Sunday, February 12, 2012

Ryle and Behaviorism

Gilbert Ryle begins the last section of The Concept of Mind (1949) with a prophecy:  "The general trend of this book will undoubtedly, and harmlessly be stigmatised as 'behaviourist'" (Ryle, 327).  He was certainly right about the stigma, but he may have grossly underestimated the threat it would pose.  Behaviorism is the pivot on which Ryle's legacy hinges.

Jason Stanley, a prominent philosopher of language, and Julia Tanney, a leading Ryle scholar, disagree on how to characterize Ryle's relationship to behaviorism.  Tanney argues that Ryle wasn't a behaviorist of any sort, not even a 'soft' behaviorist.  She says the "soft behaviorist" view is the standard (but mistaken) interpretation of Ryle, according to which "statements containing mental terms can be translated, without loss of meaning, into subjunctive conditionals about what the individual will do in various circumstances."  In chapter one of his recent book, Know How, Stanley argues for this interpretation of Ryle:  "Ryle certainly thought that mental capacities were not identical to dispositions characterized in terms of a single natural kind of behavior, like squinting.  But it is consistent with Ryle's persistent admonishments that he thought of each mental capacity as identical with a very lengthy and complex disjunction of purely physical dispositions" (Stanley, 10).  To ascribe a mental capacity to a person is just to make a series of disjunctive conditional statements about their behavior, however vaguely.

Oddly, Stanley claims that the length and complexity of such disjunctions might make it impossible for them to be described in any finite period of time. However right or wrong Stanley is for allowing infinities into his equation, it is quite a stretch to assume that Ryle made the same allowance.  More than that, Stanley's assumption goes against the letter and the spirit of Ryle's analysis.

Before continuing, we should pause to remember that Stanley's point is only about a subset of mental-conduct concepts.  While Ryle's discussion of mental capacities does lend itself to a more straightforward dispositional analysis, Ryle also discusses "semi-hypothetical" or "mongrel-categorical" statements, which he says are neither wholly dispositional nor wholly occasional.  (More on mongrel-categoricals below.)  So, even if we granted that attributions of mental capacities were analyzable as Stanley says, Ryle is very explicit that he does not take all mental-conduct concepts in this way. So Tanney is right when she says:  "Although it is true that Ryle was keen to point out the dispositional nature of many mental concepts, it would be wrong to construe him as offering a programme of analysis of mental predicates into a series of subjunctive conditionals."

Even if we restrict the discussion to mental capacities, it does not seem that Ryle is willing to reduce them to a definite series of subjunctive conditionals.  Ryle says intelligence is "indefinitely heterogeneous," and so cannot be captured in either a finite or an infinite series of propositions.  In what follows I will try to explain why Ryle takes up this point of view.

First, it's worth getting a bit clearer on Ryle's notion of mongrel-categorical statements.  In the following passage from The Concept of Mind (p. 229), Ryle employs the example of following a tune:

That a person is following a tune is, if you like, a fact both about his ears and about his mind; but it is not a conjunction of one fact about his ears and another fact about his mind, or a conjoint report of one incident in his sensitive life and another incident in his intellectual life. It is what I have called a 'semi-hypothetical', or 'mongrel-categorical', statement."
By "ears," I take Ryle to mean the auditory apparatus.  When we refer to a person as following a tune, we are saying something about how their body is functioning (ears, brain, etc.), but we are also saying something more, but it is not a reference to some other happening, function or entity.  But then what is this "something more?"  Apparently it is close to a subjunctive conditional, but it is not identical to one.  For, if it were identical to one, then Ryle's mongrel-categorical statements would simply be conjunctions of occasional and dispositional statements.  But Ryle's point is that they are not conjunctions, but something else.  That "something else" might look like a great mystery, as if Ryle were postulating some bizarre ontological or epistemological category.  But if you're tempted at looking at it that way, then you're misunderstanding Ryle's method.

Ryle is talking about the difference between behavior and action.  The ears function--they behave.  But following a tune is an action, and not a simple behavior.  It is intelligent behavior.  Ryle (p. 44) discusses intelligence in terms of "dispositions the exercises of which are indefinitely heterogeneous. . . .  Epistemologists, among others, often fall into the trap or expecting dispositions to have uniform exercises. For instance, when they recognise that the verbs 'know' and 'believe' are ordinarily used dispositionally, they assume that there must therefore exist one-pattern intellectual processes in which these cognitive dispositions are actualised."  For Ryle, capacities for intelligent behavior do not correspond to any single pattern, nor to a definite set of behaviors.  Stanley says that we might just lack the time or capacity to identify them.  But Ryle suggests a much stronger view:  They simply don't correspond to a particular cognitive pattern or definite set of behaviors.

Ryle's point is not arbitrary.  It is central to the view of intelligence which he is developing.  According to Ryle, action always involves improvisation.  He makes the point several times, such as here, when he discusses the capacity to reason:
[The reasoner] has to meet new objections, interpret new evidence and make connections between elements in the situation which had not previously been co-ordinated. In short he has to innovate, and where he innovates he is not operating from habit. He is not repeating hackneyed moves. That he is now thinking what he is doing is shown not only by this fact that he is operating without precedents, but also by the fact that he is ready to recast his expression of obscurely put points, on guard against ambiguities or else on the look out for chances to exploit them, taking care not to rely on easily refutable inferences, alert in meeting objections and resolute in steering the general course of his reasoning in the direction of his final goal.
The need for innovation, for performing tasks which have never been performed before, implies an inability to correlate the competence with a definite set of dispositions.  To improvise is not to do one of a set number of possible things, but to do something which one has not planned or intended to do and yet which still meets some criteria for correctness.  But the criteria need not (and indeed cannot) account for every possible correct application.  So it is a mistake to suppose that the capacities we attribute to people somehow include every fact which would (or could) make them correct.  This is why they cannot be analyzed in terms of a series of subjunctive conditionals.

Stanley is philosophically committed to a certain view of analysis:  Every meaningful statement must be analyzable as a statement of fact, and every fact is a true proposition.  So, if there is such a thing as a mongrel-categorical statement, Stanley insists that it must be propositional:  It must ascribe a property to an object.  Yet, as Stanley indicates, Ryle was quite critical of this philosophical program.  It's not that Ryle was "unreflectively and immediately hostile to analysis and reduction of any kind," as Stanley says on page 10.  On the contrary, The Concept of Mind can be seen as a certain sort of analysis of mental-conduct expressions.  It's just not the kind of analysis Stanley favors.  Ryle's is a pragmatic form of analysis, in which the meaning of our expressions is analyzed in terms of how they are used.

As I've said, Ryle analyzes mental-conduct concepts in terms of dispositional and mongrel-categorical statements.  To understand the nature of this analysis, we have to understand his notion of inference tickets.  Ryle (1949, 124) explains it thus:
Dispositional statements about particular things and persons are also like law statements in the fact that we use them in a partly similar way. They apply to, or they are satisfied by, the actions, reactions and states of the object; they are inference-tickets, which license us to predict, retrodict, explain and modify these actions, reactions and states.
Dispositional statements, in Ryle's view, are not factual statements in which properties are attributed to objects.  They are meaningful in the way they give us license to move from one factual assertion to another, but they cannot be analyzed in terms of properties and objects.  As Ryle continues (p. 125):
But to speak as if the discovery of a law were the finding of a third, unobservable existence is simply to fall back into the old habit of construing open hypothetical statements as singular categorical statements. It is like saying that a rule of grammar is a sort of extra but unspoken noun or verb, or that a rule of chess is a sort of extra but invisible chessman. It is to fall back into the old habit of assuming that all sorts of sentences do the same sort of job, the job, namely, of ascribing a predicate to a mentioned object.
It's clear from this passage that Ryle is not against analysis simpliciter, but only against that sort of analysis which treats all sorts of sentences as expressions of propositions.  When Ryle says that some mental-conduct concepts are mongrel-categorical statements, he is not saying they are combinations of propositions of various types, but that they are partly propositional and partly law-like.  They function as propositions, but also as inference tickets, and they cannot be analyzed solely in terms of one or the other, nor can they be analyzed as a sequential conjunction of factual and law-like statements.  They function as both simultaneously.

As we have seen, Ryle takes dispositional sentences as inference tickets.  Thus, when he analyzes mental capacities in terms of dispositions, he is talking about the sorts of inferences we are allowed to make about a person to whom we ascribe mental capacities.  The crucial point is that mental capacity concepts do not ascribe particular states or functions to people, but rather confer the right to make inferences about people's behavior.  This does make mental capacities a matter of behavior, and so might qualify Ryle as a sort of behaviorist. According to Stanley (p. 10), "the view that mental capacities are nothing over and above purely physical dispositions is certainly worthy of the title of 'Behaviorism.'"  I don't take issue with that.  However, this is not the sort of "contemporary behaviorism" of which Stanley accuses Ryle.  This is not the sort of behaviorism anybody needs to be afraid of.


It's worth noting that the phrases "mongrel",  "mongrel-categorical," "inference ticket" and "semi-hypothetical" appear nowhere in Stanley's book, according to the search functions at amazon and Google books.


Stanley says Ryle disavowed behaviorism by the late '70s.  Yet, we can find Ryle criticizing behaviorism in the last section of The Concept of Mind.  He claims that early behaviorists were of two minds:  some denied that there was anything properly called "the data of consciousness and introspection", while others acknowledged the existence of mental contents, but denied that such data was scientifically knowable.  Ryle rejects both views.  He says they both rely on a "two-worlds story" about mind and body, where mental happenings are supposed to exist along side (or behind) bodily happenings.  One sort of behaviorist rejects the existence of the mind, while the other rejects its scientific knowability.  Ryle, in contrast, has spent the preceding pages of his book arguing that the mind does exist, just not in the same way that bodies exist.  Ryle does not stipulate a supernatural or non-physical thingness.  He is not a substance dualist, but he is not necessarily a property dualist, either.  Single-track dispositions, like fragility, can be considered a property of glass, but the sorts of dispositions that characterize intelligence are not neatly analyzable in terms of properties.  When we attribute a mind to a person, we are not attributing a property to a body or organism. We are speaking in a rather different sort of way.  Persons are not, in Ryle's vocabulary, identical to or definable in terms of their bodies, just as their actions are not identical to or definable in terms of their behaviors, even though there is nothing over and above bodies and behaviors.

Ryle is maybe best thought of as an eliminative materialist, since he denies that there are any literal states that correspond to our talk of mental states.  But Ryle does not eliminate (or reject) the way we ordinarily talk of the mind.  He would rather reframe the way we understand that talk.  This may explain why his work is so often misunderstood and mischaracterized.