When
Counting Conscious Subjects, the Result Needn’t Always Be a Determinate Whole
Number
Eric
Schwitzgebel
Department
of Philosophy
University
of California, Riverside
Riverside,
CA 92521
USA
Sophie
R. Nelson
Department
of Philosophy
New
York University
New
York, NY 10003
USA
December
12, 2024
Invited submission
to a special issue of Philosophical
Psychology on the philosophy of Daniel C. Dennett
When
Counting Conscious Subjects, the Result Needn’t Always Be a Determinate Whole
Number
Abstract:
Could there be 7/8 of a conscious subject, or 1.34 conscious subjects, or an
entity indeterminate between being one conscious subject and seventeen? Such
possibilities might seem absurd or inconceivable, but our ordinary assumptions on
this matter might be radically mistaken. Taking inspiration from Dennett, we
argue that, on a wide range of naturalistic views of consciousness, the
processes underlying consciousness are sufficiently complex to render it
implausible that conscious subjects must always arise in determinate whole
numbers. Whole-number-countability might be an accident of typical vertebrate
biology. We explore several versions of the inconceivability objection,
suggesting that the fact that we cannot imagine what it’s like to be 7/8 or
1.34 or an indeterminate number of conscious subjects is no evidence against
the possibility of such subjects. Either the imaginative demand is implicitly
self-contradictory (imagine the one, determinate thing it’s like to be an
entity there isn’t one, determinate thing it’s like to be) or imaginability in
the relevant sense isn’t an appropriate test of possibility (in the same way
that the unimaginability, for humans, of bat echolocation experiences does not
establish that bat echolocation experiences are impossible).
Word
Count: ~8000 words, plus 4 figures
Keywords: Artificial Intelligence; consciousness; Dennett, Daniel;
personal identity; split-brain; subjectivity
When Counting Conscious Subjects, the Result Needn’t
Always Be a Determinate Whole Number
1.
Introduction
People
typically assume that conscious subjects come in discrete, countable wholes. Either
there are no conscious subjects in the seminar room, or there are two, or six,
or twenty-three. It seems bizarre to say that there might be 9.382 subjects, or
7/8 of a subject, or 3i subjects, or that the number of subjects might
be indeterminate between five and seventeen or best represented by a nine-dimensional
non-Euclidian surface. What could such remarks even mean, unless as a wry
comment about a shy student with 30% of their head poking through the door?
We
will argue that this ordinary assumption is false. When counting conscious
subjects, the result needn’t always be a single or determinate whole number.
Ordinary counting might fail, or yield a mathematical representation other than
a whole number, or yield a set of whole numbers among which the result is
indeterminate. At best, discrete countability is an accident of typical
vertebrate biology, which needn’t apply to invertebrates, AI systems, or
atypical humans. If it’s part of our concept of consciousness that conscious
subjects necessarily come in determinate, whole bundles, then our ordinary
concept of consciousness requires repair.[1]
We
start by clarifying our notion of a conscious subject. After arguing that the
complexity of the processes underlying consciousness renders our thesis
plausible, we argue for the possibility of cases indeterminate or intermediate
between zero subjects and one, as well as the possibility of cases
indeterminate or intermediate between one subject and whole numbers greater
than one. We conclude by addressing objections concerning inconceivability.
Our
position is partly inspired by Dennett, and we believe he would have endorsed
it. According to his “Multiple Drafts” theory of consciousness, not only do
different “drafts” of experience exist at different times (generating his
famous “Orwellian” vs. “Stalinesque” dilemma); they also exist simultaneously.
He writes, “At any point in time there are multiple drafts of narrative
fragments at various stages of editing in various places in the brain” (1991,
p. 135). The narratives that constitute our “selves” “issue forth as if
form a single source” encouraging us “to (try to) posit a unified agent” (1991,
p. 418, emphasis in original). While discussing split-brain and multiple
personality cases, Dennett says that we typically want “to preserve the myth of
selves as brain-pearls, particular, concrete, countable things, rather than
abstractions… refusing to countenance the possibility of quasi-selves,
semi-selves, transitional selves” (1991, pp. 424-425).[2]
2.
Our Main Claim
Marcel
sips tea containing the crumbs of a madeleine cookie and reflects back on his
days in Combray. In that moment, he simultaneously experiences the warmth of
tea in his mouth, the taste of the madeleine, and recollections of Combray.
These three experiences (or experience-parts or experience aspects) all belong
to the same field of consciousness. They are phenomenally unified.
There’s something it’s like to have them together. Perhaps, as Bayne and
Chalmers (2003) suggest, they are subsumed within a more complex experience of
tea-with-madeleine-with-recollection. Perhaps, as Dainton (2000) suggests,
they’re united by a basic, not-further-analyzable relation of co-consciousness.
We take no stand on how to analyze phenomenal unity, but we assume that the
core idea is clear: two experiences might belong either to the same, or to
different, fields or streams of conscious experience.[3]
What
does it mean to be determinately one conscious subject? Conscious subjects,
as we’ll understand them here, are just such bundles of experiences, individuated
by maximal relations of phenomenal unity. If experience A is unified with
experience B, and if experience B is unified with experience C, and if
experience A is unified with experience C, then there is one conscious subject
with a unified experience of A-with-B-with-C.[4] If experience D is unified
with no other experiences, then there is one conscious subject experiencing
only D.
This
notion of a “conscious subject” is unusual but not unprecedented.[5] We choose it not because
we endorse it as the best approach to subjecthood from the point of view of
theories of selfhood or personhood, or because we think that “subjects” on this
conception are identical to cognitive systems, bodies, or persons, but rather
because it targets a specific phenomenon of special interest. Broader definitions
of “subject” that appeal to more than conscious experiences at a moment—for
example, diachronic relations between conscious states; or causal threads of
memory and inference; or values, personality traits, and nonconscious mental
processes; or bodies—probably make it easier to defend our thesis. Temporally extended
views, for example, appear to permit Parfitian (1984) cases of merging,
splitting, and slow change. Views that rely on synchronous nonconscious
processes appear to permit hypothetical AI or alien cases of synchronous
partial unity or overlap due to shared nonconscious mechanisms.[6] Similarly, bodies can
plausibly overlap. By focusing on “conscious subjects” in our narrow sense, we
address the countability issue in its starkest form, where skeptical interlocutors’
intuitions seem to run strongest against our view.
On
this notion of a “conscious subject”, is at least conceptually possible for
multiple conscious subjects to exist in the same body (one experiencing
A-with-B-with-C, another experiencing D-with-E-with-F), perhaps even
coordinating to operate as a single person.[7] On some accounts of
personal identity, you might be nothing but a conscious subject, perhaps
extended over time. But even if you aren’t strictly identical to a conscious
subject, you are intimately related to one. If you are, right now, experiencing
the warmth of tea, the taste of madeleine, and recollections of Combray, then
there is a conscious subject in your spatiotemporal vicinity undergoing exactly
that set of token conscious experiences, and that conscious subject is
certainly not someone other than you. Perhaps it is a part of you, an aspect of
you, a process within you, or in some other way ontologically adjacent.
Our
thesis is that conscious subjects, in the sense just articulated, needn’t
always come in determinately whole-number units. Experiences don’t always
bundle neatly.
3.
Plausibility
Considerations
For
the purposes of this argument, we assume a broadly naturalistic approach to
consciousness on which human beings have conscious experiences in virtue of
something about the complex structure and composition of their bodies and
brains. (We don’t assume materialism. Some forms of neutral monism, property
dualism, and even panpsychism are naturalistic in the relevant sense.) The
correct naturalistic theory of consciousness is far from settled. However, on a
wide range of plausible theories, it appears possible to construct cases that
defy whole-number countability, even if such cases are rare among vertebrates.
Consider
Global Workspace theories, for example, according to which information is
conscious if it is broadcast through a “global workspace” for potential use by
downstream cognitive systems.[8] How large must a workspace
be, and how much information must it be capable of feeding downstream, before
it is appropriately “global”? Is there some minimum workspace size or quantity
of connectedness below which an entity is not at all conscious and above which
an entity is conscious, with no intermediacy or indeterminacy? Could workspaces
be partial or overlapping? If information is available for 85% of downstream
processes, is that enough? 51.578103%? Two? Imagine an architecture with
downstream processes in two clusters, A1, A2, and A3 vs. B1, B2, B3, and B4.
What if some information is available to all of A and B while some is available
only to A or only to B or only to A1 and B3? Must there be a bright line
between cases of a single unified conscious subject with imperfect downstream
information uptake and two discrete subjects with limited or indirect
information sharing? The model seems to allow in principle for a diverse range
of messy cases where subjecthood and unity would either be partial or best
captured with more complex mathematical representations such as vectors or
regions.
Consider
Higher-Order theories of consciousness, according to which an experience is
conscious if it’s the target of a higher order representation of the right
sort.[9] Perhaps in typical
vertebrate cases, there’s a single higher-order system that does all of the
tracking and unifies all of the represented lower-order states. But as with
Global Workspace theory, we can imagine considerably messier architectures. The
same lower-order state might be targeted by different higher-order systems that
link to different or partially overlapping output systems. Partly overlapping
sets of lower-order states might be targeted by different higher-order systems
that are partly integrated and have different downstream influences on the
organism. It might be only in cases at the extremes that there exists either
one unified conscious subject or many sharply distinct subjects. The majority
of possibilities might lie in the middle.
Consider
Dennett’s (2005) “fame in the brain” view, according to which a cognitive
process is conscious to the extent that it is “famous”—that is, influential
on—other cognitive processes. Fame among humans is a complex phenomenon. People
can be famous in some circles, unknown in others. Fame circles can nest and
overlap. Fame relationships can correlate imperfectly, so that if Person A is
known to Person X, they are 85% likely to be known to Person Y but only 10%
likely to be known to the average member of the population. While it seems to
be typical in the human case that if a process—say a representation of the
color red in some region of the visual field—is “famous” in the centers that
govern verbal report, it will also be famous in the centers that govern
long-term planning and control of the fingers. But widespread dissociations do
occur—for instance, in split-brain patients.[10] As Dennett (1991)
emphasizes, dissociations even occur in ordinary humans. This suggests a
complex range of partial cases that defy characterization as exactly “one
conscious subject” or two, or three, or seventeen.
Similar
considerations apply to virtually all naturalistic theories of consciousness.
Almost inevitably, they ground consciousness in processes or structures that
can be implemented in fuzzy, indeterminate, or partly overlapping ways,
suggesting mechanisms for unity that can be implemented in fuzzy,
indeterminate, or partly overlapping ways. Embodied theories of consciousness
are no exception, despite the fact that we usually imagine embodied organisms
as discrete individuals. Sponges, lichen, grasses, and birch forests connected
at the root are not as readily divisible into discrete individuals,
determinately countable. Octopus arms operate partly independent of the head:
is there one conscious subject or nine?[11] The Hogan Twins have
overlapping brains and differ in personality while having the capacity to
report each other’s sensory experiences.[12] AI architectures might similarly
admit of various types and degrees of integration. Must it really always be the
case, for every possible pair of craniopagus twins, that there is either
determinately one conscious subject or two?
In
general, if consciousness has a complex, naturalistic basis, it seems
correspondingly plausible that cases could arise where the number of conscious
subjects is similarly complex, defying determinate whole-number countability.[13] Those who would argue
otherwise owe us an explanation of why conscious subjectivity must always be determinately
countable despite the complexity and multidimensionality of the processes that generate
it.
4.
Between Zero and One
Start
with a system that is not and does not have a brain. Posit a series of changes
that render it, eventually, an apparently conscious system that determinately
is or that determinately has a brain. Maybe the system begins as a fertilized
embryo and ends as a newborn child. Maybe it begins as a simple form of
artificial intelligence and ends as an extraordinarily sophisticated one. Evaluate
the changes at arbitrarily precise time scales. Theories of conscious
subjecthood then face a quadrilemma. Either (1) the system is a conscious
subject the whole way through, even when it lacks a brain entirely; (2) the
system is never a conscious subject; (3) there is a sharp, stepwise distinction
where the system suddenly becomes a conscious subject; or (4) there is a
non-sharp or non-stepwise distinction between when the system is not a
conscious subject and when it is.
For
any particular transformation of this sort, option (1) will always be available
to certain kinds of radical panpsychists, who hold that conscious experience is
present not only in all fundamental entities but also in all composites of
fundamental entities. Such a panpsychist could say that the collection of fundamental
particles that constitutes the initial, non-brain system somehow already possesses
a unified field of conscious experience. Similarly, option (2) will be available
to radical eliminativists, who hold that conscious subjects don’t exist at all.[14] However, for present
purposes, we assume that options (1) and (2) won’t apply universally to all
such gradual series. We must then sometimes choose between (3)—saltation—and
(4)—indeterminacy or intermediacy. Note that intermediacy differs from
indeterminacy. The number of conscious subjects could be intermediate (for
example, exactly 0.3589) without being indeterminate. Conversely, the number
could be indeterminate (not determinately zero, not determinately one) without
being intermediate.
Saltation
is unattractive on broadly the grounds discussed in Section 3. It would be
surprising if, in general, atop smooth gradations of gradual structural change,
conscious subjecthood suddenly jumps in, with no indeterminacy or intermediacy at
all. Are we to imagine, for example, that lizards of one genus determinately
are conscious subjects while lizards of another genus, the tiniest bit less
sophisticated, are not? Nature does admit of sudden phrase transitions. Water
freezes at exactly 0.0°. Beams suddenly snap under loads. But, except in
quantum cases, even such phase transitions aren’t perfectly sharp. Close
inspection reveals intermediate states. Must there always be an exact moment,
down to the millisecond, down to the nanosecond, at which conscious subjecthood
suddenly pops in, with no intermediate or indeterminate phase? Furthermore,
must all such sequences lack any intermediate or indeterminate
phases? This is a bold claim! Only a powerful argument could justify it, and we
are aware of no published arguments within the framework of scientific
naturalism that attempt to do so.[15]
Accepting
that the number of conscious subjects can be indeterminate or intermediate
between zero and one doesn’t settle the best numerical means for representing
that possibility. The best approach in some cases might be a real value between
zero and one. For example, if we knew that a global workspace of size 1000 constituted
exactly zero conscious subjects, a global workspace of size 2000 constituted exactly
one conscious subject, and all intermediate workspace sizes were determinately
intermediate between zero conscious subjects and one, we could use values
between 0 and 1 to represent the number of subjects present. (It might remain
open whether the number of subjects related linearly, logarithmically,
sigmoidally, or in some other way to the size of the workspace.) Similarly, a
real value could measure how close a case of indeterminacy lies to the border
of determinacy (e.g., an indeterminate case much closer to being determinately
one conscious subject than zero might be represented by the number 0.9). On the
other hand, real numbers might suggest implausible precision, unless construed
as mere approximations. A less commissive representation might simply be the
open interval (0,1) or a set of possibilities among which the number is
indeterminate {0, 1}. Or one might opt for more structure: a vector or region
that represents several independent dimensions of intermediacy or indeterminacy.
Even imaginary numbers might be applicable if consciousness involves quantum
states represented by complex numbers.
5.
Between One and N
Luke
Roelofs (2019) constructs a similar slippery-slope case between two subjects
and one. Start with two conscious brains, wholly distinct. Slowly join them,
one neural connection at a time, until they form a single, unified subject of
experience.[16]
Either (1) the same, non-zero number of conscious subjects is present the whole
way through; (2) there aren’t, in fact, any conscious subjects during any stage
of the process; (3) there’s a sudden, sharp change in the number of conscious
subjects; or (4) there’s a gradual, non-sharp change in the number of conscious
subjects.
Again,
we assume the falsity of (1) and (2), although (1) is available to radical
panpsychists while (2) is available to radical eliminativists.[17] Option (3) seems just as
implausible here as in the zero-to-one case. If each step of integration is
sufficiently tiny, the architectural and functional differences will be
correspondingly tiny, and it’s unattractive to suppose that the seemingly huge
metaphysical difference between one and two conscious subjects would be
grounded in a tiny architectural or functional difference. And recall that the
saltationist is committed to a negative universal: There cannot be any
way of slowly connecting any two conscious subjects such that there is a
single moment of indeterminacy or intermediacy. Such a bold claim
requires compelling support.
If
N > 1, the best numerical representation of indeterminacy or intermediacy is
unlikely to be a single real number value. For example, in earlier work
(Schwitzgebel & Nelson 2023), we describe a case that allows a slippery
slope between 1 and 201. Imagine a conscious AI, perhaps employing a futuristic
technology very different from silicon chips. The entity is composed of a
large, orbiting AI system plus 200 robotic bodies on a planetary surface, each
with their own local AI processors. If the entity is massively interconnected
in the right way, it is plausibly a single conscious subject with multiple
bodies or one spatially discontinuous body.[18] If the entity is sparsely
connected, or connected in the wrong way, there are plausibly 200 or 201
distinct conscious subjects in communication with one another (201 if the
orbiting system is conscious, 200 if not). Between these extremes lies
approximately a continuum of possibilities, some of which may constitute cases
indeterminate or intermediate between 1 and 201. However, it would be misleading
to numerically represent the number of conscious subjects in an AI system indeterminate
between 1 and 201 in exactly the same way we represent the number when 101
typical humans gather in a room. Furthermore, such cases can be structured to
involve multiple independent dimensions of intermediacy or indeterminacy,
depending on the degree of integration between the orbiting AI system and each
individual robot, thus inviting multidimensional mathematical representation.
We
can also construct a zero to N case as follows: Start with our orbiter
and robots, all simple enough that none is conscious. Improve them all
simultaneously, one arbitrarily small step at a time. By analogy with our non-saltationist
reasoning above, there should be a range of cases indeterminate or intermediate
between zero and N.
6.
“But It’s Inconceivable!”
As
we descend further into such unfamiliar ways of thinking, we expect that many
readers will stop somewhere short, with a worry along these lines: It is inconceivable,
and therefore metaphysically impossible, that conscious subjects exist in
anything other than determinately countable wholes. At least, some readers may
be concerned that conscious subjects lacking determinate or discrete countability
are not “positively” conceivable, constituting grounds to doubt that
indeterminate or intermediate cases of conscious subjects are metaphysically
possible.[19]
In what follows, we’ll consider three distinct inconceivability challenges for
our view.
First, however, an observation. In conversation, we find philosophical interlocutors typically more open to indeterminacy than to intermediacy or other types of non-whole-number representation, and to express particular skepticism about fractional subjects, such as 7/8 of a subject. Perhaps they are right to be skeptical. It’s not obvious what could make a cognitive system constitute some particular fractional number of subjects (e.g., 0.8) rather than some other fractional number. Here, we remind the reader that we argue only for the disjunctive thesis of indeterminacy or non-whole-numbers. That said, we have never seen a well-developed argument against fractional subjects from a (non-panpsychist, non-eliminativist) naturalist perspective.[20]
On
non-whole-number representations larger than one, imagine a cloudscape.
Sometimes clouds are straightforwardly countable. Often, however, complex
cloudscapes are best represented in mathematically complex ways. Geometrical
representations might enable accurate representation of how the clouds clump
and cluster, perhaps with high mathematical precision. One needn’t shruggingly
say only “it’s indeterminate but more than three”. Similarly, one non-whole
number possibility worth considering is a massive AI system with unity and
disunity relations best understood by complex network representations.
6.1. Between
Zero and One
To conceive of a case intermediate or
indeterminate between zero and one, we must conceive of indeterminately or
intermediately conscious mental states (see Figure 1). After all, if there’s a
single, wholly conscious mental state present, there’s at least one conscious
subject. Can we imagine indeterminate or intermediate consciousness? Return to
Marcel’s experience of warm tea, a taste of madeleine, and recollections of
Combray. 
Now, subtract the
experience of warm tea and the taste of madeleine. Marcel’s mental state is not
one-third conscious, and consequently, Marcel has not been reduced to one third
of a conscious subject. He’s still a conscious subject, just with fewer
experiences. Now, subtract Marcel’s recollections of Combray. No conscious
state remains, and therefore, no conscious subject does either (at least at
that moment, according to our definition). This constitutes a discrete jump from
one to zero. Could we somehow subtract approximately half the experience
of recollection, making Marcel somehow half
conscious? We might imagine first a vivid recollection of his bedroom in
Combray with the glimmering flame of the nightlight in its bowl of Bohemian
glass, hung by chains from the ceiling; and then we might reduce the
experience. Forget the chains, forget the glass, recall only the glimmering
flame. Still, that memory seems determinately to be an experience, and thus,
one conscious subject determinately exists. Perhaps the flame can be remembered
only vaguely—what was its shape? Its color? Was it even a flame, or only a
glowing coal? However faint we imagine Marcel’s phenomenology, it seems we
still have some phenomenology, until we remove the experience altogether and
thus conscious subjecthood altogether. No intermediate state seems conceivable.
Either there’s something it’s like to be Marcel in that moment—no matter how
simple—or there’s nothing it’s like. No conceptual space stands open between
something and nothing. A half-something is already a something, unless it is
nothing.
We
acknowledge the pull of this way of thinking. However, we reply that there’s an
implicit self-contradiction in any attempt to imagine what it’s like to be an
entity indeterminate or intermediate between zero and one. Necessarily, there’s
no one determinate thing it’s like to be a borderline case of a conscious
subject. No one, regardless of their cognitive architecture, can grasp what
it’s like to be such an entity any more than they can grasp what it would be
like to perceive a square circle. The more you try to vividly imagine the
phenomenology of something that, by nature, lacks even a single wholly
conscious, determinate experience, the worse you miss the mark.
That
said, entities who regularly enter intermediate or borderline conscious states
might have no trouble discerning or self-representing such states, and they
might develop corresponding concepts. For example, they might be able to think,
“I spend a considerable period of time in that mizzy state between being
neither determinately unconscious nor determinately conscious” (alternatively,
partly conscious, if intermediacy is possible). They might be even able to imaginatively
reconstruct mizzy states through imaginative
episodes that are themselves mizzy Nonetheless,
just like us, they would fail to conceptualize the full, determinate experience
of borderline or partial conscious subjecthood, because there is no full,
determinate experience of borderline or partial conscious subjecthood.[21]
6.2. Between
One and N
What
would be required to conceive of an indeterminate or non-whole-number of subjects
between one and N (where N is a whole number greater than one)? We see two possibilities
between which we remain neutral.
6.2.1.
Indeterminate or intermediate unity among determinately conscious experiences. First,
in order to imagine subjects indeterminate between one and N, one might attempt
to imagine determinately conscious experiences that are indeterminately
or intermediately phenomenally unified. Rather than determinately
one or several distinct, unified bundles of conscious experience—for example,
A-with-B-with-C and D-with-E-with-F—we might conceive of something in-between
(see Figure 2). Perhaps the relation of phenomenal unity admits of degrees, or there
can be cases in which its presence is vague. On this picture, if Marcel were a
cognitive system intermediate or indeterminate between one and two subjects, he
would need to have both a determinate experience of the warmth of tea and a
determinate experience of the taste of madeleine, without those experiences
being either fully (that is, determinately and not intermediately) unified into
a conscious whole nor fully disunified into two separate wholes. He would need
to feel both experiences simultaneously, but neither entirely together nor entirely
apart.
Trying to conceive of
what that would be like, we might imagine the experience of the warmth
phenomenally unified with a hazy, faint, or peripheral-seeming experience of
the taste (or vice versa). But this isn’t a case of warmth indeterminately or
intermediately unified with the more robust taste from the original scenario;
it’s just a case of warmth determinately phenomenally unified with a hazy,
faint, or peripheral-seeming taste—a determinately conscious but less vivid
experience. A worry potentially arises: However much we try to imagine
conscious experiences being only partially severed from one another, it seems
we’re always left imagining wholly unified experiences, and therefore a single
conscious subject—that is, until we sever the tie altogether and are left with determinately
multiple conscious subjects. We imagine Marcel having his experiences conjointly,
or we imagine the experiences being felt in isolation. Those are, it seems, the
only conceivable options.
But
if our hypothetical objector seeks an act of imagination that joins together,
into the objector’s own, fully unified imaginative experience, two experiences that
aren’t fully unified, then again make an implicitly self-contradictory demand. Of
course, there isn’t one fully unified thing it’s like to feel two indeterminately
or intermediately unified experiences simultaneously. If the experiences were
determinately disjoint, the incoherence of this imaginative demand would
be obvious: We can’t expect to imagine what it’s like to have Marcel’s
experience of the warmth of tea jointly with Odette’s experience of a piano
sonata. Clearly, there’s no one
thing it’s like to have these two disjoint experiences. The same holds, though
less obviously, when the two experiences are indeterminately or intermediately
conjoint.
This
isn’t to say that indeterminate or intermediate phenomenal unity is inherently unimaginable.
For example, an indeterminately or intermediately phenomenally unified
conscious subject with an indeterminately or intermediately unified
imagination might have no trouble introspecting and reflecting on such
mental states. They might be able to think, “When my tentacles with neurons at
their tips are only partly connected to my body, I start having dissy
experiences that aren’t fully unified with one another.” Although they couldn’t
imagine dissy experiences as being like one thing any better than
we can, they might be able to conceive of dissy states through
imaginative episodes that are themselves dissy. Suppose Marcel
experiences the warmth of tea in one tentacle and experiences memories of
Combray in another tentacle, and these experiences are only partly unified.
Odette—or Marcel himself later—might imagine or remember this phenomenology by
means of an imaginative structure that is itself distributed among different
tentacles and not fully unified. This would not (we assume) be much like normal
human imagination, but what is alien to us needn’t be impossible.
6.2.2.
Intransitive unity. Alternatively, subjects might be intermediate
or indeterminate between one and N because they have overlapping fields of
consciousness that share token experiences. This would require the relation of
phenomenal unity to be intransitive. Most philosophers have assumed that
if an experience A is unified with another experience B, and if B is unified
with a third experience C, then A and C are themselves unified. However, if intransitive unity—or, as many have called it,
partial unity—is possible, then A and B might be unified, and B and C might be
unified, without A and C being unified as well (see Figure 3).[22]
To
understand how subjects intermediate or indeterminate between one and N would
look if phenomenal unity were intransitive, first consider two cases: Tiny
Overlap and Massive Overlap. In Tiny Overlap, two alien brains (or AI systems)
share a tiny bit of tissue. At the target moment, each alien has a million
experiences distinctive to them—vast turbulences of vivid, unified experiential
activity—plus one tiny shared experience: the faint sound of a distant motor. Plausibly,
Tiny Overlap should be conceptualized as a case of two conscious subjects who
happen to share one token experience. Massive Overlap is the complementary
case: a million experiences are shared, but two experiences are not—a green dot
in the left visual periphery and a red dot in the right visual periphery. The green
dot is unified with everything but the red dot, and the red dot is unified with
everything but the green dot. Plausibly, Massive Overlap should be
conceptualized as a case of a single conscious subject who is ever-so-slightly
disunified. Now, imagine that, like in Section 5, Tiny Overlap gradually becomes
Massive Overlap as the aliens’ brains slowly fuse. In the middle of this
process, there would be two partially overlapping fields of experience that
would plausibly count as neither determinately one nor two subjects.
Again, intransitive unity might seem
impossible because we’re unable to imagine what having an intransitively
unified perspective would be like.[23] If we try to imagine how it
would feel to experience the warmth of tea alongside the taste of madeleine,
and the taste of madeleine alongside recollections of Combray, while the warmth
of the tea and the recollections of Combray remain disunified, we draw a blank.
Once again, however, the demand is paradoxical. This hypothetical objector seeks
to imagine, in a transitively unified way, what it would be like to have intransitively
unified experiences. They seek to bring into their unified imagination an
experience of A with B with C that is somehow simultaneously an experience of A
with B and B with C but not A with C. This self-contradictory imaginative
standard is inappropriate to the case.[24] Intransitively unified
experiences aren’t like any one thing, or part of any one phenomenal
perspective.
Still,
we see no reason to believe that a creature with intransitively unified
consciousness couldn’t imagine, in an intransitively unified way,
intransitively unified experiences. Maybe Marcel’s Arm 1 is unified with his
head, and his head is unified with Arm 6, but Arms 1 and 6 are not unified. Even
though Marcel can’t join the experiences of his head and arms in one
phenomenally unified act of imagination, he might imagine, think about, introspect,
or recall them by means of a similarly distributed imagination.
Our
response to each of the three conceivability objections is the same. The
objections, we suspect, turn on our wanting to imagine these extraordinary
cases in the ordinary, unified way. It is of course often fine to want to
imagine things in the ordinary way. That’s why the imaginative demands
superficially seem reasonable. Sometimes we feel like we can’t quite get our
head around an experience if we can’t imaginatively construct or recreate it in
the way we imaginatively construct or recreate an experience of seeing a sunlit
mound of gold while feeling its warm, smooth texture. The ordinary way we
imagine experiences is as determinate experiences (perhaps with some
indeterminate contents) that are
wholly, determinately, transitively unified. It is inappropriate and
paradoxical to apply this standard of imagination to the non-whole-number cases
at hand.
More
reasonably, one might alter the standards of imagination to appropriately match
the target cases. Mizzy, dissy, and
intransitive states might seem alien to us, but they are unimaginable only in
the same way that color is arguably unimaginable to people blind from birth and
bat echolocation seems unimaginable to ordinary humans. Human limitations, not
metaphysical or nomological impossibility, explain our inability to concretely
imagine conscious subjects that don’t come in determinate, countable whole
numbers.
6.2.3
Countability partly recovered?
If we accept the intransitivity
approach to conceiving of subjects indeterminate or intermediate between one
and N, there still might be a sense in which fields of experience are countable.
After all, overlapping entities can come in determinate whole numbers. A = {x ∣ 1 ≤ x
≤ 999,999} and B = { x ∣
2
≤ x ≤ 1,000,000} aren’t indeterminate between two sets; rather,
they’re determinately two sets that share 99.9999% of their contents. The shapes
in Figure 4 don’t occupy somewhere between one and two regions of space. They
simply occupy massively overlapping regions. If it’s possible in principle to
measure or quantify the degree of overlap between two fields of consciousness, it
might be possible in principle to count up a whole number of overlapping conscious
fields, just as it’s possible to count up a whole number of overlapping sets
and an overlapping number of bounded, continuous regions of space.
One
might worry that understanding subjects this way would once again force us to
posit a sharp, inelegant jump atop a complex physical base. However, we see no
reason for concern: This approach would still allow for gradual phenomenological
transitions alongside gradual increases and decreases in functional integration.
Consider again an analogy to sets: suppose I pull out a piece of paper and list
the sets of numbers [1, 2, 3] and [2, 3, 4]. I now have two mathematical
objects instantiated on my paper. Suppose that I then erase the number one from
the first set and the number four from the second set. Now, I only have one
mathematical object instantiated. We could, at this point, be puzzled – how
bizarre that the transition from one entity to two was so clean! But abstract
objects can be distinct, even when the differences between them are tiny, and
they can collapse into one when those tiny differences are erased. The same
goes, potentially, with “conscious subjects” if we define them in terms of
overlapping sets or fields of experience.
However,
generalizing this approach to counting subjects generates what would seem to be
intuitively, pragmatically, and functionally the wrong result in the Massive
Overlap case: The tiniest bit of intransitivity would generate two discretely
different “subjects” even if the system is functionally, practically, and
introspectively almost identical to a single subject. This would be especially
worrying if, as Dennett suggests, ordinary humans are themselves often not
fully unified. Even if it’s possible in principle to count up massively
overlapping fields of experience, calling each one a “subject” seems bizarre. Furthermore,
this restoration of countability, if it works at all, works only for the
intransitive unity case. Zero-to-one and indeterminate or intermediate unity
cases will still ruin any strict countability principle.
In Section 2, we defined “conscious subjects” as
bundles of experience, individuated by maximal relations of
phenomenal unity. If, as we’ve just suggested, unity can come in degrees or be
intransitive or indeterminate, the relevant relation might be sufficient rather
than all-or-none unity and there might not always be a single best individuation
scheme.
7.
Conclusion
Dennett
challenged readers to leave behind their presuppositions about the features of
consciousness. We celebrate this aspect of his work. Our ordinary conception of
consciousness is grounded in a specific evolutionary, developmental, and social
history, mostly focused on typical human cases and a few familiar vertebrates,
as commonly understood. Even if typical humans are normally fully and
determinately phenomenally unified,[25] typical humans are a tiny
corner of the architectural possibility space. What seems metaphysically
necessary, as viewed from this corner, may prove instead to be a matter of
contingent fact.[26]
References
Alter, Torin
(2023). The matter of consciousness. Oxford
University Press.
Antony, Michael V.
(2008). Are our concepts conscious state
and conscious creature vague? Erkenntnis, 68, 239-263.
Baars, Bernard J.
(1988). A cognitive theory of
consciousness. Cambridge University Press.
Bayne, Tim (2010).
The unity of consciousness. Oxford University Press.
Bayne, Tim &
David J. Chalmers (2003). What is the unity of consciousness? In A. Cleeremans,
ed., The unity of consciousness. Oxford
University Press.
Blackmore, Susan
(2016). Delusions of consciousness. Journal
of Consciousness Studies, 23 (11-12), 52-64, 2016
Bostrom, Nick
(2006). Quantity of experience: Brain duplication and degrees of consciousness.
Minds and Machines, 16, 185-200.
Brook, Andrew
& Paul Raymont (2001/2017). Unity of consciousness. Stanford Encyclopedia of Philosophy, Summer 2021 edition.
Builes, David
(2021). The world just is the way it ss. The Monist, 104,: 1-27.
Carls-Diamante, Sidney
(2017). The octopus and the unity of consciousness. Biology and Philosophy
32(6): 1269-1287.
Carruthers, Peter
& Rocco Gennaro (2001/2020). Higher-order theories of Consciousness. Stanford Encyclopedia of Philosophy,
Fall 2023 edition.
Chalmers, David J.
(2002). Does conceivability entail possibility? In T.S. Gendler and J.
Hawthorne, eds., Conceivability and
possibility. Oxford University Press.
Cochrane, Tom
(2020). A case of shared consciousness. Synthese, 199, 1019-1037.
Coleman, Sam (2013).
The real combination problem: Panpsychism, micro-subjects, and emergence. Erkenntnis,
79, 19-44.
Dainton, Barry (2000).
Stream of consciousness. Routledge.
Dainton, Barry
(2014). Unity, synchrony, and subjects. In D.J. Bennett and C.S. Hill, eds., Sensory
integration and the unity of consciousness. MIT Press.
Dehaene, Stanislas
(2014). Consciousness and the brain. Viking.
Dennett, Daniel C.
(1991). Consciousness explained. Little,
Brown and Company.
Dennett, Daniel C.
(2005). Sweet dreams. MIT Press.
Dennett, Daniel C.
(2007). Heterophenomenology reconsidered. Phenomenology
and the Cognitive Sciences, 6, 247–270.
Dennett, Daniel C.
(2016). Illusionism as the obvious default theory of consciousness. Journal of Consciousness Studies, 23 (11-12), 65-72.
Dominus, Susan
(2011). Could conjoined twins share a mind? New
York Times Magazine (May 25):
https://www.nytimes.com/2011/05/29/magazine/could-conjoined-twins-share-a-mind.html
Fekete, Tomer,
Cees Van Leeuwen & Shimon Edelman (2016). System, subsystem, hive: Boundary
problems in computational theories of consciousness. Frontiers in Psychology, 7. DOI:
https://doi.org/10.3389/fpsyg.2016.01041.
Frankish, Keith
(2016). Not disillusioned: Reply to commentators. Journal of Consciousness Studies, 23 (11-12), 256-289.
Gendler, Tamar
Szabó & John Hawthorne (2002). Introduction: Conceivability and
possibility. In T.S. Gendler and J. Hawthorne, eds., Conceivability and possibility. Oxford University Press.
Godfrey-Smith,
Peter (2016). Other minds. Macmillan.
Goff, Philip
(2013). Orthodox property dualism + the Linguistic Theory of Vagueness =
Panpsychism. In R. Brown, ed., Consciousness
inside and out. Springer.
Goff, Philip &
Luke Roelofs (forthcoming). In defence of phenomenal sharing. In J. Bugnon et
al. eds., The phenomenology of self-awareness and conscious subjects.
Routledge.
Heller, Mark
(1996). Against metaphysical vagueness. Philosophical Perspectives, 10, 177-185.
Hill, Christopher
S. (2018). Unity of consciousness. WIREs
Cognitive Science, 9 (5), e1465.Hirstein, William (2012). Mindmelding:
Consciousness, Neuroscience, and the Mind’s Privacy. Oxford University Press.
Hurley, Susan L. (2003).
Action, the unity of consciousness, and vehicle externalism. In A. Cleeremans,
ed., The unity of consciousness. Oxford Academic.
Jackson, Frank
(1986). What Mary didn’t know. Journal of
Philosophy, 83, 291-295.
Kang, Shao-Pu
(2022). Shared consciousness and asymmetry. Synthese 200, (5), 1-17.
Lockwood, Michael
(1994). Issues of unity and objectivity. In C. Peacocke, ed., Objectivity, simulation and unity of
consciousness. Proceedings of the British Academy, 83.
Lycan, William G.
How far is there a fact of the matter? Journal of Consciousness Studies,
29 (1-2), 160-169.
Mallozzi,
Antonella, Anand Vaidya, & Michael Wallner (2007/2023). The epistemology of
modality. Stanford Encyclopedia of
Philosophy, Summer 2024 edition.
Masrour,
Farid. The phenomenal unity of consciousness. In U. Kriegel, ed., Oxford handbook
of the philosophy of consciousness. Oxford University Press.
Nagel, Thomas
(1971). Brain bisection and the unity of consciousness. Synthese, 22, 396-413.
O’Brien, Gerard
& Jonathan Opie (1998). The disunity of consciousness. Australasian Journal of Philosophy, 76, 378-95.
Oizumi, Masafumi,
Larissa Albantakis, & Giulio Tononi (2014). From the phenomenology to the
mechanisms of consciousness: Integrated Information Theory 3.0. PLoS Computational Biology, 10 (5),
e1003588.
Parft,
Derek (1984). Reasons and persons.
Oxford University Press.
Roelofs, Luke
(2016). The unity of consciousness, within and between subjects. Philosophical
Studies, 173, 3199-3221.
Roelofs, Luke
(2019). Combining minds. Oxford
University Press.
Roelofs, Luke (in
draft). What’s wrong with one-and-a-half minds?
Rosenthal, David
M. (2003). Unity of consciousness and the self. Proceedings of the Aristotelian Society, 103, 325-352.
Salisbury, Jenelle
Gloria (2023). The unity of consciousness
and the first-person perspective. Doctoral Dissertation, University of Connecticut.
Schechter,
Elizabeth (2014). Partial unity of consciousness: A preliminary defense. In
D.J. Bennett and C.S. Hill, eds., Sensory integration and the unity of consciousness.
MIT Press.
Schechter,
Elizabeth (2018). Self-consciousness and
“split” brains. Oxford University Press.
Schwitzgebel, Eric
(2007). No unchallengeable epistemic authority, of any sort, regarding our own
conscious experience—contra Dennett? Phenomenology
and the Cognitive Sciences, 6, 107-113.
Schwitzgebel, Eric
(2014). Tononi’s Exclusion Postulate would make consciousness (nearly)
irrelevant. Blog post at The Splintered
Mind (Jul 16).
Schwitzgebel, Eric
(2015). If materialism is true, the United States is probably conscious, Philosophical Studies, 172, 1697-1721.
Schwitzgebel, Eric
(2023). Borderline consciousness, when it’s neither determinately true nor
determinately false that experience is present. Philosophical Studies, 180, 3415-3439.
Schwitzgebel, Eric
(2024). The weirdness of the world. Princeton
University Press.
Schwitzgebel,
Eric, & Sophie R. Nelson (2023). Introspection in group minds, disunities
of consciousness, and indiscrete persons. Journal
of Consciousness Studies, 30 (9-10), 288-303.
Searle, John R.
(2000). Consciousness. Annual Review of
Neuroscience, 23, 557-578.
Simon, Jonathan A.
(2017). Vagueness and zombies: Why ‘phenomenally conscious’ has no borderline
cases. Philosophical Studies, 174, 2105–2123.
Shani, Itay &
Heath Williams (2022). The incoherence challenge for subject combination: an
analytic assessment. Inquiry: https://doi.org/10.1080/0020174X.2022.2124541
Sotala, Kaj & Harri
Valpola (2012). Coalescing minds: Brain uploading-related group mind scenarios.
International Journal of Machine Consciousness, 4, 293-312.
Strawson, Galen (2003).
What is the relation between an experience, the subject of experience, and the content
of experience? Philosophical Issues, 13, 279-315.
Tye, Michael
(2003). Consciousness and persons. MIT
Press.
Vogel, Jonathan
(2014). Counting minds and mental states. In
D.J. Bennett and C.S. Hill, eds., Sensory integration and the unity of
consciousness. MIT Press.
Volz, Lukas J., &
Gazzaniga, Michael S. (2017). Interaction in isolation: 50 years of insights
from split-brain research. Brain: A Journal of Neurology 140(7): 2051-2060.
Wundt, Wilhelm
(1897/1897). Outlines of psychology,
trans. C. H. Judd. Wilhelm Engelmann.
Yetter-Chappell,
Helen (in draft). The view from everywhere: Realist idealism without God.