To: Christian M.
From: Geoffrey Klempner
Subject: Truth tracking and the closure principle
Date: 12 February 2008 13:15

Dear Chris,

Thank you for your email of 28 January, with your University of London Epistemology essay in response to your own question, 'Why a truth tracker does not need to reject closure.'

This is an excellent piece of work, well up to the standard of the best essays that my UoL students have sent me.

On the basis of this I do think that you have a good chance of achieving a First in at least some of your papers. Even if, given the time constraints, your final result is 2/i, it does make a significant difference how this overall mark is achieved. So my advice is, 'Do your level best and don't worry.'

There are some points in your argument where one might raise objections, but the overall impression is that you have a very good grasp of this issue, and have something of your own to contribute to the debate.

You have given a clear exposition and critique of Dretske's account of knowledge and his 'independent' arguments in favour of rejecting the closure principle. You go on to develop a plausible response to BIV type arguments which cleverly concedes the 'metaphysical' point to the sceptic while at the same time defending knowledge. I like this a lot. In fact, I am almost convinced.

Let's look at a couple of points of detail, where I was not fully persuaded.

1. Dretske's regret analogy. What is the argument here? Dretske is saying that we intuitively think that closure holds for regret, but looking closer we discover that it does not. So the *intuition* that closure holds for knowledge might, similarly, turn out to be unwarranted.

My problem here is that, as you yourself admit, the example you give to illustrate closure is very unpersuasive. I regret getting drunk, and getting drunk entails that I drank something. Therefore, I regret that I drank something? Obviously not! Could this really be what Dretske meant?

Surely, the intuition about regret is that when we do things that we regret, those actions have *causal consequences* that we also regret. Causal consequences are not the same as logical implications. I got drunk at the Faculty party and made a pass at the Professor's wife. I also did an extremely insulting impression of one of the female lecturers and when reprimanded, I dropped my pants and showed my bare bottom.

These are things to regret. However, following my being ignominiously being fired from the university, I found a much better lecturing job elsewhere at twice the salary where I am happy and content. Now, I no longer regret my getting drunk, although I have resolved never to test my luck by getting drunk in similar circumstances.

This is an argument in favour of regarding regret as closed under causal consequences. When we do things we regret, our actions have regrettable consequences. When, later, the consequences are no longer regarded as regrettable, we cease to regret the action that led to them.

However, Dretske would (presumably) argue that not all the consequences have to be regrettable. It suffices that the bad consequences outweigh the good. For example, suppose I succeeded in seducing the Professor's wife. That is not a regrettable consequence, but it is insufficient compensation for the ignominy of being fired from the department and as a result being unemployed.

So, I regret getting drunk and what this led to, even though there is one consequence of getting drunk which I don't regret.

I can't remember what Dretske says about his regret analogy so I don't know whether he would agree or not with my take.

2. Towards the end of your essay, you belatedly consider examples which do not fit the stratification of worlds model, such as Dretske's painted donkey. I was wondering when you were going to come to this as it does seem a serious limitation on your response to the closure question.

I don't agree that it is sufficient to say, 'in principle I could tell if it is a zebra or a mule. It is not out of the theoretical epistemic reach of us as humans.'

There are many cases, such as those David Lewis describes in his argument for a Contextual theory of knowledge, where we say we know that P, then are prompted to withdraw that claim when asked, of some Q where P entails Q, whether we know that Q. This applies to the mule/ zebra. You are going through the safari park and you see a zebra running alongside your car. Do you *know* that it is a zebra? Isn't this a paradigm case of knowledge?

If, in fact, it is a zebra then you do know, would be the response given by the majority of theories of knowledge. You don't need to possess sufficient evidence to rule out a donkey painting hypothesis, if the context is such that this is not a hypothesis that one would normally consider. But this is, arguably, a typical case where closure fails. It is irrelevant that you *could* find out. We habitually make knowledge claims without waiting for the final results of a painstaking and exhaustive investigation.

What this *shows* about the concept of knowledge is another matter.

All the best,

Geoffrey