Psyche 9(05): 'The Introspectibility Thesis' by Cody Gilmore
Patrick Wilken
PSYCHE: AN INTERDISCIPLINARY JOURNAL OF RESEARCH ON CONSCIOUSNESS
-----------------------------------------------------------------
THE INTROSPECTIBILITY THESIS
Cody S. Gilmore
Department of Philosophy
Princeton University
1879 Hall
Princeton, NJ 08544
U.S.A.
Copyright (c) Cody S. Gilmore 2003
PSYCHE, 9(05), February 2003
http://psyche.cs.monash.edu.au/v9/psyche-9-05-gilmore.html
KEYWORDS: unity of consciousness, co-consciousness, introspection,
introspectibility, headache argument, shrub argument, split-brain
argument
COMMENTARY ON: Dainton, B. (2000). *Stream of Consciousness: Unity and
Continuity in Conscious Experience.* London: Routledge.
ABSTRACT: According to what Barry Dainton calls the 'Strong
Introspectibility thesis', it is a necessary truth that mental states
S and S* are co-conscious (experienced together) if and only if they
are 'jointly introspectible', i.e., if and only if it is possible for
there to be some single state of introspective awareness that
represents both S and S*. Dainton offers two arguments for the
conclusion that joint introspectibility is *unnecessary* for
co-consciousness. In these comments I attempt to show, first, that
Dainton's arguments fail, and, second, that joint introspectibility is
actually *insufficient* for co-consciousness. (As to whether it is
also unnecessary, I take no stance.)
1. INTRODUCTION
Barry Dainton's fascinating and insightful book on the unity of
consciousness explores a huge swath of heretofore poorly charted
territory. In these comments I shall confine my attention to just one
small corner of this territory, the so-called 'Strong Introspectibility
thesis':
Co-consciousness is constituted by introspectibility: experiences
are co-conscious [i.e., experienced together] *because* they are
introspected or introspectible. A group of token experiences are
co-conscious if and only if they are either the actual or potential
objects of a single introspective awareness (2000: 35, Dainton's
emphasis).
The I-thesis (as I shall call it) is a *reductive account* of
co-consciousness; i.e., it purports to give informative, individually
necessary and jointly sufficient conditions for a pair of mental states
to be co-conscious. It says: *joint introspectibility* is necessary and
sufficient for co-consciousness.
Since Dainton regards the I-thesis as one of the main rivals to his own
(non-reductive) account of co-consciousness, and since his defense of
his own account consists entirely of a series of objections to its main
rivals, much rests on Dainton's case against the I-thesis. Accordingly,
he aims for redundancy. As I read him, he attempts to refute the
I-thesis twice over: he offers two arguments each of which purports to
establish that a pair of mental states can be co-conscious despite
failing to be jointly introspectible, hence that joint introspectibility
is *unnecessary* for co-consciousness.
In these comments I shall attempt to show, first, that both of Dainton's
arguments against the I-thesis are unsuccessful and, second, that joint
introspectibility is actually *insufficient* for co-consciousness. (As
to whether it is also unnecessary, I take no stance.)
2. THE I-THESIS AND ITS VARIANTS
Before we begin our examination of the I-thesis itself, we ought to
contrast this thesis with two of its close relatives. First is what we
might call the *Introspection* thesis:
Necessarily, for any (token) mental states S and S*, S and S* are
co-conscious if and only if they are jointly introspected, i.e., if
and only if there is some single state of introspective awareness
that represents both S and S*. (See, e.g., Parfit 1984: 250-51)
The Introspection thesis says, in other words, that 'joint
introspectedness' is both necessary and sufficient for co-consciousness.
Dainton has very little patience for this claim. He takes it to be
obvious that we often have pairs of co-conscious perceptual experiences
(e.g.) that fail to be jointly introspected, hence that joint
introspectedness is unnecessary for co-consciousness.<1>
But won't it always be true in such cases that the relevant co-conscious
experiences at least *could have been* jointly introspected, even if
they were not actually so? It is this suggestion that gives rise to the
Introspectibility thesis, which takes it to be a necessary truth that
experiences are co-conscious just in case they are at least jointly
introspect*ible*.
There are, however, a number of different things one could mean by
saying that a pair of experiences 'could have been' jointly
introspected. To see this, consider an attempt by Christopher Hill to
refute the I-thesis:
There are lower animals who appear to have unified sensory fields
but who do not display the sort of conceptual sophistication that a
being must have in order to be capable of having introspective
beliefs. Consider raccoons. (1991: 232).
Call this the 'raccoon argument'. Hill is probably right on the
following counts: (i) raccoons have co-conscious experiences, (ii)
raccoons lack the 'cognitive capacity' to be introspectively aware of
their experiences; i.e., raccoons do not actually have the right sort of
physical or causal-functional make-up to engage in this sort of
introspection. Therefore, it seems to me that Hill's raccoon argument
succeeds in refuting the following version of the I-thesis:
For any mental states S and S*, any possible world w and any time t
in w, S and S* are co-conscious at t in w if and only if: (i) S and
S* occur at t in w, and (ii) the owner of S and S* in w has a
physical or causal-functional make-up at t in w that gives it the
cognitive capacity to jointly introspect S and S* at t in w.
Call this the 'ICC-thesis'. It states that joint introspectibility
*understood in terms of actual cognitive capacities* is both necessary
and sufficient for co-consciousness. As the raccoon argument
demonstrates, the ICC-thesis is implausible: this sort of joint
introspectibility seems to be *unnecessary* for co-consciousness. A
raccoon could have co-conscious experiences despite lacking the actual
cognitive capacity to introspect them jointly.<2>
Nevertheless, for any co-conscious raccoon experiences e and e*, there
is probably *some* sense in which e and e* 'could have been' jointly
introspected. Suppose, for example, that e and e* are co-conscious
experiences that belong to raccoon R. Then presumably there is a
possible world in which, prior to the occurrence of e and e*, R
undergoes some brain-enhancing procedure and thereby *acquires* the
cognitive capacity for the relevant sort of introspection. When e and e*
occur, R can and does jointly introspect them. For this reason it seems
to me that the raccoon argument does *not* refute the following version
of the I-thesis:
For any mental states S and S*, any possible world w, and any time t
in w, S and S* are co-conscious at t in w if and only if (i) S and
S* occur at t in w, and (ii) S and S* are jointly introspectible;
i.e., *there is some possible world in which S and S* are jointly
introspected.*
Call this the 'IP-thesis'. It says that two mental states are
co-conscious just in case it is *possible* (somehow or other) for these
states to be jointly introspected. In what follows, I shall construe
Dainton's I-thesis as the IP-thesis.
3. THE HEADACHE ARGUMENT
We turn now to the first of Dainton's two arguments against the
I-thesis. The argument has three main premises: (i) some pairs of
co-conscious experiences have at least one member that is not
introspected, (ii) introspection affects the phenomenal character of its
objects,<3> and (ii) each (token) mental state has its phenomenal
character essentially. From this appealing starting point, Dainton
reasons as follows:
The idea that we could have introspected experiences we did not in
fact introspect only makes sense if . . . introspection . . . does
not alter the character of the relevant experiences. Since the
identity of an experience depends in part on its phenomenal
character, if introspection affected the character of an experience
we clearly could not introspect experiences that we did not in fact
introspect. It is also clear that . . . introspection *does*
influence the character of experience. . . . Suppose that five
minutes ago I was not . . . introspecting my slight headache.
According to the . . . I-thesis, I could have been attentively aware
of this sensation if I had chosen to be. But if I had so chosen my
headache would very probably have intensified, and so the pain I
would have been reflectively aware of would not have been
numerically the same pain as I actually had. . . . Since our
experience does exhibit this dependency, it will often be the case
that experiences we *did not* introspect *could not have been*
introspected" (2000: 37-38, my emphasis).
Now, to complete the argument, consider Dainton's headache experience e
and suppose that it is co-conscious with some other experience e*: e
could not have been introspected, and a fortiori e and e* could not have
been *jointly* introspected. So e and e* are co-conscious despite
failing to be jointly introspectible, hence the I-thesis is false.
Call this the 'headache argument'. Is it successful? No, not even if we
grant its premises. The flaw lies in Dainton's inference from (1) to
(2):
(1) If it *had* been the case that
(A) Dainton was engaged in the relevant sort of
headache-directed introspection at time t,
then it *would not have been* the case that
(B) the object of this introspection was e.
(Rather, the object of his introspection would have been some
phenomenally distinct, hence numerically distinct, experience, e'.
If we let '[]-->' be the counterfactual conditional sign and 'not-'
be the negation sign, then we can express (1) symbolically as
follows: A[]--> not-B.)
(2) Therefore, it *could not* have been the case both that:
(A) Dainton was engaged in the relevant sort of
headache-directed introspection at t, and
(B) the object of this introspection was e.
(Letting 'POSS' be the possibility operator, we can express (2)
symbolically thus: not-[POSS(A&B)].)
It is uncontroversial that the counterfactual conditional p[]-->q does
not entail the denial of compossibility not-[POSS(p&q)]. Equivalently,
it is uncontroversial that the counterfactual conditional is weaker than
the strict conditional.<4> This can be seen very easily by considering a
different pair of propositions with the relevant logical forms:
(1*) If Gore had won the 2000 election, Laura Bush would not have
become First Lady in 2001. Symbolically: A*[]-->not-B*.
(2*) It is not possible that both Gore won the 2000 election and
Laura Bush became First Lady in 2001. Symbolically: not-[POSS(A*&B*)].
(1*) clearly does not entail (2*) since, despite the probable truth of
the former, the latter is plainly false. After all, it *could* have been
the case, to take just one specific course of events, that Gore won the
election, became a widower, and shortly thereafter married a recently
divorced Laura Bush, who then ascended to First Lady-hood as Gore's wife
upon his inauguration. There are possible worlds in which this occurs.
But it seems likely that the *closest* possible world in which Gore won
the election is a world in which Laura Bush did not become First Lady,
hence that (2*) is true. (I adopt possible worlds semantics here only
for the sake of vividness; my claim that the relevant counterfactuals do
not entail the corresponding denials of compossibility is independent of
this semantics.)
Similarly, we can formulate a number of more or less specific hypotheses
that entail (1) while entailing the *negation* of (2). Here is one.
H1 The principle that introspection affects the phenomenal
character of its objects is a *counterfactual supporting* but
*contingent* truth, much like a law of nature. Since this principle
is counterfactual supporting, it is true that if Dainton *had* been
engaged in headache-directed introspection at t, his headache *would
have* had a different phenomenal character. And since the principle
is *contingent*, there are possible worlds in which introspection
does *not* affect the character of its objects. In some of these
worlds, Dainton can and does engage in headache-directed
introspection at t without thereby intensifying his headache
sensations, hence without preventing his actual headache, e, from
occurring. In some such world, e occurs and is introspected.
Since Dainton has given us no reason to believe that it is a *necessary*
truth that introspection affects the phenomenal character of its
objects, he has given us no reason to reject H1; and consequently he has
given us no reason to accept (2) or to reject the I-thesis.
Even if he *had* established the necessity of the relevant principle,
however, he still would have failed to make his case. This can be seen
by considering H2.
H2 It is a necessary truth that introspection affects the
phenomenal character of its objects. In the actual world, w, Dainton
had a highly stressful morning, and his psychological tension built
throughout the afternoon. By 6:00 p.m. (time t), Dainton had a
headache, e. On a scale of 1 to 10, 10 being the most intense, e was
a 'level 3' headache; and, at 6:00 p.m., Dainton was *not*
introspectively aware of e.
However, in nearby possible world w*, on the very same day, things
turned out somewhat differently. Dainton's day was less stressful
and his headache sensations were less severe throughout the
afternoon. At 5:58 p.m., Dainton had a 'level 2' headache, of which
he was not introspectively aware. At 5:59 p.m., Dainton happened to
turn his introspective attention toward his headache sensations,
with the result that they intensified somewhat. At 6:00 p.m.,
Dainton had a 'level 3' headache, e, of which he *was*
introspectively aware.
The main idea behind H2 is straightforward. As things actually happened,
e was a non-introspected, 'level 3' headache sensation preceded by a
long, monotonous series of the same. If we confine our attention to
those possible worlds in which these preceding conditions are held
*constant*, we may thereby rule out the possibility of e's being
introspected. How so? Any possible world in which these preceding
conditions obtain *and* in which Dainton begins to introspect just in
time for e will be a world in which the monotony of the series is
disrupted, the headache sensations intensify, and e (which is
*essentially* a 'level 3' headache) is thereby kept from occurring. (Or
so it might be argued by one who accepted Dainton's premises.)
But if we *vary* the preceding conditions in the manner indicated by H2,
we can make it turn out *both* that e occurs *and* that introspective
awareness is directed its way: e could be the result of Dainton's
introspecting, and thereby intensifying, a series of 'level 2' headache
sensations.
Each of the two foregoing hypotheses entails (1) while entailing the
negation of (2). Since Dainton has done nothing to cast doubt on either
H1 or H2, I conclude that the headache argument gives us no reason to
think that joint introspectibility is unnecessary for co-consciousness
or that the I-thesis is false.
4. THE SHRUB ARGUMENT
The following passage presents a second argument against the I-thesis:
Suppose you do introspect some part of your current experience. . .
. This introspected experience remains co-conscious with the
remainder of your experience. . . . Given this, what is responsible
for the unity of the introspected experience with the
non-introspected experience? One thing seems certain: it cannot be
any form of introspection. When you focus your attention onto the
shrub [e.g.,] your . . . auditory experiences [etc.] all remain
co-conscious with your visual experiences. These experiential
relationships cannot be explained in terms of introspectibility, for
they are not even *potential* objects of introspection. If you were
to try to . . . introspect these relationships you would have to
stop introspecting your experience of the shrub. As this example
makes plain, the co-consciousness of experiences which are not being
. . . introspected with experiences which are is not something which
can be . . . introspected. . . . This provides another reason for
rejecting the . . . I-thesis: there is at least one form of
co-consciousness that is in principle non-introspectible (2000:
36-37, Dainton's emphasis).
Let us suppose for a moment that the relevant "experiential
relationships" are not even "potential objects of introspection." Does
it follow that the I-thesis is false? Not without some additional
premises. The I-thesis claims only this: that when two *experiences* are
co-conscious, this is in virtue of the fact that *they* are jointly
introspectible, that they could be among the objects of a single state
of introspective awareness. It does not claim that when two experiences
are co-conscious, this is in virtue of the fact that *their
co-consciousness* is introspectible; indeed, the I-thesis is perfectly
consistent with Dainton's claim that certain instances of
co-consciousness cannot themselves be introspected.
To convert the passage quoted above into a logically valid argument
against the I-thesis, we will have to fill in some tacit assumptions. I
suggest that the following reconstruction strikes the best balance
between fidelity to Dainton's words and plausibility of premises:
(1) c_av is an instance of co-consciousness that holds between e_a, which
is a non-introspected auditory experience, and e_v, which is an
introspected visual experience of a shrub.
(2) Necessarily: for any experiences e and e* and any instance c of
co-consciousness, if c holds between e and e*, then: (i) c is an
experience in is own right, and (ii) c is an experience that is
co-conscious both with e and with e*.<5>
(3) Therefore, by (1) and (2), c_av is an experience that is co-conscious
both with e_a and with e_v.
(4) For any instance c of co-consciousness, if c holds between two
experiences only one of which is introspected, then, necessarily, c is
non-introspected.
(5) Therefore, by (1) and (4), necessarily, c_av is non-introspected; and
a fortiori it is necessary that c_av is not jointly introspected together
with e_a (or e_v). In other words, c_av and e_a are not jointly
introspectible.
(6) Therefore, by (1), (3), and (5), c_av and e_a are co-conscious
experiences despite failing to be jointly introspectible; and
consequently the I-thesis is false.
For the sake of argument, I will grant (1) and (2). In particular, I
will concede the following questionable moves: the decision to posit
such entities as *instances of co-consciousness*, the decision to treat
these entities as *experiences (or mental states, or phenomenal items)
in their own right*, and the assumption that these entities are always
co-conscious with the experiences they relate.<6>
The only remaining premise is (4), which states that if c is an instance
of co-consciousness that holds between an introspected and a
non-introspected experience, then c itself is "in principle
non-introspectible". What does Dainton have to say in its defense? "The
co-consciousness of experiences which are *not* being introspected [such
as your auditory experience] with those which are [such as your visual
experience of a shrub] is not something which can be introspected." This
is made "plain", Dainton thinks, by an example: "if you were to try to
introspect these relationships, you would have to stop introspecting
your experience of the shrub."
But why *couldn't* you simultaneously introspect both (i) the relevant
instance of co-consciousness and (ii) your experience of the shrub? Is
Dainton assuming that it's impossible to be introspectively aware of
more that one experience at a time? If so, I deny his assumption. But in
any case he has failed to provide us with a good reason for accepting
(4).
Here is an alternative attempt at defending (4). Consider the following
principle, which I find plausible:
(A) It is impossible to introspect a given instance of co-consciousness
without simultaneously introspecting the two experiences *related* by
the given instance; i.e., necessarily, for any experiences e and e*, any
instance c of co-consciousness, and any mental state S, if c holds
between e and e*, then, if S introspectively represents c, S also
introspectively represents e and e*.
If (A) is true, and if you were to try to introspect the instance of
co-consciousness that holds between your introspected visual experience
and your non-introspected auditory experience, then, far from having to
*stop* introspecting your visual experience (as Dainton claims), you
would actually have to *start* introspecting your previously
non-introspected *auditory* experience. Therefore, if (A) is true, it is
impossible to introspect an instance of co-consciousness that holds
between an introspected experience and a non-introspected experience.
The problem with this line of reasoning is that (A), though plausible,
does not entail (4); rather, it entails
(4*) Necessarily, for any instance c of co-consciousness, if c holds
between two experiences only one of which is introspected, then c is
non-introspected.<7>
Contrast this with (4):
(4) For any instance c of co-consciousness, if c holds between two
experiences only one of which is introspected, then, necessarily, c is
non-introspected.<8>
Let H be the property *being an instance of co-consciousness that holds
between two experiences only one of which is introspected*; and let N be
the property *being introspected*. Now it is plausible that for any
possible world w, anything that has H in w fails to have N in w; i.e.,
(4*) is plausible. But it is not plausible that anything that has H in
the actual world has H in every possible world; on the contrary it seems
that there are some entities that have H in the actual world despite
existing in other possible worlds where they do *not* have H. Consider,
for example, our instance c_av: in the actual world it holds between an
introspected and a non-introspected experience (hence, by (4*), we must
conclude that c_av is not introspected in the actual world); but it is
plausible to suppose that in other possible worlds c_av holds between
*two introspected experiences*. Moreover, it seems plausible to suppose
that in *some* of these other worlds, c_av itself is introspected.
Therefore (4) is apparently false.
Even with all of our concessions in place, then, the shrub argument does
nothing to suggest that joint introspectibility is not necessary for
co-consciousness; hence this argument remains powerless to harm the
I-thesis.
5. THE SPLIT-BRAIN ARGUMENT
I would now like to suggest a very different line of attack against the
I-thesis. The foregoing arguments both purport to show that some pairs
of mental states succeed in being co-conscious despite failing to be
jointly introspectible, hence that joint introspectibility is *not
necessary* for co-consciousness. But if my criticisms are correct, each
of these arguments fails. A better way to challenge the I-thesis, I
think, is to find a possible counterexample to the *sufficiency* claim,
i.e., to find a possible case involving a pair of mental states that
*fail* to be co-conscious despite *succeeding* in being jointly
introspectible.
Here is one such case. Suppose that I have had a futuristic device
installed in my brain that allows me to connect or disconnect my
cerebral hemispheres at will (as described by Parfit (1984: 246)). When
the hemispheres are connected, each mental state supported by any part
of my brain is co-conscious with every other such state. But when the
hemispheres are disconnected, each of them supports its own set of
mental states, i.e., (i) each mental state supported by the right
hemisphere is co-conscious with every other such state, (ii) each mental
state supported by the left hemisphere is co-conscious with every other
such state, and (iii) no mental state supported by the left is
co-conscious with any mental state supported by the right. We might also
add that the following sort of situation cannot occur: experiences e and
e* are jointly introspected despite being supported by distinct,
disconnected hemispheres.
Now consider a possible world w at which the following is true. During
T, a brief period of disconnectedness, my right hemisphere supports
(inter alia) a visual experience, e_v, and my left hemisphere supports an
auditory experience, e_a. Then, since e_v and e_a are supported by
distinct, disconnected hemispheres, they fail to be co-conscious, and
they also fail to be jointly introspected.
But it is plausible to suppose that e_a and e_v* could have been jointly
introspected*, since their respective hemispheres *could have been
connected* throughout T, the period when e_v and e_a occur. That is to
say, it seems likely that there is a possible world w* at which my
hemispheres are connected throughout T and at which there is a single
higher-order state S that introspectively represents both e_v and e_a.
(And if it is not possible that all of this be true of *me*, then surely
it is possible that it all be true of *some* sentient being. Let us
waive this complication.)
There is no good reason to deny that *these very experiences* occur at
some possible world w* where my hemispheres are connected throughout T.
This is true even if each (token) mental state is so 'modally fragile'
that it has its spatiotemporal location, its physical basis, its
phenomenal character, and its owner *essentially*, i.e., in every
possible world in which it occurs. After all, in light of the set-up of
the case, it is clear that all these features can remain constant from
possible world w (where my hemispheres are disconnected throughout T),
to a possible world w* where my hemispheres are connected throughout
T.<9>,<10> Nor is there any good reason to deny that at *some* such
world where my hemispheres are connected throughout T, the relevant
experiences are jointly introspected.
Thus it seems that we have our possible counterexample to the
sufficiency clause of the thesis that joint introspectibility is
necessary and sufficient for co-consciousness: e_v and e_a are jointly
introspectible in w despite failing to be co-conscious there.
It should be pointed out that this 'split-brain argument' does *not*
count against the ICC-thesis, which says that mental states S and S* are
co-conscious at time t in possible world w if and only if they occur at
t in w and their owner has, at t in w, the cognitive capacity to
introspect S and S* jointly. In the world w described in our case above,
my hemispheres are disconnected throughout T. Hence it is plausible to
say that throughout T in w I lack the cognitive capacity to introspect
e_v and e_a jointly.
6. CONCLUSION
Our results can be summarized informally as follows. My *actually having
the cognitive capacity* to jointly introspect my experiences e and e*
may be a *sufficient* condition for these experiences to be
co-conscious; but, as the raccoon argument apparently shows, this is not
a *necessary* condition. And my being such that *it is possible* for e
and e* to be jointly introspected may be a *necessary* condition for
these experiences to be co-conscious (given the failure of the headache
and shrub arguments); but, as the split-brain argument apparently shows,
this is not a *sufficient* condition.<11>
NOTES
<1>See, e.g., Dainton (2000: 34) and Hill (1991: 231-2).
<2>Dainton has suggested in correspondence that while raccoons probably
lack the concepts to form introspective *thought*-like representations
(HOTs) of their experiences, they may have the capacity to form
(non-conceptual) *perception*-like representations (HOPs) of these
experiences. (On the distinction between HOTs and HOPs, see Block et.
al. (1997: sect. X).) Perhaps this is so. But surely there at least
*could be* creatures that have co-conscious experiences despite lacking
the cognitive capacity to form *any* sort of higher-order
representations. Indeed I would guess that there *actually are* such
creatures: frogs might be a good example.
<3>Actually Dainton distinguishes two kinds of introspection - *active*
and *passive* - and restricts his claim that introspection affects the
character of its objects to the case of active introspection. I ignore
passive introspection in what follows for two reasons: (i) taking it
into account would complicate the discussion, and (ii) we should be
suspicious of it. Insofar as one is willing to concede the
character-affecting powers of *some* form of introspection, one should
be unwilling to *depend* upon the inertness of *any* form of
introspection. In particular, it seems reasonable to suppose that if
some forms of introspection are more active and character-affecting than
others, this is because there is a whole *continuum* of such forms,
ranging from the most active and character-affecting to the least active
and least character-affecting. And if *any* form of introspection is
character-affecting, then it seems likely that all such forms are at
least *somewhat* character-affecting.
<4>See, e.g., the entry on 'Counterfactuals' in Audi (1995: 163).
<5>The following two claims, both of which Dainton seems to accept,
might provide some motivation for (2): (a) necessarily, for any
experiences e and e*, e and e* are co-conscious iff e and e* have a
fusion that is an experience in its own right; (b) necessarily, for any
experiences e and e*, if e is a part of e*, then e and e* are
co-conscious. See Dainton (2000: ch. 9) and Bayne (2001).
<6>Do these assumptions lead to an infinite regress of experiences, and
if so is the regress vicious? On these questions see Bayne (2001).
<7>Letting 'NEC' be the necessity operator and 'A' be the universal
quantifier, we can express (4*) symbolically as follows: NEC[Ax(Hx -->
not-Nx)].
<8>Symbolically: Ax[Hx --> NEC(not-Nx)].
<9>I assume here that the relevant experiences can fail to be
co-conscious in w without having different owners there.
<10>In correspondence, Dainton has noted that certain forms of
interexperiential holism would prevent e_a and e_v from occurring both in
worlds where they are co-conscious and in worlds where they are not. In
particular, if we say that each experience has all of its phenomenal
properties essentially, and if we count such relational, haecceitistic
properties as *being co-conscious with the particular experience e_v* as
phenomenal, then of course e_a either has this property essentially or
lacks it essentially. In my view, the strongest form of holism that is
not obviously absurd is what Dainton calls 'Weak C-holism' (2000: 223),
which I state in my own words as follows: each experience has its
(intrinsic) phenomenal *character* essentially, and each experience has
its *context type* (but not its context token) essentially. (The context
token of an experience e can be defined as the set of token experiences
co-conscious with e; we can say that two context tokens C and C* are of
the same context type iff the members of the sets C and C* can be put
into a one-to-one correspondence in such a way that each member e of C
is paired with a member of C* that has the same intrinsic phenomenal
character as e.) Although I find Weak C-holism implausible, it may be
worthwhile to show that it *does not* stand in the way of the
split-brain argument. To see this, just suppose that the set of
experiences supported by my left hemisphere throughout T in w is of the
same context type as the set of experiences supported by my right
hemisphere throughout T in w, and that each of these sets is of the
same context type as the set of experiences supported by my entire
(connected) brain throughout T in some world w*.
More specifically. Let e_a be a particular auditory experience, and let
e_v be a particular visual experience. Let e_a* be a particular auditory
experience that has the same intrinsic phenomenal character as e_a but
that is numerically distinct from e_a. Let e_v* be a particular visual
experience that is numerically distinct from e_v but that has the same
intrinsic phenomenal character as e_v. Let "I(e)" mean: e is
introspected. Assume any two co-conscious experiences e and e* have a
unique fusion that is an experience, and let "e+e*" refer to the fusion
of e and e*. Then world w can be represented as follows:
My left stream: I(e_a*+e_v) My right stream: I(e_a+e_v*)
And here is world w*:
My single unified stream: I(e_a+e_v)
Here's the basic idea: In world w, where my hemispheres are split, e_a is
co-conscious with and jointly introspected with e_v*, an intrinsic
duplicate of e_v; and e_v is co-conscious with and jointly introspected
with e_a*, an intrinsic duplicate of e_a. So my right and left streams are
qualitatively exactly alike in w. Moreover, each of these two streams is
qualitatively exactly similar to the single stream that I have in w*,
where e_a and e_v are co-conscious and jointly introspected.
Assuming the possibility of this set-up, it seems to follow that one can
accept both the split-brain argument and Weak C-holism.
<11>I would like to thank Barry Dainton for his helpful comments. A
version of this paper was written with support from a Princeton
University Graduate Fellowship, which I gratefully acknowledge.
REFERENCES
Audi, R., ed. (1995). *The Cambridge Dictionary of Philosophy*.
Cambridge: Cambridge University Press.
Bayne, T. (2001). "Review of Barry Dainton, *Stream of Consciousness*"
*Journal of Consciousness Studies, 8,* 79-92.
Block, N., Flanagan, O., and Guezeldere, G., eds. (1997). *The Nature
of Consciousness: Philosophical and Scientific Debates*.
Cambridge, MA: MIT Press.
Dainton, B. (2000). *Stream of Consciousness: Unity and Continuity in
Conscious Experience*. London: Routledge.
Hill, C. (1991). *Sensations: A Defense of Type Materialism*.
Cambridge University Press: Cambridge.
Parfit, D. (1984). *Reasons and Persons*. Oxford: Oxford University
Press.