Self-Consciousness and Psychological Subjects: I

advertisement
Self-Consciousness and Psychological Subjects:
I-thoughts and I-thinkers in the Split-brain Case
You think you’re yourself, but there are other persons in you.
Barth, 1968, “Lost in the funhouse”
Abstract: The paper concerns self-consciousness in split-brain subjects. I argue that even if a split-brain
subject has two minds, there is something about the operation of self-consciousness in split-brain subjects
that acts to make each such subject more like one of us than like, say, a pair of conjoined twins. Two distinct
accounts of how split-brain self-consciousness might play this role are contrasted. According to the latter
account, endorsed here, although there may be two “I-thinkers” in the split-brain subject, they don’t think
about each other in the way that two persons do.
1
Introduction: Individuating Individuals
Under experimental conditions, split-brain subjects often behave in ways eerily suggestive of two conscious
minds in one body. Call this impression the duality intuition. A split-brain subject nonetheless like no more
than a single person. Call this the unity intuition. These two intuitions are generally taken to be in some tension
with each other. That the tension is generally viewed as a serious one is reflected in how rarely philosophers
have attempted to reconcile the duality intuition and the unity intuition. Philosophers have very rarely
defended a reconciliation model of the split-brain phenomenon, that is, in which a split-brain subject is one
person but with two minds. Philosophers have instead generally argued for some or other version of an
illusion model, according to which either apparent duality of mind or apparent unity of the split-brain person is
an illusion.
Here is one reason why a reconciliation model might look like a non-starter. Suppose that persons
are what we think about when we think about ourselves. Characteristically human minds or thinkers are selfconscious: they have the capacity to think about themselves. If this is correct, though, then it follows either that
right and left hemisphere are associated with distinct minds and thus distinct persons, or that a split-brain
subject is a single person and thus has a single mind.
At least one philosopher, however, has argued that self-consciousness is not the major obstacle to
but in fact the basis of reconciling unity of the split-brain person with duality of the split-brain mind.
According to Davis, the “disconnected” right and left hemispheres are associated with distinct thinkers of
self-conscious thoughts or “I-thoughts,” and yet all of these thoughts are about the split-brain subject as a
whole—relative to whom the two hemisphere systems are mere subpersonal parts, and not in fact genuinely
self-conscious at all.
This paper defends a distinct view. I will argue that regardless of the referent/s of I-thoughts in the
split-brain subject, there is something about the operation of self-consciousness in split-brain subjects that acts
to make each such subject more like any one of us than like a pair of conjoined twins.
1
The paper will proceed as follows. In the next section I will explain the conceptual and empirical
basis of the claim that a split-brain subject has two minds and streams of consciousness, so that there are
indeed two distinct thinkers of I-thoughts in the split-brain subject, whom I call “R” and “L”. Section 3 will
look at how “I-statements” are used in split-brain subjects, taking these to reflect the operation of I-thoughts in
such subjects as well. I will consider some possible arguments for the claim that only the split-brain subject as
a whole, S, is the referent of R’s and L’s I-thoughts, and explain the central difficulties that I think all such
arguments will face. In Section 4 I will argue that even if we had a compelling argument for the claim that S is
the referent of R’s and L’s I-thoughts, the argument would face some really serious challenges. Ultimately I
will suggest that I-thoughts in the split-brain subject may have no single, stable referent.
Section 5 will turn from questions about the referent/s of I-thoughts in the split-brain subject to
questions about self-conception in the split-brain subject. I will argue that neither hemisphere system recognizes
the existence of a second self-conscious thinker—other than the one with which each identifies—also sharing
the subject’s body. I will then explain the role that lack of mutual recognition plays in making a split-brain
subject like one of us, in an interesting psychological sense.
2
Split-Brain Psychology
This section deals with the reasons for identifying the two hemisphere-systems with distinct thinkers and the
reasons for identifying them with distinct thinkers of apparently self-conscious thoughts.
2.1
Psychological subjects
In brief, the conceptual and empirical basis of the “mental duality claim” for split-brain subjects is simply that
mental appear to interact with each other in ways most characteristic of the kinds of mental states they are
only within one hemisphere system or the other, rather than within the split-brain subject as a whole.
Suppose you blindfold a split-brain subject and place a familiar object, such as a pipe, in his left hand,
from which only the right hemisphere receives fine-grained tactile information. Via his verbal left hemisphere,
the subject says he does not know what he’s holding. A minute later, however, using the same left hand,
which is under dominant (though not exclusive) motor control of the right hemisphere, the subject easily
selects the pipe from a box of objects. In this case, the process of fine-grained sensation or perception, as
well as recognition, recollection, and selection, and so on, seem to occur most clearly within a system that
excludes the left hemisphere. This is to say that, within a system that does include the left hemisphere, the
processes don’t clearly look like fine-grained perception, recognition, recollection, or selection at all.
Take a simple example. Early experiments performed with split-brain non-human primates showed
that the hemispheres could learn to make reverse perceptual discriminations with no interference (Sperry,
1958). This finding thus disconfirms a pretty basic generalization about learning—unless such generalizations
do hold in the split-brain subject—but only relative to systems that don’t include both hemispheres.
2
Take another experiment (Sperry, 1968) in which different familiar symbols were visually presented
to the two hemispheres at once, a question mark to the left hemisphere (right visual field) and a dollar sign to
the right hemisphere (left visual field). The subject is asked to draw what he’s seen using his left hand (again,
predominantly controlled by the right hemisphere). When asked, the subject says that he’s drawn a question
mark, the left hemisphere stimulus—when in fact he has just drawn a dollar sign, the right hemisphere
stimulus. The kinds of causal connections that ordinarily hold between perception, recognition, action, and
self-ascription of action, do not appear to hold within the split-brain subject as a whole, but only within two
more circumscribed parts of the split-brain animal, neither of which includes both hemispheres.
In a more complex experiment, different halves of a compound word (e.g., “hot” and “dog”) were
presented on each trial to the two hemispheres, one in each visual field. Then the subject was asked to draw
the referent of the word he’d seen using the left hand (again under dominant but not exclusive control of the
right hemisphere). Interestingly, while the subjects sometimes created a drawing that combined the referents
of both halves of the compound words, they never drew the single referent of the compound words. In this
case, perceptual experiences of both words did in a sense interact—they combined in the drawing of some of
the pictures—but they did so quite indirectly, via a chain of physical events involving or including overt
behavior (and observation of that behavior). We ordinarily expect, however, that conscious experiences
should be able to interact more directly than this. And they do interact more directly than this even in the
split-brain subject—just not across the two hemispheres. For example, when the two halves of the compound
word were both presented to the left hemisphere, the subject (usually) drew the single referent of the
compound word (Kingstone and Gazzaniga, 1995).
Is it possible that the visual percepts of the two halves of each compound word were not truly
conscious? After all they were not integrated interhemispherically in the way that conscious experiences can
characteristically be integrated. On the other hand, one can’t normally draw the referent of a word one didn’t
perceive consciously. So either, e.g., right hemisphere percepts were not conscious, and we must amend our
understanding of non-conscious processing and capacities generally, or right hemisphere percepts were
conscious—but only relative to a system that excludes the left hemisphere.
I have been speaking throughout of a right hemisphere-associated system and a left hemisphereassociated system. Call these systems R and L, respectively, where R is the whole subject minus the left
hemisphere (S – LH) and L is the whole subject minus the right hemisphere (S – RH). Note that, at least
relative to mere hemispheres alone, R and L have a good deal in common with whole animals. Indeed, there
is a sense in which, following e.g. left hemispherectomy, R is all that is left—and although hemispherectomy
(particularly left hemispherectomy in adulthood following basically normal development) is a devastating
procedure, “No one thinks that people with either right or left hemispherectomies lack minds comfortably
characterizable as human” (Marks, 1981, p. 47 fn 18).
3
The upshot of the foregoing is that the conceptual basis of the mental and conscious duality claims is
that mental processes like seeing, feeling, recognizing, and learning seem to go on or within L and within R
but not “across” them, within S as a whole. Psychological generalizations thus hold and mental kinds are thus
defined within R and within L. And this explains how difficult it is to describe a split-brain subjects’ behavior,
in familiar psychological terms, without speaking of what the right hemisphere (or R) thinks and what the left
hemisphere (or L) does (Bogen, 1977): because acts of thinking (feeling, recognizing, remembering) are as it
were relativized to one of these systems or the other. But this just makes it R and L, rather than S, that stand
in the thinking (feeling, recognizing, remembering) relation to mental states, the relation that we ordinarily
think whole animals stand in to those states. After all, ordinarily we can identify a particular act of e.g.
remembering by attributing it to an animal or person—we don’t need to say anything about the particular lobe
relative to which it’s remembered.
In the remainder of this paper I will use the term “thinker” to refer to that entity within which
mental states act like the kinds of mental state that they are, and which thus stands in the individuating relation
to particular (token) mental states, events, and property instantiations. For the most part I will in what follows
take myself to have adequately (albeit briefly) presented the case for identifying R and L with distinct thinkers.
I don’t begin by assuming that R and L are persons standing in all the relations that we think persons stand in
to mental states, however. The paper in fact centrally concerns whether there is an important relation to at
least one kind of mental state—a self-conscious mental state—that R and L don’t stand in.
2.2
I-thinkers and I-thoughts
Davis (1997) assumes that I-thoughts in the split-brain subject refer either to R and to L on the one hand or
to S as a whole. I mean also mean to leave open the possibility that I-thoughts refer to entities constituted by L,
R, or S, and even that I-thoughts in fact refer to merely intentional entities. Readers, however, may decide
that my arguments work better against the background of one of these very general metaphysical positions on
persons or selves or the other.
The previous section presented the case for identifying R and L as distinct thinkers; this section
presents the case for identifying R and L as distinct I-thinkers. One might preemptively object, however, that
whether R and L are thinkers of I-thoughts hinges on whether or not these thoughts are really about R and L—
if not, then these thoughts are not genuine I-thoughts at all. I will assume going forward that we have some
preliminary way of identifying I-thoughts that doesn’t presuppose knowing their referents. Perhaps Ithoughts are those intended to refer to their thinkers, for instance. In what follows, when I refer to R and L as
thinkers of I-thoughts, I do not mean to presuppose the referent of those thoughts. I will sometimes refer to
what e.g. L thinks of himself*, or what R thinks that she* is doing, where the astrix functions to signify, again,
lack of prior commitment about the actual referent of R’s and L’s self*-conscious thoughts.
What reason is there to think that R and L are not just distinct thinkers, but thinkers of I-thoughts in
particular? First note that in fact no one doubts that the left hemisphere is associated with an I-thinker. After
4
all, L makes plenty of I-statements, and we just assume (and I will assume here myself) that these statements
reflect L’s I-thoughts as well. As usual, skepticism arises only with respect to poor silent R, who cannot make
I-statements.
It nonetheless looks very likely that R, too, is a thinker of I-thoughts. Consider a study by Schiffer et
al. (1997) designed to test self-consciousness in the two hemispheres (or in R and L). Each (split-brain)
subject in this experiment was seated with his hands before him but obscured from his view (so as to avoid
giving to R and L visual feedback about his manual responses). A block, with five pegs rising out of it, was
placed before each of his two hands. Subjects were then trained to use the pegs to give quantitative answers
to questions, tapping the rightmost peg on each block to signify “extreme” and the leftmost peg to signify
“none”, and so on. The two subjects in the experiment were then asked a series of questions asking for their
thoughts and feelings about various topics. Some of these topics were personal or emotional (e.g., subjects
were asked about a childhood experience of having been bullied, and about how they saw themselves, how
often they felt sad or lonely, etc.) and others were general or impersonal (e.g., subjects were asked whether
they believed in God, and how they felt about sex, work, the President, and so on). Actually, on many trials R
and L were asked different questions (so that L might receive the question “How do you feel about work?”
while R received the question “How do you feel about God?”). For our purposes the relevant findings were,
first, that left handed responses to the more personal and emotional questions appeared to be the responses
of R in particular (as they conflicted with both right handed and verbal responses), and that these responses
were intelligible and systematic rather than random; that is, R appeared to have a self-image that was
internally consistent but that differed from the self-picture offered by L. Many of the questions asked of the
split-brain subjects (or R and L) in this experiment were of a kind that we would consider to be
straightforwardly introspective—questions about the subjects’ (or R’s and L’s) thoughts and feelings,
including their thoughts and feelings about themselves.*
Some of R’s emotional responses also seem to suggest self-consciousness. For example, when Sperry,
Zaidel, and Zaidel (1979) presented to the right hemisphere (to R) of subject N.G., in the midst of a series of
visual recognition tasks, four pictures of herself, N.G. exclaimed (or, R exclaimed?) “Oh no! . . . Where’d you
g. . .” She then laughed loudly and exclaimed again, “Oh God!” Although she (or L) was unable to say what
she was looking at—the authors hypothesize that the initial exclamation may have come from R in the
moment before L regained control of speech—her (or R’s) immediate emotional reaction not only showed
that R obviously recognized the face, but that she recognized it as her own: she (or R) responded with the
kind of shyness or embarrassment that many people would experience in such a situation. My own view,
then, is that philosophers and neuropsychologists who have attributed thought and consciousness to the right
hemisphere have erred in denying it self-consciousness (Eccles, 1981; Popper and Eccles, 1977; DeWitt, 1975).
Of course, R is not strictly self-conscious unless its I-thoughts are about R. Davis (1997), for one,
considers the question of self-consciousness in R and L to hinge totally upon whether or not it is R and L or
5
instead S that is the referent of R’s and L’s (respective) I-thoughts. At a minimum, though, I believe that R at
least tries to self-refer, by thinking I-thoughts. So far that is all we know L—or any of us!—can do, anyway.
2.3
I-thoughts and I-statements
The paper is ultimately interested not in I-statements, but in I-thoughts—thoughts in whose content the
mental analogue of the word “I” appears. (Although we might use the symbol “Im” to refer to this mental or
conceptual analogue, I’ll just use the word “I” again, for readability.) Since thoughts are private, however, the
best evidence we have for how I-thoughts work in split-brain subjects is, of course, uses of the word “I” by
split-brain subjects. I am therefore forced to assume that whatever the referents of I-statements in split-brain
subjects, I-thoughts have the same referents. This is already a substantive assumption.
Worse yet, looking at a split-brain subject’s I-statements in practice amounts to looking at the Istatements of L in particular; while R can speak in some split-brain subjects, R doesn’t speak fluidly, in whole
I-statements. I take this seriously as an evidential problem, but I don’t know how to proceed except to ignore
it. I am thus forced to assume that L’s I-statements reflect as much about I-thoughts and self*-consciousness
in R as they do about I-thoughts and self*-consciousness in L.1
Even with these two simplifying assumptions, the data give us much to grapple with. The split-brain
literature cites and describes numerous instances in which L appears to claim responsibility for a response
that (it is obvious to the experimenter) has its origins in the RH. That is, in these cases L strongly identifies with
the agent responsible for the action (or expression of emotion, and so on): L not only uses the word “I” to
refer to this agent, but offers (confabulates) a reason for having done whatever R in fact did. In what is
probably the most commonly cited case, different images were shown to the split-brain subject P.S.’s two
hemispheres, and then P.S. was asked to point to one of a number of possible images that went with the one
he’d seen. On one trial, an image of a snow scene was presented to the RH (left visual field), an image of a
chicken’s foot to the LH (right visual field). With his left hand (RH) the subject pointed to a shovel, with his
right hand (LH), to a chicken. He did this in free vision, so that both R and L could see his choices. When
asked, “What did you see?” P.S. replied, "I saw a claw and I picked the chicken, and you have to clean out the
chicken shed with a shovel" (Gazzaniga 1983, 534). Rather than saying, “I saw a chicken’s foot, so God
knows why I pointed to a shovel” or even “I saw a chicken’s foot and so pointed to the chicken—and I
didn’t (or didn’t mean to) point to the shovel at all”, L invents some reason why he* (or, the person with
Puccetti (1993) once suggested that R might have a very different, actually more accurate sense of self than L. The
suggestion that R is more self-aware than is L runs counter to the usual order of things in the split-brain case—but then
again many authors have noted that anosognosia is more common following RH than LH damage, and have postulated
hemispheric differences of various sorts to account for the LH’s greater propensity to self-blindness (e.g. Ramachandran,
1995). On the other hand there is more recent work questioning whether the RH is really less inclined to self-blindness
than the LH, or whether anosognosia is in fact equally likely to follow LH as RH damage, but is just harder to test in
LH-damaged patients because of their concomitant linguistic impairments (Cocchini et al., 2009).
1
6
whom L identifies) might have pointed to a shovel, consistent with what L knows he* saw. Again, I will say
then that in this instance L has appropriated R’s response. 2
In other instances, rather than confabulating an explanation for why he* has done this or that, L will
say something like “I don’t know why I did it” or “I just guessed” or “Well, I must have done it
unconsciously” (Sperry, 1968). Still, even in such instances, L appropriates the response L makes no attempt
to rationalize.
Such instances of appropriation could be interpreted in two different ways. If there is an accepted
interpretation of instances of what I’m calling appropriation in the split-brain subject, it is that in such
instances L is speaking, incorrectly, about what L has just done—incorrectly because it is R who did it. A
distinct interpretation, however, is that L is speaking, correctly, about what S has done… Even if L is wrong
about why S has done it. That is, when L appropriates a response of R, subjectively identifying with the
person who responded in whatever way, L is not wrong in his or its subjective identification, although L may
often provide an inadequate or inaccurate explanation of the response—an error of the sort we can readily
understand in light of the nature of split-brain subjects’ neurological abnormality.3
We should note however that there are a number of instances in which the more natural
interpretation of L’s I-statements is that L is making a correct statement about L, as when L says, “I don’t
know the answer” or “I can’t feel my left hand” (Mark, 1996) when R clearly does, or “I can’t do this task”
(Levy, 1969) though as it turns out R can. Such instances could of course still be interpreted as ones in which L
is speaking for S, and is simply wrong about what S can feel or do. This, presumably, has to be Davis’s
interpretation, in order to defend the S-as-I-referent claim.
This way of interpreting such instances is not totally appealing, however. I don’t think it’s the
principle of charity—entirely or even primarily—that makes it more natural to interpret L as making a correct
statement about L. It’s that we know a little bit about the causal origin of such a statement; we know that the
psychological processes that lead L to say, “I can’t feel my left hand” are apparently the same as those that
would lead me to say such a thing. This provides at least some reason to think that L’s I-thoughts are, at least
in such instances, part-referring—that is, that they refer to L—rather than whole-referring—that is, referring to S.
Still, such reasons might be outweighed by others. Since the data are, on the surface of things, mixed,
is there a compelling case to be made for interpreting all of L’s I-statements and thus all of R’s and L’s Ithoughts as referring to S?
It actually does seem potentially important that L did not explicitly say, “I pointed to a shovel because”, but instead just
offers a general rationalization for pointing to a shovel after having seen a chicken. I think it is unclear, that is, that L (as
I will say) strongly identifies with R’s response here. But I will set this aside to follow the usual interpretation of this
experimental result, which is that P.S. has confabulated a reason why he (or, L has confabulated a reason why he*)
pointed to the shovel.
3 Logically, I guess, there is a third option, which is that L is using “I” to refer to R. But the only two alternatives I
consider are that L speaks for L, or that L speaks for S—something that includes R but that includes L as well.
2
7
3
Making the Positive Case for Co-Self-Reference
The first thought one might have is that L’s I-statements should be interpreted consistently with the principle
of charity. Of course, this assumes a theory of meaning (and of mental state attribution and psychological
explanation) that many people, myself included, would reject. Because the data are varied, however, even a
friend of the principle of charity would have difficulty applying it in the split-brain case in order to argue for
S-as-I-referent. For while it is true that many of L’s I-statements will be true only if they refer to an entity that
includes the RH, others will be true only if they refer to L alone, or to something that excludes the RH. Now,
some errors are more serious than others, and I think it is intuitive to worry that if we interpret L’s Istatements as being about L then we will have to understand L as making errors that are more serious than
those L is making if L is speaking about S. If L is speaking about S, L will often be wrong about why S has
done what he’s done, or even what S has or hasn’t done. If L is speaking about L, however, then L will often
be in error about who has done something, since L often appropriates responses that have their origins in the
RH. This is a natural thought. But why? What makes the second error more serious? To see this, we have to
move beyond the principle of charity.
Fortunately, there are possible arguments not from charity but from causal role for the S-as-Ireferent claim. Perhaps the simplest or at least the most obvious argument for the claim that S is the referent
of R’s and L’s I-thoughts—I am calling this the S-as-I-referent claim—would appeal directly to R’s and L’s
sharing a body. Note that R and L share a body individuated not just in terms of e.g. spatial boundedness but,
largely anyway, functionally. That is, R and L bear, to one and the same purely physically individuated body, the
special causal relations that each of us bears to her own body alone.
We might appeal here to Shoemaker’s notion of sensory-and-volitional embodiment. According to
Shoemaker, “a person is ‘sensorily embodied’ in a certain body to the extent that the interactions of that body
with its surroundings produce in the person veridical perceptions of aspects of those surroundings”
(Shoemaker 1976, p. 112). Note that a person is sensorily embodied in a certain body also, crucially, in the
sense that events of or in that body tend to produce veridal sensations (and percepts) of that body.
Meanwhile, “a person is ‘volitionally embodied’ to the extent that volitions of the person produce in that
body movements that conform to them or fulfill them” (Shoemaker 1976, p. 112). R and L are largely, as I
will say, sensorily-and-volitionally (SV) co-embodied.
SV co-embodiment has at least the following significance: there are special psychological relations that a
person ordinarily stands in to only her own body, that R and L stand in with respect to each others’ body as well.
We can very broadly sort these into the categories of upstream and downstream causal relations. Davis’s
argument for S-as-I-referent focuses on the upstream causes of I-thoughts. He writes (using the letter “W”,
rather than “S”, to refer to the split-brain human animal as a whole):
8
In an ordinary human being, beliefs of this kind [“I am F”, where “F” refers to
some kind of occurrent action] arise from the operation of sense organs,
introspective processes, and other mechanisms of self-awareness. These organs,
processes, and mechanisms…. will operate to give rise to belief expressible as “I am
F” only when that human being is in fact F. In W [that is, S], “I” plays this same role.
Whereas in R and L, “I” plays a different role. For example, certain beliefs of the
form “I am F” regularly arise in L when it is the case that R, and not L, is F. So as
regards this kind of belief, “I” plays the role of a self-referring expression referring
to W [S]; it does not (quite) play the role of such an expression referring to L or to
R.
(Davis, 1997: 213)
Although I think Davis’s argument here is actually slightly unclear, let us suppose that he means to
refer to the fact that some sensory-perceptual processes are as I will say proprietary: they are mechanisms of
self-knowledge par excellence, since their function, in an obvious sense, is to tell one about oneself; they do not tell
one about anyone else (or, not via the same route, not as “directly”). These proprietary sensory-perceptual
processes yield I-thoughts that Shoemaker would say (1968) are immune to error through misidentification. Sensoryperceptual processes that we ordinarily think of as proprietary are not proprietary in R or L, however;
proprioception, for instance, tells L no more about the position of L’s left arm than about the position of R’s,
since they in fact share a left arm. So perhaps I-thoughts in the split-brain subject—at least to the extent that
they are formed on the basis of proprietary processes—refer to S because proprietary processes like
proprioception are proprietary only within S, and not within something “smaller” than S, it is S that these
processes inform R and L about.4
Interestingly, in his argument for S-as-I-referent, Davis appeals only to the upstream causes of Ithoughts, especially I-beliefs, and not to the downstream effects of I-thoughts and, especially, I-intentions.
Citing Perry (1979), Davis concedes that “the connection with action” is also an essential part of the role of Ithoughts, but says that “on this front, it appears that L and R would qualify as well as W [S] as the referent of
‘I’” (Davis, 1997, p. 213. That is, I suppose, that there is nothing in the downstream effects of I-thoughts on
action, in split-brain subjects, to distinguish between those thoughts being part- or whole-referring. Supposing
that this is right, however, one might think that there is in fact an argument for S-as-I-referent, just here, in
this very ambiguity. Let us say that some downstream effects of I-thoughts are privileged: one’s intentions act
I’ve cited Shoemaker (1968) here because I think there’s an important connection between “proprietary” states as I’ve
defined them and the principle of immunity to error through misidentification, but I admit that the connection is not
totally clear. Both visual and pain experiences are introspectible, and self-attributions of both pain and visual experiences
are immune to error through misidentification (well, they are equally immune; perhaps neither are totally immune). But if
pain experiences tell you about anyone’s body, they must tell you about your own, whereas this is not true for vision.
4
9
not (typically) on but through one’s body and not anyone else’s. Once again however this is not true for R or
for L: only within S as a whole are many of these downstream effects privileged, just as only within S as a
whole are (many of) their upstream causes proprietary.
Of course, some upstream causes of I-thoughts in the split-brain subject will still be proprietary within
R and within L, and so too will some of their downstream effects be privileged. Essentially, and again
following Shoemaker, we can distinguish between psycho-psycho and psycho-physical connections between
mental states, in this case I-thoughts. Psycho-psycho causal connections involve direct interactions between
mental states. Psycho-physical causal connections involve interactions between stimuli and mental states or
between mental states and movements (behaviors or responses). R and L have distinct minds because (or if)
there are psycho-psycho connections that characterize the workings of a single mind, and these connections
hold within R and within L, but not across them, within S as a whole. R and L are SV co-embodied, however,
because the psycho-physical causal connections that typify (that may literally type-define) I-thoughts hold
within S as a whole: they hold between R’s and L’s I-thoughts indirectly, via the mediation of nonpsychological, physical events, because R and L are co-embodied. The suggestion is that because some
ordinarily proprietary causes and privileged effects of I-thoughts are not so within R or within L, but only
within S as a whole, these I-thoughts refer to S.
There are two obvious and related problems with these arguments for S-as-I-referent, however. First
of all, why not just think that R and L are two strictly self-referring subjects who in this usual case are coembodied, and for whom, then, ordinarily proprietary causes and privileged effects of I-thoughts are not
wholly proprietary or privileged? After all—and this is the second problem—with respect to the psychopsycho connections that typify I-thoughts, though, these must hold largely intrahemispherically—or anyway,
they must fail to hold interhemispherically—or else R and L wouldn’t be distinct thinkers of I-thoughts in the
first place. So even if S is a single network, where psycho-physical relations are concerned, R and L are
distinct causal networks where their (or, the split-brain subject’s) psycho-psycho relations are concerned.
Which network should be identified as the referent of any particular I-thought? Which matters more to
determining the referent of such a thought—its psycho-physical relations or its psycho-psycho relations? It
would be a fun and not particularly difficult exercise to construct arguments for one position or the other, but
it’s hard to see what general grounds there could be for finding any such argument decisive.
There is a further difficulty that any argument for the referents of I-thoughts in split-brain subjects
will face. This difficulty is rooted in the fact that, any way you slice it, the split-brain phenomenon will be
characterized in part by failures of self-knowledge: that is, regardless of your account of mind and
personhood (and self-reference) in the split-brain subject, an unusual number of I-statements (and
presumably I-thoughts) in the split-brain subject will be false.
The best evidence we have for the referents of I-thoughts in split-brain subjects are I-statements in
such subjects. But of course these, too, must be interpreted: with I-statements no less than with I-thoughts
10
we must determine whom it is that L is referring to. We could do this without much trouble if we had some
independent grounds for knowing which of these I-statements were true and which false. But of course their
truth or falsity depends crucially upon their referents. But that is what we are trying to determine in the first
place. If there is no way to break out of this circle then it is unclear how to proceed. Note, or instance, that
Davis’s argument for S-as-I-referent implicitly appeals to the veridicality of I-thoughts: “These organs,
processes, and mechanisms…. will operate to give rise to belief expressible as ‘I am F’ only when that human
being is in fact F” (Davis, 1997: 213; emphasis added). But, again why not just conclude that L’s I-thoughts are
often wrong? In fact, again, any way you look at it, many of them must be wrong.
Well, actually, this is not quite right. Many of L’s I-thoughts must be wrong only if all of those Ithoughts have the same referent. In light of the somewhat messy variety of ways in which I-statements are
used in the split-brain subject, however, one possible conclusion is that such statements (and, by assumption,
I-thoughts as well) have no stable referent or perhaps no determinate referent. (These are of course distinct
possibilities, but I will refer to both of them as the unstable referent view.) This is actually an appealing
proposal for at least two reasons, one related to the split-brain phenomenon in particular and one more
general. First, R and L substantially physically overlap, meaning that the sequence of mental events leading up
to, and the kind of machinery contributing to the formation of, I-thoughts, will substantially causally overlap.
(I will refer to the upstream processes leading to I-thoughts.) Just as the physical overlap between R and L is not
total, however, neither is the causal overlap in the processes leading to I-thoughts. Similarly, some I-thoughts
will contribute to downstream processes that are more bi-hemispheric, and some to downstream processes that
are more uni-hemispheric. So perhaps when an I-thought is formed on the basis of upstream processes that
overlap, to a greater extent, or when an I-thought contributes to downstream processes that are more bihemispheric, that I-thought refers to S; perhaps when an I-thought is formed on the basis of upstream
processes that are unshared, or when the thought contributes to downstream processes that are more unihemispheric, the thought is part-referring.
A second potentially appealing feature of the unstable referent view is just that the opposing, “stable
referent” view seems to presuppose that there’s a unity that pre-exists the formation of I-thoughts, a self that
exists prior to one’s attempt or intention to self-refer, and that is therefore unaffected by or independent of
the nature of such attempts. Now of course there are plenty of things, like animals, whose unity doesn’t
depend upon any act of self-reference.5 It’s not clear, though, that that’s what we’re thinking about when we
try to think about ourselves. Indeed on the face of it people do seem to use “I” sometimes to refer to animals
and sometimes to, e.g., immortal souls. (That’s just an interpretation, and a controversial one; I just mean to
say that it wouldn’t be hard to make a prima facie case for it.) In other words, it’s actually not clear that I-
Actually I just assume, throughout, that animals are easy to individuate and that their unity isn’t grounded in
psychological features. But of course there are difficult questions about the identity of animals as well.
5
11
statements have a single stable and determinate referent even in the ordinary case, and, among other things,
the referent of a particular use of “I” may depend on the speaker’s intentions.
Now of course many people don’t believe that “I” is a demonstrative whose referent depends on
one’s intentions at all, but rather than indexical that can only refer to its thinker. In fact this is presumably the
commonsense view. But that view just assumes that it is trivial to identify the speaker (utterer) of a token of
“I”—and this is just what the split-brain case calls into question.
4
Objections to S as I-Referent
Even if we had a really compelling argument for the S-as-I-referent claim, the claim might still prove
untenable. This section considers three challenges to the claim that although R and L are the thinkers of Ithoughts in split-brain subjects, S is the referent of those thoughts. Although I conclude that the first two do
not ultimately pose an insurmountable problem, the last really may.
4.1
The objection from inference
The first objection to the S-as-I-referent claim is that, if R and L are distinct thinkers of I-thoughts, then their
I-thoughts cannot co-refer, because they cannot interact with each other directly in thought or consciousness
as co-referring thoughts can.
Suppose that I think, “I am wearing socks” and “I am not wearing shoes”. Having these two Ithoughts puts me in a position to conclude “I am wearing socks but [and] no shoes.” But if L thinks “I am
wearing socks” and R thinks “I am not wearing shoes”, neither L nor R (nor of course S) will on the basis of
those I-thoughts be put in a position to conclude, “I am wearing socks but no shoes.”
We might begin by conceding at least a part of the objection: R’s and L’s I-thoughts won’t play exactly
the role that mine do in me—or, rather, that they will play this role only if they refer to L and to R
individually rather than to S. But the question is whether that difference in actual role is somehow essential to
the referent of I-thoughts.
There are reasons to think that it isn’t. Note, first, that thought contents in general can’t be (directly)
conjoined in split-brain subjects across the hemispheres. If L thinks, “Socks are cheap” and R thinks, “Shoes
are expensive”, neither is thereby put in a position to conclude “Socks are cheap but [and] shoes are
expensive.” Or suppose L thinks, about the subject’s father, “He’s getting older,” while R thinks, about the
subject’s father, “He looks healthy.” Again, neither will be in a position to think, “He’s getting older but [and]
he looks healthy.” Yet we still think that these thoughts refer to a single person, the subject’s father. This is
because the thoughts bear the right causal relationship to what is in fact a single person, the subject’s father.
12
But now we are supposing that we have grounds for having accepted the S-as-I-referent claim—so by
hypothesis the same is true here, as well.6
Assuming that I am a single person, and a single thinker or psychological subject, that is identical
with (or constituted by) a whole animal, then if R and L are distinct psychological subjects, their I-thoughts
are not going to play quite the role mine do in me, whether their referents are R or L or S. Callosotomy just
seems to come with some kind of disruption in self-conscious thought. In this case, the question is whether
the disruptions are of such a seriousness or kind that R’s and L’s thoughts couldn’t possibly refer to S. I
myself am inclined to think that they aren’t—that unity of the person and indeed co-reference to that person
in self-conscious thought can survive a good deal of psychological disintegration and fragmentation. In that
case, certain very banal inferences won’t be capable of being drawn on the basis of I-thoughts in the splitbrain subject—but this is because there are two drawers of inference in the split-brain subject.
4.2
Objection from intrinsic structure
According to one prominent way of thinking, the ability to self-refer is necessary for self-consciousness
(Strawson, 1959 and 1966; Evans, 1982; see Smith, 2004, for discussion). Accordingly, if R and L do not
strictly self-refer via their I-thoughts, then R and L are not self-conscious. This is just Davis’s conclusion:
because R and L do not strictly self-refer, they are not self-conscious; because they are not self-conscious,
they are not persons.
One might object, however, that if this is right, then the capacity for self-consciousness turns out not
to wholly depend upon one’s intrinsic structure and features. For, one might say, L is all that remains after
right hemispherectomy, and R is all that remains after left hemispherectomy. And no one denies that
hemispherectomied people are self-conscious (see for instance the case discussed in Smith, 1966). Why
should the mere presence of a functioning right hemisphere, for example, make the difference between L’s
achieving and failing to achieve self-consciousness? And, if self-consciousness is required for personhood,
between L’s being and not being a person?
This objection is, of course, only an objection on the assumption or intuition that an entity’s capacity
for self-consciousness, and its meeting or failing to meet the criteria for personhood, should supervene on
that entity—should be fully determined by its intrinsic or non-relational entities. I want to make two points
about this objection. The first point is that the objection may equivocate on the meaning of “selfconsciousness.” On one understanding self-consciousness does and on another understanding it does not
reduce to successful explicit self-reference—where self-reference is itself taken to be an objective matter. If
we decline to understand self-consciousness wholly in terms of self-reference, then we can allow that R and L
Suppose persons (or, the referents of I-thoughts) don’t in fact exist. Then it might seem that, because of
interhemispheric inferential isolation of I-thoughts, R’s and L’s I-thoughts cannot co-refer. But in fact even this is unclear.
The referent of “I” would be determined by the role it played within a system, in that case. But the relevant system
might still be S, the whole animal—even if the particular thoughts in whose contents the concept “I” figured, got their
identity as thoughts from the role they played in R and in L individually.
6
13
have the capacity for and even are self-conscious, whether or not their I-thoughts refer to them or to S.
Essentially, any being that was capable of trying to refer to the thinker of its thoughts would be self-conscous.
If, on the other hand, self-conscious thinking is a success verb, then perhaps it should not be surprising that
whether an entity achieves self-consciousness turns out to depend upon facts beyond itself and its intrinsic
capacities, for this will be true of successful reference generally.
Of course the obvious difference is that in the case of self-reference, one is referring to oneself—so of
course we assume that successful self-reference supervenes on facts about oneself. But precisely what we are
considering is the truth of the claim that the referent of an I-thought is strictly identical with its thinker. Any
compelling argument for S-as-I-referent is an argument against that claim. If the reflexivity claim is false it will
not be surprising that whether a subject strictly self-refers (and thus is strictly self-conscious) turns out to
depend on more than just facts about that subject. For we are imagining a case in which the self the subject
either succeeds or fails to refer to is not strictly identical with the subject.
Really I don’t think it would be so surprising if the capacity to strictly self-refer turned out to be a
relational capacity in some sense, or in some odd cases. We might well be designed or constructed—and I
refer here to the design of the whole nervous system and body but also to the architecture of our conceptual
schema—to think of whole animals when we try to think about ourselves. And whether L is an animal depends
upon the presence or absence of the right hemisphere. Or perhaps even if we are not built to identify with
animals per se, we are built to identify with the “only one home.”7
4.3
The objection from error
The third objection to the claim that R and L are the thinkers of I-thoughts that refer to S is that if this is so,
then all psychological self-attributions in the split-brain subject are false. For any thought with the content [I
think X], the “I’ refers to S, but S isn’t a thinker at all. (It is possible to extend the objection to selfattributions of intentional action as well, but I’ll focus on self-attributions of mental states in what follows.)
Call this the objection from error.
The most obvious response to the objection from error is to deny that R and L are the exclusive
thinkers of at least I-thoughts in the split-brain subject, and to insist that S is the co-thinker of all such
thoughts, instead. This is Davis’s own strategy: immediately after arguing for S-as-I-referent, he argues for Sas-I-thinker, as well—though all he says is:
In laying out and responding to the objection in the way that I have I have in fact neglecting an important and
uncertain issue, which is the kind of designators “R” and “L” are. I have merely assumed, here, that after left
hemispherectomy, R = S, and that after right hemispherectomy, L = S. But if “R” and “L” essentially designate mere
proper parts of animals, then actually after hemispherectomy neither R nor L exist. (See Garrett, 1998, for discussion.)
But I think that both the objection and the response could be phrased, albeit more cumbersomely, to accommodate this
possibility. Note that if “L” essentially designates a proper part of an animal then it is a relational fact about L that it
exists.
7
14
“Just as much as its two (overlapping) parts… [S] is an entity interacting with its
environment in ways we must acknowledge as involving perceptually guided
behavior and practical reasoning” (Davis, 1997: 212).
While correct, what Davis says here is not plainly sufficient to show that S is a thinker. A boy band is
also an entity interacting with its environment in ways we must acknowledge as “involving” perceptually
guided behavior and (some) practical reasoning, and yet most of us would deny that a boy band was thinker
(or perceiver, or practical reasoner). Analogously, S is plainly an entity qua something—but is S an entity,
singular, qua thinker, perceiver, and practical reasoner? Or is S merely constituted by thinkers, perceivers, and
practical reasoners? Do the mental states that we must appeal to in order to explain S’s interactions with its
environment genuinely belong to S, in the ordinary sense—in the sense in which L’s mental states belong to L?
Another difficulty with this response is that, prima facie, the truth of the S-as-I-thinker claim
undercuts the original claim of mental duality and is undercut by whatever empirical evidence motivated that
claim. If there is no difficulty identifying S as a psychological subject—the thinker of thoughts, subject of
experiencers, intender of actions, and so on—then why was it necessary to identify R and L with
psychological subjects in the first place?
Perhaps, however, there is a way to knock out these two worries with one stone. Perhaps if S is the
referent of R’s and L’s I-thoughts, no additional argument is needed to show that S is also a co-thinker of those
thoughts. Perhaps by attributing their mental states to S, R and L thereby constitute S as a psychological
subject (of a certain kind) to whom those mental states thus equally belong. That is, basically, S is the thinker
of R’s and L’s I-thoughts because R and L make S so.
Since I-thoughts are the only thoughts that refer to S, these are the thoughts for which there is the
most pressing need to identify S as their thinker (to avoid the objection from error). I think that, for present
purposes, we may restrict S’s thoughts to I-thoughts. Of course this substantive restriction means that S’s Ithoughts won’t play the same role in S that they do in R or in L. This might be all right though, for it seems
most plausible that S doesn’t have its own token mental states (I-thoughts), but rather than S is just the cothinker of R’s and L’s token I-thoughts. Obviously there is ordinarily an incredibly tight relationship between
token-individuating thoughts and token-individuating their thinkers, but in this case the co-thinkers of token
thoughts—R and S on the one hand and L and S on the other—stand in an unusual (constituting-in-part)
relationship.
The most appealing feature of the proposal, from my perspective, is that, if it worked, it would
suggest an interesting way of understanding the so-called personal-level and subpersonal-level distinction in
psychology. In theory this distinction is supposed to be one between psychological states properly attributable
to whole persons, and those attributable only to parts of persons. But there is, first of all, controversy about
which kinds of psychological states fall into which category (popular proposals include: propositional versus
15
sub-propositional states, conscious versus non-conscious states, and states familiar from folk versus from
scientific psychology). And the distinction is itself controversial: proponents believe that it captures some
deep fact about our conceptual schema and about the difference between scientific and ordinary thought and
practice; skeptics ask why we should think that there is any deep distinction that can be drawn here at all.
Certainly, skeptics think, we can’t state a priori which psychological claims are true of us and which are true
only of parts of ourselves: that itself is an empirical question.
Suppose we somehow discovered, however, that our I-thoughts referred to whole animals, rather
than to parts of animals. Then certain psychological attributions would be true only of whole animals if they
were true at all. These would be attributions of self-referring mental states. So the personal/subpersonal-level
distinction would be drawn between explicitly self-referring mental states and all others. Some might call this
construal of the personal/subpersonal distinction deflationary (which I wouldn’t disagree with, really):
according to this construal of the distinction, certain mental states are properly attributable only to, say, me as
a whole, because I attribute them to me as a whole. But I think that, construed this way, the distinction still
has some of the significance defenders want it to.
Unfortunately I’m not sure that the proposal is ultimately workable in the split-brain case. Return
again to a simple tactile recognition experiment, in which S is blindfolded, R palpates and (it soon becomes
clear) feels and recognizes a pipe, while L says, “I don’t know what I’m holding.” Suppose that R first forms a
belief that the object that S (or R) is holding is a pipe:
R:
BEL [This is a pipe].
Because R’s belief here is not an I-thought, it’s unclear whether S is the co-thinker of this thought;
let’s set this question aside to focus on the cases the proposal is clearly designed to deal with. Suppose that R
forms the following belief:
R:
BEL [I am holding a pipe]
The proposal under consideration here is that because R is attempting to refer to the thinker of this
thought (or belief) by using a concept (I) that in fact refers to S, S is thereby constituted as the believer of this
content as well:
S:
BEL [I am holding a pipe]
Now at the same time that R is thinking that he* is holding a pipe, L is simultaneously thinking (let
us say) that he* doesn’t know what he* is holding:
16
L:
BEL [I don’t know what I am holding]
S is thereby constituted as the thinker of this thought as well:
S:
BEL [I don’t know what I am holding]
One might worry, however, that the proposal therefore implies that S will now form (or at least be
capable of forming) the following belief:
S:
BEL [I am holding a pipe, but I have no beliefs about what I am holding]
This thought seems obviously self-contradictory somehow. Call this the Moore’s paradox objection to the
S-as-I-thinker claim.
On the face of it, the objection isn’t as serious as it looks. Although S may believe that S was holding
a pipe while simultaneously believing that S has no beliefs about what he was holding, S will not thereby be in
a position to form a single belief with the conjoined content, “I was holding a pipe, but I have no beliefs
about what I was holding.” On the proposal that we are considering, S does not have any independent
inferential powers; S can only think a thought with the conjoined content, “I was holding a pipe, but I don’t
know what I was holding” if either R or L first thinks a thought with that very same content. But of course
they wouldn’t: as soon as L came to believe that it was a pipe the subject had been holding, L would cease to
believe that he* had no idea what he* had been holding.
The problem is that, upon reflection, this response to the Moore’s paradox objection undercuts the
case for identifying S with a thinker to start with. First, the fact that S has no independent inferential
powers—no powers beyond those of R and L—itself emphasizes that S is a thinker only in a very derivative
way. Second of all, Moore’s paradox of belief is a paradox only against the background of assumptions about
psychological integration that simply aren’t true of S. These assumptions about integration do hold, in the
split-brain case, for R and for L, especially if R and L strictly self-refer in I-thought. Again, this undercuts the
case for identifying S, too, rather than R or L alone, as a thinker.
By this point in the dialectic I imagine that most readers will regret having gotten into this mess in
the first place. It is admittedly tempting to step back and reconsider our much earlier preliminary
identification of R and L with distinct psychological subjects. The problem is that there is really only so far in
this direction that we can realistically retreat, and it may not be far enough to allow us to escape the
difficulties we have stumbled into. As we have seen, R and L are, at a minimum, sufficiently complex in terms
of their intrinsic structure and properties to be capable of thinking I-thoughts, when one or the other
17
hemisphere is missing. And, in each other’s presence—that is, in the split-brain case—RH and LH I-thoughts
are sufficiently causally isolated from each other to so as to produce the paradoxes we are considering, even if
we identify S as their exclusive thinker. Again I hasten to emphasize that there is absolutely an obvious
(psychological and ultimately physical) explanation for S’s ability to think, at one and the same time, “The X
was Y” and “I have no beliefs about the X”, without being able to infer, “The X was Y and I have no beliefs
about the X.” The problem is that this explanation essentially constitutes the positive case for identifying R
and L with thinkers.
Are there other possible responses to the objection from error? Of course one might just bite the
bullet and accept that all self-attributions of psychological states in the split-brain subject are strictly false.
This position is an uncomfortable one. Although I have acknowledged that, any way you look at it, the splitbrain phenomenon will be characterized in part by an impairment of self-knowledge, it would still be
surprising if psychological self-attributions were routinely and in fact entirely in error in the split-brain
subject. Among other things, so many of the means and mechanisms by which any person ordinarily comes
to form beliefs about her mental states are unchanged and intact in the split-brain subject. Why should a splitbrain subject’s psychological self-attributions always be wrong, when a “normal” person’s psychological selfattributions are often right? On the other hand, I suppose someone willing to bite the bullet on this one could
suggest that the kind of pervasive error that R and L commit is one that we can’t really take credit for not
making ourselves; that we don’t make this error is contingent upon our factual identity with animals, or on the
fact (if it is a fact) that we aren’t multiply psychologically embodied.
Alternatively, instead of biting the bullet, and accepting that R and L are the exclusive thinkers and S
the exclusive referent of I-thoughts, perhaps we should simply reject the S-as-I-referent claim. Perhaps L’s Ithoughts simply refer to L, and R’s to R. The only trouble with this possibility—which in other ways is
actually the least problematic—is that it just seems incredible. In fact, I myself find it basically impossible to
accept.
How much weight should we put on this intuition (what we might call the “incredulous stare
objection”)? This is impossible to answer without getting a sense of the basis of the intuition. Someone might
propose that the idea of two self-conscious beings in one body is in and of itself incredible. I’m not sure that
this is right, though; people seem capable of belief in e.g. demonic possession, for instance. And then it isn’t
so easy to individuate bodies either; by some criteria, Brittany and Abigail Hensel share a body, and yet
(although there are certain puzzles of personal identity there) it seems very natural in that case to think of
each girl strictly self-referring via her I-thoughts and I-statements. (See Blatti, 2007.)
I think the reason it is so difficult to imagine that R and L strictly self-refer is that self-consciousness
in split-brain subjects really does seem importantly different from that of Brittany and Abigail Hensel.
Regardless of the referent/s of I-thoughts in split-brain subjects, there is something about the operation of self-
18
consciousness in the split-brain subject that makes such a subject more like one of us than like a pair of
conjoined twins.
5
Self and Other in Split-brain Subjects
It’s difficult to figure out just who, or what, R and L think about when they try to think self-consciously. By
comparison, however, I think it is fairly clear that neither thinks about the other in the way that, say, conjoined
twins do. And this says something about what R and L take themselves* to be.
To preempt any confusion on this point I should note that L clearly does not think thoughts with the
content like [I don’t have a right hemisphere] or [I am the left-hemisphere-associated system]. L does not
think, [I am L], in other words. L.B., for instance, may sometimes say things like, “‘I cannot name objects in
the left visual field because the information is conveyed to my right hemisphere which has no language
centers and, because of surgery, that information cannot reach my left hemisphere” (Levy and Trevarthen,
1976: 301). Levy and Trevarthen explicitly refer here to L.B.’s “intellectual understanding” of his situation,
distinguishing this intellectual knowledge from his (or from L’s) actual sense of self* or self*-consciousness.
Here, however, L.B. (or L) is no different from anyone else; as Evans suggests, there is nothing in our (pretheoretic) self-conception to support identification with nervous systems or parts of nervous systems (Evans,
2001:136). I know I have a brain, but don’t think of myself as a brain—even though I am intellectually open
to the possibility (however unlikely) that I may ultimately be my brain, and thus that my I-thoughts in fact
refer to it. So although L does not conceive of itself as L, its I-thoughts could still be about L; furthermore, L’s
I-thoughts could be about something that was, as it turned out, constituted or realized by L alone.
Against this claim, it is commonly said that split-brain subjects don’t think of themselves as multiple
persons, psychological subjects, or subjects of experience. So phrased, however, the point begs the relevant
questions. Is it really the split-brain animal as a whole that thinks that animal is a single person? Or is it, say,
L, that thinks L is a single person? The latter is of course totally consistent with claims of mental, conscious,
and personal duality.8
Still, though not phrased aptly above, there is something significant about the observation. I will say
that there is no mutual recognition in the split-brain subject: that is, neither R nor L appears to recognize the
existence of a second entity, of the same psychological kind as that with which it identifies, also inhabiting the
In the same paper quoted above, for instance, Levy and Trevarthen comment that “One has a very strong impression
that his [L.B.’s] intellectual understanding [of his mental and neural architecture] is completely separate from his ordinary
self-consciousness. In conversation, it appears that his [L.B.’s] ‘I’ refers to the same ‘I’ he referred to prior to surgery.
His left hemisphere experiences itself as a whole, unified stream of consciousness with no awareness of being split away
from any part of itself…. intellectual knowledge had no effects on the left hemisphere's consciousness of itself” (1976:
301). The first two sentences apparently suggest that L.B.’s (or L’s) I-statements refer to L.B. (presumably they did
before his surgery?)—but then they speak of the left hemisphere’s (L’s) experiencing itself as whole, and of its
consciousness of itself.
8
19
subject’s body. Whatever the entity to which e.g. L’s I-thoughts refer, L thinks that this entity is the only such
psychological subject present in L’s body.
5.1
Lack of mutual recognition in the split-brain subject
Although the split-brain literature as a whole supports this contention, an experiment by MacKay and
MacKay (1982) provides a particularly neat example. In the first of three stages in the experiment, one
experimenter would show to the split-brain subject J.W., in free vision (so that both R and L could
presumably see it), a number between zero and nine. A second experimenter would then guess what the
number was. Placed before J.W., also in free vision, was a card on which were written the options “Go up,”
“Go down,’ and “o.k.” J.W. was trained to provide feedback to the “guesser” by pointing, after each guess, to
one of the options written on the card.
In the second stage of the experiment, subject and experimenter traded roles, so that the
experimenter was now the “knower” of the correct answer and J.W. was the “guesser.” The experimenter said
he was thinking of a number between zero and nine, in other words, and J.W.’s task was to guess the number,
using feedback from the experimenter (to “go up” or “go down”) to narrow in on the proper answer (“o.k.”).
In the third, crucial stage of the experiment, J.W. played both roles at once. The experimenter
presented a number in his left visual field only (to R, in other words), and then J.W. was ask to verbally guess
(L) what the number was, using his left hand (R) to provide hints (to L) that would help him (or, L!) narrow
in on the right answer. All of this, J.W. (or, R and L) could do.
The experiment has several noteworthy features. First of course it provides some support for the
mental duality claim. J.W. was in fact able to play, with himself, a game that would ordinarily require multiple
people; a single psychological subject can’t play the game with himself because knowing the right answer would
preclude his playing the role of guesser. Now MacKay and MacKay themselves took pains to try to avoid
describing any of their results by attributing distinct mental states to the two hemispheres.9 As a result,
though, they end up needing to attribute them to, e.g., the subject’s left hand (e.g., “The left hand proved
immediately capable of providing reliable feedback in this mode”; subject’s “mouth guessed ‘1’ and his left
hand pointed to ‘go down’”… (MacKay and MacKay, 1982: 691). This illustrates why it has been so difficult
for experimenters to describe results of studies with such patients in ordinary psychological terms without
eventually speaking in terms of what a hemisphere (or R or L) thinks, hears, decides, and so on (see Bogen,
1977)—without implicitly identifying the two hemisphere systems, that is, with distinct psychological subjects.
At the same time, another aspect of this study’s findings strongly suggests that L didn’t recognize the
existence of a second psychological subject, distinct from the subject with whom L identified: during the third
stage of the experiment, L had to be repeatedly reminded to his left hand for clues, rather than asking the
Possibly the authors wanted to avoid describing the results in ways that implied anything about split-brain personal
identity at all. But since they also sought to (and perhaps could not but) describe the results in familiar psychological
terms, this is not really possible. Acts of pointing require pointers; one can only say that the split-brain human being
pointed or that e.g. R pointed. The authors at least attempted to go the former route.
9
20
experimenter. The experimenter had to keep saying, “Ask your left hand!” But this isn’t the way we relate to
parts of our bodies; if I decide to make a sandwich, I don’t ask my hands for help.
Is there any evidence that speaks in favor of mutual recognition? Some cross-cuing behaviors look
like instances of intentional communication. For instance, if L is asked to respond to a question to which only
R knows the answer, R may attempt to “spell out” the answer, on the back of the subject’s right hand, using a
finger from the left (Sperry, Zaidel, and Zaidel 1979). But while those kinds of cross-cuing behaviors look
conscious and deliberate, and like cases of “personal-level” communication of information, each of us
“interacts with” and treats herself in some ways similar to how she interacts with and treats others (as
philosophers like Dennett have emphasized)—such as when one gives oneself a pep talk before undertaking
some onerous task. In the cross-cuing case just described, R and L may be distinct psychological subjects
working harmoniously together while thinking of themselves—or, of the person with whom each identifies—
as struggling alone to come up with an answer he or she knows on some level. When I struggle to recall
someone’s name or to articulate an inchoate idea, I don’t think of myself as the “seeker” struggling to extract
information from a distinct “knower”; I think of myself as both seeker and knower.
One might think that certain cases of interhemispheric conflict speak in favor of mutual recognition in
the split-brain subject. The split-brain literature contains cases in which L complains about the “bad” or “outof-control” left hand (which is under dominant control of R). In the following cases, for instance, it is
presumably always L speaking:




“I get so disgusted with myself. I know what I want to do but the hand won’t
cooperate with the mind.” (Henninger, 1996 : 303)
“I open the closet door. I know what I want to wear. As I reach for something with
my right hand, my left comes up and takes something different. I can’t put it down
if it’s in my left hand. I have to call my daughter” (Ferguson et al., 1985: 504).
“My right hand has always been under control…. My left hand—I have to fight with
it… If I’m reading I can hold the book in my right hand it’s a lot easier to sit on my
left hand than to hold it with both hands than [sic] fighting it. I compensate for
everything that’s wrong or feels wrong. One hand that fights you. I cannot use it a
lot. If I had to lose an arm I’d rather lose my left arm than my right one. The right
one knows what I want it to do and it does it…. With your right hand you correct
what the left has done.” (Dimond 1980: 342)
“Intermanual conflict has been noted in split-brain patients by a number of people,
and when one of the patients first told me about it I just could not tell anybody else
because I did not think they would believe it. He and his wife came to the office and
I said, ‘How are things?’ He said, ‘All right except I'm having a little bit of trouble
with my left hand.’ I said, ‘How's that?’ He said, ‘Well, I picked up the paper to read
it and my left hand took the paper away and set it down. So I picked it up again and
my left hand came up and set it down again. So I picked it up and this time the left
hand came up and picked up the paper and threw it on the floor.’ I never reported
that because it was just too exotic.” (Bogen, 2003: 94)
21

One “patient told me he was pretty good because he had gone to the ball game
[with his wife] and this was something he had not done in years because of the
convulsions he was having so often…. But then he said, ‘Well, what was unusual
was she bought some licorice, which I don't like, and we had this shopping bag
between us on the way home and my left hand reached in the bag and pulled out the
licorice, which I don't like.’ I said, ‘Well, what happened?’ He said, ‘The left hand
brought it up to my mouth so I ate it but I didn't like it.’ You cannot put that in a
professional journal but it was a true story.” (Bogen 2003: 94)
It is certainly possible that L is speaking metaphorically in these cases—that L doesn’t really think that
mental states or intetional actions can be attributed to the left hand. That said, it also seems possible that L is
speaking literally (or that the distinction isn’t totally clear). People seem willing to identify all kinds of things as
at least primitive agents; why not hands?
When I deny that there is mutual recognition in the split-brain subject, however, I mean that neither
R nor L recognizes the existence of a second entity of the same psychological kind as that with which it identifies.
And there is little or no evidence that L identifies the left and as the (presumably shared) hand of an agent
distinct from but of the same psychological kind or complexity as the agent with which (or with whom) L identifies.
L does not seem to view the psychological subject and person with whom it identifies as one of two people
whose left hand its left hand is, that is. Thus while L may well resort to sitting on or slapping the left hand to
keep it under control, L never attempts to reason with the left hand (or with an agent sharing that hand) in the
way that I might attempt to reason with another adult (or even a young child).
Note that, throughout, L seems, at most, to identify the (left) hand as a primitive kind of agent. When
L complains, “That hand does whatever it wants,” the “it” apparently refers to the left hand, and not to some
other person whose hand it is. Similarly, in the case quoted by Henninger (1996), above, the subject prefaces
her complaint (or, L prefaces its complaint) by saying that she (or, L) gets so frustrated with herself*—not with
the jerk with whom she has to share her left hand.
This distinction—between L’s regarding the left hand as an agent (at least of a primitive kind) rather
than as the hand of another agent—is in and of itself significant. Partly because of the apparent sensory basis
of alien hand (or limb) syndrome, it is tempting to think that lack of mutual recognition in split-brain subjects
is heavily rooted in R’s and L’s highly similar somato-sensory representations (and/or similar bodily
phenomenology). Even when L does not have a sense of agency for the left hand’s responses, L does not
conclude that the hand is being controlled by someone else—much less that the hand equally belongs to
someone else. We might say, indeed, that L has not just a sense of ownership (Gallagher, 2000) for the left
hand but a sense of exclusive ownership. Indeed it is tempting to consider whether the sense of exclusive
ownership (e.g. this leg is mine and no one else’s) is an implicit part of the ordinary sense of bodily
ownership.
22
Lack of mutual recognition in split-brain subjects suggests that we are able and (under some
circumstances) willing to think of ourselves as in some sense containing multitudes, able to recognize and
concede that knowledge and control are not totally centralized within us. Interestingly, that the split-brain
phenomenon is so gripping to nearly everyone who encounters it has often been taken to show just the
opposite. It is commonly said (at least since Nagel, 1971), that the split-brain phenomenon grips us because
we naturally think of ourselves as highly psychologically unified. While I don’t think that this common
explanation gets things wrong, I don’t think that it’s wholly complete, either. Split-brain subjects themselves
seem to roll with the punches that their own “disunified” behavior, under experimental (and for that matter
non-experimental) conditions, presents them. Rather than insisting, “I am wholly unified and in control, so
someone else must have been responsible for that,” L concedes, “I don’t know how or why I did that” or “I
must have done it unconsciously” or “Well, I just guessed.”
Perhaps, then, it is better to say that L recognizes the left hand as an independent locus of agency—as
driven by goals and desires that L does not view as its* own, per se, but that L also denies are the goals or
desires of any third (or second) party. L does not distinguish itself* from a second psychological subject of
the same psychological kind and complexity as that with which (or with whom) L identifies, also associate
with the subject’s body.
This distinction—between recognizing the left hand as an independent locus of agency or even an
independent agent, of a primitive kind, versus viewing the left hand as the hand of a second, sophisticated
psychological subject—is important in part because to be a person requires that one be able to recognize the
existence of other persons and to recognize them as persons—as psychological subjects that can themselves,
among other things, deliberately self-refer. L’s failing to distinguish itself—or, the psychological subject with
whom it identifies—from anyone else standing in the same relationship to L’s body that L does at least
counts as a strike against identifying R with a person distinct from L.
Note that it does not count against identifying R or L with any person whatsoever. It would be
different if split-brain subjects seemed unable to recognize persons generally. But they don’t. Social
perception and engagement seems more or less normal in the split-brain subject. It is just that R and L don’t
perceive or engage with each other as distinct social subjects. Consider again that some split-brain subjects (or
L) begin to deal with the wayward left hand (or R’s behavior!) by sitting on it. This isn’t how well-socialized
adults handle interpersonal conflict—by sitting on each other.
Nor do R and L interact with third parties independently, in the social realm. When a subject (or L)
complains that her left hand (or R) keeps grabbing dresses she (or L) doesn’t want to wear, no one says, “Well
you’ve chosen your outfit the last two days; it’s Righty’s turn to choose an outfit.” Lack of mutual recognition
thus makes R and L genuinely unlike a pair of conjoined twins, and like any one of us, in this sense.
Of course this is just to gesture at the possibility of an account of unity of the person, qua social
subject, in the split-brain case. I have not really offered such an account. And any such account would face an
23
obvious and interesting challenge. This is that we “relate to” ourselves in a lot of the same ways we relate to
other people. Indeed some philosophers (perhaps Dennett most notably) believe that this is essential to being
a person, at least of the sort any of us are: we rebuke ourselves, praise ourselves, eye ourselves critically, bribe
ourselves, and so on. Conversely a lot of social interaction needn’t involve (at least continual)mutual
recognition. If that’s right—if a lot of interpersonal interaction doesn’t involve viewing other people as
persons, and if a lot of intrapersonal interaction involves viewing or treating myself as a third party—then lack
of mutual recognition in the split-brain subject doesn’t seem very important. But if that’s right then maybe
being a single person—as opposed to a pair of persons—isn’t very important either.
References
Blatti, S. 2007. Animalism, dicephalus, and borderline cases. In Philosophical Psychology 20: 595-608.
Bogen, J.E. 2003. Joseph E. Bogen. In L. Squire (Ed.), The History of Neuroscience in Autobiography, Vol. 5, pp.
47-122. Elsevier.
Bogen, J.E. 1977. Further discussion on split-brains and hemispheric capabilities. British Journal for the
Philosophy of Science 28: 281-286.
Cocchini, G., Beschin, N., Cameron, A., Fotopoulou, A., and Della Salla, S. 2009. Anosognosia for motor
impairment following left brain damage. Neuropsychology 23: 223-230.
Davis, L. 1997. Cerebral hemispheres. Philosophical Studies 87: 207-22.
DeWitt, L. 1975. Consciousness, mind, and self: The implications of the split brain studies. British Journal for
the Philosophy of Science 26: 41-47.
Dimond, S. 1980. Neuropsychology: A textbook of systems and psychological functions of the human brain. London:
Butterworths.
Eccles, J. 1981. Mental dualism and commissurotomy. Brain and Behavioral Science 4: 105.
Evans, G. 2001. Self-identification. In A. Brook and R. DeVidi (Eds.), Self-Reference and Self-Awareness, pp. 95142. Amsterdam and Philadelphia: John Benjamins Publishing Company.
Evans, G. 1982. In J. McDowell (Ed.), The Varieties of Reference. Clarendon Press.
Ferguson, S., Rayport, M., and Corrie, W. 1985. Neuropsychiatric observations on behavioural consequences
of corpus callosum section for seizure control. In A. Reeves (Ed), Epilepsy and the Corpus Callosum, pp.
501-514. Plenum Press.
Frankish, K. 2004. Mind and Supermind. Cambridge University Press.
Gallagher, S. 2000. Philosophical conceptions of the self: Implications for cognitive science. Trends in Cognitive
Sciences 4: 14-21.
24
Garrett, B. 1998. Identity and Self-Consciousness. Routledge.
Gazzaniga, M. 1983. Right hemisphere language following brain bisection: A 20-year perspective. American
Psychologist, May 1983: 525-537.
Henninger, P. 1996. Inkblot testing of commissurotomy subjects: Contrasting modes of organizing reality. In
S. Hameroff, A. Kaszniak, and A. Scott (Eds.), Toward a Science of Consciousness: The First Tucson Discussions
and Debates, pp. 203-222. MIT Press.
Kingstone, A. and Gazzaniga, M. 1995. Subcortical transfer of higher order information: More illusory than
real? Neuropsychology 9: 321-328.
Korsgaard, C. 1989. Personal identity and the unity of agency: A Kantian response to Parfit. Philosophy and
Public Affairs 18: 101-132.
Levy, J. 1969. Information Processing and Higher Psychological Functions in the Disconnected Hemispheres of Human
Commissurotomy Patients. Unpublished doctoral dissertation. California Institute of Technology.
Levy, J. and Trevarthen, C. 1976. Metacontrol of hemispheric function in human split-brain patients. Journal of
Experimental Psychology: Human Perception and Performance 2: 299-312
MacKay, D., and MacKay, V. 1982. Explicit dialogue between left and right half-systems of split brains.
Nature 295: 690-691.
Mark, V. 1996. Conflicting communicative behavior in a split-brain patient: Support for dual consciousness.
In S. Hameroff, A. Kaszniak, & A. Scott (Eds.), Toward a Science of Consciousness: The First Tucson Discussions
and Debates (pp. 189-196). Cambridge, MA: MIT Press.
Marks, C. 1981. Commissurotomy, Consciousness and Unity of Mind. Cambridge: MIT Press.
Moran, R. 2001. Authority and Estrangement: An Essay on Self-Knowledge. Princeton University Press.
Nagel, T. 1971. Brain bisection and the unity of consciousness. Synthese 22: 396-413.
Perry. 1979. The problem of the essential indexical. Noûs 13: 3-21.
Popper, K. and Eccles, J. 1977. The Self and Its Brain. Berlin, Heidelberg, London, New York: Springer
International.
Puccett, R. 1993. Mind with a double brain. The British Journal for the Philosophy of Science 44: 675-691.
Ramachandran, V. 1995. Anosognosia in parietal lobe syndrome. Consciousness and Cognition 4: 22-51.
Schiffer, F.; Zaidel, E.; Bogen, J.; Chasan-Taber, S. 1998. Different psychological status in the two
hemispheres of two split-brain patients. Neuropsychiatry, Neuropsychology, and Behavioral Neurology 11: 151156.
Shoemaker, S. 1968. Self-reference and self-awareness. The Journal of Philosophy 65: 555-567.
Shoemaker, S. 1976. Embodiment and behavior. In A. Rorty (Ed.), The Identities of Persons, pp. 109-137.
University of California Press.
25
Smith, J. 2004. On knowing which thing I am. Philosophy 79: 591-608.
Smith, A. 1966. Speech and other functions after left (dominant) hemispherectomy. Journal of Neurology
Neuroscience and Psychiatry 29: 467-471.
Sperry, R. 1968 Hemisphere deconnection and unity in conscious awareness. American Psychologist 23: 723-733.
Sperry, R. 1958. Corpus callosum and interhemispheric transfer in the monkey (Macaca mulatta). The
Anatomical Record 131: 297.
Sperry, R., Zaidel, E., Zaidel, D. 1979. Self recognition and social awareness in the deconnected minor
hemisphere. Neuropsychologia 17: 153-166.
Strawson, P. 1959. Individuals: An Essay in Descriptive Metaphysics. Routledge.
Strawson, P. 1966. The Bounds of Sense: An Essay on Kant’s Critique of Pure Reason. Methuen.
26
Download