To: Eric G.
From: Geoffrey Klempner
Subject: Objections to knowledge as justified true belief
Date: 9 March 2007 11:18

Dear Eric,

Thank you for your email of 2 March, with your University of London essay in response to the question, '"Knowing that P is at least a matter of having a belief that P which is both true and justified." Is this an adequate definition of knowledge? If not, how should it be improved.'

This is another well researched answer, which shows a good grasp of the issues.

If an examiner was being picky, they might raise a question over your ready acceptance that knowing that P is 'at least' a matter of having a belief that P'. The case of the nervous schoolboy has been cited as an example where knowledge does not entail belief.

Again, being picky, from the point of view of the logic of the question, 'X is at least a matter of ABC' cannot be an adequate definition, irrespective of X or ABC. A definition must give sufficient as well as necessary conditions. You would get some credit for recognizing this.

I do like the way you have avoided launching straight into the tri-partite account and considered instead the question whether justification is fallible or infallible. This sets the context for this question. However, given this context, it is not so clear that the central motivation for a definition of knowledge is to provide ammunition for someone challenged with 'How do you know?'

'I know because I believe, my belief is true and I have justification for my belief, and etc. etc.' (or, insert any account you like, e.g. Nozick's)? No. This is not how it works. If challenged, you can't appeal to the truth of your belief because that is one of the things being challenged. You are being asked by the challenger to provide a persuasive argument for your case; which might involve citing evidence, proof of your own authority as someone who has the right to make this kind of judgement, or etc.

Plato's interest (in Theaetetus, and also in the Meno) was very different from that of modern epistemologists. In his view, if you don't have a proper 'account', you are liable to be persuaded to give up your belief by dubious counter arguments, the belief has a tendency to 'run away'. It turns out that only philosophy is able to give a sufficient 'account' for its claims. As Plato argues in 'Republic', there is no such thing as empirical 'knowledge'. In other words, it is hard to see 'fallible justification' as having any merit at all from Plato's exacting perspective.

It is worth asking, why do we need a definition of knowledge? Assuming a fallible view of justification, we are clearly not seeking to refute scepticism; that task is conceived as already having been done, or something we can tackle separately. Human fallibility was the centrepiece of the sceptic's case, but we are not too bothered by that.

Knowledge presents a challenge to philosophical analysis. Like 'person' or 'cause' we feel we ought to be able to provide necessary and sufficient conditions, on pain of admitting that we don't really know 'what knowledge is'. It is essential to this exercise that the request for a definition is framed in the third person.

Given this framework, it is not at all clear however to what degree the subject whose knowledge we are considering needs to be aware of what we, from our superior vantage point, are aware of. Hence the clash between internalist and externalist accounts. Is it, in fact necessary for the subject to be able to provide any justification for his/her belief? why isn't truth enough? (provided the route is 'reliable'). This is another reason for questioning the 'at least' claim made in the question.

Another theory which fits the externalist view, apparently originally suggested by Russell (although I don't have the reference), is to require simply the absence of false assumptions. This view has been advocated by Gilbert Harman. Obviously the subject can't demonstrate this. Maybe we can't either. However, from an imaginary God's-eye standpoint, case of knowledge just are cases where there are no false assumptions anywhere along the line that leads to the subject's belief. In other words, belief is naturally knowledge - unless something goes wrong.

Remember, we are not trying to refute the sceptic but merely seeking to give adequate necessary and sufficient conditions. This is the analytical game we have chosen to play.

This kind of exam question is tricky, in that it seems to be fishing for how much you know about this issue. In order to fully answer the question, 'How should it be improved?' it is not enough to give your favourite account or to offer one or two alternatives. It looks like you need to sketch all the theories that have been put forward. I've suggested a couple that you have missed (reliabilism and the no false assumptions theory). I have to confess I didn't fully understand your explanation of Dretske's position. If I was answering the question I probably would have forgotten Dretske, which I suppose shows that one can't be expected to say everything, especially when you've only got one hour.

All the best,

Geoffrey