs to others we try to make them rational by avoiding to ascribe contents
which violate some strong principle like PSI. But then it seems that this must also be
true of self-ascriptions. In a case like (6) it would be quite
"self-uncharitable" for the subject to judge that she believes a proposition
which violates a principle she admittedly accepts, unless she would knows the external
conditions which determine the content of the belief she intends to ascribe to herself.
Let me try to clarify this point. Suppose that after making judgment (6), the subject
was brought to the window and shown that there is only one ship, then it does not seem
incoherent that she would react in the following way. About her non-reflexive thought she
may say: "Gee! I did believe something that entails that the ship is not
self-identical". But if pressed to say whether in her reflexive judgment about that
belief she judged that she was believing something that entails the falsity of PSI, then
she could coherently answer that she did not, that she reflexively judged that she was
believing something which does not entail that the ship is not self-identical, perhaps
something like (5*), but that she was mistaken in her reflexive judgment. Nobody's
perfect! But it is plausible that we are charitable in belief ascriptions, particularly in
one's own case. So, if what I have argued so far is correct, then error through
misidentification of content in reflexive judgments is indeed possible. And if that's
correct, it suffices to show that RTEC is false. But, as we may recall, RTEC is precisely
the fundamental assumption underlying the objection to the incompatibility worry about
externalism and self-knowledge.
For the that-clause in a reflexive judgment like (6) to express the same externally
determined content as the one expressed by the non-reflexive judgment like (5), which
violates PSI, it does seem that the subject would need to know the external conditions
that determine the content of the latter. Unless the subject knows those external
conditions, it seems rather implausible that she would self-ascribe a belief which
violates a principle she admittedly accepts, even though in her reflexive judgment (6) she
may use "I think" intending to refer both to (5), which violates PSI, and to
(6)'s that-clause, intending the latter to express a proposition like (5*), which
does not violate PSI.(9)
Thus, if the argument of this paper is correct, then the worry that externalism and
self-knowledge are incompatible is more than a worry: those theses are indeed
incompatible. If my argument is correct, it indicates that the a priori intuitions
underlying the compatibilist claim that we do have some kind of special first-person
epistemic access to the content of our intentional states is probably mistaken, unless one
is ready to deny externalism. But this is an issue I shall leave for another occasion.
|
