2.1 Introduction
One lesson from Chapter 1 was that the regresses Ryle famously tried to pin on his intellectualist opponent had bite only if intellectualism is characterized so that it’s committed to an ‘absurd datum of phenomenology’. The absurd datum of phenomenology, recall, was that an agent knows how to do something only if she actively contemplates some proposition. This thesis, we saw, should be rejected outright. What matters for all parties to the debate is that there is a more reasonable way to state the intellectualist position:
Intellectualism: S knows how to φ in virtue of S’s having some propositional attitudes vis-à-vis φ-ing.1
Intellectualism, thus articulated, is not ‘regress-bait’ – and this is good news for the prospects of intellectualism. But a thesis needs more to recommend it than the credential that it ‘avoids regresses’. So, then, why be an intellectualist?
This will be the guiding question of this chapter, and in what follows we’ll be exploring the question in an organized way. But before doing so, some quick preliminary points are in order. Firstly, since one notable strategy of argument for intellectualism takes the shape of demonstrating anti-intellectualism to be an unacceptable alternative, it’s important to get clear what the rival thesis is actually saying. Pared down to the very core: the key idea separating the positions is whether certain propositional attitudes are sufficient for knowledge-how. It is the denial of the intellectualist’s claim that certain propositional attitudes are sufficient for knowledge-how that makes one strictly an anti-intellectualist.2 Though, in merely denying this claim, one hasn’t yet offered any alternative. Thus – and this is important – anti-intellectualists have almost categorically maintained (beyond just the denial of the intellectualist’s thesis about propositional attitudes) a further positive thesis that knowledge-how is grounded in the possession of abilities or dispositions. Consider Bengson and Moffett’s remark that ‘one of Ryle’s most important contributions was to uncover a general, theoretically significant fault line in the theory of knowledge, mind and action, to which [the terms “intellectualism” and “anti-intellectualism”] rightly apply’.3 The core contention of the intellectualist side of this line is that states of intelligence and exercises thereof are grounded in propositional attitudes. The core contention of the anti-intellectualist side, by contrast, is that ‘states of Intelligence and exercises thereof are grounded in powers, abilities, or dispositions to behave, not in propositional attitudes’.4 Thus anti-intellectualism, understood as a positive thesis about what it is in virtue of which a given agent S knows how to φ, says:
Anti-Intellectualism: S knows how to φ in virtue of S’s having some ability to φ.
Thus, intellectualism and anti-intellectualism each imply the following claims about necessary and sufficient conditions for knowledge-how.
Intellectualism–sufficiency: If S has some propositional attitudes vis-à-vis φ-ing, then S knows how to φ.
Intellectualism–necessity: If S knows how to φ, then S has some propositional attitudes vis-à-vis φ-ing.
Anti-Intellectualism–sufficiency: If S has the ability to φ, then S knows how to φ.
Anti-Intellectualism–necessity: If S knows how to φ, then S has the ability to φ.
Keeping these differences in mind, there are admittedly several ways one might attempt to organize the philosophical case for intellectualism. One natural thought would be to do so along the most well-worn division line for evaluating the intellectualist thesis in the literature – arguments from cognitive science on the one hand, and linguistic arguments on the other.5 However, dividing the case for intellectualism into these two categories of argument would be misleading. For one thing, cognitive science-based evaluations of the intellectualist thesis, drawing from a claimed difference between declarative and procedural knowledge (i.e. knowledge-that can be easily articulated and knowledge exercised in tasks which cannot be easily articulated), are almost entirely anti-intellectualist in spirit.6 As Glick remarks:
The case from cognitive science, in essence, is that empirical data shows that know how can be possessed and exercised even when there is reason to doubt that propositional knowledge is present.7
While arguments from cognitive science play an important role in evaluations of the intellectualist position, such considerations do not play a substantial role in accounting for its motivation. Moreover, the linguistic/cognitive science divide fails to capture a range of important arguments for intellectualism that take the dialectical shape of arguments against anti-intellectualism. We thus propose to divide the case for intellectualism into the categories of linguistic strategies and broadly non-linguistic strategies.
2.2 Linguistic arguments for intellectualism
Intellectualism, as we just saw, is the thesis that one knows how to φ in virtue of having some propositional attitudes about φ-ing. Plausibly, propositional attitudes less than knowledge (e.g. mere unjustified belief) will not suffice, and so the intellectualist view is that knowledge-how is a species of propositional knowledge. Propositional knowledge is knowledge of some fact, and it is also indicated as ‘knowledge-that’ where the that-clause is some sentence expressing a true proposition. For instance, ‘John knows that 33 = 27’ indicates that John has propositional knowledge of the fact that 3 cubed is 27. So, knowledge-that is knowledge of a true proposition. An intellectualist about know-how claims that knowledge-how is knowledge of a true proposition. On an intellectualist account, one knows how to perform some action if and only if one knows that this is a way to perform the action. In the following we examine arguments for intellectualism that stem from semantic accounts of embedded questions.8
2.2.1 The development of formal semantics
As we saw in the first chapter, the modern debate over the nature of knowledge-how goes back to Gilbert Ryle. A more recent development to this debate appeals to the relatively recent field of formal semantics. This field develops as an extension of the logical study of language. Gottlob Frege and Bertrand Russell introduced and developed a formal language to analyse the logical significance of language. For example, Russell’s treatment of the nature of definite descriptions used logic to analyse puzzles relating to ordinary sentences such as ‘The present king of France is bald.’ This sentence has a meaning, but, because there is no present king of France, it fails to refer to any existing person. The grammatical subject of the sentence is empty; it is as if words are being used without any cognitive significance. This conflicts, though, with the fact that we do understand the sentence. The puzzle was to understand the meaning of a sentence that appeared to refer to an individual that doesn’t exist. Reference is a real relation that requires existing relata and so the apparent meaning of such sentences poses a philosophical puzzle.
Bertrand Russell’s theory of definite description uses the tools of modern logic to dissolve this puzzle. On his view, the sentence ‘the present king of France is bald’ has a more elaborate logical structure. Its logical structure is revealed by the following:
(∃x)(x is the king of France & (∀y)(if y is the king of France, then x = y) & x is bald).
This is read as follows. There exists an x such that x is the king of France, and for all y if y is the king of France, then x is identical to y and x is bald. Thus, on Russell’s analysis, this sentence implies that there is a king of France, there is only one king of France, and he is bald. Thus, given Russell’s analysis the sentence does not require reference to a non-existent individual. The logical structure of the sentence does not require any singular terms. Rather the sentence’s structure is revealed by quantification over propositional functions (e.g. ‘x is the king of France’). On this analysis, the sentence is meaningful but false since it implies that there is a king of France.
The general approach of using the resources of formal logic to analyse the structure of ordinary sentences has a wide appeal. In the 1950s Richard Montague, a student of Alfred Tarski at Berkeley, extended the model-theoretical approach to an analysis of natural language. It was widely held by logicians of the day that natural languages were too rich to be usefully analysed by model-theoretical semantics. Montague showed otherwise, developing a formal semantics known as ‘Montague grammar’.9
2.2.2 Stanley’s master argument
Jason Stanley’s linguistic argument for intellectualism grows directly out of this tradition.10 He uses formal semantics to develop a sophisticated analysis of sentences involving ‘know-how’. Stanley argues that an account of knowledge-how must fit in a general account of the semantics of embedded questions. An embedded question is a clause that begins with a question word: ‘who’, ‘what’, ‘when’, ‘why’ and ‘how’. The sentence ‘John knows how to ride a bike’ contains an embedded question ‘how to ride a bike’. Similarly, the sentence like ‘John knows where to ride a bike’ contains the embedded question ‘where to ride a bike’. Thus, Stanley reasons, an account of knowledge-how needs to grow out of a more general account of knowledge-wh, which itself will grow out of a general theory of the semantics of embedded questions. He aims to show thus that a good semantic theory for knowledge-wh and embedded questions supports the view that one knows how to φ in virtue of having some propositional attitudes about φ-ing. Let us turn now to a more detailed presentation of his arguments.
Stanley’s main thesis is that ‘knowing how to do something is the same as knowing a fact’.11 A fact is a true proposition. It is something that can be believed and asserted. On his view a subject, S, knows how to φ if and only if S knows that, for some way w, w is a way to φ. We can formulate Stanley’s main argument thus:
Stanley’s master argument
1 Sentences that ascribe knowledge-how contain embedded questions.
2 If a sentence ascribing knowledge-how to a subject S is true, then S knows an answer to the embedded question contained in the ascription of knowledge-how.
3 If S knows an answer to the embedded question contained in the ascription of knowledge-how, then S knows that p, where p is a correct answer to the embedded question.
4 Thus,
5 If a sentence ascribing knowledge-how to a subject S is true, then S knows that p, where p is a correct answer to the embedded question.
Defence of 1
The first premise situates a treatment of the semantics of knowledge-how within a general semantics for embedded questions. Stanley provides the following list of sentences with knowledge verbs that all contain embedded questions.
1 (a) John knows whether Mary came to the party.
(b) John knows why Obama won.
(c) Hannah knows what Obama will do in office.
(d) Hannah knows who Obama is.
(e) Hannah knows what she is pointing to.
(f) Hannah knows how Obama will govern.
(g) Hannah knows why to vote for Obama.
(h) Hannah knows how to vote.12
Each of these constructions contain an embedded question. In (a) the embedded question is ‘whether Mary came to the party’; in (b) it is ‘why Obama won’; in (c) it is ‘what Obama will do in office’; and so on. Knowledge-how is not special in this respect. It too contains an embedded question. Notice that both (f) and (h) contained embedded questions: (f) ‘How Obama will govern’ and (1h) ‘how to vote’. The difference between (f) and (h) is that the former contains a finite clause, a clause marked for tense, and the later contains an infinitival clause, one not marked for tense. This difference, while significant, does not affect whether constructions involving ‘knowledge-how’ contain embedded questions.13
Defence of 2
Premise 2 relates (i) sentences that ascribe know-how to (ii) knowing the answer to the embedded question ‘how to φ?’ The primary defence for 2 comes from reflection on all the other cases of knowledge-wh, that is, cases (a)–(g). Consider (a): John knows whether Mary came to the party. If John knows whether Mary came to the party, then John knows the answer to the question ‘did Mary come to the party?’ Similarly, with (b), if John knows why Obama won, then John knows the answer to the question ‘why did Obama win?’ Similar considerations hold for other cases of knowledge-wh. Moreover, the occurrence of the finite clause with ‘know-how’ in (f) fits the pattern of the other cases. If Hannah knows how Obama will govern, then Hannah knows the answer to the question ‘how will Obama govern?’ Thus, reflection on (a)–(g) supports the claim that if a sentence ascribing knowledge-wh to a subject S is true, then S knows an answer to the embedded question continued in the ascription of knowledge-how.
A principle of simplicity supports the thought that this pattern continues to all cases of knowledge-how. Unless there are significant reasons to think that knowledge-how differs significantly from other kinds of know-wh, one should infer that properties of know-wh are also properties of know-how. In Chapter 5 we return to this issue when we discuss the difference between knowledge-how with a finite clause such as ‘knows how Obama will govern’ and knowledge-how with an infinitival clause such as ‘knows how to vote’. For now, though, we will rest the intellectualist’s case on the parallel between (a)–(g) and (h).
Defence of 3: Karttunen’s semantics
Premise (3) states that ‘If S knows an answer to the embedded question contained in the ascription of knowledge-how then S knows that p, where p is a correct answer to the embedded question.’ This premise connects knowing the answer to a question with propositional knowledge. The defence of this premise requires some working knowledge of the semantics of embedded questions. To grasp the motivation for premise (3) one needs to understand why the best semantic accounts of knowledge-wh imply that intellectualism is true. The complete argument for premise (3) would involve more space that we can allot. We refer the reader to Jason Stanley superb book Know How.14 In place of a complete argument for premise (3), we shall develop Lauri Karttunen’s (1977) early semantic theory of embedded questions and observe how it naturally extends to provide an intellectualist account of knowledge-how. This will involve smoothing out some rough edges but the overall picture we offer is not far from Stanley’s presentation.
Karttunen’s account of embedded questions begins with the idea that questions have semantic objects.15 A semantic object can be thought of as a meaning. Declarative sentences have semantic objects – that is, propositions. Consider the sentence ‘John knows that snow is white.’ This sentence relates John via a knowledge relation to the proposition 〈snow is white〉. Similarly, questions have semantic objects. The sentence ‘John knows whether it rained last night’ relates John via a knowledge relation to the question ‘whether it rained last night’. Since we have a semantic object for propositional knowledge, we should also have a semantic object for questions.
Karttunen’s specific account of questions develops in response to perceived inadequacies with C. L. Hamblin’s (1973) early account. Hamblin argued that the semantic content of a question is its set of possible answers. For a yes/no question about p, its content is the set {p or ¬p}. For example, the content of ‘Is it raining?’ is {it is raining or it is not raining}. For an open-ended question like ‘who came to the party?’ its content is the set of propositions of the form: x came to the party, where ‘x’ ranges over individuals.
This set can be more concisely expressed using lambda notation. This notation is widespread in linguistics because of its use as a λ-determiner. A determiner like ‘every’, ‘most’ and ‘some’ binds a variable that would otherwise be free. For example, the universal quantifier, ∀, binds the variable x in the sentence ‘(∀x) x runs’ to express the sentence that ‘everyone runs’. Using lambda notation we can express this as the set of all people who run λx, x runs. That expression is equivalent to the more familiar set theoretical notation: {x: x runs}. To express the content of the question ‘is it raining?’ we write:
Is it raining? = λp (p = the proposition that it is raining or p = the proposition that it is not raining)
This is read as be read the set of propositions, p, such that p=it is raining or p=it is not raining.
We can express the content of the question ‘who came?’ thusly: λp(∃x(p = the proposition that x came)). This reads: the set of propositions, p, such that there exists an x such that sentences of the form ‘x came’ are either true or false. So the question ‘Who came?’ has as its content the set of possible answers to that question.
We are interested in the implications of this account for knowledge-wh. The sentence ‘Tim knows whether John walks’ involves a knowledge relation Tim stands in to the question ‘whether John walks’. On Karttunen’s semantics this involves knowledge relation to set {John walks or John does not walk}. More specifically, we can say that the property expressed by ‘knows whether John walks’ is the following.
[knows whether John walks] = λx(knows(x,λp (p & (p=the proposition that John walks) or (p=the proposition that John doesn’t walk)))).16
In this case the question embedding verb ‘knows whether’ is a function from individuals to a composite function from propositions to truth values. This implies that Tom knows whether John walks if and only if if John walks, then Tom knows that and if John doesn’t walk, then Tom knows that.
Stanley observes that Karttunen’s semantic account of embedded questions can be generalized to apply to ‘know why’ and ‘know-how’. Karttunen’s account of ‘know who’ requires quantification over persons. Stanley claims that an account of ‘why’ questions will involve quantification over reasons, and an account of ‘how’ questions will involve quantification over ways of doing things. 17 Stanley provides the following two examples.
[How does Bill swim] = λp ∃w (p & p = the proposition that Bill swims in way w).
[Why does Bill swim] = λp ∃r (p & p = the proposition that Bill swims for reason r).18
We can see a clear basis for intellectualism in Karttunen’s account. John knows how Bill swims if and only if John stands in the knowledge relation to λp ∃w (p & p= the proposition that Bill swims in way w).19
Why did we say ‘stand in the knowledge relation to λp ∃w (p & p = the proposition that Bill swims in way w)’ rather than knows that φ, where φ is the correct answer of the form ‘Bill swims in way w?’ The reason is that on Karttunen’s account knowledge-wh and knowledge-that belong to different semantic categories. Knowledge-wh relates a knower to a set of possible answers whereas knowledge-that relates a knower to a proposition. As Stanley observes this posed a problem with Karttunen’s account because his account didn’t validate the following inference: (a) Bill knows whether John walks; (b) John walks; so, (c) Bill knows that John walks.20 Karttunen met this problem by a meaning postulate relating knowledge-wh to knowledge-that.21 We shall ignore this complication and assume that Karttunen’s account predicts that if John knows how Bill swims, then John knows that Bill swims in way w, for some way of swimming.
The last remaining part of an argument for intellectualism is to apply Karttunen’s account to an infinitival use of ‘know-how’. Karttunen’s semantics for ‘how to φ’ predicts the following:
[how does one ride a bike] = λp ∃w (p & p = the proposition that one rides a bike in way w).
Thus, we would expect that John knows how to ride a bike just in case John stands in the knowledge relation to λp ∃w (p & p = the proposition that John rides a bike in way w). This does involve some self-knowledge on John’s part since John is the subject of the knowledge. We ignore this detail. Karttunen’s account suggests that John knows how to ride a bike in and only if John knows that w is a way to ride a bike.
There are a few more steps to get from Karttunen’s semantics to the specific account that Stanley offers, but one can clearly see the basis in formal semantics. Thus, Stanley gives us the following intellectualist account.
(INT) A subject S knows how to φ if and only if there is some contextually relevant way w such that S stands in the knowledge-that relation to the proposition that w is a way for S to φ, and S entertains this proposition under a practical mode of presentation.22
What’s new here is the idea that S entertains this proposition under a practical mode of presentation. But, like Karttunen’s account, the core idea is that the semantics for know-how require that S stand in a knowledge relation to some proposition that w is a way to φ .
2.3 The non-linguistic case for intellectualism
In this section, we’ll evaluate the broadly non-linguistic case for intellectualism in two parts, negative and positive. What we’re calling the ‘negative case’ for intellectualism emerges from alleged defects in the anti-intellectualist thesis, which we’ll outline and evaluate in Sec. 2.3.1. In Sec. 2.3.2, we evaluate briefly three positive (non-linguistic) strands of argument in favour of the intellectualist’s position: (i) Snowdon’s ‘substantive knowledge’ argument; Stanley’s argument from cognitive science; and (iii) Bengson and Moffett’s argument for non-propositional intellectualism.
2.3.1 The negative case
Bengson and Moffett maintain that one good reason to be an intellectualist is that ability possession is neither necessary nor sufficient for knowledge-how.23 Their argument strategy is by counterexamples to both the necessity and sufficiency claims. If their combined argument is right, then the thesis that one knows how to φ in virtue of possessing the ability to φ is off the table, leaving (ceteris paribus) intellectualism looking like the viable alternative. We’ll now engage with this negative case for intellectualism in some detail.
Against the anti-intellectualist’s necessity condition
Let’s look first at the necessity leg of the anti-intellectualist’s position, according to which ability possession is necessary for know-how:
Anti-intellectualism – Necessity (AI-N): S knows how to φ only if S has the ability to φ.
Notice that something like AI-N is surely what Ryle himself had in mind when he diagnosed the Lewis Carroll case of the pupil and student (see Section 1.2.3). Recall that Ryle insisted that the student failed to know how to draw the inference, despite all the propositional knowledge Ryle granted that the student possessed about the premises and conclusion. Moreover, the explanation Ryle advanced for why the pupil lacked know how was that the student lacked the ability to draw the inference. Present in Ryle’s thinking is that ability possession is a necessary condition for know-how. And indeed, to the extent that anti-intellectualists are committed to the core insight that when one knows how to φ it will be in virtue of one’s possessing the relevant φ-abilities, an endorsement of some version of AI-N looks unavoidable for the intellectualist.
That said, Bengson and Moffett offer the following counterexample to AI-N:24
ski instructor. Pat has been a ski instructor for twenty years, teaching people how to do complex ski stunts. He is in high demand as an instructor, since he is considered to be the best at what he does. Although an accomplished skier, he has never been able to do the stunts himself. Nonetheless, over the years he has taught many people how to do them well. In fact, a number of his students have won medals in international competitions and competed in the Olympic games.25
Bengson and Moffett’s diagnosis is that, in ski instructor, Pat (i) knows how to do the stunts, despite (ii) lacking the ability to do them. If this diagnosis is correct, then AI-N is false.
A natural reply to their diagnosis of the case, on behalf of the anti-intellectualist, proceeds as follows:
1 SKI INSTRUCTOR is a case where one merely knows how one φs while lacking the ability to φ.
2 A genuine counterexample to (AI-N) must be one where the agent both (a) lacks the ability to φ and (b) knows how to φ.
3 Knowing how one φs does not entail knowing how to φ.
4 Thus, ski instructor is not a counterexample to (AI-N).
Call this the one-φs-reply. Is the one-φs-reply a viable strategy on behalf of the anti-intellectualist? Bengson and Moffett are not convinced. In an attempt to counter this move, they offer a kind of counterreply via comparison. Their strategy grants that knowing how one φs does not entail knowing how to φ. However, they insist that, despite the intuition relied on in the anticipated one-φs-reply to ski instructor, Pat actually does know how to do the stunts after all, and so doesn’t merely know how one does them.
In support of this claim, they invite us to compare Pat with Albert, a scientist who doesn’t ski, but who knows how one does the stunts by knowing what muscles contract in what ways. The suggested line of thinking here is that we should draw three key conclusions from the comparison of Pat’s situation with Albert’s situation.
• Both Pat and Albert know how one does the stunts.
• Neither is able to do the stunts (by stipulation).
• Only Pat, but not Albert, knows how to do the stunts.
Granted, this is not a counterintuitive way to think about the situation. But, to the extent this diagnosis looks viable, it is presumably because – by way of comparison – Pat seems ‘mor e engaged’ with the activity in question than Albert, given that Pat is a skier himself and Albert merely reads books, despite the fact that neither can do the stunts in question. A rationale for Bengson and Moffett’s diagnosis thus relies on a kind of pairing insight – namely, one which pairs Pat (more engaged) with knowing-how to do something and Albert (less engaged) with knowing how one does something.
Table 2.1: Pairing Insight
An objection to relying on such a pairing insight, as a rationale for dismissing the one-φs-reply is that we can just as easily run an argument by comparison that appeals to an equally intuitive alternative pairing insight, to reach a very different conclusion concerning Pat, one that is compatible with the one-φs-reply. To appreciate this point, consider the case of Kamil:
kamil. Kamil is able to do all of the jumps Pat teaches with ease (and knows the relevant physiology). Given Pat’s inability to actually do the jumps Kamil can do, it seems just as natural to say that Kamil knows how to do the jumps but Pat merely knows how one does the jumps. That is, it is just as natural to say this as it is to say in Bengson and Moffett’s counter-reply that Pat (the coach) knows how to do the jumps but Albert (the scientist) merely knows how one does the jumps.
When Pat is compared with Kamil (who can actually do the stunts) rather than with Albert (who does not even ski), the following pairing insight seems intuitive:
Table 2.2: Alternative Pairing Insight
The alternative pairing insight seems every bit as intuitive as Bengson and Moffett’s pairing insight. But if that is the case, then their counterreply by comparison to the one-φs-reply is not successful. It does not provide a good reason for thinking that the one-φs-reply is off the table, as a way for the anti-intellectualist to defend AI-N against the ski instructor counterexample. Thus, we have no good reason to regard ski instructor as a decisive counterexample to (AI-N).
Against the anti-intellectualist’s sufficiency condition
Anti-intellectualists maintain that having the ability to φ is not just necessary for knowing how to φ, but also sufficient for doing so, though this point needs a quick adjustment.26 It is obviously false that someone – say, Jack – knows how to build a campfire with flint and steel were he to possess an unreliable ability to do so; say, he succeeds 1 out of an average of 50 attempts. In such a case, Jack doesn’t know how to build a fire, even though he gets it right rarely by happenstance.
Accordingly, following Bengson and Moffett, we can restrict the anti-intellectualist’s claim about the sufficiency of ability for know-how as follows, so that reliability is part of the formula:
Anti-intellectualism – sufficiency (AI-S): If S is reliably able to φ, then S knows how to φ.
Again, the core anti-intellectualist claim that one knows how to φ in virtue of possessing the relevant abilities rather than propositional attitudes looks to stand or fall with AI-S no less than with AI-N. Consider now a case – aimed at undermining AI-S – that has received considerable recent attention:27
salchow. Irina, who is a novice figure skater, decides to try a complex jump called the salchow. When one performs a salchow, one takes off from the back inside edge of one skate and lands on the back outside edge of the opposite skate after one or more rotations in the air. Irina, however, is seriously mistaken about how to perform a salchow. She believes incorrectly that the way to perform a salchow is to take off from the front outside edge of one skate, jump in the air, spin and land on the front inside edge of the other skate. However, Irina has a severe neurological abnormality that makes her act in ways that differ dramatically from how she actually thinks she is acting. So despite the fact that she is seriously mistaken about how to perform a salchow, whenever she actually attempts to do a salchow (in accordance with her misconceptions), the abnormality causes Irina to unknowingly perform the correct sequence of moves, and so she ends up successfully performing a salchow. Although what she is doing and what she thinks she is doing come apart, she fails to notice the mismatch.
Bengson and Moffett’s line on the case is ‘it is clear that Irina is reliably able to do a salchow. However, because of her confusions regarding how to execute the move, she cannot be said to know how to do a salchow.’28 It’s hard not to agree with the observation that Irina fails to know how to do a salchow, given her confusion. Also, given that ‘whenever she actually attempts to do a salchow (in accordance with her misconceptions), the abnormality causes Irina to unknowingly perform the correct sequence of moves’, it also looks like Irina’s possessing a reliable ability to perform the salchow is equally uncontroversial. But the second point requires some closer inspection.
To get a feel for why something might be amiss with Bengson and Moffett’s diagnosis of the salchow case, it will be helpful to briefly consider a famous case in the literature on reliabilist epistemology – namely, Plantinga (1993a)’s ‘brain lesion’ case, which challenges the reliabilist’s core thesis that one knows that something is so if one has a reliably formed, true belief that it is so.
In Plantinga’s brain lesion case, we are to imagine that our hero, Al, has a strange brain lesion that reliably causes him to believe that he has a brain lesion. Suppose, further, that Al has no other item of evidence that supports this conclusion. Now, one popular explanation for why Al’s reliable brain lesion-caused belief fails to qualify as knowledge proceeds as follows:29 knowledge requires that a belief’s correctness be attributable to the agent’s exercise of cognitive ability;30 and it’s hard to see how Al’s correctly believing the target proposition is attributable to any ability to which we can credit Al. As John Greco puts it, beliefs attributable to the brain lesion belief generating process a re in an important sense not attributable to any abilities belonging to Al’s own cognitive character.31 We’ll consider shortly the implications of this idea for Bengson and Moffett’s diagnosis of salchow. But first, compare the brain lesion case with an even more radical case in the reliabilist literature: Keith Lehrer’s (1990) case of ‘TrueTemp’ who has (unbeknownst to him) a temperature-detecting device implanted in his head that regularly produces accurate beliefs about the ambient temperature. A natural reaction will be to point out that despite TrueTemp’s reliability, his correctness is nonetheless not a production of his own efforts or faculties, but rather, the production of a cause external to TrueTemp’s own cognitive agency. Now, granted, it is possible that external devices might be incorporated into one’s cognitive character after a long period of calibration, perhaps as an instance of what Clark and Chalmers call extended cognition.32 However, unless the case were supplemented in such a way a great deal further, it remains a case where one’s reliably attained success is not attributable to one’s ability.
Now to the extent that we should resist, in the brain lesion and TrueTemp cases, the attribution of Al’s and TrueTemp’s correctness to Al’s and TrueTemp’s ‘abilities’, we have a precedent for denying an individual an ability to do something even when she reliably can do the thing in question successfully. Moreover, given Irina’s ‘severe neurological abnormalities’ which cause a mismatch between what Irina does and what she thinks she is doing, there is cause to think that we should not attribute the successful salchow to Irina’s ability any more than we should attribute Al’s correct brain lesion belief to an ability of Al’s (or for that matter, credit TrueTemp’s correct temperature belief to some ability TrueTemp has himself). In each case, we have instances in which what causes the relevant success is not appropriately integrated in the agent’s cognitive psychology to warrant an attribution of the success in question to the agent’s own ability.
Now, granted, even if the foregoing is right, then although salchow fails as a counterexample to AI-S, one might point out that the case remains a counterexample to a stronger version of the sufficiency leg of the anti-intellectualist’s ability claim, which articulates abilities in terms of mere doings. This stronger thesis is formulated thusly:
Anti-intellectualism (Sufficiency-Strong) [AI-SW] If S reliably φs, then S knows how to φ.
It doesn’t much matter, however, that salchow counts against this stronger version of the sufficiency leg of the anti-intellectualist thesis – given that no anti-intellectualist writing today opts for the implausibly inclusive AI-SW over the more reasonable AI-S. Unlike AI-S, AI-SW is manifestly false. Whereas crediting an individual with cognitive or action-relevant abilities entails at least minimal agency, attributions of the form ‘S reliably φs’ do not. Accordingly, S reliably φs indicates only that there is a reliable connection between S and φ-ing. For instance, Inspector Jacques Clouseau, in the Pink Panther series, regularly solves crimes. Yet Clouseau is so inept and incompetent that it’s practically a miracle each time that he uncovers the crime. As it happens with Clouseau, the world works out in his case to ‘correct’ each mistake such that he reliably solves crimes. Clearly, though, Inspector Clouseau does not know how to solve crimes. Clouseau has no relevant ability. It is, in fact, a defining characteristic of anti-intellectualism that the mark of knowledge-how is ability possession – something AI-S preserves, but AI-SW doesn’t.
Although Bengson and Moffett have relied in several papers on the salchow case,33 it is not their only case against AI-S, and it is worth considering whether their other alleged counterexample could succeed where salchow appears not to. Consider kytoon:
kytoon. Chris forms the desire to build a kytoon – a lighter-than-air kite that may, like a balloon, be filled with gas (e.g. hydrogen, hot air, or helium). She has never built a kite before, let alone a kytoon. But she is very good with her hands and thus is confident in her ability to make one. Seeking information about how to build a kytoon, information she currently lacks, Chris goes online and performs a Google search for ‘building a kytoon.’ She finds a website with instructions. The instructions are long, but she is able to understand and follow each step with a modest amount of effort. Over the course of the next few days, she succeeds in executing the steps. The result of her efforts is her own personal kytoon, which she then proceeds to learn to fly.34
Bengson and Moffett’s diagnosis of the kytoon case is interesting. They write that although the information Chris has at the time of her decision to build the kytoon is inadequate to build a kytoon, ‘there is a clear sense in which her situation is not hopeless. Her current information state, coupled with the information she will encounter once she performs a Google search, will together be sufficient to reliably build a kytoon.’35 Bengson and Moffett reason from this observation to the conclusion that, consequently, ‘Chris is, at the time of her decision, reliably able to build a kytoon’ even though, at the time of her initial decision ‘she does not know how to φ (build a kytoon).’36
As with the salchow case, kytoon is one where we should be happy to grant that the protagonist lacks know-how. However, we think there is a respect in which salchow actually does better than kytoon. In salchow, Irina lacked the ability to do a salchow despite reliably doing a salchow. This was why salchow, though not effective against AI-S, was at least effective against t he implausibly formulated AI-SW. In kytoon, by contrast, we submit that Chris not only lacks the ability to make a kytoon at t 1 (the time of her decision) but moreover, it is problematic to say she reliably can do so, at t 1. More carefully: we have no reason to think she reliably can build a kytoon at t 1 on the basis of the rationale Bengson and Moffett offer. Their diagnosis of the case relies on a background assumption to the effect that: an agent A can reliably φ at time τ provided that the following two conditions hold: A’s current information state at τ coupled with the information that at τ + 1 A will encounter, will be sufficient to reliably φ. But brief reflection shows this general principle to be lacking. Just consider the following case: (the time of her decision) but moreover, it is problematic to say she reliably can do so, at t 1. More carefully: it’s not apparent, on the basis of the rationale they offer, that Chris reliably can build a kytoon at t 1. Consider that their diagnosis of the case relies on a background assumption to the effect that: an agent A can reliably φ at time τ provided that the following two conditions hold: A’s current information state at τ coupled with the information that at τ + 1 A will encounter, will be sufficient to reliably φ. But brief reflection shows this general principle to be open to counterexample. Just consider the following case:
speech: Wesley is supposed to recite, as part of a school production, a key paragraph from Winston Churchill’s famous ‘Iron Curtain’ speech. Wesley’s present information at t 1 includes the line: ‘From Stettin in the Baltic to Trieste in the Adriatic an Iron Curtain has descended across the continent.’ Wesley, due to a poor memory, can’t remember the rest. However, at t 1, given antecedent events and conditions in conjunction with laws of nature, Wesley will (at t 2) acquire a piece of paper blowing in the wind, which contains the remainder of the speech.
Because Wesley is such that, at t 1 his present information state plus the information he at t 2 will acquire are sufficient for reliably reciting the ‘Iron Curtain’ speech37 , the rationale Bengson and Moffett rely on in the kytoon case implies that Wesley can reliably recite the Iron Curtain speech at t 1, although that is absurd. kytoon is a case where the protagonist (at the time of the decision to make the kytoon) not only lacks an ability to do so, but moreover, it’s implausible that Chris reliably can build a kytoon at t 1 given the explanation they advert to. Of course, we leave it open that perhaps a better explanation could be formulated for why Chris is reliably able to build a kytoon at t 1. Though even armed with such an explanation, kytoon would then progress only to the same position as salchow – a case that counts against only an implausibly inclusive formulation of the anti-intellectualist’s sufficiency thesis, and not one that counts against the reasonable articulation of the view in terms of ability possession.
2.3.2 The positive case
In this section, we examine three non-linguistic positive considerations that count in favour of the intellectualist thesis due to Snowdon (2004),Stanley and Krakauer (2013) and Bengson and Moffett (2011b).
Snowdon’s ‘substantive’ knowledge argument
Paul Snowdon (2004), like Bengson and Moffett, wants to reject both the necessity and sufficiency claims made by the intellectualist. In service of challenging the sufficiency claim, Snowdon raises a series of examples that differ in a very important respect from the salchow and kytoon cases. Snowdon’s cases aim to establish that ability possession isn’t sufficient for know-how by demonstrating that some propositional attitudes are necessary for know-how. Here are three cases he offers to support this suggestion:38
chess: For example, I am thinking about a chess puzzle and, as we say, it dawned on me how to achieve mate in three. Surely, the onset of this knowledge consisted in my realizing that moving the queen to D3, followed by moving the knight to … etc., will lead to mate in three.
train: S knows how to get from London to Swansea by train before midday. S’s knowing how to do that surely consists in knowing that one first catches the 7.30 a.m. train to Reading from Paddington, and then one … etc.
word: Finally, if someone knows how to insert footnotes using Word, then they know that the way to insert footnotes is to click on Insert and then on Reference, and so on.
Snowdon reveals his formula for generating these cases. He comments that ‘One way to think about such examples is to note that in the construction ‘S knows how to G’ the place filled by ‘G’ can be occupied by descriptions of actions of very different kinds and levels.’39 By ‘levels’ Snowdon means that: just as ‘G’ can pick out a basic physical activity such as scratching one’s nose, ‘G’ might also pick out a less physical, complex activity such as ‘applying to Oxford University’ or ‘applying for a bank loan’ where propositional knowledge seems front and centre.40 Snowdon’s point in short is that even if ability possession seems sufficient for commanding certain basic physical tasks, propositional knowledge seems require d to know how to do other more complex kinds of activities.
The pattern of cases Snowdon offers would not favour intellectualism over anti-intellectualism if the propositional knowledge at issue in each case were thought to be just an accidental feature of cases where one knows how to do something. Snowdon is thus relying on a stronger suggestion: that in the cases he offers, it looks like knowing how to do something entails having certain items of propositional knowledge. Problematically though, for Snowdon, φ entails ψ is not an argument against φ in virtue of χ. Tiger Woods might not be able to win a golf tournament without having arms; this does not mean it’s not the case that, whenever Tiger wins a golf tournament, it is in virtue of shooting the lowest score. And this is true even if, to be clear, shooting the lowest score is not something Tiger can ever do without having arms.
To illustrate our point, consider a case of the metaphysical grounding relation noted by Fine: ‘the particle is accelerating in virtue of increasing its velocity over time.’41 Let’s call Fine’s particle P. We can easily imagine a range of cases, C 1, C 2 and C 3, where P would not have accelerated had it not been hit with a bat. We might then claim that in at least three cases, it’s true that being hit by a bat is necessary for P’s acceleration, in that P would not have accelerated had P not been hit with a bat. Such examples however would not count against the metaphysical grounding claim that the particle is accelerating in virtue of increasing its velocity over time. Even if knowing how to do something in some cases entails propositional knowledge, this much is compatible with the claim that when one knows how to do that thing, it is in virtue of possessing certain abilities. While the observation in each case – that particular instances of know-how entail propositional knowledge – is (obviously) clearly compatible with the claim made by the intellectualist, it is nonetheless not incompatible with the claim made by the anti-intellectualist.
Intellectualist arguments from cognitive science
As we saw in the previous section, Jason Stanley is the leading contemporary exponent of intellectualism on linguistic grounds. Recently, Stanley has collaborated with cognitive neuroscientist John Krakauer to advance an intellectualist case on cognitive neuroscientific grounds. Stanley and Krakauer (2013)’s collaborative work, published in the journal Frontiers of Neuroscience, takes a focal point a classic case in cognitive neuroscience – the case of an anterograde amnesia patient ‘HM’ (Milner 1962) – which has been important in the discovery of multiple memory systems.
HM was a patient with intractable epilepsy who underwent bilateral temporal lobectomy and was subsequently found to have persistent and pervasive anterograde amnesia – he would rapidly forget events soon after they occurred. ... The psychologist Brenda Milner had HM perform a mirror drawing task in which he had to trace the outline of star with a pencil through a mirror with vision of his own arm obscured (Milner 1962). HM showed improvement over 3 days on this task even though on each day he had no explicit memory for ever having encountered the task before nor even a feeling of familiarity with it.42
Now, the HM case at least seems to show that motor learning can be retained even though one’s knowledge about the activity in question is not. Unsurprisingly, the HM case has been taken to support a distinction between knowledge and skill that would appear prima facie unreconcilable with an intellectualist construal of knowledge-how.
But despite received thinking about the case vis-à-vis knowledge and skill, Stanley and Krakauer (2013) think that suitable appreciation of the case’s details reveals that the case points in the other direction. In particular, they write:
Here is a fact about HM. Each time HM performed the task he received explicit verbal instruction, and was able to use that knowledge each time. HM of course forgot that he had used explicit knowledge. But that of course does not entail he did not require the knowledge at the time.43
They argue further that in similar studies of patients with medial temporal lobe lesions like HM’s, where the objective was to demonstrate that improvement in motor performance does not positively correlate with an ability to (explicitly) remember aspects of the task, it was clear that ‘the amnestic patients could not perform any of the tasks unless instruction was provided on each day’.44
Let’s distinguish two issues here. First, are Stanley and Krakauer correct that the case of HM fails to establish anti-intellectualism? The rationale for anti-intellectualism found in the cognitive science literature has an intermediary step. The line of thought is that HM shows that there is a distinction between declarative and procedural knowledge. And further, that because, as it is widely thought by cognitive scientists working in AI, declarative and procedural knowledge map on to propositional and ability knowledge, the HM case is taken to support a distinction between theoretical and practical knowledge that is inconsistent with intellectualism. We do not take a stance on whether this line of thought is ultimately successful. Stanley and Krakauer, and Stanley (2011, ch. 7), raise several important problems with the tacit identification of the declarative/procedural distinction with the knowledge/ability distinction. Our aim is rather to assess whether the HM case does provide positive support for intellectualism. So, second: are they correct that this case provides strong support for intellectualism? We think it does not.
The relevant datum Stanley and Krakauer (2013, 503) cite is that ‘the amnestic patients could not perform any of the tasks unless instruction was provided on each day.’ These instructions would be soon forgotten but the instructions were required for the performance of the task. Thus, the performance of the tasks depends on having the relevant propositional attitudes when the explicit instructions were received. To evaluate whether this supports intellectualism, we need to think about the relationship between grounding and dependence. In particular, Stanley and Krakauer need the claim that (i) the performance depends on possessing propositional attitudes to support the claim that (ii) the performance is in virtue of the propositional attitudes. We argue that (i) doesn’t support (ii).
Theodore Sider (2017, 1) notes that there is one sense of metaphysical grounding that is picked out by the dependence relation. For instance, for two facts F1 and F2, we might say that F1 depends on F2 in a way that is meant to be equivalent to other grounding expressions such as: F1 holds in virtue of F2, F1 is grounded in F2. Another sense of dependence is captured by the idea that, for F1 and F2, F1 depends on F2 only if were F2 not to hold, F1 would not hold. In this respect, F1’s holding depends on F2’s holding in the sense that F2 is necessary for F1.45 While the metaphysical grounding relation is thought to imply necessity, it’s not the case that opposite holds. That is, it’s not the case that F2’s being necessary for F1 entails that F1 holds in virtue of F2. Consider this example, offered by Fine to illustrate the point: it is necessary that if the ball is red and round, then it is red but the fact that the ball is red does not obtain in virtue of its being red and round.46
The relevance of this point is as follows: even if Stanley and Krakauer’s observation that the patients could not perform any of the tasks unless instruction was provided on each day could establish dependence in the sense of necessity, given that necessity doesn’t entail metaphysical grounding, it’s not the case that Stanley and Krakauer’s observation entails the further point about metaphysical grounding. But since intellectualism is a thesis about what grounds what, the argument comes up short. This is of course not to say that they could not appeal to the data to attempt to defend the stronger claim; it suffices to say that without a further positive argument to this effect, intellectualism isn’t positively supported over anti-intellectualism by their observations.
That said, we can envision less interesting intellectualist positions by simply weakening the relevant relation between know-how and knowledge. For instance, Stanley and Krakauer’s observations do support a position Weatherson (2017) has called weak causal intellectualism according to which ‘the possession of [knowledge-how] is, often, caused by the possession of a piece of knowledge’. But most anti-intellectualists will be happy to accept weak causal intellectualism.
Non-propositional intellectualism
Though Bengson and Moffett are avowed intellectualists about knowledge-how, they are not intellectualists of the standard form. We see the debate between intellectualism and anti-intellectualism as over the question of what is it in virtue of which an action is intelligent? Bengson and Moffett’s (2011b) aim to distinguish this question – which they call the question about what grounds knowledge-how – from the nature question. Namely: what is the nature of knowledge-how – namely, is it a propositional attitude relation, or something else?
Even though the debate has proceeded as if these two questions are the same, Bengson and Moffett think that one can give separate answers to these two questions. So they resist orthodox intellectualism, which claims that the grounds of knowledge-how and the nature of knowledge-how are the same: propositional knowledge.
The basic contour of their proposal is that knowledge-how is grounded in propositional attitudes, though its nature is not a propositional attitude relation (between agent and proposition) but rather an objectual attitude relation between an agent and a way of φ-ing.
Objectualism is, as they see it, a kind of further alternative about the nature of knowing how that is neither ‘propositionalist’ (a matter of standing in a propositional attitude relation) nor ‘dispositionalist’ (a matter of possessing a certain disposition or dispositions). According to objectualism, the nature of knowledge-how consists in ‘a non-propositional, non-behavioural-dispositional objectual attitude relation (e.g. a knowledge-of relation), and the relatum is a non-propositional item (e.g. a way of φ-ing)’.47 As they remark, their objectualist intellectualism ‘combines intellectualism with a non-propositionalist view of the nature of knowledge how’.48 There are further details to the position, but first it will be helpful to look at their guiding rationale for opting for a non-propositional variety of intellectualism over the orthodox variety, which is, in short, that: ‘no extant theory has seemed capable of respecting all three of the following attractive but prima facie incompatible theses about knowing how:
(i) Knowing how is not merely a kind of knowing that.
(ii) Knowing how is practical: it bears a substantive connection to action.
(iii) Knowing how is a cognitive achievement: its status as a piece of practical knowledge is not merely coincidental.”49
As Bengson and Moffett see it, orthodox intellectualism, like Stanley’s version we’ve considered, can’t accommodate (i) and (ii). Of course, reductive intellectualists deny that (i) is a genuine desiderata of an account of knowledge-how.
But that leaves (ii). The worry is that orthodox intellectualism does not capture the practical character of knowledge-how. However, and by contrast, Bengson and Moffett claim that anti-intellectualists of any stripe cannot accommodate (iii) – the idea that knowledge-how is a cognitive achievement. Intellectualism paired with objectualism, they insist, can accommodate (i), (ii) and (iii).
The details of Bengson and Moffett’s proposal are rich, though we’ve not the space to engage with them all here. Rather, we want to raise, with respect to this general argument strategy, a potential oversight: if anti-intellectualism paired with objectualism would do just as well vis-à-vis (i)–(iii), then their intellectualist’s invoking strategy of objectualism wouldn’t (at least on the basis of the rationale offered) support intellectualism over anti-intellectualism.
A simple way to explore this suggestion will be to consider some available responses to the question: in virtue of what does one know how to φ? Here we can envision:
Answer 1: S knows how to φ in virtue of S’s having some propositional attitudes vis-à-vis φ (Intellectualism)
Answer 2: S knows how to φ in virtue of S’s having some ability to φ, rather than some propositional attitudes (Anti-intellectualism)
Answer 3: S knows how to φ in virtue of S’s having some objectual attitude vis-à-vis a way of φ-ing (?)
To be clear, Bengson and Moffett, in taking objectualism as an alternative to propositionalism and dispositionalism, do not consider how it might feature as an alternative to the standard dichotomy of intellectualism and anti-intellectualism. It is appropriate to ask, then, what kind of position is Answer 3? Further, can the kind of position represented by Answer 3 accommodate each of (i-iii) of Bengson and Moffett’s desiderata for a theory of knowledge-how?
In response to the first issue, it’s plausible to think that Answer 3 could be given either a propositional attitude interpretation, on which Answer 3 would be a version of intellectualism, or – importantly – an ability interpretation, on which Answer 3 is a version of anti-intellectualism.
That Answer 3 should be given an ability-based interpretation, for one thing, gains support from recent work on objectual understanding in mainstream epistemology. A standard line, advanced by Kvanvig (2003), Grimm (2014), Riggs (2009b), Gordon (2014), Carter and Gordon (2014) and Hills (2009) is that for an agent and a body of information φ, for S’s understanding φ is essentially constituted by S’s grasping of relevant coherence- or explanatory-making relations between the relevant propositions constituting φ. Notice that on this model of objectual understanding in mainstream epistemology, while what is grasped when one bears the objectual understanding attitude is relations between propositions, one actually counts as understanding the subject matter in virtue of possessing a certain ability.
The literature on objectual understanding in epistemology thus motivates the suggestion that Answer 3 could very plausibly be characterized along anti-intellectualist lines, and in a way that would (via the invoking of objectualism) handle (i)–(iii) as well as the proposal Bengson and Moffett offer, provided at least that the kind of anti-intellectualist objectualism shown to be a live option here could accommodate (iii) – the idea that knowing how to do something constitutes a kind of cognitive achievement, or success through ability.50 We think it can and shall consider this point in further detail in a later chapter on knowledge-how and cognitive achievement.
Objectualist intellectualism: Further issues
Perhaps a deeper strand of objection to Bengson and Moffett’s intellectualist objectualism is that the distinction between a grounding claim and a nature claim, at least as this proposal relies on it, is a distinction without a difference. Consider the relationship between (i) what it is in virtue of which a thing is the thing it is and (ii) the nature of a thing. Let us take (i) as picking up a grounding claim and (ii) as picking up a nature claim. Arguably, the nature of a thing explains the thing. But note that both grounding and explanation are asymmetric dependence relations. If x grounds y, then y depends on x but x doesn’t depend on y. Explanation is likewise an asymmetric dependence relation; if x explains y, then an account of y depends on x but an account of x doesn’t depend on y.
Let us see how grounding and explanation work together in a simple example. What grounds the fact that a porcelain cup is fragile? Its micro-physical structure. What explains the fact that porcelain cups are fragile? Its micro-physical structure. Grounding facts and explanations often track one another. In the case of knowledge-how, Bengson and Moffett aim to distinguish between facts about what grounds knowledge-how and facts about the nature of knowledge-how. As they have it, what grounds knowledge-how is a relation to propositional attitudes. But the nature of knowledge-how is an objectual knowledge relation to a way of acting.
It becomes unclear on further consideration whether this objectual intellectualist view is coherent. For on the proposal being advanced, what explains one’s possession of knowledge-how is that one stands in an objectual knowledge relation to a way of acting. This may involve some propositional attitudes. But suppose we fix those propositional attitudes and vary the objectual knowledge relation to a way of acting.
Then, on their view, in the case in which one lacks the objectual knowledge relation to a way of acting one does not know how to φ. So, knowing how is not depend on some propositional attitudes. On our view, the objectualist part of Bengson and Moffett’s view is unproblematic; it’s the intellectualism that runs in to trouble.
2.4 Hetherington’s reductivism
We want to conclude the chapter by considering an interesting way in which the conceptual geography we’ve been working with can be expanded. For just as there is space for multiple kinds of intellectualism, there is also a way to envision altogether a different kind of ‘reductivism’ about the relationship between knowledge-how and knowledge-that. Stephen Hetherington has in recent work (e.g. 2011; 2012) defended such a position which he calls practicalism. This is the view that knowing that p always involves knowing one or more aspects or constituents of how it is that p.
To sharpen this idea, consider the following example Hetherington offers:
Your knowing that you are in a particular room = Your knowing how to believe accurately that you are in the room, and/or your knowing how to process the relevant data accurately (such as visual data), and/or your knowing how to describe the situation accurately, and/or your knowing how to use relevant concepts accurately, and/or your knowing how to answer questions accurately about the situation, and/or your knowing how to reason accurately about the situation (such as how to link your belief, about being in the particular room, accurately with other beliefs), etc. (2012, 42)
If, as Hetherington puts it, knowing that something is so just is a certain kind or kinds of know-how, then we’ve arrived at a position that looks like the mirror-opposite of Stanley’s position – namely, a view that reverses the order of reduction between knowledge-that and knowledge-how. It is debatable, however, whether we should think of Hetherington’s practicalism as a theory of knowledge-how per se, or alternatively, as a theory of knowledge-that. Compare: We think of Stanley’s thesis that knowledge-how is a kind of knowledge that as, principally, an analysis of knowledge-how, rather than as an analysis of knowledge-that. And this is so even though Stanley regards knowledge-how to be a species of knowledge-that.
This point aside, Hetherington’s is a very interesting position and one that forces among other things a new perspective on the question of whether propositional knowledge might come in degrees. As a matter of clarification, we do not take challenges of reductivism considered thus far to be interpreted as challenges to Hetherington’s particular variety of reductivism. And the reason for this is straightforward: we’re not purporting to provide (or for that matter to assess the material adequacy of) particular substantive accounts of knowledge-that. On the matter of what knowledge-that involves, we’re very happy to remain neutral on (most of) the controversial question that divide mainstream epistemologists51 , and so we take no stance on the matter of whether knowledge-that always involves knowing one or more aspects or constituents of how it is that p.
2.5 Conclusion
Let us take stock. As we’ve seen in this section, the broadly non-linguistic strategies we’ve considered in this section as support for the intellectualist thesis have each come up short. In this section we presented the powerful linguistic argument for intellectualism. As we noted, a complete evaluation of the intellectualist argument depends on the strength of parallel between cases of knowledge-wh with the case of knowledge-how to. We return to this crucial issue in Chapter 5: ‘Knowledge-how and Testimony.’
2.6 Further reading
• Bengson, J. and Moffett, M. A. (2011b). Nonpropositional intellectualism. In Bengson, J. and Moffett, M. A., editors, Knowing How: Essays on Knowledge, Mind, and Action, pages 161–211. Oxford: Oxford University Press
• Bengson, J., Moffett, M. and Wright, J. (2009). The folk on knowing how. Philosophical Studies, 142(3):387–401
• Carter, J. A. and Czarnecki, B. (2016). (anti)-anti-intellectualism and the sufficiency thesis. Pacific Philosophical Quarterly, pages 1–24
• Heim, I. and Kratzer, A. (1998). Semantics in Generative Grammar. London: Blackwell
• Stanley, J. and Williamson, T. (2001). Knowing how. The Journal of Philosophy, 98(8):411–44
• Stanley, J. and Krakauer, J. W. (2013). Motor skill depends on knowledge of facts. Frontiers in Human Neuroscience, 7:1–11
2.7 Study questions
1 Stanley thinks that knowledge-how ascriptions contain embedded questions. How is this observation relevant to Stanley’s conclusion that knowledge-how is a kind of knowledge-that?
2 Why do Bengson and Moffett think the case of Pat, the ski instructor, counts against anti-intellectualism?
3 What is the ‘one-φs-reply’ to the ski instructor case?
4 Explain the salchow case. Is this case meant to challenge the necessity or sufficiency of ability possession for know-how? Does the case succeed as a counterexample against anti-intellectualism?
5 What is ‘objectualism’?
6 Explain Stanley and Krakauer’s diagnosis of the case of HM. Do you agree with it?