Philosophy, Film, Politics, Etc.

Monday, February 15, 2010

Stanley and Williamson's "Knowing How", Revisited

Last September, I defended Gilbert Ryle against Stanley and Williamson's "Knowing How" (2001). After reviewing Ryle, reading a good many papers on the topic, discussing some issues with a few professional philosophers (including an in-depth and spirited email exchange with Jason Stanley about how to interpret Ryle), I've come to the following conclusions, which I divide into two parts. The first addresses S&W's misrepresentation of Ryle. In the second, I develop extensive, original arguments against S&W's formulation of knowing-how as a species of knowing-that.

1.0) Stanley and Williamson (S&W) profoundly misinterpret Ryle. I stand by my original critique of the way they present Ryle's regress argument. Similar critiques are found in Hetherington (2006) and Sheiber (2003). Others (Noe 2005; Wiggins 2009) point out other ways in which S&W get Ryle wrong, though I do not think anyone has yet to identify the full extent to which S&W misinterpret Ryle's knowing-how/knowing-that distinction. I will briefly draw attention to some points which I think have been overlooked and are crucial for understanding Ryle and S&W's misunderstanding of his philosophy.

1.1) S&W wrongly take Ryle to suppose that "knowing how to X" amounts to "being able to X." For Ryle, knowing-how is not a single-track function and does not correspond with a particular action type. Knowing-how entails an indefinitely heterogenous set of dispositions and competences to intelligently apply criteria, where intelligent performance entails creativity and active learning.

1.2) Ryle distinguishes between applying criteria (following a rule) on the one hand, and grasping a proposition on the other hand. I think S&W overlook this distinction, and that is at least partly why they misunderstand Ryle. Interestingly, I've just found Millikan (1990), in which she argues for the same distinction (without observing any connection to Ryle) on the grounds that the former case--in which rules are not expressed as such--is a matter of what she calls "biological purposiveness." She is critical of a straight dispositional analysis, though I think her analysis complements Ryle, whose view of intelligence is not a matter of simple dispositions, but of complex dispositions involving creative development--and I do not think Ryle would object to defining creative development in biological terms.

1.3) Pace S&W, Ryle regards knowing-that in terms of abilities; however, Ryle does not thereby suppose that knowing-that is a variety of knowing-how--for, again, knowing-how is not simply abilities. Pace S&W, Ryle's knowing-how/knowing-that distinction is not a distinction between abilities and something else. Rather, it is a distinction between intelligence on the one hand, and abilities pertaining to the jobs of didactic discourse on the other. Neither is a species of the other, and the two are not mutually exclusive. According to Ryle, all but the most primitive varieties of knowing-how entail knowing-that.

1.4) S&W confuse the nature of intellectualism. They think intellectualism is the thesis that knowing-how is a species of knowing-that, when in fact intellectualism is the thesis that intelligent behavior is the result of intellectual acts of grasping true propositions. We should rather say that, in arguing that knowing-how is a species of knowing-that, S&W are propositionalists, not intellectualists. Jason Stanley (in correspondence) confirms that, when framed in these terms, he is a propositionalist, and not an intellectualist. (I haven't run any of this past Timothy Williamson.)

1.5) Jason Stanley (in correspondence) claims that Ryle wrongly supposes that propositionalism entails intellectualism. The veracity of this claim depends on how we understand knowing-how and knowing-that, and therefore propositionalism. In Ryle's view, demonstrations of propositional knowledge (knowledge-that) entail the explicit acknowledgment or statement of a proposition, where propositions are linguistic entities. On this view of knowing-that, it appears that propositionalism does entail intellectualism, because it would mean that any intelligent action (any demonstration of knowing-how) would entail stating or acknowledging a proposition, and this in turn would mean that the proposition would have to be intelligently interpreted in order for the intelligent action to occur. Perhaps Ryle is wrong to associate propositional knowledge with the explicit statement or acknowledgment of linguistic entities. (Stanley thinks he is, and Stanley is representative of the Russellian tradition here; for a discussion of Ryle's challenge to the traditional notion of propositions, see Julia Tanney's introduction to The Concept of Mind: 60th Anniversary Edition [2009].) In that case, some other view of propositions and propositional knowledge might allow for a propositionalism which did not entail intellectualism. Such a propositionalism would not be a threat to Ryle's anti-intellectualism, however; it would only suggest that Ryle's formulation of the knowledge-how/knowledge-that distinction was misleading. In any case, as I will now show, I do not think S&W make a compelling case for any variety of propositionalism.

2.0) Apart from getting Ryle wrong, S&W fail to make a compelling case for knowing-how as a species of knowing-that. Many criticisms of S&W's positive account of knowing-how are in print, and some are compelling, though I do not think the full extent of S&W's faulty reasoning has been exposed. I will briefly describe some original criticisms I have developed.

2.1) S&W's and similar arguments for propositionalism utilize the claim that knowing-how is a variety of knowing-wh (knowing-what/where/who/when/why). However, the reduction of knowing-how to knowing-wh is not persuasive. Given the linguistic similarities between know-how and know-wh expressions, we could just as easily claim that the latter are variations on the former as we could claim that the former is a species of the latter. In fact, at least some uses of know-wh expressions entail knowing-how (e.g., "knowing where to find good pizza" entails knowing how to find the right place; "knowing what to do" entails knowing how to act appropriately), so I am highly skeptical of the suggestion that knowing-how is just a kind of knowing-wh. I rather think that many cases of know-wh entail non-propositional know-how.

2.2) According to S&W, if a knowledge ascription can be interpreted in terms of an embedded question, such that the knowledge ascribed is knowledge of the answer to that question, then said knowledge must be propositional. Yet, it is not the case that every answer to a question is a proposition. The question "how could you swim?" can be answered with a performance, and not with a proposition. Thus, when we ascribe knowing-how, we might be ascribing non-propositional performative competences, as Ryle and others have argued. S&W may be correct that "knowing how" expressions (of the relevant sort) do contain embedded questions of the form, "How could you X?"; however, it does not follow that the ascribed knowledge is propositional in nature.

2.3.0) S&W overlook important linguistic differences between knowing-how and knowing-that expressions. As Ryle notes, it is common to ask for reasons for one's propositional knowledge, but not for knowing-how. This suggests that knowing-how is not analyzable in terms of justified true beliefs. To make this clearer, consider the difference between asking "How do you know how to X?" and "How do you know that X?"

2.3.1) In asking "how do you know that X?", we are asking for a justification for the belief in X. If we accept the justification, then we may believe that the person knows that X. Furthermore, if we believe that a person knows that X, it follows that we believe that X. If we are justified in accepting their justification, then we know that X. In contrast, asking "how do you know how to X?" is not asking for a justification for a belief. It rather asks for an account of a skill or competence. Furthermore, we can accept the answer, and so accept that somebody knows how to X, without believing that any particular w is a way to X. We can be justified in accepting the account without knowing of any way to X that it is a way for anybody to X. For example, I can be justified in believing that Irina knows how to perform a complex ski jump called the quadruple salchow, even though I have no idea what that is and have never seen one performed. Being justified in believing that a person knows how to ski does not entail knowing how to ski. However, being justified in believing that a person knows that Obama is POTUS entails knowing that Obama is POTUS. If knowing-how were just a case of knowing-that, this asymmetry should not appear. It is not accounted for in S&W's linguistic analysis, and it suggests that knowing-how is not propositional.

2.3.2) A related asymmetry also occurs. In some cases, an answer to "How do you know how to X?" can give a person knowledge how to X. For example, if asked "How do you know how to find good pizza?", one could answer by describing in detail how one found a good pizza place. The information required to find the pizza place may be communicated in the answer; and, presuming that the interlocutor has enough know-how to put that information into practice, the interlocutor may thus learn how to find the good pizza in question. This shows that, in this and similar cases, know-how involves propositional knowledge. (This is not a mark against Ryle, however, as Ryle argues that all but the most primitive cases of knowing-how entail propositional competence.) For other cases of knowing-how, no answer to the question "How do you know how to X?" can give the interlocutor knowledge how to X. For example, no correct answer to the question "How do you know how to sex chicks" (to use one of Sheiber's examples of non-propositional know-how) can produce knowledge how to sex chicks. Yet, any correct answer to the questions "How do you know that X?" can produce knowledge that X.

2.4.0) The inadequacy of S&W's linguistic analysis is further evidenced by their appeal to practical modes of presentation (aka "practical guises"). I submit that no such mode of presentation exists; or, rather, that all modes of presentation are practical, in so far as modes of presentation entail dispositions to behave, and that S&W have not presented a compelling argument for any particular kind of mode of presentation. S&W claim that the argument for practical guises is the same as that for first-person guises. This is not true. The argument for first-person guises is significantly different from their argument for practical guises. In the argument for first-person guises, we are introduced to John, who sees his pants burning in a mirror but thinks he is looking out a window. He therefore does not believe that his pants are on fire, though he believes that that man's pants are on fire. Yet, "his pants are on fire" and "that man's pants are on fire" are said to express the same proposition, for John is that man. Thus, the complement clauses in the following two belief attributions are said to express the same proposition:

a) "John believes his pants are on fire."
b) "John believes that man's pants are on fire."

While (a) is false, (b) is true. To avoid contradiction, it is supposed that John entertains the same proposition under different modes of presentation. The different modes of presentation are practical in nature, in that they determine how John acts intelligently. It is particularly significant that the notion of modes of presentation explains why the same clauses can be used to make the same statement about John, even though John would not use the same sentences to make the same statement about himself. That is, while somebody in a privileged position would knowingly use "that man's pants are on fire" and "his pants are on fire" to refer to John's pants, John would not. This is precisely why we reject (a), and not (b). To create a similar situation, S&W introduce Hannah, who cannot ride a bicycle, but who sees a cyclist on television and is told that that is a way for her to ride a bicycle. She is still sitting on the couch, so it is granted that she does not know how to ride a bicycle in the performative sense. Yet it is supposed that Hannah does know of a way to ride a bicycle such that it is a way for her to ride a bicycle. According to S&W's linguistic analysis, the complement clauses in (c) and (d) express the same proposition:

c) "Hannah knows that that is a way for her to ride a bicycle."
d) "Hannah knows how to ride a bicycle."

As in the case of John's burning pants, we are willing to accept (c) but not (d). S&W claim this means that the proposition in question is entertained under different modes of presentation. Thus we are to suppose that Hannah's lack of performative competence is just a matter of her not entertaining the same proposition under the right mode of presentation. It is hard to compare beliefs about burning pants with capacities to ride a bicycle, and it does not seem that gaining competence at cycling is a matter of learning how to entertain a proposition under a new mode of presentation--at least, not if entertaining propositions constitutes belief. It is hard to see why we should think of beliefs and competences in the same way, when they seem so dissimilar. We might be willing to give S&W the benefit of the doubt here, or at least to suspend judgment, if there were no evidence of any significant linguistic differences between the knowing-how and knowing-that expressions, and if it were true that the case of Hannah the cyclist was not significantly different from the case of John's burning pants. However, as I argue above, there is a significant linguistic difference between knowing-how and knowing-that. Furthermore, as I will now show, the cases of Hannah and John are significantly different. Unlike the complement clauses in (a) and (b), the complement clauses in (c) and (d) cannot be used to make the same statement. Recall that, in the case of John, a person in a privileged position could knowingly use "his pants are on fire" and "that man's pants are on fire" to refer to John, while John could not. There is no analogous situation in the case of Hannah. Hannah's relevant verbal behavior is the same as anybody else's. Nobody is in a position to make justified references which Hannah, herself, cannot make. So the analogy to the case of John's burning pants fails. The argument for first-person modes of presentation does not support practical modes of presentation as presented.

2.4.2) Perhaps an adjusted argument for practical guises could work, though there is good reason to suppose this is not the case. Practical modes of presentation are supposed to be practical in some way over and above the way in which other modes of presentation are practical. S&W have a particular sort of practicality in mind; specifically, it is a practicality that does not entail the explicit representation of a proposition. S&W want us to think that non-representational purposiveness entails entertaining a proposition under a special mode of presentation; however, entertaining a proposition under a mode of presentation entails representing that proposition. The sort of mode of presentation they require is a logical impossibility.

2.4.3) By arguing for a "practical mode of presentation," S&W seem to be acknowledging the sense of Ryle's distinction. As Noe (2005) claims, S&W want to give us Ryle's knowing-how/knowing-that distinction in a new garb. Indeed, the idea that knowing-how does not entail the explicit representation of a proposition is precisely Ryle's point--it is the essence of anti-intellectualism. This is also what Millikan means when she says (following Wittgenstein) that purposiveness (at its most basic, biological level) is not representational. Yet, S&W fail to observe the linguistic differences between knowing-how and knowing-that and insist that a proposition is entertained in the relevant cases of knowing-how. This makes their position untenable.

3.0) S&W note that, depending on the communicative purposes, knowledge ascriptions can either be interpreted as mention-some or mention-all, depending on whether the ascribed knowledge is of every correct propositional answer or only of a subset of all correct propositional answers. If communicative purposes suggest that some know-how expressions ascribe non-propositional competences, as the preceding analysis suggests, we should allow for a mention-none reading, such that the ascribed knowledge is not of any propositional answer, but rather of a non-propositional capacity to perform intelligently. This preserves those aspects of S&W's linguistic analysis which are unproblematic.

See also:


Hetherington, Stephen (2006). “How to Know (That Knowledge-that is Knowledge-how)” in Stephen Hetherington (ed.) (2006). Epistemology Futures. Oxford University Press., pp. 71-94.

Millikan, Ruth G. (1990). "Truth, Rules, Hoverflies, and the Kripke-Wittgenstein Paradox." Philosophical Review 99 (3):323-53.

Noë, Alva (2005). "Against Intellectualism." Analysis 65 (288):278–290.

Ryle, Gilbert (1949). The Concept of Mind. Chicago: University Press.

Shieber, Joseph (2003). “What our Rylean Ancestors Knew: More on Knowing How and Knowing That” in Pre-Proceedings of the 26th International Wittgenstein Symposium. Austrian Ludwig Wittgenstein Society, Kirchberg am Wechsel, pp. 328-330.

Stanley, Jason and Williamson, Timothy (2001). “Knowing How.” The Journal of Philosophy, 98.

Wiggins, David
(2009). “Knowing How To and Knowing That.” In Wittgenstein and Analytic Philosophy: Essays for P.M.S. Hacker, Hans-Johann Glock & John Hyman, eds. (Oxford University Press).