Sunday, October 24, 2010

The Heterogeneity of Mind

In 1949 Gilbert Ryle published The Concept of Mind, one of the most important books of philosophy of mind of the last century and probably the best manifesto of philosophical behaviorism. Although today few would endorse Ryle’s strictly behaviorist semantics of psychological predicates the book continues to be persuasive as a sustained attack on what Ryle calls “Cartesian” theories of mind. Specifically Ryle challenges the ancient intuition that the word “mind” refers to some one, unanalyzable thing. He does this more thoroughly (and in a grander style) than anything I can do here, but he wrote at a time when the practice of metaphysics was out of favor in the English-language philosophical world. Today we enjoy the benefits of the “language” philosophy that was done by the early 20th century empiricists and the benefits of the revival of metaphysics, which has been to some extent motivated by the emphasis on philosophy of mind, of the past several decades.

Imagine, Ryle asks, that a visitor has asked to be shown the university. One walks the visitor through campus: “There is the Student Center, and there is College Hall, and those young people sitting around the fountain over there are students, and there is old Professor Whiskers, you can set your watch by his walks across campus, and say hello to my friend Imelda here, she’s our new dean, now come I’ll show you the library,” and so on. At the end of the day the visitor is asked what he thought of the university. “But,” he protests, “You didn’t show me the university. We only saw buildings, people, books and things like that.” Ryle argues that a similar “category mistake” is made when we posit, behind or above or in addition to specific, observable behaviors, a “ghost in the machine,” a “mind.” He further argues, echoing Hume, that there is no “inner mental space” where mental events occur. His very title, “the concept of mind,” telegraphs his view that “mind” is a heterogeneous concept.

A heterogeneous concept is one that turns out, under analysis, to consist of multiple, distinguishable things. Ryle points out that the grammatical behavior of nouns is such that we can be led to think that there is something that exists when there is nothing (Dickens’s Mr. Pickwick, for example), but that this is practically speaking the same thing as thinking that only one thing exists when in fact the concept involves many things (Dickens, one of his novels, the tradition of fiction; football players, uniforms, equipment). All I mean by "analysis," that I am not using in any sort of technical manner, is thinking about the referents of the term (semantics and metaphysics often come to the same thing). Examples of heterogeneous concepts from outside of philosophy of mind are value terms like "ethics" or "beauty," or for that matter very many abstract nouns such as (opening the dictionary randomly) "reservoir." Wittgenstein famously explained the heterogeneous nature of the concept of “game.”

Heterogeneous words are common (really, I don’t like to use the word "concept," although it is hard to avoid. It comes with a treacherous load of academic baggage. I'm thinking about the uses of the word; the nature of the “concept” is what is at stake, after all). We can understand the continuity of meaning between "That man's reservoir of good will" and "The city's reservoir of water" (the first use started as a simile of the second) but if we are thinking about what the word refers to the two uses are different enough that it makes most sense to say "'Reservoir' is a heterogeneous word," meaning that it is a word that can refer to multiple, distinguishable things.

If we stay alert to the fact that individual nouns, and particularly abstract nouns, routinely turn out to refer to distinguishable things we can sometimes clear the smoke away a bit from philosophical arguments. For example ethical theorists (perhaps not the best ethical theorists, but quite a few ethical theorists) might see themselves as involved in some sort of partisan contest: are the “rights theorists” correct (or better or what have you), or are the “consequentialists” the ones who are giving us the best account of things? Or maybe virtue theory is preferable to both? Certainly philosophers working on ethical theory are frequently identified as “rights theorists” or as “consequentialists”: “I’m a consequentialist” is taken to mean not only that one endorses consequentialism but also that one declines to endorse the other types of ethical theory on offer.

But wait: actual people are "ethical" on a formal, logical sort of level (respecting others' rights through applying the logic of universality) and "ethical" on a situational, emotional sort of level (minimizing felt harm through the capacity for empathy) and they appreciate "good" people who they estimate to be salutary examples of a well-realized person (a “gentleman of Athens”). In fact real ethical people (that is, people when they're actually trying to act ethically rather than merely trying to do ethical theory) use Kantian-style "golden rule" reasoning and Millian outcomes-oriented strategies and they make Aristotelean evaluations of themselves and others all at the same time. “Ethics" turns out to be a heterogeneous concept: the intentions of rational beings, the qualitative experiences of conscious beings and the health or pathology of living beings are all different things, such that there turn out to be not so much differences of opinion among "ethical theorists" as there are changings of the subject. Confusion (and sound and fury) is generated by a presumption that ethical thinking must be one kind of thinking and so there must be one “theory” that gives an account of it. The misleading grammar in this case is the use of a singular abstract noun “ethics,” which creates the strong impression that there is only one topic when in fact there are several that come under that rubric.

The alarmed ethical theorist might speak up at this point: “Too fast.” When David Hume says “Reason is the slave of the passions,” he is making the substantial claim that logical operations are secondary and merely instrumental and that qualitative experience is the primary explanans of “ethics.” When Kant argues that all and only rational beings constitute a “kingdom of ends” he is making a substantial claim that the physical universe portrayed by science (the “phenomenal world”) is valueless qua physical, and that transcendental logical necessity is that explanans. These look to be mutually exclusive claims, and neither is compatible with Aristotle’s view that fulfilling the telos of a living human being is ultimately the aim of “ethical” behavior.

And mutually exclusive they are. But the claim that experience is the only thing we know, or the claim that there are no values in the physical world studied by science and that therefore they must come from somewhere else, are metaphysical and epistemological claims. All philosophy is about metaphysics and epistemology, as unfashionable as it may be to say so these days. And Hume (about whom I will have a good deal more to say in Chapter Three) points out the curious fact that no amount of discussion of physical experience produces any account of programmatic duty, while Kant is moved by his sense of the amorality of the physical world to make the radical claim that the phenomenal world is not, could not be, all that there is. The penultimate difference between Hume and Kant is a difference about the nature of the human mind; like all of the best philosophers their views on both ethics and psychology are systematically motivated by more central positions on epistemology and metaphysics. So if there is a persuasive argument that mind is a heterogeneous concept that argument will extend to the claim that ethics is a heterogeneous concept.

The deeply-internalized intuition that there is some one thing that is the “mind” reflects the plain fact that there is one thing that is the body. For each person the body is singular (at least in our experience!), and once the idea emerged that the mind existed separately from the body (or, at least, that the mind was metaphysically distinct from the body) it was natural to think that there was a one-to-one correspondence between bodies and minds (or “souls”). But the burden of proof is surely on those who would maintain that psychological predicates refer to some one, unanalyzable thing. The metaphysical dualist points to the difficulty we have in providing a naturalistic semantics for psychological terms as a justification for accepting dualism, but we have already seen that the intentional terms and the phenomenal terms resist naturalization in different ways: we might eventually be forced to accept a dualist account of the intentional mind but not of the phenomenal mind, or vice versa, so even a convincing argument for dualism wouldn’t entail that psychological predicates refer to something homogeneous.

As for phenomenal arguments about the unity of perception, apperception, consciousness or what have you, “unity” is exactly what one would expect if one held that in the final analysis psychological predicates referred to embodied beings in physical environments. Kant, one of the greatest and richest philosophers in this field, has to work hard on his account of the unity of mind because he does endorse just the distinction between the rational mind and the conscious (that is physical-world-experiencing) mind that I am stressing here, he doesn’t think that the rational mind can be naturalized and he does think (he fears) that the conscious mind can be. (Strictly speaking Kant’s famous distinction between the “noumenal” and the “phenomenal” worlds is epistemological – the world of experience is that part of the world-in-itself that our minds can feature in a representation – but if rationality is assigned to the noumenal and sensory experience is assigned to the phenomenal then the distinction is equivalent to the one I am making here.) If there were persuasive natural semantics available for both types of psychological predicate (contra Kant who thinks there can be none for intentional predicates) then the “unity of mind” would have been shown to be simply the unity of body: to claim that mind is unanalyzable prima facie is to beg this question.

There is one more objection that cannot be avoided, this one from familiar arguments in the area of personal identity. A defining argument in the area of personal identity is that between advocates of physical continuity and advocates of psychological continuity. At least since the time of Locke the majority view has been that psychological theories of personal identity are more persuasive than physical theories. Imagine (the story goes) that one’s mind has been switched with another (physical) person’s: mind A in body B and mind B in body A. Where (one asks the students) are you now? Most people have the intuition that they go where their mind goes, that is, that they are their mind as opposed to their body if forced to make the choice. It is significant that it does seem possible to conceive of one’s mind separated from one’s body. Isn’t that a problem for any physicalist theory of mind? I think it is, and I will take up the issue of what it is actually possible to conceive, and what that possibility might show, in Chapter Three in the discussion of the “absent qualia” arguments, the possibility of “zombies” etc.

But what is at issue in this section is not the mind/body problem itself but the ground-preparing question of whether there are two problems rather than one. Consider the “memory theory” owed to Locke himself. On this view shared memories are the psychological link that establishes the continuity of self across the passage of time (the old general remembers the brave officer’s battle, the brave officer remembers the young boy stealing the apple and so forth). But if the operationalist holds that memory is a representational system that gains, edits and stores information, this functional ability is not sufficient to constitute selfhood: two beings with the same database are not thereby the same person. And if the phenomenologist is right that no amount of functional description will ever capture the quality of conscious experience then there can be no purely functional account of memory itself, let alone of personal identity.

If, on the other hand, we have a phenomenal account of memory continuity – that would have to be something like “having memories with the identical qualities” – then we get the problem in reverse, since we cannot establish the causal role of consciousness (which is just another way of putting the phenomenologist’s point that we cannot provide a functional account of consciousness). So a phenomenal account of memory (whatever that might be) would also not be sufficient if used to try to establish personal identity. Identity of representational content and identity of qualitative experience are both necessary, but neither is sufficient, for personal identity. Since the reason that neither account of memory is sufficient is that each leaves the other one out that establishes that they are two different things.

To summarize, my claim is that there are two metaphysical problems for the naturalization of psychology. My method is to look at the metaphysical commitments – the semantics – of the vocabulary of psychological predication. This vocabulary divides into two sets of words. First there is the intentional vocabulary. This consists of words like “belief,” “desire,” “hope,” “fear” and so on. Use of these words appears to commit us to the existence of rationality and mental representation; I will use the word intelligence to refer to the intentional mind in toto. The other set is the phenomenal vocabulary. This consists of words like “sensation,” “pain,” “taste,” “texture” and so on. Use of these words appears to commit us to the existence of consciousness. Operationalist theories are theories about intelligence; phenomenal theories (which are rather thin on the ground, for reasons I will discuss in Chapter Three) are theories about consciousness.

Once one sees that there are two mind/body problems, not one, it is possible to address each problem in turn. Chapter Two breaks down the problem of intentionality further, developing the distinction between the problem of mental representation and the problem of rationality, and offers two respective arguments to naturalize the semantics of intentional predicates. Chapter Three offers arguments to the effect that the problem of phenomenology is a pseudoproblem and then explains how phenomenal predicates can be naturalized as well. The arguments in the two chapters are different responses to different metaphysical problems, but taken together they may work towards a naturalistic semantic for psychological predicates. In the more speculative Chapter Four an account of the nature of the relationship between intelligence and consciousness is proposed that reflects the conclusions of the earlier chapters.

1 comment:

  1. Ryle taught Dennett. Anyway, Ryle's real problem was that he had no idea of the nature of time and considered that his "regress" showed that mind did not exist. A scientist would have realised that a theory that leads to a regress when explaining a phenomenon is false, not the phenomenon. See New Empiricism for a fuller explanation.