I have just come to the end of teaching an epistemology module (just a revision session to go). More so than other subjects I teach, at the end of the module, no particular analysis of knowledge and no particular response to scepticism seems totally compelling. The concept of knowledge seems particularly resistant to analysis. My attempt to frame a overall view - and to grapple with why it is so tricky - focuses on five key questions.
1: What more is there to knowledge than true belief?
Epistemology starts with the extra content to the concept of knowledge over and above true belief. (I am ignoring, without argument, the thought that knowledge and belief are incompatible because to say ‘I believe that p’ implies that one does not know it. But it does seem to me that this has more to do with pragmatics than semantics.) But why is there excess content to knowledge?
After all, mere true belief gets one what one wants. Eg. I want a coffee and I believe that there is coffee in the café. The truth of this belief (and subsidiary related beliefs) is what enables me to satisfy that want.
But: mere true belief is insufficient for knowledge. Knowledge can be undermined by luck. This seems to be a pre-philosophical datum. But recognising that isn’t the end of the issue. We can ask ‘why?’, what is the point of the concept? Edward Craig is a rare philosopher who asks this question but whether his hypothesis of knowledge’s purpose can be independent of a prior grasp of what knowledge is is moot.
2: Is scepticism natural?
(Natural by contrast with artificial / an artefact.)
We do not normally take any account of sceptical possibilities ie. it is normally correct / appropriate to ascribe everyday knowledge independent of them.
But, also and normally, a change of context can undermine a knowledge ascription. Eg the revelation that Smith has a local twin undermines the claim that we know Smith is in town. Unless, that is, we also have a way of distinguishing the twins.
Scepticism seems to grow from the way the context of a claim to know can suggest ringers both:
i) relative to evidence so far (the twins) and
ii) relative to all possible evidence (brain in a vat).
If scepticism is merely implicit in this structure, a natural implication of it, ie if there is no difference of kind between i) and ii), then scepticism is natural.
So both Wittgenstein and Williams attempts to undermine the sceptical ringers, specifically.
Wittgenstein suggests that we do not understand them as they threaten the preconditions for the language game of ascribing and claiming knowledge and doubt.
Williams argues that the sceptical ringer depends on the idea of ‘knowledge as whole’ which presupposes a false theory of knowledge (cf Bacon on heat).
2b: If scepticism is natural, can it be reconciled with everyday knowledge ascriptions?
Nozick: concedes that we do not know that sceptical hypotheses are not true but denies that this has implications for everyday knowledge. For Nozick this follows from his tracking conception of knowledge.
But it seems odd to maintain everyday knowledge in the face of active contemplation of an undefeated sceptical ringer.
American contextualists such as DeRose and Cohen argue that everyday knowledge ascription is correct in our everyday context but that that scepticism changes the context. There, we do not know that sceptical ringers are not true.
But it is hard not then to think that the everyday context is simply ignoring a possibility that should be taken into account.
Preliminary conclusion to question 2:
Either concede that scepticism trumps everyday knowledge and convict everyday ascription of error,
Or deny that scepticism is natural. But if so, we need to know why, and we need to know why it isn’t natural before the sceptical argument is off the ground.
3: Can ‘knowledge’ be analysed?
· The analysis begs the question by assuming notions a grasp of which follows that of knowledge.
· The analysis would be substantive but is not sufficient for knowledge.
The second horn applies to reductionist accounts of knowledge. But the dilemma as a whole applies applies to the Justified True Belief account in the face of Gettier cases. Gettier presents his cases as of JTB but nevertheless insufficient for knowledge (the second horn). One response is to argue that Smith is not, in fact, justified in his beliefs. But this threatens to equate our understanding of justification with whatever is sufficient (combined with true belief) for knowledge.
So could 'knowledge' be a basic concept? And if so, what does that do to epistemology?
A concept may be complex, in the sense that its philosophical elucidation requires the establishing of its connections with other concepts, and yet at the same time irreducible, in the sense that it cannot be defined away, without circularity, in terms of those other concepts to which it is necessarily related. (Strawson 1992, 22-3)
4: Is the quest for generality the problem?
Consider the problem for reliabilism that a method must be reliable (eg, not less than 100% else luck undermines knowledge and also not merely 100% but by luck, so, perhaps, 100% in this and nearby pws).
But no general method can be that reliable. Even perception of medium sized objects from close by in afternoon light is not that reliable. But in this context, looking at that apple juice carton, there may be no possibility of error. So could local context, rather than general method, be the right response to the worry about luck?
Why might general analysis be difficult? Perhaps it has to do with the connection between knowledge and other possibilities that might have been.
5: Is knowledge natural*?
*Where natural = part of nature, this world.
The cat is on the mat in this world, whatever the relation of cat and mat in other possible worlds (pws).
But whether Smith knows that the cat is on the mat depends not just on her relation to that fact in this world, but also in some other pws. If it had not been, she would not have believed it. (Or, in nearby pws where it is not on the mat, she does not believe that it is.)
We may, with McDowell, want to say that although the capacity to have knowledge is fallible, when all goes well, Smith does know that p. This connects to what McDowell says about perception: when all goes well, Smith is directly experiencing the fact that p even though, in other cases, she can be mistaken. Perception is a disjunction: either a genuine seeing (of the cat on the mat) or a mere appearance.
But unlike ‘directly experiencing’, knowing seems to depend on what might have been, not just what is. For knowledge, the difference between ‘good’ and ‘bad’ disjuncts is not just what actually obtains but what might have been.
(The contrast is made more complicated by the fact that an analysis of perception in either causal or counter-factual terms also presupposes how things would have been in other cases. If Smith sees that p then had p not been the case then, all other things being equal, Smith would not have had the experience as of p. But in the case of knowledge, the other possibilities, and hence the range of potential ringers, seems particularly badly behaved.)
But the lack of a plausible, non-circular, general account of the possibilities that need to be taken into account, the extra content of knowledge over true belief, makes the prospects of a general analysis of knowledge dim and throws attention back on particular circumstances. Hence epistemology helps unpack some of our commitments here and the connections between knowledge, justification, luck, etc but does not offer a reductive account of the concept in general terms.