To: Alistair L.
From: Geoffrey Klempner
Subject: Counterexamples to knowledge as justified true belief
Date: 17th April 2009 11:40

Dear Alistair,

Thank you for your email of 3 April, with your essay for the University of London Epistemology module, in response to the question, 'How should one respond to apparent counterexamples to the view that knowledge is a form of justified true belief?'

I am somewhat puzzled by this question. I can't believe that the examiner seriously wanted you to consider three sets of possible counterexamples -- to the view that belief is necessary, or that justification is necessary, or that justified true belief is sufficient. Each of these would more than justify an essay to itself. Nor can I see any useful general question of methodology concerning how one might approach counterexamples to definitions of knowledge, without going into specifics.

Considered as three essays, you have done well to say as much as you have, given constraints of space.

I wonder: in discussing the third question (whether justified true belief is sufficient for knowledge) you mention theories which add a fourth condition in order to meet Gettier counterexamples. Surely, any theory which *adds* a fourth condition accepts that knowledge is a 'form of justified true belief' (namely, justified true belief which meets some as yet unspecified fourth condition) in which case Gettier arguments are not 'counterexamples' to this view, but rather to the view which holds that knowledge *is* justified true belief.

(I seem to have mislaid my copy of the 2007 examiners reports. I don't normally check on these but it would have been useful in this case.)

My initial reaction was that this question is about Gettier. However, given what I've just said, I'm not so sure now.

Things are complicated because philosophers who opt for more radical solutions to Gettier counterexamples (e.g. Nozick on 'truth tracking') *do* question whether knowledge is a 'form of justified true belief'. In other words, in tackling the problem of the sufficiency of the three conditions combined, one is led to question the necessity of the first and/ or second condition.

On an externalist, reliabilist view for example, justification (conceived as an argument you could give to an interlocutor if challenged) is not necessary. All that is required is that the route through which the relevant information was acquired is one that accurately transmits truths. Your chess example is a possible candidate, although for reasons which you give it is somewhat questionable. If you are good enough to make accurate judgements of a position, surely you are good enough to give the required analysis if challenged. (That's what chess players do.) Otherwise, the strong suspicion is that you are playing a hunch. In chess, a single overlooked line can destroy what would otherwise have been considered a 'well judged' move.

A better example might be someone who is a good judge of character who can't explain why they thought a particular individual had something to hide -- appearance, voice, body language all play a part. The standard (somewhat humorous) example is chicken sexers, who can't explain how they are able to reliably sort male and female chicks (the theory is that it has something to do with the smell of the chick, not the appearance as the chicken sexers seem to believe).

Is belief necessary for knowledge? You give a decent answer here. The question we need to answer is just what sort of thing a 'belief' is. It can't be correct to say that a belief is necessarily something you are in a position to avow as a belief. Leaving aside odd cases like the nervous schoolboy, we do talk ourselves into things, giving specious arguments and justifications. And yet when push comes to shove we show by our actions that we never really believed what we said we believed.

On the 'Davidsonian' model (from Donald Davidson who has done seminal work on the analysis of action statements) an action is explained by a combination of the agents beliefs and desires. Knowing what the agent wanted, and what they did, we can infer their beliefs. We normally expect beliefs so inferred to correspond with what an agent would avow; but there is no logically necessary connection.

I would go further and point to a fundamental difference in the 'point' of the concepts of knowledge and belief. The point of the concept of belief is in the explanation of action. Whereas the point of the concept of knowledge is to identify subjects as reliable sources of truth. Occasions will arise when we will confidently act on the testimony of a subject who (for whatever reasons) would not consider themself a reliable authority. On this view, individual human subjects are just links in the chain of transmission of knowledge.

All the best,

Geoffrey