I want to clarify and expand on a point I recently raised against Ryle. A friend has suggested that I moved a little too quickly through the points, especially concerning the relevance of Wittgenstein, so I'll try to make it a bit more cogent. I'll draw some connections to Dennett, Millikan and Kripke in the process. It begins with Ryle's distinction between museum-possession and workshop-possession of factual knowledge (see Ryle, "Knowing How and Knowing That," 1946).
Let MUSEUM(p) = museum-possession of knowledge that p.
Let WORKSHOP(p) = workshop-possession of knowledge that p.
Ryle claims that MUSEUM(p) is impoverished with respect to knowing-how, where "knowing how" is a general term for intelligence associated with the application or expression of knowledge. In order to intelligently exploit one's propositional knowledge that p, one needs WORKSHOP(p). Ryle is also clear that, in order to know that p, one must establish or derive the fact that p.
If we consider the examples Ryle gives of single-track functions--a seal performing tricks, a clock telling the right time--we don't get a cogent picture of what is involved. Clocks don't know anything. Seals might, but it's not clear that they lack intelligence or that their performances are not multi-track. Ryle also mentions parrots, and even uses the example of a person who dutifully recites the rules of chess but does not know how to play. Yet, parrots do not know the facts they recite. They simply repeat sounds. So we shouldn't say a parrot has MUSEUM(p). Similarly, imagine a person who can repeat the sounds we identify as sentences which state the rules of chess, but who does not understand what those sentences mean. That person does not have MUSEUM(p), since that person has not derived or established any facts. They have only learned to repeat sounds they don't understand. If, on the other hand, they do understand what they are saying, then they must have some intelligent way of applying their understanding. Knowing the rules of chess does involve some understanding of those rules, and understanding implies intelligence. Ryle says as much at the end of his 1946 paper, "Knowing How and Knowing That," just at the spot where he makes the museum/workshop distinction. He says that if a person cannot intelligently exploit their knowledge of a fact, then they don't really know the fact. So MUSEUM(p) is not really knowledge at all. It therefore appears that Ryle's museum/workshop distinction is inconsistent.
The problem relates to a more general issue concerning intentionality and rule-following that Wittgenstein brings out in Philosophical Investigations. A rule-following dilemma occurs whenever we try to define an action as "doing the same as before." Situations are never exactly the same as before, and any rule can always have multiple interpretations. Thus, we cannot deduce any single-track function merely by looking at the rule in question. To apply a rule is to act intentionally and this means following how the rule is to be applied in particular cases, but that application cannot be contained in the rule itself. Wittgenstein concludes that not every act of following a rule requires an act of interpretation, because acts of interpretation are themselves instances of following rules. So intentionality at some point bottoms out in the mere fact that one acts in accordance with a rule. But what counts as acting in accordance with a rule if any action can be made to conform to the rule? Wittgenstein says it's just our way of life. That our understanding depends on our doing things a certain way, and that if we didn't have this shared way of doing things, we would never be able to communicate or talk about rules in the first place. Dennett and Millikan, for example, have developed this view, arguing that intentionality ultimately comes down to free-floating rationales (Dennett) or biological purposiveness (Millikan) which are the result of evolutionary mechanisms.
The perhaps startling result of this approach is that there is no unshakable reason for ever saying that an action is or is not in accordance with a rule. There is just a point at which questioning the application of rules becomes too annoying or tedious to warrant attention. It's not that rule-following is completely arbitrary or a matter of social convention. It's that we just live with certain ways of doing things, and there are no rules for us to follow apart from what emerges naturally from our way of life.
I think this approach is more or less correct. The alternative would be to say that all intentionality is an illusion or a fiction, and that there is no such thing as following a rule. If that is your attitude, then Ryle's museum/workshop distinction certainly falls apart, since it depends on there being a distinction between single- and multi-track ways of following rules. But why should we abandon the notion of intentionality, or try to claim it as a useful fiction?
It's not that there are no facts about whether or not we are following a rule in the same way we have followed it before. It's not that rule-following is just a matter of social convention. (In other words, I strongly disagree with Kripke on Wittgenstein; see here for a descending trilogy of posts on that topic.) We should resist the temptation to say that intentional behavior is just acting systematically according to some function. That would throw the net far too wide. And we should not say that intentional behavior just is the behavior of certain sorts of organisms which act with purposiveness or goal-orientation. That would beg the question, since we need to define "purposiveness" and "goal-orientation" with reference to intentionality.
Intentional action is behavior in which an organism takes up their own goals as such. To act intentionally is not just to have a way of doing something, but to act with an intentional attitude towards one's way of doing it. This requires learning how to adjust one's behavior through action. There is thus no principled distinction between an organism which acts intentionally "in the same way as before" and one which applies rules in different ways depending on the situation. There is no intentional application of a rule that does not involve learning how to apply it in a novel situation. We can distinguish between degrees or depth of learning, but that's it. There is no categorical distinction between applying a rule and applying it intelligently. So there is no such thing as a single-track function, when we are talking about intentional behavior.
To make this clearer, we can look at Ryle's examples again. A functional clock does not intend to tell the right time. We might say it exhibits derived intentionality, in that it exhibits the intentions of the clock-maker. The clock does not act, but carries out the action of its maker. A well-trained seal, however, might be said to act intentionally, but in this case, it is because the seal applies rules in novel situations. The seal, unlike the clock, is capable of learning how to follow rules. That is the mark of intelligence.
There is a separate issue about whether or not intentional action requires a representation of one's goals as such. There is a difference between having an intentional attitude towards one's goals and having a representation of one's goals. So the view I am advocating does not necessarily amount to propositionalism. I am not claiming that all intentional attitudes involve relations between persons and propositions, if we take "proposition" to indicate some kind of (re)presentation of facts. Jason Stanley would here warn against confusing representationalism with propositionalism. He holds that all intentional attitudes involve relations between persons and propositions, but that such relations need not be representational. My question for Jason is, what is a proposition, if not a linguistic or otherwise symbolic presentation of a fact?