Tuesday, July 27, 2010

Ryle On Rules And Creativity

Introduction

In his introduction to Creativity, Cognition, and Knowledge: An Interaction (2002), the late Terry Dartnall challenges his fellow cognitive scientists to pay attention to Wittgenstein in order to understand the central role of creativity in human intelligence. I want to explain why I think he is right, and why I think just as much attention should be paid to Gilbert Ryle. The main issue is the relationship between rules and behavior, an issue which may seem simple, but which implicates our very conception of what it means to be a person.

It is sometimes said that, whereas people are creative, computers can only do what they are programmed to do. The idea is that creativity cannot be wholly rule-based. Our intelligence cannot simply be a matter of following rules.

The truth may be more complicated. Contemporary wisdom has it that the right kinds of programs can lead to creative behavior. Take genetic algorithms. They involve controlled processes of random mutation and selection based on the principle of natural selection, and lead to novel solutions to pre-defined problems. A complex set of genetic algorithms can allow a computer to define new problems for itself to solve, new rules for itself to follow.

When we are creative, we are enjoying a controlled spontaneity which involves rules, but which is not wholly determined by them. For example, musical improvisation involves a delicate balance between control and spontaneity, so that we develop melodies, rhythms and harmonies as we go. This may be the key to understanding intelligence.


Wittgenstein

Wittgenstein famously argues that games and language are not wholly circumscribed by rules. In Philosophical Investigations (1953, section 68), he writes, "[The use of the word 'game'] is not everywhere circumscribed by rules; but no more are there any rules for how high one throws the ball in tennis, or how hard; yet tennis is a game for all that and has rules too." His point is as much about what it means to play a game as it is about our use of the word. Part of playing a game means grasping the rules of the game; yet, if all you know are the rules, then you still have to learn how to apply them. You have to understand them in an executive manner, with creativity. Intelligent performance cannot be derived from the rules.

Wittgenstein not only argues that the performance cannot be derived from the rules, but also that the rules cannot be derived from the performance. Any intelligent act of rule-following is conceivably consistent with mutually exclusive sets of rules, and there is no way to establish a connection between rules and behavior without relying on another set of rules which, in turn, would themselves require yet another set of rules to be established, ad infinitum. The question is, how do we know when we have correctly followed a rule?

Wittgenstein's answer is that we cannot always have a rule for that, and that we do not always need rules for that. If it were otherwise, every rule would need another to establish its correctness, and we would end up with an infinite regress.

It might seem that Wittgenstein is asking us to abandon the notion of "correct." This is why Saul Kripke (1982) thought Wittgenstein was courting a radical skepticism. Ruth Millikan (1990) suggests otherwise, drawing a distinction between following a rule and grasping a proposition, where the former is more biologically primitive. Biologically primitive systems follow rules without representing them, so there is no act of interpretation required for their application. We do not always need rules to tell us how to follow rules. What is "correct" or "true" is ultimately a function of living organisms or, as Wittgenstein puts it, ways of life.


The Concept of Mind

Wittgenstein once said that only two people understood his philosophy, and that one of them was Gilbert Ryle (Monk 1991, p. 436). It is therefore unsurprising that Ryle follows a similar line of thought in The Concept of Mind (1949), where he distinguishes between following a rule (or applying criteria) and grasping a proposition. This may be expressed as the distinction between knowing how (to do something) and knowing that (something is the case). Facts can be parroted; intelligent performance cannot. Performances can be aped, and they can be aped creatively; but any creativity thereby exhibited is creativity in the aping, and not in the procedure being aped. The point is that, in some cases--such as assembly-line procedures which are done routinely, out of habit--performance is clearly not creative. (There can be a gray area between the clearly intelligent and the clearly unintelligent; creativity is not always so easy to spot.)

Ryle writes, "To be intelligent is not merely to satisfy criteria, but to apply them; to regulate one's actions and not merely to be well-regulated. A person's performance is described as careful or skillful, if in his operations he is ready to detect and correct lapses, to repeat and improve upon successes, to profit from the examples of others and so forth. He applies criteria in performing critically, that is, in trying to get things right" (pp. 28-29). To follow a rule is to regulate one's actions, to show intelligence in the performance of a task, and not to simply do the same thing one has done before. It is to do something with the readiness to improve and improvise, to learn as one acts and through one's action. The mark of the intellect is active learning. As one of Ryle's students observes, "it is not repeated behaviors but changing behaviors that are the sign of creative intelligence, and this can only be observed in the long term" (Dennett 1983, reprinted in 1998, p. 335).

This is Ryle's challenge to what he calls "the dogma of the ghost in the machine," the tradition of Cartesian dualism which says that minds and bodies are distinct entities, each with their own causal properties. Ryle says this is a category error. When we speak of minds, we are not speaking of entities, states, or events. The language we use to talk about minds employs a logic of dispositions, and not simply of occurrences. Rather than think of the mind as a particular place or thing, Ryle asks us to imagine it as a complex set of abilities, capacities, skills, and so on. These are observed "in the long term," as Dennett puts it, and not as discrete entities, states, or causes.

For example, we do not describe a particular occurrence when we refer to the brittleness of glass. Rather, when we say a glass is brittle, we are saying something about how the glass will act in certain circumstances. Similarly, to say that a person has a mind is to say that they act in special sorts of ways--ways which exhibit intelligence. Of course, minds, like brittleness, may be explained in terms of causes and effects. However, neither minds nor brittleness are particular structures or particular sets of causes and effects.

Ryle's view has been associated with behaviorism, since it defines the mind in terms of observable behavior. (It is sometimes called "logical" or "philosophical" behaviorism, to distinguish it from the psychological behaviorism commonly associated with B. F. Skinner.) This association, while justified, may be a little misleading. Ryle's argument is not that we do not have private thoughts, or that we do not imagine, think, or feel. He does not ignore the richness and potency of experience. Rather, he says that the marks of the mental are not intrinsically private. Sometimes they are public, such as when we speak or write, or otherwise perform publicly. Thoughts are only sometimes private, and then only by convention or circumstance. Furthermore, such private acts do not confer intelligence to outward, public acts. Ryle's insight is this: Intelligent behavior is not the product of intelligence; it is intelligence itself.

Ryle's view shares some features with a view in contemporary philosophy of mind known as the extended mind hypothesis, which says that mental states are not all confined by the physical boundaries of the body. Ryle might agree with the extended mind hypothesis--he argues that judgments and ideas are as present in our published works as they are anywhere else--though he would have reservations about the word "state." He rejects the idea that the mind can be reduced to discrete states of any sort. When we attribute mental states to individuals, we are saying something about how they are likely to behave. We are not postulating particular states which exist in the brain or anywhere else. This is not to say that the brain is irrelevant; it is rather to say that so-called "mental states" are not states, even if they depend in some way on brain states. We may postulate functional brain states in our attempt to understand minds; we should just be wary of confusing those functional states with what we are trying to explain.

That said, Ryle observes that sometimes we do directly implicate the brain when we speak of the mind: for example, when we say that a person is imagining, feeling, or perceiving something at some particular time. In such cases, Ryle says, we are creating a mongrel category which cannot be analyzed as simply dispositional or simply occasional. Perhaps there would be less disagreement about consciousness and perception if we had better tools for analyzing this mongrel category. The idea of discrete perceptual feels (qualia), for example, may result from failing to appreciate the dispositional elements in our reports of perceptions and perceptual knowledge.


Intellectualism

Ryle contrasts his view with another view which he calls intellectualism, which is one face of the dogma of the ghost in the machine. Intellectualists attempt to explain intelligent behavior by postulating a special faculty which exists behind or antecedent to our intelligent acts. On this view, all intelligent behavior is caused by antecedent acts of intellection; intelligence itself is just a matter of grasping truths in the mind, and to be a person is to have such a truth-grasping faculty.

Ryle says no, to act intelligently is not to first consider an action in private, and then to execute it in public, according to a plan. Intelligence can and often does involve planning, but the relationship between an action and a plan is not the relationship between a copy and an original. Plans are active components of our intelligent behavior; they are not necessary prerequisites of our intelligent action. Much like the regress argument which would later appear in Wittgenstein's Philosophical Investigations, Ryle presents a reductio against intellectualism: If we think of planning or contemplation as a prerequisite for intelligent action, we end up with an infinite regress, because planning and contemplation are varieties of intelligent action.

Imagine listening to a song on the radio and tapping the beat on your knee. In this case, your body is a sort of metronome. If you needed an internal metronome to keep steady your external metronome, then why wouldn't you need another metronome to keep your internal metronome steady? And why wouldn't you need yet another metronome to make sure your external metronome was properly correlated with the internal one? Nothing is solved by postulating an internal metronome which helps your hands keep with the music. It only creates more problems, more acts of beat-keeping which need to be explained. There is no reason to postulate a special sort of internal beat-keeping to explain our public displays of percussion.

Of course, when we learn how to keep a beat, we might start by doing it "in our heads"--that is, quietly, to ourselves. After we've gotten the beat down privately, we are more comfortable learning to synchronize our hand-on-knee movements. Once we've learned how to do that, we don't have to do both--we can just keep the beat with our hands. We don't need an internal metronome to monitor our external acts of percussion. I don't think we need the private, quiet performance in the first place; it just gives us a chance to learn one way of keeping the beat without revealing our mistakes.

The metronome example is my own. Ryle elucidates his view of intelligence with a number of other examples. In one, he discusses a clown who trips on purpose. The clown has some intention to trip, but the trip itself is, to one extent or another, spontaneous. The act of planning does not confer intelligence to the trip. Rather, the clown's active involvement in the tripping is what makes it intelligent. Indeed, a clown could plan to trip in a precise manner at a precise time, and then accidentally trip at that very time, and in a manner very similar to the way she had intended. An intelligent plan can be followed by an involuntary performance which looks just like an intelligent act, but which is not intelligent. Thus, the plan does not account for the intelligence of the act.

Ryle's critique of intellectualism has recently been challenged by Jason Stanley and Timothy Williamson (2001), who claim that Ryle gets the knowledge-how/knowledge-that distinction wrong. They present a linguistic analysis aimed at demonstrating that knowing-how is actually a species of knowing-that. Their work is responsible for a resurgence of interest in Ryle's distinction, though unfortunately they do not give Ryle a fair representation. They misconstrue the knowledge-how/knowledge-that distinction as a distinction between abilities and something else. In fact, Ryle regards all knowledge in terms of abilities and capacities, observing that "to know is to be equipped to get something right" and that the word "'know' is of the same family as skill words" (Ryle 1949, p. 134). This misconception leads Stanley & Williamson to misidentify intellectualism itself. On top of that, they misrepresent Ryle's reductio argument. In short, they do not connect with Ryle very well, if at all. (For a fuller discussion, as well as criticisms of Stanley & Williamson's linguistic analysis, see these previous blog posts: here and here.)

Psychologists have recently found evidence that children are able to attribute knowledge-that before they are able to attribute knowledge-how (Tardif, Wellman, Fung, Liu, and Fang 2005). Knowledge-how appears to be more opaque, or it requires different skills to identify, than knowledge-that. This is consistent with Ryle's view, which is that knowledge-how is more complex and heterogeneous than knowledge-that. For some reason, however, Stanley (forthcoming) suggests that this research corroborates his and Williamson's view that knowledge-how is really just a species of knowledge-that. On the contrary, this evidence seems to work against them. If knowledge-how were just a species of knowledge-that, then children should not be able to identify knowledge-that first.


Misinterpreting Ryle

Misinterpreting Ryle is a tradition which began long before Stanley & Williamson. One notable act is perpetrated by Jerry Fodor, who gives a woefully inadequate treatment of Ryle in his seminal The Language of Thought (1975). Fodor focuses on Ryle's clown example, but he leaves out everything about creativity. He claims that Ryle says the clown's behavior is clever only because of circumstantial facts, such as the time and place of the performance, and whether or not it was expected by the audience. This does a disservice to Ryle, and may partially explain some of the anti-Rylean prejudice which has permeated philosophy and cognitive science over the past several decades.

Unfortunately, Dennett does not catch the error in his 1978 response to Fodor. Worse, he echoes Fodor's accusation that Ryle harbors a “groundless anti-scientific bias” (Dennett 1978, reprinted in 1989, p. 45). I see no grounds for this accusation. Ryle does not privilege “conceptual” over “causal” accounts of behavior, as Fodor and Dennett say. He does not reject scientific accounts of behavior, nor does he minimize or devalue their efficacy. Rather, Ryle’s aim is to map out the logic of psychological explanations, a project he refers to as “philosophical psychology” (Ryle 1949, p. 319). He argues that psychology proper is a mixture of causal and non-causal explanations, and that it can only pave the way for “the establishment of precise functional correlations or causal laws” (Ryle 1949, p. 327). This suggests that psychology may one day be replaced by a more rigorous science of human behavior. Far from showing bias against science, Ryle embraces its potential.

Other philosophers have misconstrued Ryle in one way or another. Paul Snowdon (2003) makes many of the mistakes Stanley & Williamson make. Alva Noe (2005) defends Ryle against Stanley & Williamson, but mistakenly claims that Ryle was not an ordinary language philosopher. As it happens, Ryle was arguably the first ordinary language philosopher: His "Systematically Misleading Expressions" (1932) was a seminal work in ordinary language philosophy, and his "Ordinary Language" (1953) leaves no room for doubt about his views on the topic. In fact,
The Concept of Mind exemplifies Ryle's ordinary-language approach: He begins each new avenue of thought by identifying the common use of words and he consistently focuses on the logical behavior of our ordinary concepts. As he says, "this book as a whole is a discussion of the logical behavior of some of the cardinal terms, dispositional and occurrent, in which we talk about minds" (Ryle 1949, p. 126).


Conclusion

Though Wittgenstein is regarded as one of the greatest philosophers of the 20th century (perhaps the greatest), his contribution to philosophy is often debated. There is vast disagreement over what he said and whether or not it was worth saying. (This may partially be because Wittgenstein never fully formulated his later views, though he believed even the work he had published during his lifetime was generally misunderstood. Not even Bertrand Russell and Gottlob Frege understood it to Wittgenstein's satisfaction [Monk 1991, pp. 160-166].)

The situation with Ryle is worse.
Ryle is one of the most influential 20th-century philosophers. His early work initiated the ordinary language philosophy movement. Then, with The Concept Of Mind, he made an enormous impact on the philosophy of mind and psychology. He is rightly famous for his critique of Cartesian dualism as well as his knowledge-how/knowledge-that distinction, even if the details and nature of this work are widely misunderstood. Yet, when asked who were the greatest philosophers of the century, contemporary philosophers are not inclined to even include Ryle on the list. Ryle has become unfashionable.

There are a number of possible explanations for this state of affairs. It may be due to a widespread and misinformed prejudice against ordinary language philosophy, which T. P. Uschanov (2001, 2002) blames on the controversial work of Ernest Gellner. We might also blame the fact that Ryle and Wittgenstein were working against the mainstream, challenging some of the basic ideas of analytic philosophy, particularly the dominant Russellian and Fregean conceptions of propositions and semantics (Speaks 2010; Tanney 2009). Furthermore, for decades philosophers and cognitive scientists have been strongly prejudiced against any form of behaviorism. (David C. Palmer [2006] cogently argues that even Skinnerian behaviorism has been given a raw deal.) The influence of such figures as Chomsky and Fodor might be blamed here.

My primary concern (shared by Dennett [2002] and Tanney [2007/2009, 2009]) is that Ryle is too often misrepresented and overlooked. Thanks to Stanley & Williamson, Ryle's work is getting a little more attention, even if amidst a sea of confusion. Hopefully the tides will soon turn in his favor.


See also: