Theories of Consciousness

Theories_of_Consciousness_An_Introduction_----_HOT_Theory_The_Mentalistic_Reduction_of_Consciousness.pdf

Home >Education homework help >Theories of Consciousness

HOT THEORY: THE MENTALISTIC REDUCTION OF CONSCIOUSNESS

Box 3.1 • Preview

The higher-order thought (HOT) theory of consciousness asserts that a mental state is conscious if it is the object of a thought about it. Given that we have some naturalistically acceptable understanding of thoughts independent of the problem of consciousness, HOT theory promises a mentalistic reduction of consciousness. Then, the naturalistic account of non-conscious mind – which is presumably relatively easy to attain – solves the whole mind–body problem. HOT theory makes substantial assumptions . It assumes that the mind’s contents divide into the intentional (or representational) and the non-intentional (qualia, sensations). It assumes that consciousness requires conceptual thought, and what is more, requires apparently pretty sophisticated concepts about mental states as such. It assumes that no mental state is essentially a conscious state. It comes dangerously close to assuming that consciousness is always and only of mental states. Not all these assumptions are plausible, and they lead to many objections (e.g. can animals, to whom the ability to engage in conceptual thought may be doubted, be conscious; what is an unconscious pain, etc.). Some objections can be deflected, but problems remain that engage the generation problem and prevent the mentalistic reduction from going through successfully.

Philosophers have always been attracted by projects aiming to reduce consciousness to Something Else, even if this reduction might require a more or less radical reconception of our understanding of consciousness. They have been motivated by the hope that, as compared to consciousness, the Something Else would prove more tractable to analysis and would fit more easily into the physicalist world view (here it is perhaps encouraging that, compared to consciousness, almost anything else would possess these relative virtues). In the tradition of Descartes, consciousness was supposed to exhaust the realm of the mind, which itself thus became something immediately apparent and open to the mind’s own self inspection (inasmuch as conscious states of mind were somehow essentially self- intimating). There is of course something intuitively appealing to such a thesis

Seager, William. Theories of Consciousness : An Introduction, Taylor & Francis Group, 1999. ProQuest Ebook Central, http://ebookcentral.proquest.com/lib/oculwlu-ebooks/detail.action?docID=169266. Created from oculwlu-ebooks on 2020-04-07 21:28:23.

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

but we have long since lost any sense that it must be true and are now happy to countenance legions of unconscious mental states and hosts of cognitive processes existing beneath or behind our conscious mental states. As we saw in chapter 1, even Descartes ended up endorsing a form of the view that finds cognition, or cognition-like phenomena outside of consciousness. A second traditional idea, one stemming from the empiricist heritage, is that there are basic or ‘atomic’ elements of consciousness which are pure sensory qualities and from which all ‘higher’ states of consciousness are constructed, either by complex conjunction or mental replication, or both. Hume, for example, calls these atomic elements the simple impressions.1 The impressions are the truly immediate objects of consciousness and their occurrence is supposed to be entirely independent of thought. The radical proposal of the HOT theories is to deny this last claim. What if consciousness were in fact dependent upon certain sorts of thoughts which themselves were part of the now admissible zone of unconscious mentation?

The appealing possibility is that consciousness is somehow a definable relation holding between certain mental states, where the latter do not already essentially involve consciousness and, of course, are in themselves less puzzling than consciousness itself. A mentalistic reduction of consciousness would have several virtues. The explanation of consciousness in terms of mentality would avoid the direct explanatory leap from consciousness to the physical, a leap which has always seemed somewhat to exceed philosophy’s strength. If consciousness can be reduced to anything at all, it is evidently more plausible that it be to something already mental than directly to brute matter. Yet mental states which do not intrinsically involve consciousness can be seen as ‘closer’ to the natural, physical world, and so this sort of reduction promises to build a bridge across our explanatory gap, supported by intermediate mental structures which can be linked to both sides with relative ease.

In order to evaluate such a project we require a precise specification of, first, the relevant non-conscious mental states and, second, the relation between them that is to account for consciousness. One such reductive theory, distinguished by its clarity and detailed presentation, has been advanced by David Rosenthal, first in ‘Two Concepts of Consciousness’ (1986) and then in a series of papers that have appeared over the last decade (see for example 1993a, 1993b, 1995). My aim here is to review Rosenthal’s theory and to argue that, in the end, it fails to reduce consciousness successfully. I will not claim outright that any theory of the sort we are considering must similarly fail, but I confess that the wide scope and extensive development of Rosenthal’s theory makes me doubt whether there are other theories of this sort which differ significantly from it. Thus I hope my objections will possess a quite general applicability.2

Rosenthal begins by dividing mental states into the two traditional, and presumably exhaustive, classes: intentional mental states (e.g. beliefs, hopes, expectations, etc.) and phenomenal or sensory mental states (e.g. pains, visual sensations, etc.).3 For now I’ll follow Rosenthal in this distinction, but it is in fact a substantial assumption which I shall doubt for much of the rest of this book, and

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

one that is curiously unsupported by the details of the HOT theory. Rosenthal understands this distinction in terms of a division of mentalistic properties, so:

All mental states, of whatever sort, exhibit properties of one of two types: intentional properties and phenomenal, or sensory, properties. . . . Some mental states may have both intentional and phenomenal properties. But whatever else is true of mental states, it is plain that we would not count a state as a mental state at all unless it had some intentional property or some phenomenal property.

(1986, p. 332)

The first demand of theory specification is then met by asserting that no mental states are intrinsically or essentially conscious. This sweeping assertion would appear to be necessary to ensure the completeness of the theory, for otherwise there would remain a species of consciousness – the essential, non-relational sort of consciousness – for which the theory would offer no account. The claim that mental states are not intrinsically conscious is most plausible for the intentional states and least plausible for the phenomenal states, but there are some intuitive grounds for both. It is undeniable that we frequently ascribe intentional states of which we claim the subject is not conscious, even as we also claim that these intentional states are part of the causes and explanation of the subject’s behaviour. As for phenomenal states, Rosenthal offers this:

Examples of sensory states that sometimes occur without consciousness are not hard to come by. When a headache lasts for several hours, one is seldom aware of it for that entire time. . . . But we do not conclude that each headache literally ceases to exist when it temporarily stops being part of our stream of consciousness, and that such a person has only a sequence of discontinuous, brief headaches.

(1986, p. 349)

Of course, this is contentious, for one naturally wants to draw a distinction between the headache and the persistent condition that underlies it. The ache, which is the mental component, is indeed discontinuous but we allow the persistence of the underlying cause to guide our speech, even though the underlying cause is occasionally blocked from having its usual effect on consciousness. One wants to say that the ache is a sensing of this underlying condition and this sensing is not continuous. By analogy, if we are watching a woodpecker move through a dense wood for an extended time we will not actually be seeing the bird throughout that time. We nonetheless say that we watched the woodpecker for an hour. However, on Rosenthal’s side, I should point out that in cases where the headache can be felt whenever attention is directed towards it we are, I think, rather more inclined to say

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

that the headache itself persisted even during the time it was not being consciously experienced. This sort of neglected but continually accessible sensation is quite common. If, even upon introspection, nothing was felt we would be reluctant to say that the ache might still ‘be there’, whether or not the underlying condition persisted. Of course, such considerations do not sever the relation between certain mental states and consciousness, but they do make that relation more complex.

Box 3.2 • Essential HOT Theory

For α to be a conscious mental state, the subject must have a higher-order thought about α. But not just any sort of thought, brought about in any sort of way, will do. Roughly speaking, we can say that for α to be conscious one must have the ‘properly’ acquired belief that one is in α. So HOT theory defines consciousness as follows:

α is a conscious state of S if and only if (iff) (1) S is in the mental state α, (2) S has an ‘appropriate’ thought about α (we’ll call having

this thought ‘being in the state T[α]’; the content of T[α] is something like ‘I am in state α’),

(3) S’s being in α causes S’s being in T[α], (4) S’s being in α does not cause S’s being in T[α] via inference

or sensory information. Each clause is necessary to avoid potential objections. It follows from HOT theory that to be conscious of anything is to be conscious of it as something- or-other. Every state of consciousness is ‘aspectual’. This follows from the fact that every thought must be, so to speak, structured from concepts. But it does not follow from HOT theory that anything has an essential conceptual aspect under which one must be conscious of it. It also follows from HOT theory that one can’t be conscious without having beliefs (i.e. the appropriate higher-order thought). But it does not follow that when one is conscious of a mental state that one is conscious of a belief. To be conscious of such beliefs requires yet higher-order thoughts about them.

In any case, I don’t want to press this point since HOT theory may offer an explanation of why we tend to think that consciousness is intrinsic to certain mental states. This involves the second specification task, the delineation of the relation between non-conscious mental states that accounts for consciousness. Rosenthal explains it so:

. . . it is natural to identify a mental state’s being conscious with one’s having a roughly contemporaneous thought that one is in that mental state. When a mental state is conscious, one’s awareness of it is,

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

intuitively, immediate in some way. So we can stipulate that the contemporaneous thought one has is not mediated by any inference or perceptual input. We are then in a position to advance a useful, informative explanation of what makes conscious states conscious. Since a mental state is conscious if it is accompanied by a suitable higher-order thought, we can explain a mental state’s being conscious by hypothesizing that the mental state itself causes that higher-order thought to occur.

(1986, pp. 335–36)

Thus it is possible to maintain that if we tend to think of certain sorts of mental states as essentially involving consciousness this can be explained as the mistaking of a purely nomological link for a ‘metaphysical’ one. It might be, for example, that pains are normally such as to invariably cause the second-order thought that one is in pain and that abnormal cases are exceptionally rare (and, needless to say, rather hard to spot). In fact, this does not seem at all implausible. The machinery of philosophical distinctions mounted above is then seen as merely a case of philosophical error forcing us into an unnecessarily complex view of pains. It is literally true, according to the HOT Theory, that a pain – in possession of its painfulness – can exist without consciousness of it, but in fact almost all pains will be attended by consciousness of them, in virtue of causing the appropriate state of consciousness. One might even hope to account for the strength and constancy of this nomological link by appeal to its evolutionary usefulness. Rosenthal comes close to making this point (while actually making another) when he says: ‘. . . people cannot tell us about their non-conscious sensations and bodily sensations usually have negligible effect unless they are conscious. So non-conscious sensations are not much use as cues to [bodily] well being . . .’ (1986, p. 348). Nature would not likely miss the chance to entrench a causal connection between sensations, whether of pleasure or pain, and consciousness that is of such obvious biological benefit. Still, I believe that there remain serious difficulties with this view of the consciousness of phenomenal mental states, but it will take some effort to bring out my worries clearly.

Before proceeding let me introduce a piece of notation. We will frequently need to consider both a mental state and the second-order thought to the effect that one is in the former mental state. I will use Greek letters for mental states and form the second (or higher) order mental states as follows: the thought that one is in mental state α will be designated by T[α]. If necessary, we can allow this construction to be iterated, so the thought that one is in the mental state of having the thought that one is in the mental state α gets formally named T[T[α]], and so on. This notation allows a succinct characterization of HOT theory:

For any subject, x, and mental state, α, α is a conscious state iff (1) x is in α, (2) x is in (or, more colloquially, has) T[α],

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

(3) x’s being in α causes x’s being in T[α], (4) x’s being in α does not cause x’s being in T[α] via inference or

sensory information.

Note that for a to be a conscious state, the subject, x, must be in T[α], but x will not normally be conscious of T[α] as well. This would require x to be in the still higher-order state T[T[α]]. Such higher-order thoughts are entirely possible but relatively rare; we are not usually conscious that we are conscious (of some particular mental state) and HOT theory’s explanation of this is quite satisfying. HOT theory has many other virtues which are well remarked by Rosenthal himself.

Still, the definition as it stands fails to mark a crucial distinction the neglect of which can lead to confusion. We must distinguish between α’s being a conscious state of the subject x and x’s being conscious of α. Sometimes HOT theorists as well as objectors appear to be conflating the idea that the subject has a second- order thought about α which makes α a conscious state with the idea that the subject is conscious of α in virtue of having the second-order thought. I think it would be an unfortunate consequence if HOT theory entailed that one could be conscious only of mental states. Most conscious states have an (intentional) object; a conscious perception of a cat has the cat as its object and the subject in such a state is conscious not of his state of consciousness but rather of the cat, that is, the intentional object of the state of consciousness. In fact, it is very rare for anyone to be conscious of a mental state, at least if it is a mental state with its own intentional object, and despite the philosophical tradition it is entirely mistaken to define consciousness as an apprehension of one’s own mental states. So in a spirit of improvement and to forestall confusion, we can emend the definition as follows. If α is a conscious state and the intentional object of α is ∈ then we say that the subject is conscious of ∈ (in virtue of being in the conscious state α). There may be, and Rosenthal assumes that there are, conscious states that have no intentional objects. In such cases, saying that a is a conscious state is equivalent to saying that the subject is aware of α. For example, if we suppose that pains are ‘purely phenomenal’ states with no intentional objects then to be conscious of a pain is just the same thing as the pain being conscious. But even here we must be cautious. To be conscious of a pain in this sense is not to be conscious of a pain as such. This is a much higher level affair demanding a state of consciousness whose intentional object is the pain, conceived of as a pain. We shall shortly see how attention to these distinctions can be important and can fit rather nicely into the HOT theory.

It is worth digressing here to consider a line of objection to HOT theory which I think ultimately fails. But the objection is interesting in at least three ways: it endorses its own radical transformation of our notion of consciousness and the reply to it reveals some subtle strengths of the HOT theory as well as bringing out certain features crucial for the defence of a representational view of consciousness. The attack is mounted by Fred Dretske (1993). Dretske’ s objections

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

fundamentally depend upon a distinction between an experience’s being conscious and someone’s being conscious of that experience, and the claim that the former does not imply the latter. If Dretske is right about this we have not only a powerful challenge to HOT theories, but also a substantial and, I would say, very surprising extension of our knowledge about consciousness. However, I will try to show that Dretske’s objections cannot be sustained, revealing on the way some subtle strengths of HOT theories of consciousness.

Dretske follows Rosenthal’s use of some key concepts in setting forth his objections. Some states of mind are conscious and some are not: state consciousness is the sort of consciousness which conscious states enjoy. Conscious states are always (we think) states of some creature which is conscious: creature consciousness marks the difference between the conscious and the un- or non-conscious denizens of the universe. Creature consciousness comes in two flavours: transitive and intransitive. Transitive creature consciousness is a creature’s consciousness of something or other; intransitive creature consciousness is just the creature’s being conscious. Dretske allows that transitive creature consciousness implies the intransitive form, or

(1) S is conscious of x or that P ⇒ S is conscious. (1993, p. 269)

Furthermore, transitive creature consciousness implies state consciousness:

(2) S is conscious of x or that P ⇒ S is in a conscious state of some sort. (1993, p. 270)

A further crucial distinction is evident in (1) and (2) – the distinction between what Dretske calls thing-consciousness and fact-consciousness or the distinction between being conscious of an object4 and being conscious that such-and-such is the case.

Dretske’s basic objection to HOT theories, although articulated in a number of ways, can be briefly stated in terms of some further claims involving these distinctions. The most significant is that, in a certain sense, state consciousness does not require creature consciousness. That is, Dretske allows that states can be conscious without their possessor being conscious of them or conscious that they are occurring. Consider, for example, someone who is consciously experiencing a pain. By hypothesis, this is a conscious experience. Dretske’s claim is that it is a further and independent question whether this person is conscious of the pain or is conscious that he or she is in pain, and one which need not always receive a positive answer. If Dretske is correct, then HOT theories would appear to be in trouble, for they assert an identity between a state’s being a conscious experience of pain and the possession of the belief than one is in pain.

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

We must, however, re-emphasize a subtlety of the HOT theory here. The belief that one is in pain, which according to HOT theories constitutes one’s consciousness of the pain, does not itself have to be and generally will not be a conscious state. One would be conscious of this belief only via a third-order state, namely a belief that one believed that one was in pain. Thus one cannot refute the HOT theory by claiming that it is possible for one consciously to experience pain without consciously believing that one is in pain, that is, without being conscious of a belief that one is in pain. HOT theories cheerfully embrace this possibility. This is important because Dretske does not seem sufficiently to appreciate this subtlety. He claims that HOT theories must make a negative answer to the following question: ‘can one have conscious experiences without being conscious that one is having them? Can there, in other words, be conscious states without the person in whom they occur being fact-aware of their occurrence?’ (1993, p. 272). But, plainly, HOT theories allow an affirmative answer to this question. To have a conscious experience is, according to the theory, to believe that one is having it but not necessarily to consciously believe that one is having it. To put the point more generally in terms of the notation introduced above, to be conscious of a is to be in the state T[α]; this says absolutely nothing about whether one is in the state T[T[α]] or not, and it is the latter state that is required for T[α] to be conscious. So, according to HOT theories we have, roughly,

S is conscious of pain = S believes that he is in pain,

so the correct analysis of fact-awareness must be along these lines:

S is conscious that he is in pain = S believes that he is in f(he is in pain),

where f is some self-ascription function. I would suggest that f(he is in pain) should be cashed out as something like ‘. . . is in a state characterized by I am in pain’.5 Of course, normally we are rapidly carried from the conscious pain to the fact-awareness that we are in pain but this is a feature of our cognitive machinery, not an analytic truth constraining HOT theories of consciousness. If one considers animal consciousness the need to separate these states is apparent. HOT theories must assert that an animal’s being conscious of something is the animal’s having an appropriate thought. While this is a real difficulty for HOT theories of consciousness, for there are many who would deny to animals the ability to have thoughts of any kind and even more who would deny that they have thoughts about their own mental states, this is not the difficulty Dretske advances.6 It is natural to say that animals can be conscious of pains but that they cannot be conscious that they are in pain. However, given that animals can have some, perhaps quite ‘primitive’, thoughts (and the HOT theory simply must address

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

animal consciousness in this way), the distinction is successfully accounted for within HOT theories by the above analysis.

The worry that Dretske may not be taking this subtlety into account is strengthened by his remark that: ‘HOT theories . . . take an experience to be conscious in virtue of [its] being the object of some higher-order-thought-like entity, a higher-order mental state that . . . involves the deployment of concepts. My concern . . . therefore, was to show that conscious experience required no fact- awareness . . .’ (1993, p. 279). Since HOT theories allow that experiences can be conscious in the absence of fact-awareness of these experiences, this line of attack is, strictly speaking, misguided. It may be that Dretske meant to assert no more by ‘fact-awareness of p’ than ‘belief that p’, without any implication that these beliefs are themselves conscious. Such an interpretation would not be foreign to common usage and would lead immediately to the objection against HOT theories considered below. But Dretske actually says that ‘consciousness of a fact [which must surely be fact-awareness] . . . requires a conscious belief that this is a fact’ (1993, p. 272, my emphasis). HOT theories do not require this, and would consider it an unnecessary leap to third-order thoughts.

Box 3.3 • Dretske’s Objection

Since HOT theory makes every conscious state the object of a thought about it, every conscious state has an associated conceptualization of it, as given in the thought that ‘makes it’ conscious. Dretske objects that it is possible for there to be conscious experience without any of what he calls fact awareness. Fact awareness is consciousness of facts, which are conceptual entities; an example would be an awareness that snow is white. One can be aware of white snow without being aware that snow is white (one can even be aware of the whiteness of snow without being aware that snow is white). But HOT theory does not require any consciousness of facts for there to be conscious experience; it only demands that there be some conceptual categorization of the experience which is itself generally not conscious. Dretske’s basic objection can thus be countered. Dretske can, however, further deny that every conscious experience requires some conceptualization of it. However, while one can plausibly argue that no conscious experience has a mandatory conceptualization, it is very difficult to show that some conscious experience has no conceptualization. HOT theory asserts rather that every consciousness is a consciousness as of. . . . Contrary to Dretske, this seems entirely plausible.

In any case, HOT theories do assert an intimate connection between conscious experiences and beliefs about those experiences. Dretske must show that some states can be conscious in the absence of any such beliefs. Put another way, he needs to

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

show that states can be conscious in the absence of any conceptually articulated characterizations of them.

I do not find any explicit argument for such a thesis in Dretske’s article. The nearest thing is his defence of the following principle:

(3) For all things x and properties F, it is not the case that, S is conscious

of x ⇒ S is conscious that x is F. (1993, p. 266)

It would be easy to produce an argument against HOT theories based on this principle if we could identify ‘S believes that x is F’ with ‘S is conscious that x is F’, but as we have seen this is an identification that HOT theories need not endorse in general. But there is a closely connected claim which is made by HOT theories.

HOT theories must endorse the transition from state consciousness to transitive creature consciousness. For suppose some state, α, is a conscious state (i.e. possesses state consciousness) of subject S. HOT theories analyse this as S’s believing that he is in a (or, in the notation introduced above, having the thought T[a]). But this is identified with the state of S’s being conscious of α.7 Thus HOT theories identify transitive creature consciousness of α with α’s being a conscious state. Thus Dretske’s line of attack is indeed well motivated and he is right to say that HOT theories must deliver a negative answer to the question: ‘can there be conscious states in a person who is not thing-aware of them?’ (1993, p. 272). S’s belief that he is in a, or S’s consciousness of α, must then characterize α in some way via the deployment of concepts. I take it that this is HOT theory’s way of claiming that, as well as explaining why, all consciousness is consciousness as. . ., where the ‘. . .’ is to be filled in by the conceptual characterization of a occurring in S’s belief. It is possible, then, to interpret Dretske’s defence of (3) as a defence of a slightly different principle, namely the denial that a consciousness of is always a consciousness as. We could write this version of (3) as

(3*) For all things x and properties F, it is not the case that, S is conscious

of x ⇒ S is conscious of x as F.

If (3*) is correct we have the basis of a powerful objection against HOT theories of consciousness. However, this is a big ‘if’.

In the first place, the exact import of (3*) (or (3) for that matter) is not altogether clear. Dretske’s arguments for the principle may help to clarify it. He begins by noting the obvious truth that one can be conscious of x, which as a matter of fact is an F, without being conscious of x as an F (his example, which certainly rings true for me, is the possibility of (consciously) seeing an armadillo while having only the faintest idea of what an armadillo is). Such cases, however, only support a much weaker version of (3*), which would read as follows:

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

(3**) For all things x and some properties F, it is not the case that, S is

conscious of x ⇒ S is conscious of x as F.

Dretske then goes on to argue that there is no property, F, such that if one sees an armadillo one must characterize it as an F. This is also true but exposes a critical ambiguity in (3*).

To see this clearly we must note that a modal component lurks within our principles. The ‘⇒’ in (3*) cannot be regarded merely as material implication on pain of the ridiculous logical consequence that S is conscious of everything. Dretske means to assert that it is possible to be conscious of x without being conscious of x as an F. The proper understanding of (3*) crucially depends upon the scope of this possibility operator.

Approaching this point somewhat obliquely, consider the following explicitly modal principle:

where ‘Aw(S,x)’ represents ‘S is conscious of x’ and ‘CON(S,x,F): stands for ‘S is conscious of x as F’. This principle asserts that there is at least one distinguished or essential conceptual characterization which any consciousness of x must ascribe to x. This principle is clearly false, as Dretske ably shows. Thus we can take it that:

(Strictly speaking, this is stronger than the mere denial of (EX) but there is no reason to suppose that essential characterizations exist for any object.) After some logical manipulation, this becomes something close to (3*), viz.

or, equivalently,

This states that for any characterization, F, of x, it is possible to be conscious of x but not to be conscious of x as F. This seems to be true. However, (POS) is not the correct rendition of (3*) for (POS) is compatible with a weaker version of (EX) stating only that it is necessary that some characterization apply to x whenever one is conscious of x. Precisely,

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

The arguments that Dretske offers for (3*) are all compatible with (EX*), for they are all to the effect that no particular ascribed characterization of x is necessary for one to be conscious of x in any given situation. Nonetheless, all these situations are such that x is characterized by some F. Thus, these arguments can only support the weaker principle. But it is fallacious to infer from (POS) the stronger form which does express the intended meaning of (3*), namely,

The fallacy here is the well known modal fallacy – here applied to attributes – of inferring from the fact that something is possible of each thing to the fact that something is possible of all things.

Since (3*) is unsupported, it cannot be used in an attack on HOT theories. What is more, the correct principle, (EX*), can be invoked to disarm Dretske’s final objection against these theories. This objection begins from an unexceptionable premise, namely that experienced differences require different experiences. Dretske asks us to imagine attentively examining some complex scene and then shifting our attention to a second, very similar scene which we then also attentively examine (as in those common puzzles that ask you to spot the difference between two very similar pictures). One might not consciously notice that there was any difference between the scenes but nonetheless it may be true that one was conscious of every element of each scene. Thus the experience of scene 1 must have been different from the experience of scene 2 (for example, it could be that scene 2 lacks an element of scene 1 and so one consciously experienced that element when scanning scene 1 but of course had no such experience during examination of scene 2). Dretske concludes that we are thus committed to the ‘possibility of differences in conscious experience that are not reflected in conscious belief’ (1993, p. 275). Although we have seen that this is an infelicitous way to put the objection, Dretske wishes us to take his point to show that there can be ‘internal state consciousness with no corresponding (transitive) creature consciousness of the conscious state’ (1993, p. 275). This would clearly threaten HOT theories given their inescapable contention that state consciousness entails transitive creature consciousness.

But here, I think, Dretske is equivocating between what is, in essence, a de re and a de dicto characterization of consciousness. Would HOT theories demand that S be conscious of the difference between any two distinct experiences as a difference? Clearly the answer is no, for S may simply have never consciously compared them. In such cases – quite common I should think – S need not be conscious of the difference at all. Well, should HOT theories require that if any two of S’s conscious experiences are different and S is actually conscious of the difference (i.e. conscious of what is different between the two experiences) then S must be conscious of this difference as a difference? This also calls for a negative answer. To say that S is conscious of the difference in this sense is to say that there

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

is something different about the two experiences of which S is conscious; this puts no, or very few, restrictions on how that experience will be characterized in S’s belief about it which, according to the HOT theory, constitutes S’s consciousness.

That is, to say that S is conscious of the difference is, on the HOT theory, to say that S believes that he is experiencing the difference. But in the case envisaged this is true only on a de re reading of this belief. A more precise specification of this belief that brings out its de re character is this: of the difference (between the two experiences) S believes of it that he is experiencing it. It does not follow that S is conscious of the difference as a difference. To find out how S is experiencing the difference (that is, how to fill in the relevant as. . .) one must discover the correct de dicto characterization of S’s belief. Our principle, (EX*), guarantees that there is some such characterization but certainly does not demand that S should end up experiencing the difference as a difference. I can see no good reason to deny HOT theories access to de re characterizations of the beliefs that underwrite conscious experience. Of course, such characterizations do not help to specify the state of consciousness as it is to the subject himself but that is quite typical of de re belief constructions. They function, as illustrated above, to provide identification for outsiders of what a belief is about or, through the use of the HOT theory, to explain what someone is conscious of without a commitment as to how that person is conscious of that thing.

In short, the HOT theories of consciousness can admit the phenomena that Dretske points out without succumbing to the objections he believes they generate.

So, the HOT theory is surprisingly resilient and seems able to generate its own range of insights into the nature of consciousness. It is obvious that HOT theory is structurally similar to familiar theories of perception and thus it has certain affinities with other ‘perceptual’ theories of consciousness. By this, I do not primarily mean to connect HOT theory with those views of perception which explicitly make perceiving a kind of believing (see e.g. Armstrong 1968, chapter 10), though perhaps HOT theory could be mobilized to increase the plausibility of such views. More simply, one can see that the clauses of our definition of HOT theory quite naturally transform into a pretty conservative characterization of perception, rather as follows:

S perceives O iff (1) O exists, (2) S has an experience as of O, (3) S’s experience is caused by O, (4) S’s experience is properly immediately caused by O.

With regard to consciousness itself, HOT theory is reminiscent of both David Armstrong’s view of consciousness as one part of the brain physically ‘scanning’ another (see Armstrong 1968, pp. 92–94 and also chapter 15) and the early Daniel Dennett’s view of consciousness as a content carrying brain state that gets access to

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

the speech production centre (see Dennett 1969, chapter 6).9 The relative advantage of HOT theory is that it does not link the theory of consciousness with any attempt to model the workings or structure of the brain and its cognitive architecture. It yet remains compatible with these attempts, which can be seen as physicalist efforts to delineate the mechanisms that would be required to make clauses (3) and (4) of our formal characterization of HOT theory true within a working brain.

The analogy between theories of perception and HOT theory also suggests that according to HOT theory consciousness will suffer analogues of the various forms of misperception that philosophers of perception have appealed to, postulated or discovered. These are occasioned by considering the effects of letting one or more of the clauses of the definition of HOT theory, as given above on page 64, become false while maintaining as many of the remainder as possible. Let us catalogue the possibilities without, for now, going into either the question of their genuine possibility or their consequences.

Box 3.4 • The Four Pathologies

For each clause of the HOT theory definition of consciousness (see Box 3.2 above) there is a possible corresponding pathology of consciousness. The pathologies are generated by denying one clause of the definition while maintaining the truth of as many of the rest as possible. These are test cases with which to explore the limits of HOT theory’s plausibility. Deny clause (1) and we get an ‘hallucination’ of consciousness, e. g. one thinks one is in pain when one is in fact not. Deny clause (2) and we get a mental state that is not ‘noticed’ (this is not very pathological except in certain extreme cases, as when one fails to ‘notice’ an excruciating toothache). The denial of either clause (3) or (4) leads to interesting and problematic cases, which get to the heart of HOT theory. In both cases we have to ask whether it is in any more than at most a merely legalistic sense in which there is no conscious awareness of the lower- order mental state, a. If the subject gets into T[α], how can the subject or ‘the world’ tell how T[α] was brought about? If T[α] is the sort of state that ‘generates’ consciousness, won’t an at least as if consciousness result whenever the subject gets into T[α]?

First, let (1) be false. Then of course (3) and (4) must be false as well, but there is no reason to deny (2). This is a case of an hallucination of consciousness, the description of which is somewhat problematic, but whose possibility is a fundamental characteristic of HOT theory.

Second, let (2) be false. Again, it follows that (3) and (4) are false. This is simply the case of an unnoticed mental state, indubitably somewhat odd if the state is a pain or other highly distinct sensation. As we saw above, HOT theory can perhaps account for the sense of oddity we feel about such cases.

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

Third, let (3) be false. In such a case, while (4) must be false, (1) and (2) can remain true. The description of this case is also problematic, as we shall see.

Finally, let (4) be false. All the other clauses can nonetheless be true. Would this be another hallucination of consciousness, or in this case would there be no consciousness of a whatsoever? This is a tricky question, as we shall see below. It is also worth noting that clause (4) is unclear as to which sorts of inferences or sensory information are to be forbidden. Is any intermediary of this kind prohibited, or only conscious inference or sensory information? It could hardly be the latter for that would make HOT theory circular as an account of consciousness. As to the former, modern cognitive theories abound with hypothetical unconscious inferential, or quasi-inferential processes, particularly in the case of perception, the very case upon which HOT theory is largely modelled. Why couldn’t the link between a and T[a] be a cognitive link in this sense: that the process connecting them can be usefully described in information-theoretic terms? To put the question another way, why would a ‘cognitive link’ as opposed to one of a different, perhaps more ‘direct’ sort, fail to produce consciousness? Intuitively, we know there must be some difference. Here, as so often in philosophical analyses we wish we could simply write ‘a link of the appropriate sort . . .’. But even if we could get away with this in philosophy (which in truth we cannot), any empirical investigation into the physical differences between proper and improper linkages will bring us up against the generation problem.

We can, following Rosenthal, call these four ways of only partially fulfilling HOT theory ‘pathological cases’. According to HOT theory they are all genuinely possible. As Rosenthal says:

. . . since [the] higher-order thoughts are distinct from the mental states that are conscious, those thoughts can presumably occur even when the mental states that the higher-order thoughts purport to be about do not exist.

(1986, p. 338)

Explicitly, this passage deals only with our first pathology, but the reason Rosenthal gives for its possibility supports that of the other pathologies as well (and, of course, Rosenthal independently goes to some length to support the possibility of the second pathology).

One thing we can say about the pathologies in general is that the causal link between a and T[α] is of a much closer and more intimate sort than the causal links in perception that mediate between literally distant objects and the brain. The mechanisms of HOT theory are presumably all within the brain and in fact they will generally form a part of the more complex and extended causal sequences involved in perception. This alone suggests one reason why we feel that the link between consciousness and the mental state of which we are conscious is so

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

peculiarly intimate. According to HOT theory, it is an intimate link, but one that is at bottom causally ‘ordinary’, not metaphysically unique.

HOT theory’s willingness to countenance these pathologies allows us to answer a subtle but very important question about the theory. The general dictum that to be conscious of a is to have T[α] does not resolve the question whether possession of T[α] alone is itself sufficient for conscious experience or whether consciousness depends upon the existence of the proper link between a and T[α]. The account of pathology 1 just given suggests that HOT theory claims the former: T[α] is by itself sufficient for consciousness. The other clauses serve to mark out how a certain T[α] is a consciousness of some other particular mental state, a, or confers consciousness on a. Again we see analogies with theories of perception, which always possess components that mark out what object is the perceived object but must also include other components to account, or at least allow, for the possibility of perceptual error and hallucination. Rosenthal provides more direct textual evidence that this is his understanding of HOT theory as well. The quote above from 1986, p. 338 makes the point reasonably clearly and during a discussion of the ‘reflexivity’ of consciousness Rosenthal says:

The sense that something is reflexive about the consciousness of mental states is thus not due to the conscious state’s being directed upon itself, as is often supposed. Rather, it is the higher-order thought that confers such consciousness that is actually self- directed.

(1986, p. 346)

What is important here is the claim that it is the higher-order thought, in our terms, T[a], which confers consciousness. In addition, Rosenthal states elsewhere: ‘. . . we are not normally aware of the higher-order thoughts that, on such an account, make mental states conscious’ (1986, p. 340, my emphasis) and more recently Rosenthal says: ‘. . . a mental state’s being conscious consists in its being accompanied by a HOT’ (1995, p. 26 n.). I also observe that in 1995 Rosenthal adds: ‘. . . HOTs can presumably occur in the absence of the states they purport to be about. What would that be like subjectively? Since having a HOT makes the difference between whether there is or isn’t something it’s like to be in a mental state, it may be that having a HOT without the relevant state is, sometimes at least, subjectively indistinguishable from having both’ (1995, p. 26 n.). One has to ask: how could it be only ‘sometimes’, if the appropriate HOT occurs?

I think we can also see that this reading of HOT theory is forced by considerations of plausibility. Suppose we maintained that consciousness demanded the fulfilment of all four clauses of HOT theory – that whenever anyone was conscious at all this consciousness would satisfy all four clauses of the HOT theory. This would directly entail that, for example, no one could be consciously mistaken about their own mental states. For example, it would be impossible

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

consciously to take oneself to believe p while in fact one did not. For, on our present reading of the theory, one could not be conscious of the belief that p, or have the’ experience’10 of believing that one believed p, unless one did believe p. Otherwise clause (1) would fail, contrary to our present supposition. Yet it is, I think, overwhelmingly plausible that people can consciously take themselves to believe, desire or hope for what they in fact do not (that people can have ‘false consciousness’). This phenomenon is by its nature subtle and complex and it is difficult to trot out elementary examples. Freud offers many rather extreme cases and some are certainly quite plausible. But it is a common experience to find someone’s self-image (that constellation of beliefs, desires, etc. that one consciously takes oneself to possess) at odds with what an outside observer would rightly take to be that person’s beliefs, desires, etc. Nor is it uncommon for people to find out that they had mistaken the nature of, for example, a particular desire of theirs when suddenly confronted with its imminent satisfaction.

We might call this implausible interpretation of HOT theory the Fully Relational Reading and while it is unlikely to be correct we should note that it could immediately account for the sometimes presumed incorrigibility of our consciousness of our own mental states. If consciousness were necessarily the fulfilment of our complex of four clauses then it would indeed be impossible to be conscious of a via T[α] without being in the state α.

But the implausibility of the Fully Relational Reading of HOT theory stems from the implausibility of incorrigibility itself, and the alternative reading of HOT theory can obviously account for this failure of consciousness. Failure of self-knowledge stems from pathologies 1 and 2 (differing in the ‘direction’ of the failing). As befits pathologies, they are relatively rare, but their frequency could vary widely depending upon the type of mental state involved or the particular contents of those mental states. Such variance would be explicable in terms of the patterns of cause and effect between the relevant αs and T[α]s. The Freudian style of psychological analysis can be seen as at least a model for such causal classifications of mental state interrelationships.

We are thus driven to reject the Fully Relational Reading of HOT theory. HOT theory cannot retreat to it in the face of difficulty. This is important, since the objections I will raise stem from the acceptance that it is the higher-order thought, T[α], that confers consciousness, independent of the conditions under which T[α] is brought about. That is, should T[α] occur to someone (i.e. should someone get into the mental state designated by T[α]), that person will be in a state of consciousness indistinguishable from that of being in state α whether or not they are in α. There will be nothing ‘in’ that person’s consciousness by which to distinguish the veridical from the pathological cases. Again, this is analogous to the case of perceptual hallucination where there need be nothing ‘in’ the perceptual experience that could reveal an hallucination as such.

Nonetheless, there is nothing in HOT theory, nor in the case of perceptual hallucination, that precludes the recognition that one is the victim of a pathological state of mind. The Müller-Lyer illusion is a very simple perceptual

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

‘hallucination’ which illustrates this point. Even while we cannot help but see one line as longer than the other, we all know they are nonetheless the same length. Similarly, each of HOT theory’s pathologies is compatible with the knowledge that one is suffering it. This suggests a variety of rather traditional objections to HOT theory, of which perhaps the simplest is the following.

According to HOT theory (via pathology 1), it is possible to be as it were conscious of pain while one is in fact not experiencing any pain. In such a case, that is to say, the phenomenal property of painfulness will not be exemplified at all even though the subject ‘thinks’ that it is. The objection is straightforward. There is no difference, to the subject, between this case and the veridical case of ‘true’ consciousness of pain because in both cases the consciousness-conferring thought, T[α], occurs. There is every bit as much suffering in the pathological as in the normal case, every bit as much reason to eliminate T[α] in this case as in the normal case. Since it is the presumed painfulness of pains that provides us with the grounds to attribute and sympathize with suffering as well as giving us the reasons to try to ease the suffering, this would strongly suggest that the relevant phenomenal property of pains – painfulness – occurs in both cases, contrary to what HOT theory appears to assert. One cannot reply here that the phenomenal property goes with T[α] rather than α. Since this is a case of consciousness, HOT theory would then require an appropriate third-order thought, T[T[α]] to account for the consciousness of this phenomenal property of T[α]. We could then invoke a second-order pathology from which, in strict accord with the foregoing, it would be evident that the phenomenal property of painfulness actually belongs to T[T[α]]. We would thus generate a viciously infinite hierarchy of thoughts about thoughts about thoughts . . . . The phenomenal property in question would forever remain one step above whatever level of the hierarchy was under consideration and thus could find a home at no level of the hierarchy and therefore would not belong to any mental state, which is absurd.11

Instead of chasing an ultimately imaginary phenomenal property up through this hierarchy one could reply that the pain – α – possesses its own sort of phenomenal property, but so too does T[α]. In fact, solely within consciousness, there is no discernible difference between the experience of either. Thus, in normal cases, the appearance of T[α] will be a consciousness of a and (or via) α’s attendant phenomenal property of painfulness. In the pathological case, T[α] occurs without α but also, we may postulate, T[T[α]] occurs as well, and this latter will be a consciousness of the phenomenal property of T[α], This reply is entirely ad hoc and endlessly multiplies essentially indistinguishable phenomenal properties, but it also suffers from a worse defect. T[T[α]] is a consciousness of the thought that one is in pain (i.e. in α). Even if we grant that the thought that one is in pain has its own phenomenal properties (somehow, conveniently and miraculously, indistinguishable from the phenomenal property of painfulness that a carries), T[α] carries propositional content as well. Yet in the pathological case, there need be no conscious awareness of the thought that one is in pain. Pathology 1 only requires that one think and experience, falsely, that one is in pain. It does not

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

require that one also think that one think that one is in pain. Putative sufferers of pathology 1 would sincerely assert that they were suffering pain and there would in truth be no difference in the experience between the pathological and non- pathological cases. There is no reason whatsoever to suppose that whenever cases of pathology 1 occur there must also occur an additional, still higher-order, thought.

The danger here is that HOT theory cannot place the phenomenal properties of mental states in the proper location (and it is hard not to suspect that the reason is that phenomenal properties are in fact somehow tied intrinsically to consciousness and cannot be given a relational analysis). This difficulty is reinforced if we suppose a sufferer of pathology 1 to be informed of his condition. Let us further suppose that our sufferer is an ardent supporter of HOT theory and hence is happy to allow the possibility of pathology 1. It is pretty clear that despite the additional knowledge our subject will still report that he feels pain. In addition, he may say, in line with HOT theory, that he is not really feeling pain but he cannot deny that he is feeling something quite indistinguishable from it, for if pathology 1 were consciously distinguishable from ‘true’ consciousness then HOT theory would be falsified since in that case T[a] would not be what confers consciousness. HOT theory is thus faced with an unhappy dilemma. Either the phenomenal property of painfulness is not exemplified at all in this version of pathology 1, in which case there is no accounting for our subject’s reports and evident suffering, or else it is not the pain which exemplifies the property of painfulness, which is not only obviously implausible, but it leads to the further implausibilities I have just outlined. In fact, it seems to me that if HOT theory can seriously countenance the idea that the phenomenal property of painfulness is not exemplified in this case, then there is no reason to admit the existence of phenomenal properties at all. Their raison d’être is to account for and provide the content of the consciousness of sensations. If this very consciousness can occur without any phenomenal input, no real role remains for the phenomenal properties, which become merely a gratuitous metaphysical extravagance.12

Focussing more closely on T[α]’s ability to confer consciousness naturally brings us to pathologies 3 and 4. In both, the consciousness conferring state, T[α], occurs but, in pathology 3, α occurs but does not cause T[α] whereas, in pathology 4, α does cause T[α] though not immediately but rather through some inferential process or a process dependent upon sensory information. As we have seen above however, so long as T[α] does occur, there will be a consciousness ‘as it were’ of α. Pathology 3 is perhaps not particularly interesting – it is a case of an hallucination of consciousness, akin to pathology 1, but one in which, by some chance, the mental state which would make T[α] a case of what we might call veridical consciousness (i.e. of course, a) happens to occur alongside T[α]. In the field of perception, it is just the possibility of such coincidences of perceptual experience along with what would make them veridical that drives philosophers to impose the condition that the perceived object cause the perceptual experience. HOT theory quite properly draws the same lesson from its analogous possibility.

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

Pathology 4 is much more interesting and leads to what I take to be the most fundamental objection to HOT theory. HOT theory grants that the state T[α] confers consciousness, that is, that anyone in the state T[α] will have a conscious experience that is at least ‘as if’ one were conscious of a which is, so far as the conscious experience is concerned, completely indistinguishable from a true consciousness of a. Given this, it is hard to see what is improper, from the point of view of T[α] being a state of consciousness of α, with any causal process whatsoever getting one into T[α], so long as it does get you into that state of consciousness.

One complication can be set aside. It is possible to imagine a causal chain from a to T[α] which includes, as essential links, other phenomenal mental states the conscious experience of (at least some of) which is indistinguishable from that of a. In such a case, a principle of causal proximity would seem to require that the state of consciousness be of the final such phenomenal state in the chain. This is again rather analogous to possible examples drawn from the field of perception. Suppose that you are in a dark room looking at the place where a small light bulb will be turned on. Unbeknown to you, a mirror angled at 45° lies between you and the light bulb, but a second light bulb has been strategically placed off to the side so as to be visible in the mirror. This second bulb lights only if the first bulb lights. So the first bulb causes your perceptual experience, which is a perception of a light bulb. The causal proximity principle correctly entails that you are really perceiving the second bulb. But if the second bulb is replaced by a suitable arrangement of two mirrors (or even video monitors), you will now perceive the first bulb whenever it is illuminated even though its light (or even a representation of its light) takes a somewhat devious route to you. The causal proximity principle applies only to causal intermediaries that ‘satisfy’ the resulting perceptual experience. Returning to HOT theory, we can legitimately prohibit this kind of indirect consciousness, but of course hardly any sort of inferential process or processes relying on sensory information will interpose the particular intermediate mental states required to rule out such processes. In what follows there will be no danger of inadvertently appealing to this kind of truly illegitimate mediated consciousness.

Consider first the possibility mentioned earlier that the inferential processes that, supposedly improperly, link α to T[α] are all unconsciously buried in the sub-personal realm of our cognitive architecture. Suppose, that is, there are functional units in the brain whose cognitive task is simply to bring certain states up to consciousness. It is not implausible to suppose that there is something of a competition amongst the myriad of brain states which underlie our phenomenal and intentional mental states, some signalling distant objects of perceptions, others important states of the body, still others potentially relevant intentional states. All these states will loudly clamour for ‘consideration’ by our hypothetical functional units but only a few will become conscious. The conditions for becoming conscious could well involve some form of hypothesis generation and testing at the sub-personal level. Such cognitive mechanisms depend on various sorts of information processing, some of which are closely akin to inference, as

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

well as necessarily involving (sub-personal) sensory information.13 This is a sketch of how the brain might get from a to T[α] and the fact that this process involves essentially inferential processes and a reliance on sensory information does not seem to threaten in any way T[α]’s claim to be a consciousness of a, given that T[α] itself confers the conscious aspect of the experience, leaving only the identity of what T[α] is a consciousness of to be decided. The situation here is once again analogous to that in theories of perception, especially visual perception. Many of these theories are irredeemably inferential in nature (see, e.g. Gregory 1990 or Marr 1982). Whatever the faults of such theories, no one has ever suggested that they fail simply because the posited inferential properties of the sub-personal cognitive system by themselves preclude perceptual experience!

What really motivates inclusion of clause (4) in HOT theory is not the fear of a supposedly impossible mediated consciousness, but rather the evident fact that possession of just any second-order thought that one is in a certain first-order mental state will not, by itself, make one conscious of that mental state, even if the second-order state is caused by the first-order state. (Just as, in the case of perception, merely having the thought or belief that a candle is visible before one is not by itself sufficient for a perception of the candle, even if this belief is somehow caused by the presence of the candle.) In fact, HOT theory’s acceptance of pathology 2 makes this very clear. Recall that the second pathology involves a subject being in state a unaccompanied by the consciousness conferring state T[α]. To make the case as stark as possible and certainly pathological, let us suppose that a is the unconscious experience of a spot of white light in an otherwise jet black visual field (suppose the subject is sitting in an experimental chamber, utterly dark, in which spots of light at various locations can be turned off or on). Just as HOT theory allows that one can be in pain while not being conscious of the pain, one can have visual experiences without being conscious of them.14 In such a case, α = seeing a white spot. We are supposing that, for whatever reason and, of course, atypically, T[α] does not occur. Now imagine that our subject is told both that in actual fact he is seeing a white spot and that, let us say for some technical and neuro-experimental reason, he is suffering pathology 2. It is given that our subject fully accepts HOT theory and its consequences, has trust in the experimenter and the brain theory she employs, etc. Thus the subject comes to have the thought that he is seeing a white spot and is suffering from pathology 2, i.e. T[α] plus an independent thought about mental pathology. It is clear that this T[α] will not confer the consciousness of a small white light against a jet black background, apparently contrary to the dictates of HOT theory. HOT theory is supposed to be saved, of course, by appeal to clause (4): our imagined case is blatantly a case of inference to T[α] via sensory information.

But notice that clause (4) cannot do this job. First of all, we have seen that mere appeal to inferential or sensory informational mediation will not necessarily rule out consciousness. And second, HOT theory already accepts that T[α] is the state that confers consciousness, in the sense of ‘mere experience’ independent of questions of what the experience is an experience of. So if our subject actually

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

gets into T[α], he must be conscious ‘as it were’ of a white spot against a jet black background. It is clear that in our imagined situation the acquisition of T[α] will not confer this consciousness. So it follows that the possibility of pathology 2 along with the assumptions of our imaginary situation (all compatible with HOT theory) entail that one simply cannot get into a suitable T[α] in this sort of way.

Compare our hypothetical thought experiment with a kind of perceptual experience that actually occurs quite frequently. Say you are out bird watching; to be specific let us say that you and a friend are looking for the elusive spruce grouse, a bird given to freezing when startled, relying on its effective natural camouflage to escape detection. As you walk through the bush, you hear the tell-tale rustling sound and look in its direction. Your friend spots the grouse and quietly tells you that you are looking right at it, yet you still do not discriminate the grouse from the background. You believe your friend and thus you acquire the belief that you are looking at a spruce grouse but this belief does not yield the perceptual experience of a spruce grouse. Then quite suddenly, with no apparent change in anything else, you do see the grouse. You would not have spotted the grouse but for your friend’s information, so this is a kind of inferentially and sensory informationally mediated perceptual experience, but of course it is nonetheless a perfectly genuine perceptual experience.

More ‘scientific’ examples can easily be given as well. A well known visual illusion involves what is called the ‘transparency effect’ (see Rock 1985, pp. 112ff, 138 ff.). Consider fig. 3.1:

(Fig. 3.1)

At first glance this looks – to most people at any rate – like four distinct, variously shaded rectangles. But if one is told that it is a grey, transparent sheet placed over the two-colour rectangle underneath, one can come to see it as just that. It seems that the information about what one is looking at transforms the

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

way it looks to one, and again we have a genuine conscious experience that is inferentially, informationally mediated.

We can extend our thought experiment, by analogy with such actual events, to show that of course one can get into a suitable T[α] in all sorts of ways. We can imagine that our subject is in a ‘primed’ state such that being told that he is seeing a white spot will causally release the subject from pathology 2, just as being told that a spruce grouse is hiding right before your eyes can actually trigger its discrimination from the background. Thus an inferential and sensory informationally mediated process can yield a suitable T[α]. The crucial question is: what is the difference between a suitable and an unsuitable T[α]?

My fundamental objection to HOT theory will now, I hope, be clear and clearly damaging. It is that there is no way to delineate the suitable T[α]s from the unsuitable ones except in ways ultimately equivalent to this: a suitable T[α] is one that confers consciousness. If this is correct, then HOT theory cannot be a reduction of consciousness or an explanation of consciousness, for one must appeal, tacitly, to the very thing one wishes to reduce and explain in order to characterize HOT theory in complete detail. This does not mean that HOT theory is wrong to link consciousness and higher-order thought. It is, indeed, pretty evident that consciousness does have some intimate connections with higher- order thoughts. But it does mean that one cannot explain consciousness in terms of, or reduce consciousness to, a relation between lower and higher-order thoughts.

We can go some way towards diagnosing HOT theory’s failing. Ultimately, it lies in the acceptance that T[α] is a state which confers consciousness along with the characterization of T[α] as a content carrying higher-order thought to the effect that one is in the state a. Since HOT theory demands as well that T[α] be a separate mental state from a there is no way to rule out T[α] being caused by a wide variety of atypical causes (hence HOT theory’s acceptance of the pathologies of consciousness). At the same time, it is clear that many states that would intuitively, and correctly, be counted as such higher-order thoughts do not confer consciousness, and so the suitable ones must be separated from the unsuitable ones. But this cannot be done by any appeal to the process which generates T[α], for the separateness of T[α] means that, bluntly speaking, any such process can produce, in the respectively proper circumstances, either a suitable or an unsuitable T[α]. In the end, the only possible characterization of this distinction that conforms to the dictates of HOT theory is one that appeals to consciousness itself, or the ability to confer consciousness. Thus HOT theory cannot succeed as a reduction or explanation of consciousness.

This objection cannot be made against what I called the Fully Relational Reading of HOT theory for on that construal, T[a] alone is insufficient to confer consciousness. This is no help however, for, as we saw, the Fully Relational Reading is extremely implausible on independent grounds.

Such objections reveal that the HOT theory will not face up to the generation problem any better than did the identity theory. Suppose that we accept the division of HOTs into the two fundamental groups: those for which, to use

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

HOT THEORY

Rosenthal’s words, ‘a mental state’s being conscious consists in its being accompanied by a [suitable] HOT’ (1995, p. 26 n.) and those for which this is not true. Whatever manages to produce a HOT of the former sort (be it inference, causal mechanisms within the brain or even the direct will of God) will also produce conscious experience. Within HOT theory there can be no explanation of what this difference amounts to, for the occurrence of the appropriate HOT is by itself sufficient to ‘generate’ consciousness – the causal ancestry, or any other feature, of the HOT doesn’t matter once it has come into being. So we are left with the usual question, which is the generation problem once again, of exactly what it is about the appropriate HOTs that allows just them to confer consciousness upon certain states?

Notice that for all my objections, there is a sense in which the HOT theory could still be true. I have tried to show that one cannot reduce consciousness to a kind of thinking or explain consciousness in terms of the HOTs. Nonetheless, it might be that consciousness is, as a matter of brute fact, precisely the kind of cognitive operation to which HOT appeals. So too, the neural vector coding theory could, as a matter of brute fact, be what consciousness is (the two theories could even collapse into one if we wanted or needed to identify the neural vectors with thoughts instead of the phenomenal qualities). I noted that, on the neural identity option, the question of just how come only those creatures with the appropriate vector machinery were conscious is pressing (to which the answer ‘brute fact’ is not very comforting in the face of behavioural evidence of consciousness). Similarly, while the HOT theory resides at a higher level of abstraction than the neural vector identity theory, it too faces an analogous problem. HOT theory must claim that beings without thoughts are without consciousness. This is problematic for animals. One might wish to argue that consciousness emerged prior to thought, rather than only after there were creatures capable of having thoughts, and, worse still, having the sort of complex thoughts that manage to be about other mental states. Much more radical versions of this problem arise when we couple the HOT theory to theories of thought and the contents of thoughts. Before getting to this issue however, I want to examine Dennett’s views on consciousness, which are closely related to the HOT theory, though more radical in the reformulation of our notion of consciousness which they advocate.

Box 3.5 • Summary

HOT theory promises a mentalistic reduction of consciousness to non- conscious mental states in the hope that a naturalistic treatment of the latter will be relatively easy to find. HOT theory could then explain what makes a state conscious: a mental state, a, is a conscious state if it is the object of a thought with the content that the subject is in a.

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .

THEORIES OF CONSCIOUSNESS

Box 3.5 • Summary (cont.)

If such an explanation is successful, HOT theory might even solve the generation problem. But HOT theory suffers from several damaging objections, the primary one being the worry that HOT theory cannot distinguish those higher-order thoughts which can confer consciousness upon a mental state from the myriad of possible higher-order thoughts which do not. In the end, the only characterization of the ‘suitable’ higherorder thoughts is simply that they are the ones that confer consciousness. Unfortunately, this obviously leaves the generation problem untouched and, worse, makes the HOT theory explanation of consciousness covertly circular and hence no explanation at all.

C o p yr

ig h t ©

1 9 9 9 . T

a yl

o r

& F

ra n ci

s G

ro u p . A

ll ri g h ts

r e se

rv e d .