|
We're going to try and post an
interesting paper on consciousness at least once a week. There are
debates among those philosophers and scientists who study
consciousness about pretty much every aspect of consciousness -
especially about what consciousness actually is. Many of these are
surprisingly easy to read, given the nature of their arguments.
This first paper called 'Being Conscious of Ourselves' was
written by David M. Rosenthal and published in The Monist issue 82,
2 (April 2004) in a special issue on self-consciousness.
BEING CONSCIOUS OF OURSELVES
Abstract: I argue that we can explain how we are
conscious of ourselves by appeal to essentially indexical
thoughts we have about ourselves, in particular
about our own current mental states. I show that being
conscious of ourselves in that way doesn't require that
we are aware of ourselves in some privileged way that's
antecedent to the higher-order thoughts we have about
our own mental states. The account successfully
resists, moreover, challenges based on the so-called
immunity to error through misidentification. And an
account based on such higher-order thoughts, finally,
also does justice to the way we identify and locate
ourselves as creatures in the world.
Slightly revised from The Monist, 87, 2 (April
2004), special issue on self-consciousness,
guest edited by José Luis Bermúdez
© David M. Rosenthal
I. Consciousness of the Self
What is it that we are conscious of when we are
conscious of ourselves? Hume famously despaired of finding any
self, as against simply finding various impressions and ideas,
when, as he put it, "I enter most intimately into what I call
myself."1 And, expanding on this, he wrote: "When I turn
my reflexion on myself, I never can perceive this self without some
one or more perceptions; nor can I ever perceive any thing but the
perceptions" (Appendix, p. 634).
It is arguable that the way Hume attempted to
become conscious of the self seriously stacked the deck against
success. Hume assumed that being conscious of a self would have to
consist in perceiving that self. Perceiving things does make one
conscious of them, but perceiving something is not the only way we
can be conscious of that thing. We are also conscious of something
when we have a thought about that thing as being present. I may be
conscious of an object in front of me by seeing it or hearing it;
but, if my eyes are closed and the object makes no sound, I may be
conscious of it instead by having a thought that it is there in
front of me.
Not all thoughts one can have about an object
result in one's being conscious of that object. We resist the idea
that having thoughts about objects we take to be distant in place
or time, such as Saturn or Caesar, make one conscious of those
objects; so the thought must be about the object as being present
to one. And the thought must presumably have an assertoric mental
attitude; doubting and wondering something about an object do not
make one conscious of that object. Nor does simply being disposed
to have a thought about something make one conscious of it; the
thought must be occurrent. But having an occurrent, assertoric
thought about an object as being present does intuitively make one
conscious of that object.
Hume would presumably have argued that this
alternative way of being conscious of things has no advantage here,
since he maintained that thinking consists simply of pale versions
of qualitative perceptual states. "All ideas," he insisted, "are
borrow'd from preceding perceptions" (Appendix, p. 634). And his
problem was about finding anything other than the mental qualities
of our perceptions.
But there is good reason to reject Hume's view
about the mental nature of having thoughts. For one thing, it is
difficult to see how perceptions could be combined to yield
thoughts with complex syntactic structure. For another, though
qualitative mental states arguably do represent things,2 they do so
in a way that is strikingly different from the way the intentional
content of thoughts represents things.3 Nor is there anything in
qualitative mentality that corresponds to the mental attitudes
exhibited by intentional states. And rejecting Hume's perceptual
model of what thoughts consist in makes room for a more promising
way to understand how it is that we are conscious of ourselves. We
are conscious of ourselves by having suitable thoughts about
ourselves.
The contrast between Hume's sensory approach and
the alternative that relies on the thoughts we have about ourselves
mirrors a contrast between two views about what it is for a mental
state to be a conscious state. On the traditional inner-sense
model, a mental state is conscious if one senses or perceives that
state; this is doubtless the most widely held view about the
consciousness of mental states.4 The higher-order-thought (HOT)
model, by contrast, holds instead that a mental state's being
conscious consists in its being accompanied by a suitable thought
that one is, oneself, in that state. On the version of the view
that I have developed and defended, the thought must be assertoric
and nondispositional. And, because the thought has the content that
one is, oneself, in that state, the thought automatically
represents the target mental state as being present.5
The difference between the inner-sense and HOT
views about what it is for a mental state to be conscious sheds
light on the two models of consciousness of the self. Suppose that
one's mental states are conscious in virtue of one's sensing
those states. Sensing a state consists in having a higher-order
sensation that represents the sensed states. But nothing in one's
sensing a mental state would make reference to or in any other way
represent any self to which the target state belongs. So nothing in
one's sensing a mental state would make one conscious of the
self.
Things are different if one is, instead, conscious
of one's conscious states by having thoughts about those states.
One will then have a thought that one is, oneself, in the target
state. And that HOT will thereby make one conscious not only of
that target state, but also of a self to which the HOT represents
the target state as belonging. The HOT model explains not only how
we are conscious of our conscious mental states, but how it is that
we are conscious of ourselves as well.
But can the HOT model of how mental states are
conscious do justice to the particular way we are conscious of
ourselves? There are two main reasons to doubt that it can do so.
The way we are conscious of ourselves seems, intuitively, to be
special in a way that simply having a thought about something
cannot capture. For one thing, there is a difference between having
a thought about somebody that happens to be oneself and having a
thought about oneself, as such. HOTs presumably must be about
oneself, as such. But having a thought about oneself, as such, may
seem to require some special awareness of the self that is
antecedent to and independent of the thought.
In addition, it seems to many that we are conscious
of ourselves in a way that affords a certain immunity to error. The
special epistemological access to the self which these phenomena
seem to suggest have even been thought to provide a foundation for
the identification of all other objects. How could such special
self-awareness arise if we are conscious of ourselves simply by
having thoughts about ourselves as being bearers of particular
mental states?
There is a second kind of challenge to a view about
consciousness of the self that relies simply on such HOTs. Although
it seems that we are conscious of ourselves in a way that is
special in the respects just sketched, our consciousness of
ourselves also fits with our ordinary, everyday ways of identifying
and locating ourselves in the world. Each of us is a being with
many conscious mental states. But each of us is also a creature
that interacts with other objects in the world. And we are
conscious of ourselves in both respects. How can simply having HOTs
about our mental states explain the way we are conscious of
ourselves as located within physical reality? How could having HOTs
ground our identifying ourselves as creatures interacting with many
other physical things?
The two challenges seem to pull in opposing
directions. It may be unclear at first sight how we could be
conscious of ourselves in a way that underwrites some kind of
immunity to error and yet also captures our contingent location and
identity in the physical world.
In what follows, I argue that a model of
self-consciousness based on HOTs can meet both these challenges. In
the next section I briefly sketch the reasons why the HOT model is
preferable to the inner-sense model in explaining what it is for a
mental state to be conscious. In sections III and IV, then, I take
up the first of the two challenges just described, to explain how
self-consciousness based on HOTs squares with the way our
consciousness of ourselves seems to be special. And in section V I
conclude by showing how that account also fits with the way we
identify and locate ourselves as creatures in the world.
II. Consciousness and HOTs
There is extensive evidence from both everyday life
and experimental findings that mental states occur without being
conscious. Such evidence relies on situations in which there is
convincing reason to believe that an individual is in some
particular mental state even though that individual sincerely
denies being in the state. Such sincere denials indicate that the
individual is not conscious of being in the state in question. We
take as decisive that a mental state is not conscious if an
individual is in that state but is in no way conscious of being in
it. It follows that whenever a mental state is conscious, the
individual in that state is, in some suitable way, conscious of
being in it.
Both the inner-sense and HOT models agree thus far.
They differ in what that suitable way is of being conscious of a
state, in virtue of which that state is conscious. It has often
been emphasized that, for a mental state to be conscious, one must
be conscious of it in a way that carries some kind of immediacy.
But the way we are conscious of our conscious states need only be
subjectively immediate. It need not be that nothing actually
mediates between the state and one's awareness of that state, but
only that nothing seems to mediate.
Ordinary perceiving seems to operate in just this
way. Nothing seems, subjectively, to mediate between the things we
see and hear and our seeing and hearing of them. From a
first-person point of view, perceiving seems to be direct. This
feature of perceiving makes the inner-sense model appealing, since
higherorder perceiving can do justice to the way one's
consciousness of one's conscious states seems to one to be
unmediated.
But the HOT model is no less successful in
capturing this aspect of the way we are conscious of our conscious
states. Some of the thoughts we have about things seem,
subjectively, to rely on inference, while others do not seem to do
so. When a thought seems subjectively to occur independently of any
inference, I shall refer to it as a noninferential thought. Since
subjective impression is all that matters here, a noninferential
thought may actually arise as a result of some inference, as long
as one is wholly unaware of the way the inference figures in that
thought's occurring. So, if one is conscious of being in a state
by having a noninferential HOT that one is in that state, it will
seem subjectively that nothing mediates between that state and
one's being conscious of it. HOTs can in this way explain the
intuitive immediacy that our awareness of our conscious states
exhibits.
Thus far, HOTs and inner sense seem equally good at
explaining what it is for a mental state to be conscious. But inner
sense faces a difficulty that disqualifies it from serious
consideration. Sensing and perceiving things take place by way of
qualitative mental states. And, when sensing or perceiving is
conscious, there is something it's like for one to be in these
states, something it's like in respect of the mental quality that
these states exhibit. So, if we are conscious of our conscious
states by way of some inner sense, then there are higher-order
qualitative states in virtue of which we are conscious of our
conscious states.
But it is clear that no such higher-order
qualitative states actually occur. One way to see this is
theoretical. Every mental quality belongs to some distinctive
perceptual modality; but what modality might the mental qualities
of such higher-order qualitative states belong to? It could not be
the modality of the firstorder conscious state, since that
perceptual modality is dedicated to perceiving a particular range
of perceptible properties and the first-order qualitative state
does not exhibit those perceptible properties. Visual states, for
example, exhibit mental qualities that reflect the similarities and
differences among the commonsense physical properties perceptible
by sight; but the visual states, themselves, do not exhibit
properties perceptible by sight. So the mental qualities of
higher-order qualitative states could not simply reduplicate the
first-order mental qualities. And there is no other perceptual
modality to which such higher-order qualities could belong.
Subjective considerations point to the same
conclusion. When our mental states are conscious in the ordinary,
everyday way, we are not conscious of the higher-order states in
virtue of which we are conscious of those first-order states. But
very occasionally we are actually aware of being conscious of those
firstorder states; when we introspect, we are conscious of being
aware of the introspected states. But even when we introspect, we
are never conscious of any mental qualities that characterize the
states in virtue of which we are conscious of those introspected
mental states. The higher-order states in virtue of which we are
conscious of our own mental states are not qualitative states.
The only alternative is that those higher-order
states are simply intentional states. We have already seen that
being in an assertoric, nondispositional intentional state that
represents the thing it is about as being present makes one
conscious of that thing. And, if one is not conscious of any
inference on the basis of which one holds that assertoric attitude,
so that the awareness the attitude results in seems spontaneous and
unmediated, then one will be conscious of the target state in the
subjectively unmediated way characteristic of our conscious states.
We are conscious of our conscious states by having HOTs to the
effect that we are in those states.
As noted at the outset, HOTs have the advantage
over inner sense that, unlike higher-order sensations, a HOT makes
one conscious of its target state as belonging to a particular
self. So each HOT makes one conscious of that self. And just as a
HOT, by being noninferential, makes one conscious of its target
state in a way that seems subjectively unmediated, so that HOT will
also make one conscious of the relevant self in a way that is
subjectively unmediated. HOTs do justice to our intuitive sense
that we have special, unmediated access to ourselves.
A proponent for the inner-sense model might argue
that, whatever one thinks about the foregoing considerations, inner
sense has a decisive advantage over the HOT model. When qualitative
states are conscious, there is something it's like for the subject
to be in those states, and this is absent when qualitative states
are not conscious. It seems, however, that HOTs could not be
responsible for this difference, since HOTs have no qualitative
mental properties. We can explain why there is something it's like
for one to be in conscious qualitative states, the argument goes,
only if the higher-order states in virtue of which we are conscious
of the first-order states are themselves qualitative states.
But this argument misconceives the situation. The
higherorder states are typically not themselves conscious; they are
conscious only when we are introspectively aware of our conscious
states. The reason to invoke higher-order states in virtue of which
some mental states are conscious is not because we are normally
conscious of such higher-order states, but because invoking such
higher-order states are theoretically well-founded. The
higher-order states, whether sensations or thoughts, are
theoretical posits, which we only occasionally become subjectively
aware of.
But, since the higher-order states typically are
not conscious, their being qualitative in character could not help
explain there being something it's like for one to be in conscious
qualitative target states. There will be something it's like for
one to be in a conscious qualitative state if one is conscious of
oneself, in a way that seems subjectively to be unmediated, as
being in a state of that qualitative type. And HOTs plainly make us
conscious of ourselves in that way.
There is some indirect evidence that HOTs actually
do result in there being something it's like for one to be in
conscious qualitative states. We sometimes become conscious of more
finegrained difference among our qualitative states by learning new
words for the relevant mental qualities. Consider the way new
mental qualities seem consciously to emerge when we learn new words
for the gustatory mental qualities that result from tasting similar
wines or the auditory mental qualities that arise when we hear
similar musical instruments. We can best explain how the learning
of words for mental qualities can have that effect by supposing
that we come to deploy new concepts corresponding to those words,
which enable us to have new HOTs about our qualitative states. HOTs
with more fine-grained content result in our qualitative states'
being conscious in respect of more fine-grained qualities. And, if
the intentional content of HOTs makes a difference to what mental
qualities we are conscious of, we can infer that HOTs also make the
difference between there being something it's like for one to be
in those states and there being nothing at all that it's like.
HOTs do result in our qualitative states' "lighting up."
III. Self-Consciousness and the Essential
Indexical
A HOT makes one conscious of oneself as being in a
particular mental state because it has the content that one is,
oneself, in that state. So a HOT must somehow refer to oneself. But
as already noted, not any way of referring to oneself do.
There are many descriptions that uniquely pick me
out even though I am unaware that they do so; I might believe that
some other individual satisfies one such description, or simply
have no idea who if anybody does. Consider John Perry's now
classic example of my seeing in a grocery store that somebody is
spilling sugar from a grocery cart and not realizing that the
person spilling sugar is me. My thought that the person spilling
sugar is making a mess refers to me, though it does not refer to me
as such. Perry describes this special way of referring to oneself
as the essential indexical; classical grammarians know it as the
indirect reflexive, since it captures in indirect discourse the
role played in direct quotation by the first-person pronoun.6
For a mental state to be conscious, it is not
enough that the individual one is conscious of as being in that
state simply happens to be oneself. Suppose that I am the unique F
and I have a thought that the unique F is in pain. That would not
make me conscious of myself as being in pain unless I was also
aware that I am the unique F. Suppose I thought instead that you
were the unique F. My thought that the unique F is in pain would
then hardly result in my pain's being conscious; it would not in
any relevant way make me conscious of myself as being in pain.
Essentially indexical self-reference is one way in
which our consciousness of ourselves is special. And, as already
noted, it is sometimes argued that essentially indexical
self-reference is required for identifying everything other than
oneself.7 But we rarely do identify other objects by reference to
ourselves. We almost always use some local frame of reference in
which we figure but which we identify independently of ourselves,
by way of various objects we perceive and know about. Such local
frames of reference occasionally fail, but when they do, referring
to ourselves seldom helps. Essentially indexical self-reference
cannot sustain such foundationalist epistemological leanings.
What exactly does such essentially indexical
self-reference actually consist in? How is it that we are able to
refer to ourselves, as such? An essentially indexical thought or
speech act about myself will have the content that I am F. So we
must consider how the word 'I' functions in our speech acts and
how the mental analogue of 'I' functions in the thoughts those
speech acts express.
The word 'I' plainly refers to the individual who
performs a speech act in which that word occurs. Similarly, the
mental analogue of that word refers to the thinker of the
containing thought, the individual that holds a mental attitude
toward the relevant intentional content. The word 'I' does not
have as its meaning the individual performing this speech act; nor
does the mental analogue of 'I' express the concept the
individual holding a mental attitude toward this content. One can
refer to oneself using 'I' and its mental analogue without
explicitly referring to any speech act or intentional state.
On David Kaplan's well-known account, the
reference of 'I' is determined by a function from the context of
utterance to the individual that produces that utterance; 'I'
does not refer to the utterance itself.8 The connection between the
words uttered and the act of uttering them is pivotal. 'I' refers
to whatever individual produces a containing utterance, but not by
explicitly referring to the utterance itself. Similarly, the mental
analogue of 'I' refers to whatever individual holds a mental
attitude towards a content in which that mental analogue figures,
though again not by explicitly referring to that intentional
state.
Suppose, then, that I have the essentially
indexical thought that I am F. My thought in effect describes as
being F the individual that thinks that very thought, "in effect"
because, although the thought does not describe the individual in
that way, it still does pick out that very individual. It does not
pick out that individual because the intentional content of my
thought so describes the individual. But whenever I do have a
first-person thought that I am F, my having that thought disposes
me to have another thought that identifies the individual the first
thought is about as the thinker of that first thought. In that way,
we can say that every first-person thought tacitly or
dispositionally characterizes the self it is about as the thinker
of that thought. Nothing more is needed for essentially indexical
self-reference.
HOTs are simply first-person thoughts, and they
function semantically just as other first-person thoughts do. So,
when I have a HOT that I am in a particular state, my thought in
effect describes as being in that state the individual who thinks
that thought. Though the thought itself does not describe that
individual as thinking that thought, the thinker of the thought is
disposed to pick that individual out in just that way, by being
disposed to have another thought that does so identify the
individual the first thought is about. Because HOTs function
semantically as other first-person thoughts do, the HOT hypothesis
explains why, when a mental state is conscious, one is conscious of
oneself in an essentially indexical way as being in that state.
It is important for the HOT model that when a
thought refers to oneself in this essentially indexical way, its
content does not describe the individual it refers to as the
thinker of the thought. If an essentially indexical first-person
thought did describe the individual it is about as the thinker of
that thought, simply having that thought would make one conscious
of having it. And, since HOTs are essentially indexical
first-person thoughts, one could not have a HOT without being
conscious of oneself as having it. But we are wholly unaware of
most of our HOTs.
It is sometimes objected to the HOT model that
nonlinguistic beings, including human infants, could not have HOTs.
But this is far from obvious. Many nonlinguistic beings presumably
do have some thoughts, and the conceptual resources HOTs use to
describe their target states could well, for these beings, be
fairly minimal. These beings would not be conscious of their
conscious states in the rich way distinctive of adult humans, but
that is not at all implausible.
Still, it might be thought that the essentially
indexical self-reference HOTs make preclude their occurring in
beings without language. It is natural to suppose that such
creatures have, in any case, no HOTs about their intentional
states. And a thought can make essentially indexical self-reference
only if one is disposed to identify the individual one's thought
refers to as the thinker of that thought. But it is natural to
suppose that such nonlinguistic beings would indeed so identify the
individual their HOTs refer to if they had suitable conceptual
resources. And that should be enough for a HOT to result in the
creature's being conscious of itself in the relevant way as being
in the state in question.
As noted above, the phenomenon of essentially
indexical self-reference encourages the idea that a certain kind of
reference to oneself occurs which provides an epistemic foundation
for the identification of all other objects. And that idea may, in
turn, make it seem tempting to urge that we must have some special
access to the self that is independent of the thoughts we have
about it. Some other form of self-consciousness, antecedent to
those thoughts, might then be needed for the essentially indexical
selfreference that our HOTs involve.
The foregoing explanation of the essential
indexical helps dispel that illusion. Reference to oneself as such
is simply reference to an individual one is disposed to pick out as
the very individual doing the referring. That disposition is
independent of the thought that refers to oneself in an essentially
indexical way. And that may encourage the idea that essentially
indexical selfreference requires independent, antecedent access to
the self. But the disposition to have another thought that
identifies the individual the first thought is about as the thinker
of that thought does not rest on or constitute independent access
to the self. It is simply a disposition to have another thought.
Essentially indexical self-reference raises no difficulty for an
account of self-consciousness in terms of HOTs.
IV. Self-Consciousness and Immunity to
Error
Thereis, however, another way in which our
consciousness of our own conscious states appears to raise problems
for such an account. There is a traditional view on which our
awareness of our conscious states is both infallible and
exhaustive. When a mental state is conscious, on this view, there
is no feature we are conscious of the state as having which it
fails to have, and no mental feature the state has of which we fail
to be conscious. There is thus no distinction, on this view,
between the reality of mental states and their appearance in
consciousness.9
Few today would endorse such a strong form of
privileged access. There is doubtless much about the mental natures
of our conscious states that we are unaware of, and much that we
are wrong about. Our access to our mental states often falls short
of exhaustiveness; we are often unclear about what we actually
think about things. Nor is that access infallible; there is robust
experimental evidence, for example, that we are sometimes wrong
about what intentional states issue in our choices and other
actions. People often confabulate being in intentional states to
explain their choices in situations in which the reported
intentional states could not have been operative.10 In these cases,
we are conscious of ourselves as being in states that we are not
actually in.
Errors of both types occur not only with
intentional states, but also in connection with conscious
qualitative states. When we consciously see red, we are often
conscious of the conscious sensation in respect of a relatively
generic shade, though the sensation exhibits a fairly specific
shade of red, as subsequent attention reveals. And it may even
happen that we have one type of bodily sensation or emotion but we
are conscious of ourselves as having a type different from that.
When local anesthetic blocks any actual pain, a dental patient may
still react to the fear and vibration caused by drilling by seeming
to be in pain; in such a case, one is conscious of oneself as being
in pain, though no pain actually occurs.
The idea that our access to our conscious states is
privileged often goes hand in hand with the view that a state's
being conscious is an intrinsic property of that state. If a
state's being conscious were intrinsic to that state, that would
explain our subjective sense that nothing mediates between the
states we are conscious of and our consciousness of them. We have
that subjective sense because nothing actually does mediate. And it
may be tempting to hold that, if nothing mediates between our
consciousness of a state and the state itself, consciousness could
not be erroneous; there would be no room for error to enter. But
that picture is unfounded. Even if one's consciousness of a state
were intrinsic to that state, it could still go wrong.
But there is an echo of such privilege which
persists in a view about the way we are conscious of ourselves.
This echo pertains not to the mental nature of the states we are
conscious of ourselves as being in, but to the self we are
conscious of as being in those states. Suppose I consciously feel
pain or see a canary. Perhaps I can be wrong about whether the
state I am in is one of feeling pain or seeing a canary. But it may
well seem that, if I think I feel pain or see a canary, it cannot
be that I am right in thinking that somebody feels pain or sees a
canary, but wrong in thinking that it is I who does those things.
Such first-person thoughts would, in Sydney Shoemaker's now
classic phrase, be "immune to error through misidentification,"
specifically with respect to reference to oneself.11
Shoemaker recognizes that such immunity to error
fails if one comes to have such thoughts in the way we come to have
thoughts about the mental states of others. As he notes (7), I can
wrongly take a reflection I see in a mirror to be a reflection of
myself; I thereby misidentify myself as the person I see in the
mirror. And I might thereby think that I have some property, being
right that somebody has that property but wrong that it is I who
has the property. So such immunity to error through
misidentification does not occur every time one has a thought that
one has a particular property. It must be that one's thought that
one has that property arises from the special way we seem to have
access to our being in conscious states.
There is reason to doubt, however, that such
immunity to error actually obtains. The way we have access to being
in conscious states is a matter simply of our having noninferential
HOTs that we are in those states. We have a subjective impression
that this access is special, since it appears to arise
spontaneously and without mediation. But that subjective impression
arguably results simply from the relevant HOTs' being based on no
conscious inference, and indeed from their typically not themselves
being conscious in the first place.
As with other thoughts, we come to have these HOTs
in a variety of ways, and the process by which HOTs arise can, like
any other process, go wrong. So, however unlikely it may be that
one is ever right in thinking that somebody is in a particular
state but wrong that the individual in that state is oneself, such
error is not impossible. One might, perhaps, have such strongly
empathetic access to another's state that one becomes confused and
thinks that it is oneself that is in that state. Such strong
immunity to error through misidentification does not obtain.
Still, something like this immunity to error does
hold. I can be mistaken about whether the conscious state I am in
is pain, for example, and perhaps even about whether I am the
individual that is actually in pain. But, if I think I am in pain,
it seems that I cannot be wrong about whether it is I that I think
is in pain. Similarly, if I think that I believe or desire
something, perhaps I cannot be mistaken about whether it is I that
I think has that belief or desire.
This differs from the immunity to error that
Shoemaker and others have described. On that stronger sort of
immunity, if I think I am in pain and am right that somebody is, I
cannot go wrong specifically about whether it is I who is in pain.
On the weaker type of immunity described here, all I am immune to
error about in such a case is who it is that I think is in pain. I
shall refer to this as thin immunity.
Plainly there are ways in which we can misidentify
ourselves. Not only might I misidentify myself by wrongly taking a
reflection I see in a mirror to be a reflection of myself; I might
wrongly take myself to be Napoleon, perhaps because of delusion of
grandeur, perhaps because of evidence about Napoleon that seems to
lead to me. And, if I do misidentify myself as the person in the
mirror or as Napoleon, I will also in that way misidentify the
person who has the pains, thoughts, desires, and feelings that I am
conscious of myself as having.
How can we capture the specific kind of
misidentification that thin immunity rules out? What distinguishes
such thin immunity to error through misidentification from the ways
in which we plainly can and sometimes do misidentify ourselves? The
error I cannot make is to think, when I have a conscious pain, for
example, that the individual that has that pain is somebody
distinct from me, but I can be mistaken about just who it is that I
am. How can we capture this distinction? And how can we explain the
thin immunity that we do actually have?
When I have a conscious pain, I am conscious of
myself as being in pain. If I think I am Napoleon, I will think
that Napoleon is in pain. What I cannot go wrong about is simply
whether it is I that I think in pain, that is, whether it is I whom
I am conscious of as being in pain. The question is what this
amounts to. The earlier discussion of essentially indexical
selfreference gives us a clue. When I refer to myself as such, I
refer to the individual I could also describe as doing the
referring. Similarly, the error of misidentification I cannot make
when I am conscious of myself as being in pain is to think that the
individual I think is in pain is distinct from the individual who
is conscious of somebody's being in pain. We can readily explain
this in terms of the HOT model. The mental analogue of the word
'I' refers to whatever individual thinks a thought in which that
mental analogue occurs. So every HOT tacitly represents its target
state as belonging to the individual that thinks that very
HOT.12
Suppose, then, that I have a conscious pain. Since
the pain is conscious, I also have a HOT to the effect that I am in
pain, and that HOT tacitly represents the pain as belonging to the
very individual that thinks that HOT, itself. The HOT in virtue of
which my pain is conscious in effect represents the pain as
belonging to the very individual who thinks that HOT. But the
individual who has that HOT is thereby the individual for whom the
pain is conscious; so one cannot in that respect misidentify the
individual that seems to be in pain. I am conscious of a single
individual as being in pain and also, in effect, as the individual
who is conscious of being in pain. The reason I cannot misidentify
the individual I take to be in pain as being anybody other than me
is simply that my being conscious of myself as being in pain
involves my identifying the individual I take to be in pain as the
very individual who takes somebody to be in pain.
These considerations clarify the connection between
thin immunity to error through misidentification and essentially
indexical self-reference. The word 'I' and its mental analogue
refer to the speaker or thinker, thereby forging a connection
between an intentional content in which 'I' or its mental
analogue figures and the mental attitude held toward that content
or the illocutionary act that verbally expresses it. My essentially
indexical use of 'I' or its mental analogue to refer to myself
relies on that connection; my thought or assertion that I am F in
effect represents as being F the very individual that thinks that
thought or makes that assertion. Similarly, because I identify the
individual I take to be in pain as the individual who takes
somebody to be in pain, no error of misidentification is in that
respect possible.
This explanation leaves open all manner of mundane
misidentification, such as my taking myself to be Napoleon or the
person in the mirror. All that I am immune to error about is
whether the individual I take to be in pain is me, that is, whether
it is the very individual that takes somebody to be in pain. My
immunity is simply a reflection of the way the first-person pronoun
and its mental analogue operate. But however they operate, it is
plain that I can mistakenly think that I am Napoleon or the
individual in the mirror.
The stronger immunity to error that Shoemaker
describes trades on the special way we have access to being in
conscious states. Since that access is a matter of noninferential
HOTs, which like any other thoughts can be mistaken, such strong
immunity fails. Thin immunity, by contrast, is wholly independent
of the processes by which HOTs arise. No matter how one comes to
have a HOT, one is disposed to identify the individual it
represents as being in a particular state as the very individual
that thinks that HOT. And this amounts to representing the
individual that is in the target state as being oneself. One cannot
go wrong about its being oneself that one represents as being in
the state.
Shoemaker writes that "[m]y use of the word 'I'
as the subject of [such] statement[s as that I feel pain or see a
canary] is not due to my having identified as myself something" to
which I think the relevant predicate applies (9). But one is
disposed to identify the individual one takes to do these things as
the individual who takes somebody to do them. So one is, after all,
disposed at least in this thin way to identify as oneself the
individual one takes to feel pain or see a canary.
Shoemaker offers the mirror case as an example of a
thought about oneself that is not immune to error through
misidentification; I see somebody's reflection in a mirror and
mistakenly think that I am that person. But so far as thin immunity
goes, this case is completely parallel to that of conscious pain.
If I take the person in the mirror to be me, I can be wrong about
whether the reflection is actually of me. But even here I cannot be
wrong about who it is that I take the reflection to be of; I take
the reflection to be of the very individual who is doing the
taking. In just that way, I could be wrong about whether the person
I take to be in pain is Napoleon, but I cannot be wrong about
whether the person I take to be in pain is the individual doing the
taking.
The contrast Shoemaker sees between cases in which
immunity does and does not occur echoes Wittgenstein's idea that,
though I could be mistaken about whether a particular broken arm is
mine, I cannot be mistaken about whether a particular pain is mine.
He writes: "To ask 'are you sure that it's you who have pains?'
would be nonsensical" (67, emphasis original).
But the cases do not differ in any significant way.
The error at issue for the strong immunity Shoemaker and
Wittgenstein see may be less likely for cases of conscious pain
than for broken arms, but it is not impossible. And the cases are
parallel in respect of thin immunity. I can be wrong about who the
individual is whose arm is broken or who is in pain. But just as I
cannot be wrong about whether the individual who takes somebody to
be in pain is the individual taken to be in pain, so I cannot be
wrong about whether the person who takes somebody's arm to be
broken is the person taken to have a broken arm. Thin immunity
results simply from the way 'I' and its mental analogue function
in our firstperson thoughts and speech acts.
As noted earlier, claims of privileged access to
conscious states tend to rely on the view that a state's being
conscious is an intrinsic property of the state itself. But the way
one is conscious of a mental state could misrepresent that state
even if it were intrinsic to that state. Misrepresentation need not
be external to the thing being represented. But the idea that being
conscious of our mental states is intrinsic to those states does
shed light on why, even in respect of thin immunity, the mirror and
broken-arm cases seem to be different from the pain case.
Suppose I am in pain and the pain is conscious. Its
being conscious consists in my being conscious of myself as being
in pain. And suppose that the pain's being conscious is intrinsic
to the pain itself. It follows that my being conscious of myself as
being in pain will then be intrinsic to the pain itself. But my
being conscious of myself as being in pain means that the
individual I am conscious of as being in pain is the very
individual who is conscious of somebody as being in pain. So it
will then be intrinsic simply to my being in pain that I cannot, in
that respect, be mistaken about the individual I am conscious of as
being in pain.
When I take myself to be Napoleon or to be
reflected in a mirror or to have a broken arm, the individual I
take to have these properties is again the individual doing the
taking. But now an apparent difference from the pain case emerges.
Even if one is conscious of oneself as being Napoleon or having a
broken arm, one's being thus conscious plainly is not intrinsic to
those conditions. So, if a pain's being conscious were intrinsic
to the pain, the Napoleon and broken-arm cases would indeed differ
from the case of a mental state.
The idea that a mental state's being conscious is
intrinsic to that state even helps explain the initial plausibiliy
of the stronger immunity that Shoemaker describes. If being
conscious of a mental state were intrinsic to that state, it would
be intrinsic simply to being in a conscious state that one is
disposed to regard as being in that state the individual that takes
somebody to be in that state. Since it would be intrinsic to one's
being in a conscious state that it is oneself that one takes to be
in that state, there would be no process that leads to one's
identifying oneself as the individual that is in the state in
question. So there would be no identifying process that could go
wrong, and so no way for one to be right in thinking that somebody
is in a conscious state but wrong that it is oneself who is in the
state.
It is subjectively tempting to see consciousness as
an intrinsic feature of our mental states precisely because we are
seldom aware, from a first-person point of view, of anything that
mediates between conscious states and our consciousness of them. To
sustain this subjective impression, however, one would need some
way of individuating mental states on which our awareness of a
conscious state is not distinct from the state itself. It is hard
to see what means of individuation would have this result which
would be independent from the subjective impression under
consideration.
Indeed, the way we do actually individuate
intentional states seems to deliver the opposite result. No
intentional state can have two distinct types of mental attitude,
such as the attitudes of mental affirmation and doubt. And having a
doubt about something does not result in one's being conscious of
that thing. So when a case of doubting is conscious, our
consciousness of that doubting must exhibit an assertoric mental
attitude. And that means that the consciousness of the doubting is
distinct from the doubting itself.
There are other more general reasons to reject the
idea that being conscious of a mental state is intrinsic to that
state. States may be conscious at one time but not another, as with
minor aches or pains that last all day but are not always
conscious. If a state's being conscious were intrinsic to that
state, it would be puzzling how a particular state could at one
moment be conscious but not at another. And, if consciousness of
mental states is not intrinsic to those states, there is no reason
to hold that the stronger immunity Shoemaker describes obtains, nor
that the thin immunity that holds for conscious states differs from
that which holds for any other self-ascription.
What about an account, then, of the way we are
conscious of ourselves that appeals simply to HOTs? It seemed
possible at the outset that the phenomena of essentially indexical
self-reference and immunity to error through misidentification
might undermine a HOT account of our consciousness of the self.
That was because both immunity and the essential indexical seemed
to presuppose our having some special access to the self
independent of whatever thoughts we have about the self.
Essentially indexical self-reference, we saw,
presents no such problem. Essentially indexical first-person
thoughts refer to oneself in effect as the individual doing the
referring; they refer to oneself as an individual one is disposed
to pick out as the individual that thinks the essentially indexical
thought. So having such thoughts requires no access to the self
beyond that which we have by having thoughts about the self.
A similar conclusion holds for thin immunity to
error through misidentification. It might seem that such immunity
requires a privileged type of access to the self; how, otherwise,
could we be immune to error in referring to the self? But the error
to which we actually are immune is wholly trivial. It is the error,
when I take myself to have some property, F, of supposing
that the individual taken to be F is distinct from the
individual that takes somebody to be F.
We are immune to error through misidentification of
the self. But that immunity presupposes no special access we have
to the self. It is simply that one cannot, when one thinks that one
is, oneself, F, be wrong about whether it is the
individual doing the thinking that one takes to be F. No
independent, antecedent access to the self figures here, only a
particular form of self-reference. So nothing about immunity to
error through misidentification blocks a HOT account of the way we
are conscious of ourselves.13
V. Identifying Oneself and
Self-Consciousness
The idea that immunity to error through
misidentification occurs in connection only with the self-ascribing
of conscious states, but not in connection with broken arms and
being Napoleon, may seem to support the Cartesian view that we
identify ourselves in the first instance as mental beings. Why else
would misidentifying oneself be impossible only in connection with
mental self-ascriptions?
This picture is unfounded. For one thing, the error
we actually are immune to does not, in any substantive way, involve
the identifying of anything. It is simply the error, when I take
myself to have the property, F, of thinking that the individual
taken to be F is distinct from the individual that takes somebody
to be F. It is perhaps even a bit misleading to describe the error
we are thus immune to as one of misidentification.
Such immunity fails to support the Cartesian
conclusion for other reasons as well. Since the immunity applies
not simply to the ascribing to oneself of conscious states, but to
the selfascribing of nonmental properties as well, it cannot
sustain the idea that we identify ourselves primarily in mental
terms. How, then, do we identify ourselves? And how does our
identifying ourselves fit with the way our thoughts about ourselves
involve essentially indexical self-reference and are thinly immune
to error through misidentification?
There is no single way in which we identify
ourselves. We rely on a large and heterogeneous collection of
factors, ranging from considerations that are highly individual to
others that are fleeting and mundane. We appeal to location in time
and place, current situation, bodily features, the current and past
contents of our mental lives, and various psychological
characteristics and propensities, indeed, to all the properties we
believe ourselves to have. Contrary to the picture conjured up by
essentially indexical self-reference and immunity to error, the
factors that figure in our identifying ourselves are theoretically
uninteresting and have relatively little systematic connection
among themselves.
Each of these factors reflects some belief one has
about oneself, such as what one's name is, where one lives, what
one's physical dimensions and location are, and what the current
contents are of one's consciousness. And all these beliefs
self-ascribe properties by making essentially indexical
self-reference, and they are all immune to error through
misidentification in the thin way described above. Such thin
immunity and indexical self-reference figure in the way we identify
ourselves not because they provide or presuppose any special access
to the self, but only because the first-person beliefs on which all
self-identification relies exhibit such thin immunity and indexical
self-reference.
We can be in error about any of these beliefs about
ourselves; indeed, we could be in error about most of them. One
could be wrong about all one's personal history, background, and
current circumstances. One might even be mistaken about one's
location relative to other objects if, for example, one lacked
relevant sensory input14 or the input one had was suitably
distorted.
One can be wrong even about what conscious states
one is currently in. One may take oneself, in a distinctively
firstperson way, to have beliefs and preferences that one does not
way be wrong even about the sensations or emotions one is conscious
of oneself as having.
We identify ourselves by reference to batteries of
descriptions which our first-person thoughts and beliefs ascribe to
ourselves. And we can successfully distinguish ourselves from
others even if many of those descriptions are inaccurate. What,
then, if all identifying thoughts and beliefs of the sorts just
described are erroneous? Can one identify oneself even then?
Arguably not. We distinguish ourselves from other
beings, just as we distinguish among all other individuals, on the
basis of various properties. So, if one's beliefs about what
properties one has are all incorrect, one has nothing accurate to
go on. Our incorrect self-ascriptions would still make essentially
indexical self-reference and would still exhibit the thin immunity
to error described above. But these features of one's
self-ascriptions would not help in identifying oneself, since they
tell us only that the individual thought to satisfy a particular
description is the individual doing the thinking. Essentially
indexical selfreference and immunity to error through
misidentification cannot short circuit the need to appeal to
substantive properties in identifying oneself.
Even if I am conscious of myself as having some
thought or desire or as being in pain, I may nonetheless lack that
thought, desire, or pain. Consciousness of our mental states is not
infallible. But can I also, in such a case, be wrong even about
whether it seems that I have that thought, desire, or pain? Perhaps
there is, after all, a kind of privileged access we have not in
connection with whether our consciousness is correct, but in
connection with what our consciousness is. Perhaps appearance and
reality coincide at least in that respect; perhaps, then, it makes
no sense to talk about the way things seem to seem to one, as
against simply the way things seem.15 If so, perhaps our conscious
states do provide an unimpeachable basis for identifying ourselves,
not because we are infallible about the states we are conscious of
ourselves as being in, but because we are infallible about whether
we are conscious of being in them.
But no infallibility arises here either. One may be
wrong about any mental state one takes oneself to be in. Being
conscious of oneself as being in some mental state is itself,
however, just another higher-order mental state; on the HOT
hypothesis, it is a thought one has that one is, oneself, in that
state. So one could about whether one has the HOT in question. Such
higher-order infallibility fares no better than infallibility about
first-order states, and can provide no certain foundation for
identifying oneself.
A HOT account of the way we are conscious of
ourselves relies on a subset of the essentially indexical
first-person thoughts we have about ourselves, namely, our HOTs.
But the HOTs an individual has are about the same individual as all
the other essentially indexical first-person thoughts that that
individual has. In this respect, if not in others, the pronoun
'I' and its mental analogue function somewhat as proper names do.
When we use a proper name, we take each token to refer to the same
individual as other tokens do unless countervailing information
overrides that default. Similarly with 'I' and its mental
analogue; we assume that each token refers to the same thing unless
something interferes with that ordinary default assumption.16
The upshot is that we take all our essentially
indexical first-person thoughts and beliefs to refer to one and the
same individual. The way we are conscious of ourselves is therefore
but one aspect of the way we identify ourselves as individuals. We
are in the first instance conscious of ourselves by way of our
HOTs, but we identify ourselves by way of all our essentially
indexical first-person thoughts and beliefs.
David M. Rosenthal City University of New York
Graduate Center Philosophy and Cognitive Science Internet:
This email address is being protected from spam bots, you need Javascript enabled to view it
May 18, 2004
NOTES
1 David Hume, A Treatise of Human Nature [1739],
ed. L. A. Selby-Bigge, Oxford: Clarendon Press, 1888, 2nd edn.,
revised by P. H. Nidditch, 1978, Book I, Part IV, sec. vi, p.
252.
2 One can capture the way qualitative states
represent things by seeing each mental quality as representing the
perceptible physical property that occupies a corresponding place
in the quality space of the relevant perceptual modality. I have
defended this view in "The Colors and Shapes of Visual
Experiences," in Consciousness and Intentionality: Models and
Modalities of Attribution, ed. Denis Fisette, Dordrecht: Kluwer
Academic Publishers, 1999, pp. 95-118, and "Sensory Quality and the
Relocation Story," Philosophical Topics, 26, 1 and 2 (Spring and
Fall 1999): 321-350.
3 Pace the representationalist or intentionalist
views of writers such as D. M. Armstrong The Nature of Mind, St.
Lucia, Queensland: University of Queensland Press, 1980, ch. 9;
William G. Lycan, Consciousness, Cambridge, Massachusetts: MIT
Press/ Bradford Books, 1987, ch. 8; Gilbert Harman, "Explaining
Objective Color in terms of Subjective Reactions," Philosophical
Issues: Perception, 7 (1996): 1-17; and Alex Byrne, "Intentionalism
Defended," The Philosophical Review, 110, 2 (April 2001): 199-240.
I discuss representationalism in "Introspection and
Self-Interpretation," Philosophical Topics 28, 2 (Fall 2000):
201-233.
4 Kant first used the term 'inner sense'
(K.d.r.V., A22/B37); Locke used the similar 'internal Sense'
(Essay, II, i, 4). The view is currently championed by D. M.
Armstrong, "What is Consciousness?", in David Armstrong, The Nature
of Mind, St. Lucia, Queensland: University of Queensland Press,
1980, 55-67, and by William G. Lycan, Consciousness and Experience,
Cambridge, Massachusetts: MIT Press/Bradford Books, 1996, ch. 2,
pp. 13-43, and "The Superiority of HOP to HOT," forthcoming in
Higher-Order Theories of Consciousness, ed. Rocco J. Gennaro, John
Benjamins Publishers, 2004.
5 See, e.g., Rosenthal, "Two Concepts of
Consciousness," Philosophical Studies 49, 3 (May 1986): 329-359;
"Thinking that One Thinks," in Consciousness: Psychological and
Philosophical Essays, ed. Martin Davies and Glyn W. Humphreys,
Oxford: Basil Blackwell, 1993, 197-223; "A Theory of
Consciousness," in The Nature of Consciousness: Philosophical
Debates, ed. Ned Block, Owen Flanagan, and Güven
Güzeldere, Cambridge, MA: MIT Press, 1997, 729-753; and
"Explaining Consciousness," in Philosophy of Mind: Contemporary and
Classical Readings, ed. David J. Chalmers, New York: Oxford
University Press, 2002, 406-421. The first two will appear, along
with other papers that develop the HOT model, in Consciousness and
Mind, Oxford: Clarendon Press, forthcoming 2004.
6 John Perry, "The Problem of the Essential
Indexical," Noûs XIII, 1 (March 1979): 3-21. For reference to
oneself as such, see P. T. Geach, "On Beliefs about Oneself,"
Analysis 18, 1 (October 1957): 23-24; A. N. Prior, "On Spurious
Egocentricity," Philosophy, XLII, 162 (October 1967): 326-335;
Hector-Neri Castañeda, "On the Logic of Attributions of
Self-Knowledge to Others," The Journal of Philosophy, LXV, 15
(August 8, 1968): 439-56; G. E. M. Anscombe, "The First Person," in
Mind and Language, ed. Samuel Guttenplan, Oxford: Oxford University
Press, 1975, pp. 45- 65; David Lewis, "Attitudes De Dicto and De
Se," The Philosophical Review LXXXVIII, 4 (October 1979): 513- 543;
and Roderick M. Chisholm, The First Person, Minneapolis: University
of Minnesota Press, 1981, chs. 3 and 4.
7 See, e.g., Sydney Shoemaker, "Self-Reference
and Self- Awareness," The Journal of Philosophy LXV, 19 (October 3,
1968): 555-567, reprinted with slight revisions in Shoemaker,
Identity, Cause, and Mind: Philosophical Essays, Cambridge:
Cambridge University Press, 1984, pp. 6-18 (pages references are to
the reprinted version); Roderick M. Chisholm, Person and Object: A
Metaphysical Study, La Salle, Illinois: Open Court Publishing
Company, 1976, ch. 1, §5, and The First Person, ch. 3, esp.
pp. 29-32; and David Lewis, "Attitudes De Dicto and De Se."
8 David Kaplan, "Demonstratives," in Themes From
Kaplan, ed. Joseph Almog, John Perry, and Howard Wettstein, with
the assistance of Ingrid Deiwiks and Edward N. Zalta, New York:
Oxford University Press, 1989, 481-563, pp. 505-507.
9 See, e.g., Thomas Nagel: "The idea of moving
from appearance to reality seems to make no sense" in the case of
conscious experiences ("What Is It Like to Be a Bat?" The
Philosophical Review LXXXIII, 4 [October 1974]: 435-450; reprinted
in Mortal Questions, Cambridge: Cambridge University Press, 1979,
165-179, p. 174). If every mental state is identical with some
physical state, then every mental state has both mental and
physical properties. This thesis that we have exhaustive access to
our own mental states therefore applies only to the mental
properties.
10 The classic study is Richard E. Nisbett and
Timothy DeCamp Wilson, "Telling More Than We Can Know: Verbal
Reports on Mental Processes," Psychological Review LXXXIV, 3 (May
1977): 231- 259. A useful review of the extensive literature that
follows that study occurs in Peter A. White, "Knowing More than We
Can Tell: 'Introspective Access' and Causal Report Accuracy 10
Years Later," British Journal of Psychology, 79, 1 (February 1988):
13- 45.
11 Sydney Shoemaker, "Self-Reference and
Self-Awareness," p. 8. Shoemaker urges that such immunity applies
even when I take myself to be performing some action. See also
Ludwig Wittgenstein, The Blue and Brown Books, Oxford: Basil
Blackwell, 1958, 2nd edn. 1969, pp. 66-7, Gareth Evans,
"Demonstrative Identification," in Evans, Varieties of Reference,
ed. John McDowell, Oxford: Clarendon Press, 1982, pp. 142-266,
José Luis Bermúdez, The Paradox of
Self-Consciousness, Cambridge, Massachusetts: MIT/Bradford, 1998,
chs. 1 and 6, and Roblin Meeks, Identifying the First Person,
unpublished Ph.D. dissertation, The City University of New York
Graduate Center, 2003, chs. II-IV.
12 Tacitly, once again, because the content of a
HOT never explicitly describes the individual as the thinker of the
HOT.
13 I discuss both essentially indexical
self-reference and the thin sort of immunity to error through
misidentification in "Unity of Consciousness and the Self,"
Proceedings of the Aristotelian Society 103, 3 (April 2003):
325-352, in explaining the apparent unity of our conscious states.
Here, unlike in the earlier discussion, I stress the difference
between thin immunity and the kind of immunity described by
Shoemaker.
14 As G. E. M. Anscombe imagines; see "The First
Person," in Mind and Language: Wolfson College Lectures 1974, ed.
Samuel Guttenplan, Oxford: Oxford University Press, 1975, 45-65, p.
58.
15 See Daniel C. Dennett's contention that it
makes no sense to talk "about the way things actually, objectively
seem to you even if they don't seem to seem that way to you"
(Consciousness Explained, Boston: Little, Brown and Company, 1991,
p. 132). 16 As presumably happens with so-called Multiple
Personality Disorder (now more often known as Dissociative Identity
Disorder).
|
Consciousness 
