Myin's stimulating discussion of "The Paradox of Self-Consciousness" (Bermudez 1998, 1999) takes issue with the conception of psychological explanation and propositional attitude ascription that I use in developing my account of nonconceptual self-consciousness. He objects in particular to my appeal to internal content-bearing states. In this reply I defend my account of psychological explanation against his objections, explain its motivations and point to some problems with the alternative he offers.
2. Myin offers four different ways of understanding propositional attitude ascriptions (taking propositional attitudes to include what in my book I called proto-beliefs and proto-desires). According to the weakest of these views a correct propositional attitude ascription implies nothing more than that the subject stand in an appropriate relation to a fact (where, I take it, facts are narrowly or extensionally construed, so that nothing hangs on how the fact is described or apprehended). This is the position Myin endorses and he terms it position (A). The three further proposals are: (B) the subject is appropriately related to a content-bearing internal state; (C) the subject is appropriately related to a content-bearing internal state with a specific format, and (D) the subject is appropriately related to a content-bearing state with a propositional (i.e., conceptual) format. Myin characterises my position as affirming (B) and (C), and rejecting (A) and (D).
3. It is certainly true that I am not happy with (A), for reasons I shall come to shortly - and also that I reject (D). But the way in which Myin presents (B) and (C) is potentially confusing. The point is not that the subject should be appropriately connected to a content-bearing internal state - but rather that an internal state of the subject should be appropriately connected to a content. The point is important because there is a well-known way of thinking about propositional attitude content (the so-called Russellian conception of content), in which contents are essentially facts or states of affairs. One can accept this view of content without denying that there are content-bearing internal states. It is important to keep questions about the nature of content distinct from questions about the nature of the vehicle of that content.
4. My own view is that the content of a propositional attitude cannot be simply identified with a fact or state of affairs. Content is essentially aspectual. There are different ways of apprehending the same fact, such that one can believe a fact to obtain when it is apprehended in one way and simultaneously believe that it does not obtain under another "mode of presentation". To each of these ways of apprehending a fact there corresponds a different content. At the level of nonconceptual content, aspectuality is captured using Peacocke's notion of protopropositional content. At the level of conceptual content we need to deploy the well-known distinction between sense and reference. This alone would give me enough reason to reject Myin's position (A), because whether or not content requires a vehicle, content cannot be identified with facts or states of affairs extensionally construed.
5. It is worth at this point commenting on Myin's proposal that the difference between conceptual and nonconceptual does not need to be captured at the level of content. He suggests (section 4) that the distinction between conceptual content and nonconceptual content is essentially put forward to capture the difference between belief (and other doxastic states) and perception - and hence that, once we adopt proposal (A), we can capture these differences by specifying different cognitive ways in which a subject can be related to a fact or state of affairs, with no further need for the distinction between conceptual and nonconceptual content.
6. Although Myin has interesting things to say about the differences between perception and belief in section 5, this proposal is misguided because the distinction between conceptual and nonconceptual content does not map clearly onto the distinction between perception and belief. As I discussed in Chapter 3, the distinction between conceptual and nonconceptual is much more closely tied to the distinction between linguistic and non-linguistic and this cuts across the distinction between perception and belief. A nonconceptual content can be more like a belief than a perception (as with what I called proto-beliefs), and perceptions can have a conceptual content (as in creatures who are fully linguistic and possess concepts).
7. Turning now to the key issue, Myin in effect is asking for an answer to the question: why should contents be thought to have vehicles? My reason for maintaining that they do is not, as Myin suggests in several places, what he calls the paraplegic objection. This is the objection that, since a supporter of (A) is committed to analysing propositional attitudes in terms of behavioural dispositions, the defender of (A) is committed to denying that paraplegics can have beliefs and desires. This is a genuine problem, but not one that a resourceful defender of (A) will find insuperable.
8. The main reason for rejecting (A), as Myin recognises in section 9, is that propositional attitude ascriptions cease to be genuinely explanatory when they are cashed out purely in terms of behavioural dispositions to particular states of affairs. First, there is the potential circularity that always threatens versions of behaviourism. If part of what it is to have a particular belief is to be disposed to act in a certain way, then how can the belief explain the action? It is not much of an explanation to say that an agent acted in a certain way because they were disposed to act in that way!
9. Some philosophers of a behaviourist persuasion have responded to this difficulty by adopting a form of instrumentalism, or irrealism, about propositional attitude ascriptions (Dennett 1987). I take it, however, that this is incompatible with the idea that propositional attitude ascriptions are doing any genuine 'explanatory' work - although it might be compatible with their having a certain predictive or heuristic utility. The reason for this is the lawlike nature of explanation.
10. What makes a particular explanation (of whatever kind) genuinely explanatory is the fact that it is backed up with, or subsumed under, a law. An explanation works backward from an observed effect (the relevant piece of behaviour) by offering a set of initial conditions such that the observed effect could not but have followed - and this requires a lawlike connection between the postulated initial conditions and the observed effect. Clearly, though, this will not work unless the initial conditions and observed effects are distinct in a way that they could not be on Myin's proposal. Another way of putting this, of course, would be to say that psychological explanations are a species of causal explanation (Davidson 1980). However one understands content, content itself is not causally efficiacious (or at least not in the right way). Even if one follows Russell and thinks that the content of a belief is a state of affairs, that state of affairs is not going to cause the behaviour that one cites the belief in order to explain. The cause comes from the internal state which vehicles the content.
11. I fully agree with Myin that there is a very real issue about how, as he puts it, propositional attitude talk connects up with brain talk. It is false, however, to claim as he does that "the only option, if one accepts talk of states per se as informative, is to assume that there is some kind of mapping which connects content types to physical types" (para 10). Even if one thinks that internal content-bearing states are discrete, explicit and symbolic entities, one is not obliged to think that they are type-identical with certain classes of neural state. Even advocates of the computational theory of mind think that the representations they posit are many levels of realisation and abstraction away from the neural level. The programme of homuncular functionalism (Cummins 1983, Lycan 1987) is a clear alternative to a type-identity theory, even within the computational paradigm.
12. In any case, there seems no reason to think that talk of representations can only take place within the computational paradigm. In the book, I was at pains to stress that my position was neutral between different conceptions of sub-personal cognitive architecture. It is far from clear that connectionist and other dynamical approaches to cognitive architecture are incompatible with the broad model of psychological explanation I have defended above and in my book (see the essays in Macdonald and Macdonald 1995 for further discussion).
Bermudez, J. L. (1999) Precis of "The Paradox of Self-Consciousness". PSYCOLOQUY 10(035) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/1999.volume.10/ psyc.99.10.035.self-consciousness.1.bermudez http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?10.035
Bermudez, J. L. (1998) The Paradox of Self-Consciousness. Cambridge MA. MIT Press.
Cummins, R. 1983. The Nature of Psychological Explanation. Cambridge MA. MIT Press.
Davidson, D. 1980. Essays on Actions and Events. Oxford. Oxford University Press.
Dennett, D. 1987. The Intentional Stance. Cambridge MA. MIT Press.
Lycan, W. 1987. Consciousness. Cambridge MA. MIT Press.
Macdonald, C. and Macdonald G. 1995. Connectionism: Debates in Psychological Explanation. Oxford. Blackwells.
Myin, E. (2000) Direct Self-Consciousness. PSYCOLOQUY 11(03) ftp://ftp.princeton.edu/pub/harnad/Psycoloquy/2000.volume.11/ psyc.00.11.003.self-consciousness.2.myin http://www.cogsci.soton.ac.uk/cgi/psyc/newpsy?11.003