Reply to Murphy on Mind and Supermind

A couple of years ago, Eric and I had an exchange about my book Mind and Supermind (see here and here). Part of this revolved around passages from Dominic Murphy's review of the book on Notre Dame Philosophical Reviews. I felt that the review misrepresented the scope and aims of the book, and Eric suggested I write up a response for the underblog. I made some notes, but shelved them, thinking it would seem querulous to reply to a reviewer. However, since then I’ve occasionally been asked how I would respond to Murphy’s objections, and I felt it might be useful to dig out the comments and make them public. So, since Eric kindly kept his offer open, here they are. Aside from adding a few references to my more recent work, I've not changed them.
-----

I am grateful to Dominic Murphy for his review of my book, which I found thoughtful and provocative. However, I felt that in several places he misinterpreted the book’s aims and the nature and strength of the claims made being, and this reply is intended to set the record straight. This is not meant to imply to any criticism of Dr Murphy. If the book is open to misinterpretation, then that is squarely my fault. But I believe the issues are important enough for it to be worth clarifying exactly what I was proposing.

(1) Murphy claims that the book makes a very strong to commitment to the view that folk psychology quantifies over real entities and is almost wholly accurate about their nature.

Reply: The book does argue that folk psychology is committed to a form of intentional realism, but the form in question is a non-standard one, which does not involve claims about internal architecture. I argue that the folk concept of belief tracks personal-level states, not subpersonal ones -- behavioural dispositions in the case of what I call the 'basic mind', and behavioural commitments (specifically, premising policies) in the case of the ‘supermind’. Nor does the book argue that folk psychology is accurate. Its aim is to articulate a view of the mind that systematizes our folk-psychological intuitions about it -- a theoretical regularization of folk psychology. For the purposes of the project, I proceed on the assumption that most folk-psychological practices are sound, but (aside from a few speculative comments in the final chapter) I don’t argue that the resulting theory is in fact true. I make this clear in Chapter 1.

(2) Murphy objects that it is implausible to claim that folk psychology divides neatly into two unified sub-theories, each of which accurately characterizes one aspect of our mental nature. He suggests that the conflicts I identify cross-cut each other, and that mental states can share properties I treat as exclusive.

Reply: First, since the two-strand framework is intended as a theoretical regularization of folk psychology, it’s not essential that it should capture all our intuitions and ways of speaking (pp.8-11, 49-50). What matters it is that it should embody a reflective equilibrium, which strikes the optimal balance between intuition and theoretical economy. Second, I don’t argue that the various features of belief must go together in the way I suggest. Rather, I sketch two types of attitude (behavioural dispositions and premising policies) in which they do go together in that way, and then suggest that everyday belief-talk is best understand as referring to one or other of these two types of attitude. Third, in the book I am concerned only with folk psychology, and I make no attempt to characterize states and processes at subpersonal level -- those, that is, that support the dispositions that constitute the basic mind. These are, I argue, beyond the purview of folk psychology. Thus, nothing I say rules out the possibility that there may be subpersonal states which combine features of strand 1 and strand 2 (see pp.43-4).

(3) Murphy objects that it is just false to claim that conscious belief is binary and that we have no immediate awareness of our degrees of conscious belief.

Reply: I agree that we can have varying degrees of confidence in our conscious beliefs (the claim is in fact central to my definition of superbelief), but I deny that this affects their status as binary. Conscious belief is binary in virtue of the fact that it involves commitment to a premising policy, and this sort of commitment does not admit of degree. Typically, then, our degrees of confidence in our flat-out beliefs will not affect how we reason with those beliefs -- though they will help to determine how ready we are to abandon them in the face of conflicting evidence. I agree, too, that we have some awareness of these degrees of confidence. I deny, however, that we have direct, introspective awareness of them. We infer the strength of a belief from our reactions to relevant scenarios -- for example, to the thought of giving it up, or of making various bets on its truth.

(4) Murphy objects that attributions of unconscious beliefs seem as susceptible to a binary interpretation as attributions of conscious ones, and, more generally, that folk psychology seems to treat conscious and unconscious beliefs in the same way.

Reply: I agree that folk psychology doesn’t treat conscious and unconscious beliefs as radically different; indeed I think that for most purposes it conflates the two strands (p.10). Again, the proposed distinction is offered as a theoretical regularization of folk psychology, not an analysis of it. I claim that there are deep tensions within folk psychology, which are best understood as arising from the use of the same theoretical vocabulary for two different but superficially similar phenomena.

(5) Murphy objects to the claim that what I call ‘austere functionalists’, such as Dennett and Davidson, must hold that beliefs can only ever be sustaining causes and never dynamic (triggering) ones.

Reply: By ‘austere functionalism’ I mean the view that treats beliefs and desires as multi-track behavioural dispositions, and thus as standing states. Dynamic causation is a relation between events (e.g. between ‘occurrent’ beliefs and actions), so an austere functionalist must deny that beliefs and desires are dynamic cases. Of course, when a disposition manifests itself there will be various relevant events in the offing -- triggering events and lower-level mediating events. But on an austere functionalist view none of these can correctly be described as an occurrent belief or desire.

(6) Murphy notes that the proposed two-level framework results in a very complicated theory of reasoning on which actions may have two different explanations, and he objects that the gains don't compensate for this complication.

Reply: The account is indeed complicated, but it does have substantial compensating gains. It explains and resolves numerous tensions in folk psychology. It meets the Bayesian challenge (the challenge of explaining how flat-out beliefs can guide the behaviour of a rational agent). It explains how thought processes can constitutively involve natural language. It vindicates the intuition that we have a measure of control over our mental states and reasoning processes. It explains how we can be justified in our tacit commitments to propositional and conceptual modularity, even though we are ignorant of the functional architecture of the brain. And, as I argue in the final chapter, it offers robust explanations of akrasia, self-deception, and first-person authority. Of course, the proffered explanations may not be sound, but if they are, then they amply compensate for the complications.

(7) Murphy objects that because superbeliefs are held with high confidence I face the same problem with conjunctive closure as advocates of the confidence view. So why not stick with that view, and treat the graded conception of belief as primary -- thereby avoiding all the complicated extra machinery?

Reply: This misses the point. The problem for the confidence view is that it cannot explain why belief should be subject to conjunctive closure in the first place, and virtually requires us to deny that it is. The proposed two-level view doesn’t face this problem. If forming the flat-out belief that p involves (among other things) committing oneself to taking p as a premise in deductive reasoning, including conjunction introduction, then our commonsense commitment to conjunctive closure is vindicated. Now it’s true that on my view flat-out belief also involves high confidence, and thus that it is also subject to probabilistic norms that are incompatible with conjunctive closure. But this is not a problem, since it appears to be true. We do feel that belief is subject to conflicting norms -- which is how the paradox of the preface gets its grip. One moral of this -- which I draw explicitly in a more recent paper -- is that the folk concept of flat-out belief is unlikely to have an important role in formal epistemology.

(8) Murphy asks why we should take the view that the supermind is realized in the basic mind. Wouldn’t it be simpler to treat the basic mind and the supermind as having largely independent physical bases? He also complains that no details are given about the notion of realization.

Reply: In saying that the supermind is realized in the basic mind, I mean that it consists in the possession of a set of basic beliefs and desires -- beliefs and desires that motivate and guide an appropriate range of premising policies. This view is a corollary of the claim that superbeliefs are premising policies, since policy possession is constituted by the possession of certain beliefs and desires. (Thus, even if the supermind were realized in a separate neural subsystem, that subsystem would still have to support a hierarchy of intentional states -- those at the lower level constituting the premising policies in which those at the higher level consist.) The case for the premising view itself (a variant of the behavioural view) is made in Chapter 3, and the realization thesis is developed in detail in Chapter 4. Further support for the realization thesis comes from its role in the various applications of the theory. In particular, it is crucial to the proposed solution to the Bayesian challenge -- a challenge that can be recast in a more general form as that of explaining how two reasoning systems can constitute a unified agent. Views which posit separate reasoning systems have serious problems here. The realization thesis also has attractions from an evolutionary point of view (see the -- admittedly brief -- remarks at pp.152-3 and 228-9). More recently, I have argued that the realization thesis can also resolve problems facing dual system-theories in cognitive psychology (see my 'Systems and levels', linked in (11) below.)

(9) According to Murphy, I deal with the folk commitment to propositional and conceptual modularity by denying that modularity at the supermental level is an architectural hypothesis and treating it simply as a commitment to independently assessable contents. Given this, he argues, the proposal could have been stated as a simple Dennettian claim about the shallowness of intentional psychology, without all the machinery of mind/supermind.

Reply: I do not deny that the modularity theses are architectural hypotheses. I treat them as claims about functional organization -- about the causal roles of various mental states and components. I do deny, however, that they are claims about neural architecture. The mental states and components in question are personal-level states and skills, and the functional architecture they implement is a virtual one. This is argued for at length in Chapters 6 and 7.

(10) Murphy claims that if the supermind is realized in the basic mind, then it ultimately has whatever neural architecture the basic mind has.

Reply: If the thought is that ascriptions of supermental states entail claims about subpersonal architecture, then the objection trades on a type/token confusion. Each token superbelief will inherit the subpersonal realization base of the token basic beliefs and desires that constitute it. Every higher-level state needs some realization base. What I deny is that superbeliefs require a certain type of subpersonal realization base, and I argue at length that they don’t, since a richly structured supermind can be realized in an austere basic mind, understood as a set of multi-track behavioural dispositions.

(11) Murphy objects that there is no science in the book, and little attempt to show that the proposed two-level framework sheds light on scientific inquiries.

Reply: Again, this reflects the aim of the book, which is to systematize our everyday view of belief, not to defend that view. Now, I do in fact think that the resulting regularized theory is broadly sound (and say so in the book), but establishing its soundness is a very different project -- and one that certainly does require contact with the empirical literature. Apart from a short section on scientific psychology in Chapter 7, there is nothing on this in the book. (For some later work that seeks to apply the two-level framework to issues in scientific and clinical psychology, see here and here.) Moreover, even if the theory is sound, it will not shed light on the subpersonal organization of the human mind. The states and processes it deals in are personal-level ones: dispositions, commitments, and activities of the whole agent. Some may say that the really interesting questions about the mind are precisely ones about subpersonal psychology. I tend to agree, but it is important to disentangle those questions from those posed by folk psychology, and that is a central aim of the book.