Thursday, 20 February 2014

Adam Carter and Duncan Pritchard on knowledge-how

I’ve been reading a couple of forthcoming papers by J Adam Carter and Duncan Pritchard downloaded as pre-publications from Duncan’s website (so I assume quoting is OK). I found the two together very helpful, not least because of some general stage setting about possible responses to Stanley’s views on Gettier examples and knowledge-how (the nested dilemma in one paper is particularly helpful).

(For anyone who is not familiar with the recent literature, Ryle famously argued that knowing-how and knowing-that were distinct with knowing-how more basic. Ryle argued against what he called the ‘intellectualist myth’ - which was the idea that the intelligence implicit in knowing-how could be explained as deriving from grasping a proposition, ie a bit of knowing-that - by deploying a regress argument against the idea. But Jason Stanley and Timothy Williamson published a paper in 2001 attempting to counter the regress argument but also sketching a model of knowing-how as a matter of grasping in a proposition a way to do something, but grasping it under a practical mode of presentation or sense. Knowing-how is thus, on their picture, a particular form of knowing that. If, however, knowing-how is a form of knowing-that and if knowing-that is susceptible to Gettier inspired counter examples then so should knowing-how be. Since they – and many others afterwards – focus on examples in which the original instruction from which one derived the know-how has the mix of good and bad epistemic luck that characterise Gettier cases but which leaves intact the underlying ability, there has been some disagreement as to whether the resultant putative know-how really has been undermined: really is not knowledge.)

This is the debate into which Carter and Pritchard enter, attempting to argue that knowledge-how and knowledge-that are distinct without this escalating to forms such as knowledge-where. I’m less sure that I share enough of the basic intuitions that drive their papers to follow the detail of the argument.

Both papers (referenced below) argue that the idea of skill-like cognitive achievement doesn’t capture or codify intuitions about knowledge both because one can have knowledge (eg via testimony) without much achievement (when knowledge rubs off on one like an infectious disease, in Evans’, I think, phrase) and because cognitive achievement can survive one kind of luck (environmental rather than intervening) whereas knowledge in general cannot. This helps to motivate the idea that knowledge-how is genuinely distinct from knowledge-that which in turn is akin to all the other forms of knowledge-wh... (who, what, where, when etc).

One tool they use is a comparison between knowing-why and understanding-why where the latter is an achievement. But initially they sketch a rougher sense in which understanding-why is harder to achieve than knowledge-that. It’s here in this paragraph from ‘Knowledge-How and Cognitive Achievement’.

Campbell has a good grasp of how faulty wiring might cause a house fire, one that is sufficient for him to be credited not just with knowledge-that, but also with understanding-why. But imagine that he passes on his knowledge of what caused the house fire to his young son, and that while the son knows what faulty wiring is, he lacks a solid conception of how faulty wiring might cause a house fire. Campbell’s son can come to know that the house burned down because of faulty writing by receiving this information from his father, at least so long as he is capable of being reasonably circumspect about who he receives testimony from and what he believes on the basis of testimony (i.e., such that he doesn’t believe everyone about everything). Of course, he will be to a large extent trusting his father’s word in this regard, but in these circumstances, and about a subject matter like this, such trust seems entirely compatible with the acquisition of testimonial knowledge. But given that Campbell’s son lacks a solid grip on how cause and effect might be related, can he be rightly said to understand why his house burned down? We suggest not.

The suggestion seems to be that even if one knows why the house burned down (because it had faulty wires), if one’s grasp of the causal nexus into which this fits is poor – one lacks a solid conception of how faulty wiring might cause a house fire - then one fails to understand why it burned down. An understanding is fuller than mere knowledge. This distinction seems pretty much carried by an intuition about everyday language use although it isn’t discussed more in either paper (“He knows but he doesn’t really understand”, we might say.)

On the other hand, they connect understanding with cognitive achievement in such a way that it is also easier to gain: it is not undermined by environmental luck. (I mean that it is easier, not their word, in this sense: there is one fewer thing to worry about.)

Though the environmental epistemic luck at play in this second version of the case would be enough to undermine propositional knowledge that the house burnt down, the same doesn’t hold for understanding-why. Indeed, that Campbell could so easily been given the wrong explanation seems perfectly compatible with him having gained an understanding of why his burned down in this case.  

This might seem odd or arbitrary except for the fact that there has been a prior connection made to skill.

Imagine, for example, that our archer’s shot could so very easily have been affected by freak gusts of wind but in fact wasn’t she just happened to fire at the precise moment to avoid the freak gusts, which would have otherwise affected her shot. The archer’s success is thus lucky, in that it is a success that could so very easily have been a failure, but where this luck is of the environmental rather than the intervening variety. Here is the crux: Is this archer’s success any less of an achievement in virtue of being subject to specifically environmental luck? We claim that the natural answer to this question is ‘no’. While intervening luck can undermine a genuine achievement, a success is no less because of ability if there is environmental luck in play.

With both these intuitions in play – that understanding is more than mere knowledge of a fact and that understanding is skill-like in that it is, unlike knowledge, immune to environmental luck – we get both the differences from knowledge. Understanding needs more than mere atomic testimonial transmission – it is harder to get than that – but it can survive a merely lucky provenance. These seem independent intuitions and there doesn’t seem much argument for them. I can imagine conceding the former but then denying the latter (because I think of understanding as merely a more thorough body of uber-knowledge and thus, knowledge-like, being undermined by environmental luck). Or I can imagine contesting one spin of the former. Let’s say you cannot understand why the house burnt down unless you know more than just the fact that the faulty wires caused it. You need to know the causal steps. But why could you not learn those steps through testimony? That is, why think that the extra achievement is of a different kind rather than degree for knowledge? It is not clear to me that this follows from the everyday usage. In other words, I worry that the account of understanding has not done enough to demonstrate the connection to skill that they need in order that this can serve as an example of something - like knowledge-how to come - that is immune to environmental luck.

At the end of the paper, we get a similar easier and harder (my rough description not theirs) spin on knowledge how but here it seems to pull the other way. That is, there’s insufficient connection between knowledge-how and skill.

It is natural then to ask whether knowledge-how is present in cases of knowledge-that which fall short of a cognitive achievement. Going back to the LUCKY LIGHT BULB case, suppose that Charlie finds out how to change a light bulb by receiving this information from what he knows to be a reliable information source (and that there is, in addition, nothing epistemically amiss in his acquisition of this information), and that Charlie passes this information onto his young son. Let us stipulate that Charlie’s son exhibits the same level of cognitive ability as we saw exhibited in the testimonial case considered above. That is, while he wouldn’t have asked just anyone or believed just anything that he is told, it is nonetheless the case that for the most part he is merely trusting the word of his father. Nonetheless, his father is indeed authoritative in this regard, and the environment is epistemically friendly in all the relevant respects (in particular, it is not the case, for example, that Charlie’s son could so very easily have been deceived by his father). As we saw above, on standard proposals regarding testimonial knowledge-that, Charlie’s son will come to have knowledge-that in this case (i.e., knowledge-that such-and-such is how one goes about changing a bulb). But does he know how to change a light bulb?

So Charlie’s son does acquire knowledge-that because, as long as there is no luck in play, knowledge is easy to come by. It needn’t be more than the most minimal achievement. So, for example, he knows that such and such method is the way to change a light bulb. Still this leaves open the question of whether he knows how to change one.

At this point they say something that surprised me:

Sure, he can change a bulb...

But, hang on, in the example we don’t know this at all. For all we know, he suffers from locked in syndrome. In such a case, it is at least possible to argue that he knows how to change the bulb but can’t. (This isn’t my view – because of the action at a distance of other views – but it seems that every day usage might allow it.) It is a further assumption that – in addition to knowing the way to change the bulb - he can actually change a bulb. Still if we pause to note the assumption, we can make it and see what happens. So suppose that in addition to having all that testimonially acquired knowledge-that by listening to his father in informal theoretical lectures in the kitchen he also has the practical nous to change a bulb. Does he know-how to change the bulb?

At this point, it seems a bit weird to me to ask the question. To say that he can change the bulb is not to say that he has been successful (it’s more general than that it though to have the ability may not require past actual success). Nor is it to say that it is possible, in the luckiest of circumstances, that the son, left in a room, would end up changing a bulb though hypnotic control by others. It is to say that he has the ability, a standing reserve. It is this which explains, when all goes well environmentally, his success. But in a post-Gettier world Carter and Pritchard go on:

Sure, he can change a bulb, but there is plausibly more required for knowledge-how than that. What we are after when it comes to know-how is not merely the ability to produce a certain outcome, but rather a particular kind of epistemic relation that the agent exhibits with respect to that outcome. And sure, Charlie knows how to change a light bulb, since he can vouch for the epistemic source of his information. But the key question is whether Charlie’s son can come to acquire knowledge-how in this case, given that he is for the most part simply trusting his father’s testimony. We suggest not.

Here’s a funny thing. We get to know that Charlie (the father) knows how to change a light bulb because he can vouch for the epistemic source of his information. This looks the wrong place to look for reassurance of his practical know-how because it assumes a theoretical view. Again: perhaps Charlie suffers from locked in syndrome and cannot move from his bed. This implicit assumption is also suggested in the distrust of an ability to produce a certain outcome in favour of a particular kind of epistemic relation (I should add I like the later comment which connects ability and outcome I think it is implicit in the first, distrusted, phrase here).

It seems to me that the challenge in this area of philosophy is to locate knowledge-how in the face of two possible assimilations:
* We can connect it to ability with Ryle
* Or we can connect it to propositional knowledge with Stanley and Williamson.
A too brute connection to ability threatens to remove the mindedness or conceptuality from it and thus threaten its genuine knowledge-status. But a too theoretical or effete connection to knowledge-that threatens to undermine its practicality. For that reason, Stanley and Williamson insist that the demonstratively articulated proposition that makes up the content that is ‘known-that’ is also entertained under a practical mode of presentation. But in this discussion, that safety net seems to have gone missing so that knowing the right propositions whether practically or not counts as knowing-how to do something. And that’s surely wrong in some cases. In some cases, that would lead to saying false things.

The analogy with skill is first deployed, then played down, and then returns at the end. Carter and Pritchard point out that Gettier-related problems which result by splitting ability and success and reuniting them through luck can be avoided by stipulating that the success is because of the ability.

If one successfully Fs because of one’s ability (vis-à-vis F), then one knows how to F. If this is right, and the converse holds i.e., that if one knows how to F, then one is positioned to successfully F because of one’s ability (vis-à-vis F) then we have an account of what it would be for knowledge-how to essentially involve cognitive achievement.

This emphasis on ability and its explanatory connection to success seems right. But when it comes to the examples they offer of knowing how to do something the emphasis is too effetely cognitive and not practical enough.

I have been increasingly having Travisian thoughts about these connections. Sometimes to say that someone knows how to do something is to say, roughly, that they know, of a way of doing it, that it is a way to do it whether or not they have a practical ability to do it. Sometimes, it is to say that they can do it. If so, their success, as Carter and Pritchard stress, is their achievement even if it takes the right environment. It’s the context that makes particular Gettier style examples relevant or not. So if the context concerns practical abilities, the relevant worry wouldn’t be a purely cognitive style of Gettier example (such as the lucky light bulb case) but rather something that suggested that a skill was, by bad luck, exercised in normally unforgiving circumstances although compensated for by further luck. But if the context is theoretical, the standard kind of Gettier example is the threat to knowledge.

So whereas Carter and Pritchard seem (though they express some reservations) to think that there are firm boundaries between knowledge-how, on the one hand, and knowledge-that or knowledge-wh..., on the other, I’m less sure. Take this paragraph from ‘Knowledge-How and Epistemic Luck’ in which they discuss whether Charlie really knows where to buy a light bulb in virtue of reading a guide book which, although reliable, was randomly selected from among unreliable alternatives. So there’s environmental luck in play.

Crucially, however, we maintain that it is far from obvious that knowledge-where is compatible with environmental epistemic luck, especially when compared with knowledge-how. In order to see this, try the ‘past self’ test on this case. If Charlie were to subsequently discover that environmental epistemic luck was involved in him gaining the right directions to the light bulb store, would he still regard his past self as knowing where to go? We think this is not at all clear. At the very least, it seems right to say that Charlie will be much less inclined to judge that his previous self had knowledge-where than before. In contrast, we noted above that there seems no temptation at all for Charlie to regard his past self as lacking knowledge-how in the corresponding case.

It seems to me that context is crucial here. If Charlie has acquired a reliable navigational skill in a confusing foreign city then the ascription that he knows where the shop is survives the lucky origins. But if the focus is on knowing the address in a regular grid structure, where the further skills concerning navigation are taken for granted by all, then it does not. So it is not a property of knowledge-where as such but the ascription in context.

Carter, J.A. and Pritchard, D. (forthcoming) ‘Knowledge-How and Cognitive Achievement’ Philosophy and Phenomenological Research
Carter, J.A. and Pritchard, D. (forthcoming) ‘Knowledge-How and Epistemic Luck’ Noûs