SWIF Philosophy of Mind updated: 17 September 2001. http://www.swif.uniba.it/lei/mind/forums/carruthers4.htm
Forums Forum 2 Joseph Levine's commentary

Reply to Joe Levine

Peter Carruthers

University of Maryland, College Park

My thanks to Joe Levine for his extensive and insightful comments on my book (Carruthers, 2000). In this reply I shall focus on the main points of disagreement between us - namely, the existence of an alleged 'explanatory gap'; the supposed possibility of experiential inversion; and Levine's criticisms of my proposed dispositionalist HOT theory.

1 The 'explanatory gap'

Levine concedes some force to my attempts to mark a divide between those aspects of the phenomenon of phenomenal consciousness which can be successfully reductively explained by a naturalistic theory such as dispositionalist HOT theory, and those aspects which are best explained away. A large part of the work of the latter sort is borne by the notion of a purely-recognitional concept; and Levine uses my analogy based on the example of 'bare color' as a stalking horse on which to ground his critique. Unfortunately, he gets the example wrong.

Unlike the example of the chicken-sexer (which does some, but only some, of the same work) the bare-color example isn't supposed to involve a concept which we find ourselves applying for no apparent reason, with applications of it having no 'substantive and determinate content … over and above my response itself'. On the contrary, it is supposed to be an example of a concept whose applications are grounded in first-order (but non-phenomenal) analog intentional states. We are to imagine a creature possessing analog (or non-conceptual) intentional states representing reflective properties of surfaces, and with the capacity to apply purely-recognitional concepts grounded in the intentional character of those states, but who is altogether lacking in phenomenal consciousness of color.

(Of course, whether you think the example is even so much as possible will depend upon your prior rejection of first-order representationalist theories of phenomenal consciousness. But this assumption creates no special problem as part of an argument against mysterianism. And anyone who believes in the possibility of zombies will have no difficulty with the bare-color example.)

Now, the point is that such a creature would feel the same kind of explanatory gap as is alleged to exist in the case of phenomenal consciousness, but in this case between physical / functional / intentional facts and the facts of bare-color. For lacking any beliefs about the processes which ground its color-recognition judgments, the creature would always be capable of thinking, 'Yet all of that could be true, and still THIS could be lacking', where 'THIS' expresses a recognitional concept grounded in a first-order analog color-content. Yet this creature lacks any phenomenally conscious color-states. Moreover, most mysterians will allow that non-phenomenal intentional states and concepts admit of reductive explanation in principle. So this is a case where a reductive explanation could exist, although the subject in question would be vulnerable to the same persistent worries which dog attempts to explain phenomenal consciousness.

(Of course it is true, as Levine notes, that WE don't feel any temptation to think that there is any problematic explanatory gap present in the case of the bare-color subject. But that only goes to help my case. The point is that THE SUBJECT in this example WOULD feel a problematic explanatory gap to exist.)

At its weakest, the example shows that the so-called 'explanatory gap' for phenomenal consciousness doesn't really have anything to do with phenomenal consciousness per se. It is, rather, a gap which can arise more generally wherever there are recognitional concepts grounded in analog intentional contents. At its strongest, the example shows that the 'explanatory gap' is not really a gap at all. That certain persistent questions remain, doesn't show that anything is going unexplained.

On the stronger of the above readings, the analogy works because it gives us a case of a recognitional concept grounded in (non-phenomenal) analog intentional states. And in the case of phenomenal consciousness (according to dispositionalist HOT theory) what we have is a set of recognitional concepts whose application is grounded in higher-order analog intentional states (intentional states whose higher-order content derives from the possibility of deploying higher-order concepts in response to them - see section 3 below). So I can grant to Levine that (phenomenal) reddishness 'doesn't have the character for me of something-I-know-not-what that happens to be prompting this response'. But then the same is true for bare-color too. For the bare-color subject doesn't find himself deploying concepts in response to something-he-knows-not-what. Rather, those concepts are grounded in perceptual (but non-phenomenal) awareness of analog properties of surfaces.

2 Inverted spectra

Levine concedes that an appeal to narrow content can undermine externalist forms of argument for the possibility of inverted experiences, such as Block's (1990) example of Inverted Earth. But he worries that I have lost sight of - and have failed to respond to - Shoemaker's original intra-personal inversion case (Shoemaker, 1981). Before I get to that, however, I want to say a word about burden of proof and plausibility.

Mysterians like Levine bring up the possibility of inverted experience as part of an argument for their mysterian position. They are therefore required to argue their case - the burden of proof falls on them to show that inverted experience is possible. And, given any robust form of concept / property distinction, they also need to show that inverted experiences are genuinely - metaphysically - possible, and not just conceptually so.

As I understand it, this is why Shoemaker saw the need to concentrate, in the first instance, on developing realistic intra-personal cases of experience inversion. This is the crucial battle-ground on which the possibility of experience inversion needs to be fought. Merely pointing to the possibility of symmetrical functional roles and symmetrical external quality spaces - as Levine does in his comments - is not going to convince anyone of anything more than a conceptual possibility. For we are given no reason to believe that experiences either would or could be reversed in such a case. It is open to reductive naturalists to claim that if the symmetrical functional roles are genuinely functionally equivalent, then experiential contents will be identical also. No reason would yet have been given to shift us from such a position.

Returning now to the charge that I had neglected Shoemaker's original argument - in fact I did return briefly to such cases following my discussion of Inverted Earth, on page 86 of my book. I claim that there are two stages necessary to liberate us from feeling any force in the intra-personal inversion examples. The first is to see the possibility of appealing to some notion of narrowly-individuated intentional content. For the sake of concreteness, let us suppose that this takes the form of a functional-role semantics. Then the second step is to see that functional roles are individuated, not just by actual causes and effects, but also by counter-factual causes and effects. (Even philosophers of psychology as distinguished as Fodor are apt to forget this.)

So, consider a post-amnesial subject in a supposed intra-personal inversion case: he uses 'seems green' to describe his experiences of green grass, and all his other beliefs and behaviors are as normal; although not long ago, before his amnesia, he could still recall that seeming-green was the experience he used to get when looking at fresh blood (before the pathways in his optic nerve were switched). Is he genuinely the functional equivalent of a normal person? (And so is this a genuine case of experiential inversion with functional / intentional equivalence?) I claim not.

Take the recognitional capacity which the subject now deploys when using the term 'seems green' and ask how that capacity WOULD HAVE responded had it been present pre-optic-nerve-switching, and had it been confronted by the state normally caused by looking at green grass - would it have been activated? Surely not. On the contrary, that very recognitional capacity would have been activated in response to the state normally caused by seeing fresh blood. So the narrow intentional state which now causes him to say 'seems green' continues to have the content 'seeming-red' - and what we have here is no longer as case of experiential inversion with functional / intentional symmetry.

So although the intra-personal inversion subject is BEHAVIORALLY the same as a normal person, we can say that his internal states are functionally distinct, because different counter-factuals are true of them. And so we can explain, in functional-role terms, how it is that his experiences are different from normal too. And such cases therefore provide us with no reason to embrace mysterianism concerning phenomenal consciousness.

3 Dispositionalist HOT theory

Levine raises three difficulties for my dispositionalist HOT theory. I shall take them each in turn.

The first is that I have failed to make out the case for the existence of non-conscious experiences. Levine (like me) is inclined to deny that the cases I adduce involve phenomenal consciousness; but he thinks that they therefore don't deserve the title 'experience'. I am not going to quarrel about words. He can insist on calling them 'analog perceptual states' if he likes. But Levine has missed the dialectical position, here. The conscious / non-conscious distinction is NOT supposed to be part of any argument against mysterianism - so it is not supposed to be something which should worry Levine. By the time we get to chapter 6 in my book (where these matters are discussed), mysterianism has already been set to one side. Rather, some sort of representationalist approach to phenomenal consciousness is presupposed. And the conscious / non-conscious distinction forms the main premise in my arguments against first-order representational (FOR) theories, of the sorts defended by Dretske (1995) and Tye (1995, 2000).

Levine's second objection I have difficulty in getting straight. It has something to do with an alleged inadequacy in my appeal to recognitional concepts in defusing the conceivability arguments for mysterianism. The worry seems to be about whether dispositionalist HOT theory has provided for the 'substantive and determinate character' of the states to which our recognitional concepts are applied, and for the 'cognitive intimacy' which exists between our phenomenally conscious states and our recognitional concepts for them. But this is followed by a confused and inaccurate presentation of my view.

So here, briefly, is how I think the story should go. When I enjoy a phenomenally conscious experience as of red, I am in a perceptual state with the analog (or non-conceptual) content 'red', which also possesses the analog content 'seems red' or 'experience of red'. (The state in question is thus BOTH a first-order experience of color AND a higher-order experience of an experience of color.) I am then capable of enjoying purely-recognitional concepts for my phenomenally conscious experience, grounded in my higher-order experience of it. The perceptual state with the analog content 'seems red' provides the grounds for me to apply my recognitional concept, in much the same way that a perceptual state with the analog content 'red' provides the grounding for a recognitional application of the concept 'red'. These higher-order recognitional concepts are genuine concepts, capable of figuring in conceivability thought-experiments or in puzzling about the explanatory gap. And in particular, since they lack any conceptual connections with the various elements which go to make up dispositionalist HOT theory, it will always be possible to think, 'Dispositional HOT theory might be true of a creature, and still that creature could lack THESE kinds of states'.

It seems to me that this story provides fully for the substantive and determinate character of the states which ground our thoughts about phenomenal consciousness; and that the cognitive intimacy between such states and our judgments about them is also explained.

Levine's third objection to dispositionalist HOT theory is that he doesn't see why availability to higher-order concepts should transform the contents of our perceptual states, conferring them, at the same time, with higher-order analog contents. He uses the example of his watch-face to make the point. Surely the mere fact that Levine knows that his watch doesn't always keep time isn't enough to make the position of the hands represent 'seems 4 o'clock' as well as 'is 4 o'clock'.

The example is a misleading one, however. For watch-faces are not representations in their own right - they depend upon the minds of people to confer content upon them. Or if they don't - if they are considered representational - then the appropriate notion to apply must be an informational, or some sort of causal-covariance, conception of content. But this is not the notion presupposed by consumer-semantics. According to consumer-semantics, states which are representational-in-their-own-right acquire their content from the inferential powers of the systems which consume, or make use of, those states.

I don't believe that any sort of special case has to be made out for saying that perceptual states will acquire higher-order content on becoming available to higher-order consumer systems. For this is just what consumer semantics would predict - in general, the content of a state depends upon (and is a reflection of) the inferential powers of the systems which are the immediate and prime consumers for that state; and states which become available to new systems will thereby acquire new contents.

Consider an example from developmental psychology. Perner et al. (1994) maintain that there is a stage in development when children operate with an undifferentiated concept of 'prelief', which is a sort of amalgam of 'belief' and 'pretence'. They understand that preliefs don't always correspond to the way the world is, but don't yet have a conception of a type of state which purports to represent the world as being in a particular way, but does so incorrectly - they don't yet have a conception of 'belief' as such. But as the child's theoretical and inferential powers develop, it reaches the stage at which it can treat pretence and belief as distinct states.

Here we may have a state - e.g. ascribing a prelief to someone - which acquires a new content - it becomes an ascription of belief - as a result of the new inferential powers of the theory-of-mind system which can operate upon that state. Whether or not this particular account of this stage in child development is true, this is all standard consumer-semantic stuff. But would Levine object (echoing the remarks in the penultimate paragraph of his comments), 'It is ad hoc to say that the state in question has acquired a new content'? Would he say, 'Granted there may be some difference in content, but why say that the new content is the content "belief"? Why not say that the mere difference in causal relations is the difference in content?' Such objections would be ill-motivated in the case of belief, unless they are intended as objections to consumer semantics per se. They are equally ill-motivated as objections to dispositionalist HOT theory.

4 Finally …

One final point of clarification: Levine asks why I sometimes address questions of natural possibility: isn't metaphysical possibility always the relevant modal operator to consider? The answer is: yes, it is. But on many views of natural properties, their identity is tied to worlds in which the laws of nature remain the same. So when questions of property-identity or property-supervenience are at issue, conceivable circumstances which are naturally impossible are also very likely to be metaphysically impossible. Put differently: the relevant notion of supervenience to consider is global, where laws of nature as well as physical / functional facts are held fixed.


Block, N. (1990). "Inverted Earth.", Philosophical Perspectives, 4: Action theory and philosophy of mind, 53-80. Ridgeview Publishing.

Carruthers, P. (2000). Phenomenal Consciousness: a naturalistic theory. Cambridge University Press.

Dretske, F. (1995). Naturalizing the Mind. MIT Press.

Perner, J., Baker, S., & Hutton, D. (1994). "Prelief: The conceptual origins of belief and pretence." In C. Lewis & P. Mitchell (Eds.), Children's early understanding of mind: origins and development (261-286). Lawrence Erlbaum Associates.

Shoemaker, S. (1981). "The inverted spectrum." Journal of Philosophy, 74.

Tye, M. (1995). Ten Problems of Consciousness. MIT Press.

Tye, M. (2000). Consciousness, Color and Content. MIT Press.


© 2001 Peter Carruthers