Philosophia
DOI 10.1007/s11406-009-9187-3
Individuating Mental Tokens: The Split-Brain Case
Elizabeth Schechter
Received: 10 July 2008 / Revised: 21 February 2009 / Accepted: 9 March 2009
# Springer Science + Business Media B.V. 2009
Abstract Some philosophers have argued that so long as two neural events, within a
subject, are both of the same type and both carry the same content, then these events
may jointly constitute a single mental token, regardless of the sort of causal relation
to each other that they bear. These philosophers have used this claim—which I call
the “singularity-through-redundancy” position—in order to argue that a split-brain
subject normally has a single stream of consciousness, disjunctively realized across
the two hemispheres. This paper argues, against this position, that the kind of causal
relations multiple neural events bear to each other constrains the mental tokens with
which functionalists who are realists can identify them.
Keywords Split-brain phenomenon . Functionalism . Realism .
Multiple realizability . Individuating mental tokens .
Mind-brain relationship
E. Schechter (*)
Philosophy Department, University of Maryland, College Park, MD, USA
e-mail: lizschechter@gmail.com
Philosophia
Introduction
Many people believe that the two hemispheres of a split-brain subject1 can each
sustain a unique stream of consciousness, and in fact that even outside of
experimental conditions, when such subjects’ behavior is comparatively normal,
they have two streams of consciousness, and indeed two minds. This “duality” claim
regarding split-brain subjects’ consciousness and cognition is particularly popular
among cognitive scientists who study the brain, i.e., neuropsychologists. In fact the
neuropsychological community seemed to more or less converge upon a mental and
conscious duality claim for split-brain subjects some time ago.
Philosophers who have studied the split-brain phenomenon have not reached the
same degree of convergence. Some have urged caution in or skepticism towards
counting minds and streams of consciousness;2 others have argued forcefully against
the duality positions for split-brain subjects. Marks (1980) and Tye (2003) for
example have both suggested that claims of mental and conscious duality in splitbrain subjects rest upon an unmotivated and extravagant way of individuating mental
tokens. Essentially they have argued that because the two hemispheres of a splitbrain subject will normally generate experiences with the same contents, we should
identify them with the same stream of consciousness.
This paper examines what I call the “singularity-through-redundancy” (or STR)
position: the claim that multiple neural events of the same psychological type and
carrying the same content can be identified with a single mental token, regardless of
their causal relations to each other. I will approach this claim largely via the writings
of Marks (1980) and Tye (2003), since they have developed and defended this claim
most explicitly. But the claim enjoys broader appeal; Dennett, for example, gestures
towards a similar position (1991), and even Sperry seemed to feel its pull at times
(see his 1975 for example).
Split-brain subjects are discussed throughout much of this paper because the splitbrain studies motivated the development of the singularity-through-redundancy
position. The paper’s ultimate concern isn’t with consciousness in the split-brain
1
See Gazzaniga (2000) or Gazzaniga and LeDoux (1978) for an introduction to the split-brain
phenomenon. A “split-brain” subject is a human subject whose corpus callosum (which connects the
two cerebral hemispheres in the subjects I will call “normal”) has been surgically sectioned, and whose
hemispheres are thus (largely) divided at the cortical level. (In some split-brain subjects the anterior
commissure has also been sectioned.) The riveting behavior that these subjects exhibit occurs, for the most
part, only under the highly artificial experimental conditions distinctive of what I will call the “split-brain
experiment.” During the split-brain experiment, sensory information about a particular stimulus is directed
as much as possible only to one hemisphere at a time. When such experiments are performed on “normal”
subjects, at least some of this sensory information is made available to the non-receiving hemisphere via
the corpus callosum. In a split-brain subject, however, most of this information remains available only to
receiving hemisphere. Yet experimenters have shown that either hemisphere can respond intelligently to a
stimulus in various ways, and these responses often seem the result of conscious processing. (See, to take
just one set of examples, Sperry et al. (1979); Shallice states nicely how compelling some of the evidence
for right hemisphere consciousness is; see his 1997: p.264.) Meanwhile, the non-receiving hemisphere
appears largely or wholly unaware of what the other hemisphere has been presented with.
2
See the commentary following Puccetti’s 1981, including that of Churchland (1981), Green (1981), and
Margolis (1981); see Rey (1983) for a response to Puccetti as well. See also Nagel (1979) and Lockwood
(1989).
Philosophia
subject, however, but rather with the STR position more abstractly. This position
concerns several fundamental issues in psychology: most obviously how to
individuate mental tokens, and the relationship between psychological and neural
entities; it also concerns the proper form for a functionalist theory of the mental to
take. If defenders of the singularity-through-redundancy position are correct,
proponents of the mental and conscious duality positions have committed a
significant and fundamental error, and failed to draw the appropriate lessons from
important and well-motivated functionalist principles. This paper’s underlying
concern is whether, contra the STR position, the kinds of causal relationships that
multiple neural events bear, or fail to bear, to each other, constrain the mental events
with which we can identify them. I submit that they do.
In “Redundancy, Unity, and Streams of Consciousness” I shall locate the STR
claim with respect to the debate about the structure of consciousness in split-brain
subjects. In “Constituting a Single Token” and “Three Claims Regarding
Irrelevance” I argue against singularity-through-redundancy as a characterization of
split-brain consciousness. In “Mental Tokens” I describe a hypothetical subject for
whom I think the case of (mental) singularity-through-redundancy could be made
most strongly. I nonetheless argue that the STR claim would fail even in this case. In
fact, I suggest that a functionalism with realist commitments must use facts about the
causal dependence or independence of neural events to individuate mental tokens.
Redundancy, Unity, and Streams of Consciousness
This section explains the prima facie motivation for attributing to a split-brain
subject a single stream of consciousness, and describes the singularity-throughredundancy model, which, if tenable, may offer a way to defend this attribution. I
also clarify the element of the STR position that is the concern of this paper.
The Motivation for a Unity Claim for Split-brain Subjects
Marks and Tye say that “normal” subjects—in this context, subjects who each
possess an intact corpus callosum—behave in a unified fashion, and that we believe
that this is partly due to these subjects each having a single stream of consciousness
or a unified consciousness. In daily life, meanwhile, the gross behavior of split-brain
subjects usually resembles that of these “normal” subjects. Marks asks, “If we
account for our integrated behavior, at least in part, by assuming unity of
consciousness and can do the same for split-brain patients, why not do so?” (22)3
Marks and Tye accept that there are occasional times during the split-brain
experiment at which split-brain subjects have two streams of consciousness.4
Notwithstanding such moments they believe that a split-brain subject’s consciousness is normally singular or unified. They therefore face the challenge of showing
how the split-brain experimental paradigm could alter the structure of a split-brain
3
All references to Marks are taken from his 1980. All references to Tye are taken from his 2003.
In order to focus on the singularity-through-redundancy position in particular, this paper must assume
many (not uncontroversial) claims without argument. One such claim is that the right hemispheres of splitbrain subjects are associated with conscious mental states, as Marks and Tye contend.
4
Philosophia
subject’s consciousness, especially since it does not alter that of the “normal”
subject. (“Normals” do not exhibit the conscious dissociation under conditions of
perceptual lateralization that split-brain subjects appear to.) The solution both
philosophers take is to say that conscious unity can supervene on certain properties
of conscious contents. When present, the corpus callosum ensures conscious unity
even under conditions of lateralized perception, by allowing some kind of
interhemispheric communication of contents. But even without a corpus callosum,
the two hemispheres of a split-brain subject will normally be subject to the same (or
to highly similar) conscious contents, Marks and Tye (among others) believe,
because they will be receiving the same (or highly similar) information from the
environment and the body. The singularity-through-redundancy claim asserts that
what Marks calls the “independent duplication of information” (see pp. 17–24)
suffices for a unified consciousness. Multiple neural events of the same mental type
(conscious experience), and bearing the same content, can together constitute a
single mental token, even if their contents are produced independently of each other.
This paper’s central concern is whether neural events operating “independently”
of each other can indeed constitute a single mental token, as the defenders of STR
argue that they can. But of course a natural objection to the entire debate about STR,
as it has just been described, is that the two hemispheres of a split-brain subject don’t
generate conscious phenomena totally causally independently of each other. This is
quite correct: if via her left hemisphere, for instance, a split-brain subject says, “Oh
no, not again,” then both her right and left hemisphere will generate conscious
auditory representations of this verbal expression. This demonstrates that we need to
get clearer about the kind of causal independence that is in question.
Tye’s defense of STR refers to right and left hemisphere consciousness-realizing
events as simply being “causally unrelated” (127); Marks speaks of the absence of
“direct” causal interaction between right and left hemisphere consciousness-realizing
events in the split-brain subject, but without clarifying the distinction between direct
and indirect causal interaction. I believe that it is essential to get a little clearer than
this on the kind of causal independence whose relevance to the constitutive
conditions for mental tokens is at issue. I will postpone until “Is Causal
Independence Psychologically Relevant?”, however, a discussion of the kind of
causal relationship that I believe characterizes the two hemispheres of a split-brain
subject, and that, I argue, does pose constraints on the mental individuation of neural
events. Until that time I will just speak of the causal independence of right and left
hemisphere neural events without qualification. I do this, for now, in part because the
STR position itself, as currently defended in the philosophical literature, makes no
serious attempt to distinguish between different kinds of causal independence in the
generation of mental and conscious phenomena, casting causal independence,
without qualification, as not in and of itself relevant to psychological individuation.
The Empirical Inadequacy of the Singularity-through-redundancy Characterization
of Split-brain Consciousness
One objection to the singularity-through-redundancy characterization of split-brain
consciousness is empirical: the two hemispheres are almost certainly not subject to
identical conscious contents outside of experimental situations, and thus the right
Philosophia
and left hemisphere’s neural events cannot be considered redundant from the
psychological perspective. The hemispheres are no doubt subject to more similar
contents outside of experimental situations than they are inside of them, and this no
doubt offers at least a partial explanation for the fact that split-brain subjects behave
differently in the two types of circumstance. Nonetheless, since the hemispheres
have different patterns of perceptual access to the world, and since they also have
different processing styles and capacities, and appear to experience emotions
somewhat differently, and to have access to a somewhat different store of long-term
memories—not to mention the fact that one hemisphere can presumably generate a
fairly normal stream of inner speech and one hemisphere probably can’t—it would
be a stretch to imagine that the hemispheres are associated with highly similar, much
less identical, conscious contents.
In order to focus on the STR position considered more abstractly, however, I set
aside this empirical objection in most of what follows. For most of the paper I will
assume that the two hemispheres of a split-brain patient are subject to highly similar
conscious contents; towards the end of the paper I consider what we should say of a
subject who possessed two hemispheres that independently generated neural events
bearing truly indistinguishable psychological properties at every moment.
Disunity Versus Duality
Ordinary language generally treats “unified consciousness” and “single stream of
consciousness” as near-synonyms—as mere different parts of speech. Much
philosophical and psychological writing on the structure of consciousness similarly
equates having a unified consciousness with having a single stream of consciousness. Yet some of those who have argued that split-brain subjects have two minds
and two streams of consciousness prefer not to speak of mental or conscious disunity
so much as duality.5 After all, since the hemispheres may be associated with many of
the same psychological states, and since they share a history (in one sense), and so
forth, it might be misleading to describe them as disunified, for the term “disunity”
connotes conflict and discord.
Ascribing a dual consciousness has fewer of these potentially misleading
implications, and so, too, I submit, should ascriptions of multiple streams of
consciousness. Two streams of consciousness are highly unlikely to have
indistinguishable psychological properties. They could nonetheless be either fairly
unified or fairly disunified with each other, and how unified they were could very
well depend in part upon the degree of overlap between their contents.
This paper is concerned not with the qualitative character of mental tokens, but
with their individuality; it argues (among other things) that the two hemispheres of a
split-brain subject are each associated with an (at least largely) unique set of mental,
including conscious, tokens. For the most part I set aside the issue of how
psychologically similar to each other (and thus perhaps how unified with each other)
those tokens may be.
5
Bogen (1990) seemed to prefer this formulation, for example.
Philosophia
I speak in this paper, therefore, of whether or not the STR position gives us
reason to believe that split-brain subjects have a single stream of consciousness,
rather than of whether or not it gives us reason to believe that they have a
unified consciousness. This is why I speak of the singularity-through-redundancy
rather than the unity-through-redundancy position. In doing so I do not mean to
pre-judge the issue of whether or not semantic relations among contents might in
fact serve to meld what would otherwise be two streams of consciousness into one.
Perhaps, that is, having a unified consciousness—in the sense of a coherent set of
mental contents—is all there is to having a single stream of consciousness. But
perhaps not; there might for instance be a single stream of consciousness whose
contents were very disunified, in the sense of discordant. I therefore remain neutral
on this issue, for now, by speaking of streams of consciousness, and not of unity or
disunity.
A split-brain subject possessing two distinct streams of consciousness might still
enjoy a high degree of conscious unity (in the sense proposed just above), and a high
degree of unity might play an important role in explaining the behavior that Marks
and Tye call “unified.” Defenders of the singularity-through-redundancy characterization of split-brain consciousness do not merely believe, however, that the two
“disconnected” hemispheres are largely unified in the sense of having highly similar
conscious contents (something that many defenders of the conscious duality model
also seem to believe). They claim something stronger. Defenders of the STR position
equate having a unified consciousness with having a single stream of consciousness,
and having a disunified consciousness with having two streams of consciousness.6
They therefore claim that to the extent that the two hemispheres are associated with
the same conscious contents, they are associated with the same conscious tokens. It
is this stronger claim that I reject.
Constituting a Single Token
The STR position holds that multiple neural events may, causally independently of
each other, jointly constitute a single mental event. This section criticizes two
analogies Tye offers in support of this position: the first seems simply nonanalogous, and the second is subtly question-begging against those who would claim
that the two hemispheres of a split-brain subject are associated with distinct mental
tokens.
Tye rejects attributing to split-brain subjects “two separate streams of consciousness that remain two
from the time of the commissurotomy” in favor of saying that such subjects “are single persons whose
phenomenal consciousness is briefly split into two under certain special experimental conditions, but
whose consciousness at other times is unified” (126; emphasis added; citing Marks 1980 as well).
Likewise, as a “rough necessary condition for two simultaneous conscious experiences belonging to the
same stream of consciousness”, Marks offers, “e1 and e2 belong to the same unified consciousness only if
they are known, by introspection, to be simultaneous” (13; emphasis added), thus tying possession of a
single stream of consciousness to possession of a unified consciousness. Both philosophers meanwhile
also accept what is widely accepted, i.e. that a stream of consciousness is composed of token experiences
(or in Tye’s case, a single token experience—this is one of the main positions argued for in Tye 2003).
6
Philosophia
The Projector Analogy
Tye anticipates that some will object to the STR position by saying that, “a
single experience cannot have as its physical basis neural events in the left and
right hemispheres that are themselves causally unrelated.” But, he asks, “why
not?”
Consider the following example. Two movie projectors each project an
image onto a screen at time t. Only a single image is present on the screen at t,
since identical slides are in the projectors and they are aimed at exactly the
same part of the screen. There are two projections but only one image. One
projection is redundant. Each projection on its own suffices for the screen
image. (127)
There are three problems with this analogy. Note, first, that the generation of a
single image isn’t necessarily a function of the projectors’ projecting “redundant”
slides: one machine could project an image of a child at play, the second of a hulking
monster, and the result would be a single image of a monster lurking over an
unsuspecting child at play.
More troublingly, the two projections appear not to realize but to produce the
image. One could, for example, alter the (apparent) number of images without
manipulating the slides or projectors at all, simply by moving (or removing
altogether) the screen. But everyone will accept that a mental event can causally
depend upon two independently acting neural events. The relationship between
realizer and realized is more intimate than the relationship between cause and effect,
and Tye needs to provide an analogy more clearly involving the former.
Even if the projections did jointly realize the image, however, the doublyprojected image offers an inadequate analogy to a case of mental singularity through
neural redundancy in particular, because the individuation criteria for images and
experiences are simply too different. Clearly an audience would assume they were
seeing a single image, so long as all beams of light projected onto the same portion
of the screen. And perhaps counting images is just a matter of determining how
many images a normal viewer (e.g. a human being whose visual system is
functioning normally) would appear to see. But as Tye himself points out, the fact
that split-brain subjects don’t “see double” or experience having multiple streams of
consciousness is to be expected, regardless of how many streams of consciousness
any of them actually have. When it comes to counting images on a screen, in other
words, phenomenology may be almost everything. But when it comes to counting
conscious experiences, particularly those that do have redundant contents,
phenomenology tells us almost nothing.
If the disanalogy between individuating images and individuating experiences
isn’t immediately obvious, it is because Tye’s first analogy plays off of our naïve or
pre-theoretic notion of conscious experience as, to borrow Dennett’s (1991) oftborrowed phrase, a Cartesian theater. Once we rid ourselves of the illusion that
conscious experience is a thing we watch, it is unclear what a conscious experience
is supposed to be analogous to in this example. Maybe to the beam of light? But
there are two of those.
Philosophia
The Restaurant Analogy
In his second analogy, Tye, while seated at a restaurant, waves both of his arms at
once to catch his server’s attention. There are two arm-wavings here, he says, and
yet one event of signaling his server.
Note, however, that the two arm-wavings together seem necessary to give the
signaling event its precise character; neither arm-waving is redundant in that respect.
There are, for example, social and psychological differences between trying to catch
your server’s attention by waving one hand, and trying to catch his attention by
waving both hands, just as there are social and psychological differences between
trying to catch his attention by waving both hands wildly in the air, and trying to
catch his attention by shouting “Hey, boy!”7
Nonetheless, even if the event of signaling with one hand and the event of
signaling with two hands have a different character, each appears to be but a single
event, and individuating events (tokens), rather than providing a qualitative analysis
of their nature, is our current concern. Still, it is worth asking why Tye seems right
that waving both arms in a restaurant is one way of realizing a single event. How do
we know that waving two biological arms in the air isn’t just one way of realizing
two waving-one-arm events—waving two prosthetic arms being a different way of
realizing two waving-one-arm events?
Actually, waving two biological arms in the air may be one way of realizing two
waving-one-arm events. But Tye could agree that there are two waving-one-arm
events. His claim is that there is just one signaling event.
Events are individuated relative to a particular level of description. There is a
single signaling event in Tye’s restaurant case because signaling one’s server is a
communicative act, and we individuate such acts partly in terms of intention. Thus if
two parties seated near each other talked and decided (their situation being
somewhat desperate) to both wave their hands at the same time in the hopes of
finally catching their server’s attention, they arguably both participate in a single
signaling event. But if the same two parties both waved their hands at once by mere
coincidence, then there were arguably two, simultaneous signaling events.
Part of the reason we may find it hard to conceive of a reason we would want to
say, of the original case that Tye provided, that there are two signaling events, is
because we are just so used to thinking of a single human organism as having a
single mind and a single set of intentions (at a time) that are (in some sense)
integrated prior to behavior. In Tye’s original “restaurant” analogy, the two armwaving events aren’t initiated causally independently of each other in the way we
deem relevant to individuating communicative actions, because we attribute to Tye a
single intention to signal by waving both arms.
But of course precisely what is under consideration is whether and to what extent
we can apply this reasoning to split-brain subjects. Identifying Tye’s two armTye himself introduces the example by asking us to imagine a “case in which I am in a restaurant, and,
being anxious to leave, I signal the waiter by raising both hands in the air and waving them” (127;
emphasis added). Tye’s feeling it necessary to imply that of course he wouldn’t wave both hands in the air
to catch his server’s attention unless he were particularly anxious constitutes an inadvertent admission that
neither arm-waving is truly redundant, since both arm-wavings are necessary to give the signaling event
its precise social and psychological significance.
7
Philosophia
waving events with a single signaling event rests upon the assumption that there was
a single mental event causally responsible for both arm-waving events. Duality
advocates claim that the two hemispheres of a split-brain subject are associated with
distinct mental events that drive the subject’s (unified-seeming) behavior causally
independently of each other (in some relevant sense). Tye obviously disagrees, but
his restaurant analogy offers only question-begging support for his position.
Note that it isn’t just the number of token intentions to signal that determines how
many signaling events we locate. In the case in which two parties waved at once,
each possessed a unique token intention to wave; whether we see one or two
signaling events depends upon whether there was an interactive process leading to
the formation of the two token intentions. We ask the parties, “Did you talk to each
other, and come to a mutual decision to jointly signal your server? Or did you each
come to the decision to wave independently—was it mere coincidence that you both
waved at once?” We try to determine whether there was a certain kind of causal
relationship between the admittedly distinct token intentions that produced the
wavings.
So it is, too, I shall argue, when the events we are counting aren’t communicative
actions but cognitive acts. To determine whether two neural events realize one and the
same mental event, we must know something about their causal relationship to each
other. The next section considers more closely the causal relationship that is required
between neural events in order for them to realize a single token of a mental type.
Three Claims Regarding Irrelevance
This section treats three distinct arguments potentially supporting the singularitythrough-redundancy position, particularly as applied to split-brain consciousness.
These arguments all deny psychological significance to the causal relationships that
hold, or fail to hold, between right and left hemisphere neural events in the splitbrain subject. The first argument is premised upon a type-token confusion; the
second and third arguments fail to grasp the precise kind of causal interaction that
seems lacking between right and left hemisphere events in the split-brain subject.
Multiple Realizability and the Type-token Distinction
Fodor (1975) has argued that the kinds postulated by psychology will not reduce to
the kinds postulated by any purely physical (non-functional) science, because the
kinds of psychology—the types of entities to which it will refer in its laws—are
multiply realizable. What makes a given event a pain event, or an event of believing
something, as opposed to some other kind of mental event or no mental event at all,
isn’t a matter of its intrinsic physical properties, but the role it plays within a
functionalist story, a role connecting it to (at a minimum) sensory inputs, motor
outputs, and other mental events. Thus creatures whose physical construction was
radically different from ours could nonetheless possess the same types of mental
states that we do. All that would be required is for their physical construction to
somehow implement a functional design similar to that implemented by our own
brains (whatever design that turns out to be).
Philosophia
The principle of multiple realizability is foundational to the functionalist program,
and has been appealed to in order to support the singularity-through-redundancy
claim. Marks answers the question, “Why should neural processes unrelated by
direct causal routes not be the physical basis for a single mental state?” (23) in part
by citing Fodor (1975). Tye says that when a split-brain subject’s right and left
hemisphere, causally independently of each other, each generate a neural event of
the type experience and carrying the same content, “There is. . . . a single
experience. . . with a neurological realization or constitution that is partly redundant
and that differs from the neurological realization or constitution” that that same
experience would have “in normal subjects” (127). Same mental events, different
realizers, in other words.
By casting the neuro-functional differences between split-brain and “normal”
subjects as merely differences in how experiences are realized, rather than as
differences in how many experiences are realized, those sympathetic to the STR
position seek to obviate a simple objection. This objection states that while the
hemispheres of a split-brain patient, S, may be subject to the same types of
experiences (where contents are included in type), they are surely subject to distinct
token experiences, for S’s hemispheres realize the experiences they realize
independently of each other. This objection rests on the intuition that facts about
the causal properties of a neural event constrain the mental events with which we can
identify it.
Drawing on the language of multiple realizability, Marks and Tye respond that
such facts concern how psychological phenomena are implemented, but are not
psychological facts properly speaking. We can, and should, refer to those facts at
some times—such as when we need to explain why the consciousness of split-brain
subjects, and not that of “normal” subjects, becomes dual or disunified during the
split-brain experiment.8 Marks writes that this duality
is itself explained by the fact that the experimental controls defeat the
mechanisms which are, as a result of commissurotomy, responsible for unity
of consciousness in these patients. Similarly there is a natural explanation for the
behavioral differences between split-brain patients and normal controls. The
mechanisms which subserve unity of consciousness in the normal controls differ
in ways that make them immune to failure in the experimental situation. (22–3)
But facts about functional neuroanatomy are facts about the mechanisms subserving
various mental phenomena, not facts about which mental phenomena are actually being
subserved. Marks believes that the moral of the split-brain cases is that
bilateral neural representation is a physical basis for unity of consciousness;
but it is irrelevant how such representation is achieved, whether through the
commissures or through mechanisms for independent duplication. (22;
emphasis added)
8
It is essential to keep in mind that for Marks and Tye (and seemingly others to whom the singularitythrough-redundancy position appeals), to have a unified consciousness is to have a single stream of
consciousness, and to have two (or more) streams of consciousness is to have a disunified consciousness.
These same thinkers also view streams of consciousness as mental tokens (or composites of mental
tokens).
Philosophia
Facts about how mental phenomena are implemented in the “normal” and in the
“split” brain should not constrain the individuation of psychological entities—such
as streams of consciousness—which are, after all, multiply realizable.
If the principle of multiple realizability is correct, then creatures with a wide
variety of physical constitutions could possess streams of consciousness. This does
not mean that physical facts are irrelevant to individuating particular streams of
consciousness, however. A mental token is a physical thing—a realizer of a
particular mental type. The thesis of multiple realizability, and the fact that, as
Marks says, the “general account of mind advocated by philosophers as diverse as
Fodor and Grice. . . does not require corresponding types of neurological states for
each type of psychological state” (23; emphasis added), provides no support for the
further claim that causally unrelated neural events can form “the basis for a single
mental state” (23; emphasis added; citing Fodor (1975) and Grice (1975)). The
latter is a claim about mental tokens. And the principle of multiple realizability
doesn’t say anything about how to individuate mental tokens. It is simply silent on
this matter.
Since the principle of multiple realizability provides no guidance where counting
mental tokens is concerned, we must turn elsewhere for this guidance.
Is Causal Independence Psychologically Relevant?
The defender of the singularity-through-redundancy position must claim that
multiple neural events can constitute a single mental token even if they occur
causally independently of each other. There is a broader and a narrower way of
making this claim. Framed broadly, the claim states that the causal independence of
multiple neural events never provides reason to identify them with multiple mental
tokens. Framed narrowly, the claim states that although the causal independence of
multiple neural events usually provides reason not to identify them with a single
mental token, there is a particular kind of case that constitutes an exception to this
general rule: the case in which the neural events in question have “redundant”
psychological properties. I will consider the broad interpretation in this subsection
and the narrower interpretation in the next.
The principle underlying the intuitive objection to STR is simply that the neural
events that constitute a single token of a mental type must be themselves be causally
related in some important way. Up until this point we have postponed asking what
way that might be. The question is crucial because causal connectedness in and of
itself is cheap; my conscious neural events are connected to your conscious neural
events via your reading of this paper I have written, for example. What we really
want to know is whether there is some psychologically relevant difference between
intrahemispheric and interhemispheric causal interactions in the split-brain subject,
such that we have reason to believe that intrahemispheric neural events may jointly
constitute a single stream of consciousness, while interhemispheric neural events
cannot. What we really want to know is what kind or degree of causal distance
between two neural events should constrain our identifying them with a single
mental token.
On Marks’ interpretation, those disposed to object to STR believe that the neural
events that jointly constitute a single mental token must be “causally connected in a
Philosophia
fairly direct way, on some reasonable understanding of ‘direct’” (21). But Marks
says that this “crucial causal principle”
is not strongly motivated. Why should neural processes unrelated by direct
causal routes not be the physical basis for a single mental state? (23)
Psychofunctionalist psychology does not
require any direct causal links between the neural events which are the physical
basis for a single psychological state. It would be sufficient if causally
unrelated neural events jointly, though separately, produced effects which
were, from the standpoint of the psychology, the basis of a single mental state.
(23)
Anyone familiar with the binding problem (or problems) may be inclined to agree
with Marks. The binding problems (roughly) concern how it is that we perceive
objects possessing multiple perceptual properties, and entire scenes composed of such
objects, given the fact that perceptual information, both across and within sensory
modalities, is processed in a distributed manner in the brain (such that color and shape,
for example, are represented in different brain areas). If we state a priori, as it were,
that two neural events can only constitute a single mental event when they are directly
causally linked, then we may not find the neural correlates of many of the mental
tokens that (let’s assume) we have compelling reasons to believe we possess.
It is difficult to go further at this point without getting a little clearer on the kind
of causal connection between neural events whose relevance to psychological
individuation is being debated here. For there is a psychologically relevant
difference between the kind of causal independence between perceptual processing
areas that gives rise to the binding problems, and the kind of mutual causal
independence characterizing the two “disconnected” hemispheres.
To begin with, behavioral evidence alone allows us to conclude that whatever
neural mechanisms are responsible for phenomenal and functional binding, they
operate intrahemispherically in split-brain subjects (at least largely). For the splitbrain experiments show that in these subjects perceptual information is not bound,
either phenomenally or functionally, across the hemispheres. They also show that
although the two “disconnected” hemispheres do interact with each other in complex
ways, this interaction is generally sustained only via the mediation of environmental
and behavioral events.9
9
Due to the fact that the hemispheres are connected via numerous sub- and non-cortical structures (and
often via the anterior commissure), there may be certain types of mental states that the hemispheres
generate or experience in an interactive way, and certain ways in which the hemispheres interact absent the
mediation of behavioral and environment and bodily events. Emotional or affective states and processes
have often been seen as constituting such exceptions to the general rule of interhemispheric
“independence” (e.g. Sperry et al. 1979; Lockwood 1989). (Though there is also competing evidence
for the emotional independence of the two hemispheres; see instance Schiffer et al. 1998.) My
fundamental concern in this paper isn’t with the structure of split-brain consciousness per se however but
with the singularity-through-redundancy claim. That claim doesn’t hinge upon the possibility of any kind
of wholly neural, inter-hemispheric interaction in the generation of mental phenomena; it rather asserts that
even without such interaction, the two hemispheres can jointly generate a single stream of consciousness. I
therefore ignore many complications and subtleties regarding sub- and non-cortically mediated interhemispheric interaction in split-brain subjects.
Philosophia
Let me take a few paragraphs to demonstrate these claims. Consider an experiment
in which Kingstone and Gazzaniga (1995) showed to each hemisphere of a split-brain
subject, J.W., one half of a word pair, and then asked him to draw (the referent of)
what he’d seen. (For instance, if the word pair was “day–night”, “day” would be in J.
W.’s left visual field and thus received by his right hemisphere, while “night” would
be in his right visual field and thus received by his left hemisphere.) When not
permitted to watch himself as he drew, he usually drew only a single word in the pair
(especially when using his right hand, over which the left hemisphere generally exerts
dominant but not exclusive control); when permitted to watch himself as he drew, J.W.
was somewhat more likely to draw (the referents of) both words.
In the crucial trial type (during which J.W. was allowed to watch himself in free
vision as he drew), each hemisphere was shown a different half of a “conceptually
ambiguous” word pair: each of these word pairs could be drawn either as a literal
combination of the referents of the two words, or as an emergent object, i.e. as the
single object referred to by the compound word formed by joining the two words in
the pair. For instance, a drawing of the word pair “head-stone” could consist either in
a literal combination of the two words—e.g. a rock and a human head—or in a
single emergent object—a tombstone.
It is well known that that there is little evidence for phenomenal binding across the
two hemispheres of a split-brain subject; when asked to describe what they see, for
instance, subjects generally only describe (using the left hemisphere) left hemisphere
percepts. This study suggested that right and left hemisphere percepts also aren’t
bound together functionally somehow in the split-brain subject: when asked to draw
what he’d seen, J.W. sometimes drew the referent of one hemisphere’s word,
sometimes the referent of the other’s, and sometimes both—but he never drew the
single referent of the compound word formed by putting the two words together.10
Supporters of the STR position might point out that J.W. would once in a while
draw something like an arrow poised over a bow when given the word pair “bowarrow” (in an earlier trial type, in which the two words in a pair couldn’t always be
combined to form a compound word referring to a single emergent object). He did
sometimes draw an “integrated” picture of the referents of both words in the pair,
that is. Doesn’t this show that drawing and looking behaviors can be used to bind
right and left hemisphere percepts functionally, if not phenomenally?
But as the authors of this study pointed out, the production of this sort of drawing
does not require any internal transfer of information between the hemispheres. If
there were such internal transfer of information—if either hemisphere could
communicate to the other information about the word it had received, without
having to literally draw a picture—then we would expect J.W. to at least sometimes
draw the single, emergent referent of the compound word in the crucial trial type.
(As he easily did when compound words were presented entirely within his right
visual field, i.e., entirely within the view of his left hemisphere.) We would expect J.
W. to at least sometimes draw a skyscraper, that is, rather than a windshield scraper
lying under a sunny sky. But he never did.
10
J.W. used a single hand on each trial; the hemispheres were apparently switching control of the drawing
hand. (Miller and Kingstone 2005 found that both hemispheres can exert a degree of motor control over
the ipsilateral hand sufficient for the production of crude drawings.)
Philosophia
Instead, using one hemisphere, J.W. would draw the referent of the word that
hemisphere had received (“arrow”). Then the second hemisphere would come to
dominate performance. If the existing image (e.g. of an arrow), now visible to the
second hemisphere, could be treated as part of the background for a drawing of the
word it had received (e.g. “bow”), the first image was (sometimes) used in this way.
But if the existing image (e.g. of a windshield scraper) couldn’t be treated as part of
the background for a drawing of the referent of the second hemisphere’s word
(“sky,”), J.W. just drew the referent of that second word anyway (or in many
instances stopped drawing altogether).
In other words, when producing an “integrated” drawing would have required
some kind of internal integration of mental states across the hemispheres (integration
of the sort involved when J.W.’s left hemisphere saw the entire word pair “honeymoon,” leading J.W. to draw two touching faces, eyes closed, surrounded by hearts),
no such drawing was forthcoming. “Indeed,” Kingstone and Gazzaniga explain, the
only time that right and left hemisphere “word information. . . [is ever] integrated is
on the sheet of paper in the drawing itself” (1995: 324; emphasis added).
Although right and left hemisphere neural events interacted to produce some of
these drawings, they didn’t interact in a way that is in principle different from the
way in which two different subjects’ neural events might interact. After all, if you
and I were told to make a single drawing between us, and you were shown only the
word “bow” and I was shown only the word “arrow,” and we both took a turn with a
pencil, we might produce an arrow drawn across a bow, too—just like J.W.
The two hemispheres of a split-brain subject may interact at every moment. But
their interaction, at least by and large, does not supervene on wholly neural and
mental events. It supervenes partly on behavioral and environmental events.
Interhemispheric neural event pairs in split-brain subjects do not jointly produce
mentation. They do jointly produce behavior, in the sense that there are behaviors in
which split-brain subjects engage that involve the joint participation of both right
and left hemisphere. But then there are many behaviors in which my sister and I
engage that involve the joint participation of her and my neural events also.
Marks seems right in claiming that two neural events need not interact with each
other directly in order to constitute a single mental event. The various proposed
solutions to the binding problems (see for instance Golledge et al. 1996; O’Reilly et
al. 2003; Robertson 2003; Treisman 1996; Treisman and Gelade 1980) seem
compatible with this assertion. They are also compatible with a recognized limit to
how causally distant from each other these neural events can be, before we cease
being able to recognize them as constituting a single mental token. Right and left
hemisphere neural event pairs in split-brain subjects are for the most part located at a
point past this distance, for their interaction occurs largely via the behavior that they
cause. When duality theorists say that in a split-brain subject the right and left
hemisphere generate experience causally independently of each other, this is the kind
or degree of causal independence they have in mind.
Is Redundancy Psychologically Relevant?
There is a narrower claim that the STR position might be interpreted as making. This
is the claim that even if, as a general rule, the mutual causal independence of two
Philosophia
neural events requires identifying them with distinct mental tokens, there is an
important exception to this rule: those instances in which the two neural events in
question have indistinguishable mental properties. In such instances, either set of
events seems redundant from a psychological standpoint. Why, then, identify each
set with a unique set of mental tokens?
After all, Marks calls it “a commonplace that the neural structure of human brain
[sic] is highly redundant”; why should it, “be surprising if the redundancy is
sometimes irrelevant for the purposes of psychology” (23–24)? Tye, too, says that
There seems no obvious reason why nature should not have made us so that, in
certain circumstances, there is redundancy at the neural level in the generation
of perceptual experience. After all, it is well known that the human brain has a
neurological structure that is highly redundant anyway. Why not also here?
(128)
The purported redundancy in the split-brain case is from the psychological
perspective, but at the neural level: there are two neural events instantiating, but a
single mental token of, a mental type. The realizing events are thus psychologically
redundant, and the fact that there are two of them is psychologically irrelevant.
This analogy to the “well known” redundancy of the human brain may be subtly
question-begging, however. Two neural events need do more than bear the same
content (and instantiate the same mental type) in order to be redundant. In most
contexts, for instance, your neural event representing a rapidly approaching truck
will hardly be redundant simply because I already have neural event with the same
content. Psychological redundancy seems to occur within a mental system, where
what constitutes a single mental system seems itself to be a matter of causal
organization.
It makes sense to speak of the structure of the brain (and perhaps even of some of
the contents of the brain) as to some extent redundant, assuming that the brain is a
functional system characterized by a certain functional organization and so forth.
Thus if one area of the brain is damaged, another area can functionally compensate
for its loss.11
But those who advocate a conscious and mental duality model for split-brain
subjects believe that the two “disconnected” hemispheres are distinct mental
systems, and that right hemisphere neural events are therefore not psychologically
redundant. Indeed, it seems fairly simple to show that right and left hemisphere
neural events aren’t really redundant in the split-brain subject. If, for example,
something catastrophic suddenly happened to S’s left hemisphere visual system, his
right hemisphere visual system would not be able to compensate for it, nor would his
right hemisphere visual representations be able to serve as “back-up” copies of the
representations he’s just lost, at least for many cognitive purposes. For instance, S
wouldn’t be able to describe (via his left hemisphere) his right hemisphere
experiences.
11
Though even here there may not be exact redundancy in many instances; sometimes one area of the
brain functionally compensates for the damaged area, at least to some extent, without performing the exact
same operations. We should be careful not to individuate mental functions and capacities too grossly. (See
the criticisms of the STR position in “Functionalism, Physicalism, and the Realist Commitment”.)
Philosophia
This seems to suggest that attempting to formulate the STR position narrowly, as
stating that the causal independence of two neural events is psychologically
irrelevant so long as the events in question are redundant from a psychological
perspective, doesn’t really change the main issue. For what qualifies as genuine
redundancy itself seems to depend upon facts about causal interaction and
independence.
But the defender of STR might press that at most left and right hemisphere neural
events aren’t redundant in the split-brain subject because the two hemispheres aren’t
functionally identical, for instance, and thus right and left hemisphere neural events
won’t have truly identical functional roles. But if a subject did possess two genuinely
psychologically redundant neural events, couldn’t they jointly realize a single mental
token—no matter the manner of their causal connection to each other?
Mental Tokens
This section considers whether the constitutive conditions for mental tokens are
purely psychological/functional or partly physical (neural). I argue that they are
partly neural, for a method of individuation that is sensitive to the causal properties
of realizers is required at least by those of us who are realists about functionalist
explanation.
Purely Psychological Versus Partly Neural Constitutive Criteria for Mental Tokens
It is worth emphasizing again that the two hemispheres of a split-brain subject surely
do not have identical psychological properties. Among other things, minds—
certainly minds like ours—are so complex that, even if two hemispheres had
identical patterns of perceptual access to the world, and even if they were
functionally identical in terms of their perceptual and cognitive processing styles,
it is highly unlikely that they would really be subject to all the same types of mental
state with the same contents at every moment. (Absent some kind of integrative
process to ensure that they were, that is.)12
Furthermore, while a description of an event’s mental type and content tells us a
lot about its psychological properties, it doesn’t tell us everything. Even in a case in
which the two hemispheres of a split-brain subject each generated a neural event of
the type “belief that X”, the belief might well be put to different uses in each
hemisphere. Of course, the two uses would still have to have a lot in common in
order for both states to remain beliefs (with the same content); they would have to
share a single core or central causal role. Still, there are a great many things that can
be done with any token mental state, even keeping its functional type and content
constant.
As Gazzaniga and LeDoux once put it, “it is unlikely that the two independent mental systems (each
with its own sensory input, processing and storage mechanisms, and motor output) would maintain
equivalent attentional and motivational states over an extended period” (1978: 117). The same goes for
other types of states as well.
12
Philosophia
Thus the redundancy condition—which states that two neural events must truly be
redundant with respect to their psychological properties if they are to jointly realize a
single mental token via a causally independent processes—will be very difficult to
meet. There will frequently be a way to distinguish between even two type- and
content-identical mental tokens, even within a single creature, by attending to their
psychological properties alone, for example, by looking at the other (instantiated)
types of mental states they interact with.
The singularity-through-redundancy position may still be founded on an
important principle, however. This principle says that, whatever criteria we use to
individuate neural events, when it comes time to individuate mental events or
tokens, we should use purely psychological properties to do so. Thus if two neural
events truly do have indistinguishable psychological properties—if they are of the
same mental type, with the same content, and are functionally related to all the same
types of mental events with the same contents—then a psychological theory need not
distinguish them, but may rather identify them with a single mental token.
Perfect Parallelism: the Strong Case for Singularity-through-redundancy
We have seen that the principle of multiple realizability cannot provide support for
the claim that causally unrelated neural events may constitute a single mental token.
We have noted that the neural redundancy, sensory decomposition and binding that
exist in the “normal” brain are not analogous to the “redundancy” and causal
independence at issue in the debate about STR, and therefore not relevant to
evaluating that position. Can anything else be said in favor of singularity-throughredundancy—or can anything be said more decisively against it?
In fact I think that something more can be said in favor of it, though I also think
that ultimately, the position is still untenable. To evaluate the position at its strongest
and most plausible, consider a hypothetical case of what I will call “perfect
parallelism.” Our “perfectly parallel” subject, PP, unlike our split-brain subject S, has
two functionally identical hemispheres (or two whole brains if you prefer). Unlike
those of our split-brain subject, PP’s two hemispheres operate in perfectly
“redundant” or parallel fashion: every time PP’s left hemisphere generates a neural
event realizing the belief that X, or a desire for Y, PP’s right hemisphere does also,
and at the exact same time, of its own accord. And every time one of PP’s
hemispheres generates a motor plan, and initiates a motor impulse to, say, grab Z,
PP’s other hemisphere does also, at the exact same time and again wholly
independently, such that the set of neural events in either hemisphere alone would
have sufficed for that exact same action having been performed. (Note that because
PP’s hemispheres are both functionally identical and subject to identical psychological properties at every moment, the defender of STR will argue that they
disjunctively realize a single mental system.) But somehow (one must imagine)
PP’s two hemispheres operate in perfectly parallel fashion like this wholly
independently of each other in the following sense: there is not a single activity of
either of PP’s hemispheres throughout PP’s life that depended upon PP’s other
hemisphere’s actual activities; each hemisphere would have behaved identically had
the other hemisphere been totally absent. So how many beliefs that X does PP have?
How many streams of consciousness? How many minds?
Philosophia
Perfect parallelism is clearly a fiction, but one that may suggest that those
sympathetic to the STR claim still have an important theoretical point against duality
theorists, who have tended to take facts about the causal independence of neural
events as in and of themselves relevant to psychological individuation. The defender
of the singularity-through-redundancy position would presumably press, here, that
while PP’s neuroanatomy may be striking and fascinating from some neuroscientific
standpoint, the fact that PP’s beliefs, desires, experiences, intentions, memories,
motor commands, and so forth, are realized “disjunctively” by two causally
independent sets of neural events, is now without question psychologically
irrelevant. For surely we lose nothing, from an explanatory perspective, by
attributing to our perfectly parallel subject, PP, only a single belief that X, a single
desire that Y, and so forth, at every moment, and indeed a single mental system,
itself disjunctively realized across two hemispheres. And surely we gain nothing by
attributing to PP two sets of mental tokens with indistinguishable psychological
properties at every moment. Occam’s razor alone would suggest that we attribute the
less extravagant number of mental tokens to PP. In which case the STR position,
even if not tenable with respect to split-brain subjects for purely empirical reasons, is
premised upon a sound principle for individuating mental tokens: utilizing their
purely psychological properties, and not their neural ones. In which case many
members of the neuropsychological community have made a significant error.
Functionalism, Physicalism, and the Realist Commitment
Contra this position, I believe that the causal relations a neural event stands in, with
respect to other neural events, always matter to its individuation as mental token. At
least, they matter to those functionalists who are also realists.
In the abstract, functionalism doesn’t say much about the nature of mental tokens;
functionalism proper is neutral with respect to what sorts of things can occupy the
roles tagged by mental state terms, and is therefore in principle compatible even with
ontological dualism.13 Of course the vast majority of functionalists are physicalists,
who believe that all existing phenomena are actually physical (or have physical
properties).
This paper is written from the standpoint of a psychofunctionalism with realist
commitments; so far as I can tell, Marks and Tye are both realist psychofunctionalists as well. Psychofunctionalism states that the best account of the constitutive
conditions for mental phenomena will come from the functional analysis of a
developed psychological theory. To commit to psychological realism is just to
believe that there really are mental phenomena doing the things, playing the roles,
that our best psychological theory will describe. The realist believes that there really
are things out there instantiating the mental types implicitly defined by our best
psychological theory.
13
Though this is not to say that a functionalist psychological theory cannot refer to any non-functionally
defined terms. My point here is merely that functionalism can be neutral with respect to the kinds of things
that can realize functional types, not that all forms of functionalism are so neutral, nor that the best form
will be.
Philosophia
The functionalist who is also a realist thus has an additional reason for being a
physicalist. To paraphrase Rey (1977, p. 57; also citing Wiggins 1977), causation
just seems in fact to require physical stuff. Identifying mental tokens, the realizers of
mental types, with physical things, offers a way for functionally defined entities to
have causal powers.
For the psychological realist, neuroscience has the potential to vindicate a
psychological theory, to show that the theory is not just empirically adequate or
useful with respect to predicting behavior, but that the story the theory tells is also
true. Of course the realist runs greater risks as well; neuroscientific discoveries can
falsify the psychological realist’s best theory, in a way that they can’t falsify the best
theory of the non-realist.
The realist expects and in fact requires that at multiple points in the development of
a psychological theory, neuroscientists will hunt for physical things doing the things
that the theory describes mental phenomena doing. If the realist is lucky, neuroscientists will find such a thing. If she is less lucky, no such thing will be found, and
she will either have to repeatedly revise her theory or (perhaps, ultimately, following
enough failures of the right sort) drop the theory altogether and start anew. And
sometimes neuroscientists may not only find something playing the role her theory
described—they may find more of those things than anyone expected.
From the perspective of the functionalist who is also a realist, causal relationships
to (instantiated) types of mental state are essential to a neural event’s mental type
identity. But are causal relationships to instantiated types of mental event all that
matter to a neural event’s (mental) token identity, too? Or are causal relationships to
particular mental tokens also essential to a neural event’s (mental) token identity?
It seems quite clear that relationships to particular tokens matter. That one of your
neural events and one of my neural events are identical with respect to the types of
mental state to which they are causally related provides good reason to accept that they
instantiate the same mental type. But clearly it provides no reason for identifying these
two neural events with a unique mental token. For the two neural events are causally
related to distinct actual occupiers of these mental types. The point can be put in the
following way. A mental token’s type identity is a matter of its bearing certain
relationships to any mental tokens of particular types. But a mental token’s token
identity is a matter of it bearing certain relationships to particular mental tokens. The
relationships that matter are of course causal ones: a realist functionalism, again, is
distinguished by a commitment to the reality, to the causal efficacy, of the entities
characterized in (some) functionalist psychological theory.
While it is admittedly difficult to say at this point at exactly what level of
abstraction away from the physical a psychofunctionalist story will be pitched, it
nonetheless seems very likely that how any of these mental tokens are individuated
will be in part a physical matter. A mental token is an occupier of a given
functionally defined role; counting tokens is a matter of determining how many
occupiers of that role there are. Occupiers are physical things; their physical
properties are relevant to determining how many of them exist. Admittedly, the
physical, causal properties that an occupier of a psychological role possesses may
not all be relevant to its individuation as a mental token; indeed, a great number of
them (e.g., the number of dendritic spines synapsed upon) are no doubt irrelevant.
For that matter, not all of a neural event’s psychologically defined, actual causes and
Philosophia
effects may matter to its identity qua mental token. Perhaps a token percept that
gives rise to a certain belief, for instance, might have been that same percept even if
it hadn’t given rise to that belief. So not all of the causal relationships that a mental
token does and does not bear to other mental tokens need be essential to its token
identity.
But certain counterfactual relationships such as the causal relationships a mental
token can and can not bear to other role-occupiers is clearly the sort of physical,
functional property that matters to its token identity. And, more broadly speaking,
the larger web of causal and potential interactions that it is embedded in matters. At
least, it matters from the perspective of a functionalism committed to the view that
mental tokens are real causal actors.
I submit that the realist must admit that the perfectly parallel subject PP has two
sets of mental tokens with indistinguishable psychological properties at every
moment. Granted, by offering a complete psychological characterization of one of
PP’s hemispheres, you will have offered a complete characterization of PP’s other
hemisphere as well, since PP’s two hemispheres possess the same mental
architecture and all the same psychological properties. Nevertheless, a characterization of the structure of PP’s cognitive life would not be complete without the
acknowledgment that there are in fact two of everything heretofore mentioned in this
characterization: two of every type of mental state, two of every type of
psychological process, two streams of consciousness, and two minds.
Granted, you would never, by stipulation, see any evidence of these two sets of
tokens with indistinguishable psychological properties in PP’s behavior.14 Granted a
psychological theory for PP would therefore never have to mention the existence of
two sets of mental states with indistinguishable mental properties in order to be
empirically adequate with respect to even a single one of PP’s perfectly unified
behaviors. Instrumentalists, therefore, need not recognize two such sets. But as
Ramsey et al. (1990) point out, the realist commitment means that consistency with
behavioral data isn’t all that is relevant to evaluating a functionalist story. Whether
that functionalist story correctly identifies the events that actually caused the
behavior is also relevant.
I take it that scientific psychology isn’t just looking for the readily observable
evidence for a thing, but for something more subtle, for the underlying causes of
what we observe. (Putnam’s (1975) disease examples are still relevant here.) And
(by stipulation) there are two sets of such causes in the perfectly parallel subject.
There are two sets of neural events in PP, and each set of neural events meets the
criteria for various types of mental state, and does so causally independently of the
other. We should say that each set of neural events in PP constitutes a unique set of
mental tokens, then, simply because each set of neural events does constitute a set of
mental tokens. All that ultimately turns on the acceptance or rejection of the STR
position with regard to our perfectly parallel subject, then, is realism itself.
14
Although surely you could see evidence for these two sets of mental states under certain counterfactual
circumstances: if PP suffers a massive stroke to one entire hemisphere while giving a lecture, PP won’t
pause mid-sentence for even a moment. But we are imagining that nothing of this sort ever happens, that
PP’s two hemispheres in fact operate in perfectly parallel fashion throughout PP’s entire life.
Philosophia
When we set aside the fantastical case of PP, and imagine instead applying the STR
model to any subject we will ever actually encounter, the stakes of the debate become
sharper, more immediate. As I mentioned in “The Empirical Inadequacy of the
Singularity-through-redundancy Characterization of Split-brain Consciousness” and
“Purely Psychological Versus Partly Neural Constitutive Criteria for Mental Tokens”,
the STR model of split-brain subjects is straightforwardly empirically inadequate. It is
only by classifying mental phenomena very grossly—not the direction that a
developing psychological science seems to be moving in—that one could claim that
the two hemispheres are associated with psychologically redundant neural events. The
hemispheres don’t have identical perceptual access to the world or to the body. They
also aren’t identical with respect to their perceptual processing styles. They also aren’t
identical with respect to their cognitive capacities. Indeed, the only way to make STR
plausible as a characterization of the split-brain phenomenon is to not look very
closely at it. And this is just what the STR position encourages: to ignore as merely
“implementational” details that in fact matter from a psychological perspective. This is
the creeping verificationism of the STR position, and realists must resist it.
Defenders of the “singularity-through-redundancy” view might worry, however, that
by attributing distinct mental tokens to PP’s two hemispheres, we actually lose
something of explanatory value with respect to behavior: we lose an explanatory role
for streams of consciousness to play. For now having a single stream of consciousness
can’t be an explanation for unified behavior, and having multiple streams of
consciousness can’t be an explanation for disunified behavior. But this misses the
point that perfect parallelism would be extraordinary. To the degree that an overlap in
conscious contents ever plays a role in producing unified behavior, then we should
expect two streams of consciousness with identical contents to produce unified
behavior. It is just that it is highly unlikely that two streams of consciousness with
identical contents will ever exist, for reasons discussed at the beginning of this section.
Conclusion
Even if split-brain subjects had two hemispheres associated with highly similar or in fact
identical conscious contents, this would not mean that such subjects had single streams of
consciousness. Mental tokens are causal actors. Causally distinct mental actors—i.e.
neural events that have psychological properties and that are causally independent in the
strong sense described in “Is Causal Independence Psychologically Relevant?”—are
distinct mental tokens. Well motivated functionalist principles give no reason to think
otherwise. In fact, while functionalism says little about individuating tokens, a
functionalism with realist commitments should use some physical properties to
individuate mental tokens. Doing so respects the role that mental state types and contents
play in functionalist explanations, and the status of mental tokens as causal actors.
References
Bogen, J. (1990). Partial hemispheric independence with the neocommissures intact. In C. Trevarthen
(Ed.), Brain circuits and functions of the mind, pp. 215–230. London: Cambridge University Press.
Philosophia
Churchland, P. S. (1981). How many angels... ? The Behavioral and Brain Sciences, 4, 103–104.
Dennett, D. (1991). Consciousness explained. Boston, MA: Little, Brown.
Fodor, J. (1975). The language of thought. Cambridge, MA: Harvard University Press.
Gazzaniga, M. (2000). Cerebral specialization and interhemispheric communication: does the corpus
callosum enable the human condition? Brain, 123, 1293–1326.
Gazzaniga, M., & LeDoux, J. (1978). The integrated mind. New York: Plenum.
Golledge, D., Hilgetag, C., & Tovée, M. (1996). Information processing: a solution to the binding
problem. Current Biology, 6, 1092–1095.
Green, M. (1981). May we forget our minds for the moment? The Behavioral and Brain Sciences, 4, 107–
108.
Grice, P. (1975). Method in philosophical psychology (from the banal to the bizarre). Proceedings and
Addresses of the American Philosophical Association, 48, 23–53.
Kingstone, A., & Gazzaniga, M. (1995). Subcortical transfer of higher order information: more illusory
than real? Neuropsychology, 9, 321–328.
Lockwood, M. (1989). Mind, brain and the quantum. Oxford: Blackwell.
Margolis, J. (1981). Mental ascriptions and mental unity: Molar subjects, brains, and homunculi. The
Behavioral and Brain Sciences, 4, 110–111.
Marks, C. (1980). Commissurotomy, consciousness, and unity of mind. Montgomery, Vermont: Bradford
Books.
Miller, M., & Kingstone, M. (2005). Taking the high road on subcortical transfer. Brain and Cognition,
57, 162–164.
Nagel, T. (1979). Mortal questions. Cambridge, UK: Cambridge University Press.
O’Reilly, R., Busby, R., & Soto, R. (2003). Three forms of binding and their neural substrates: alternative
solutions. In A. Cleeremans (Ed.), The unity of consciousness (pp. 168-190). Oxford University Press.
Puccetti, R. (1981). The case for mental duality: evidence from split-brain data and other considerations.
Behavioral and Brain Sciences, 4, 93–123.
Putnam, H. (1975). Mind language and reality: Philosophical papers, Vol. 2. Cambridge, MA: Cambridge
University Press.
Ramsey, W., Stich, S., & Garon, J. (1990). Connectionism, eliminativism, and the future of folk
psychology. Philosophical Perspectives 4: Action Theory and Philosophy of Mind: 499–533.
Rey, G. (1977). Survival. In A. Rorty (Ed.), The identities of persons, pp. 41–66. Berkeley, CA: University
of California Press.
Rey, G. (1983). The lack of a case for mental duality. The Behavioral and Brain Sciences, 4, 733–734.
Robertson, L. (2003). Binding, spatial attention and perceptual awareness. Nature Reviews Neuroscience,
4, 93–102.
Schiffer, F., Zaidel, E. B., & Chasan-Taber, S. (1998). Different psychological status in the two
hemispheres of two split-brain patients. Neuropsychiatry, Neuropsychology, and Behavioral
Neurology, 11, 151–156.
Shallice, T. (1997). Modularity and consciousness. In N. Block, O. Flanagan & G. Güzeldere (Eds.), The
nature of consciousness, pp. 255–276. Cambridge, MA: MIT.
Sperry, R. (1975). Mental phenomena as causal determinants in brain function. Process Studies, 5, 247–
256.
Sperry, R., Zaidel, E., & Zaidel, D. (1979). Self recognition and social awareness in the deconnected
minor hemisphere. Neuropsychologia, 17, 153–166.
Treisman, A. (1996). The binding problem. Current Opinion in Neurobiology, 6, 171–178.
Treisman, A., & Gelade, G. (1980). A feature integration theory of attention. Cognitive Psychology, 12,
97–136.
Tye, M. (2003). Consciousness and persons. Cambridge, MA: MIT.
Wiggins, D. (1977). In A. Rorty (Ed.), The identities of persons (pp. 139–174). University of California
Press.