Tuesday, May 10, 2011

Xenophanes on the limits of human knowledge

To: Serguei R.
From: Geoffrey Klempner
Subject: Xenophanes on the limits of human knowledge
Date: 2 May 2002 08:50

Dear Serguei,

Thank you for your e-mail of 24 April, with your second essay for the Ancient Philosophy program, on Xenophanes, in response to the question, 'What is it to 'know' that something is the case? Can the truth of a scientific theory ever be known? Illustrate your answer by reference to Xenophanes' reflections on the limits of human knowledge.'

With regard to theory, you say, 'According to Xenophanes humans are unable to know the truth... But he never doubt[ed] existence of truth as such.' We are assuming that Xenophanes and ourselves are agreed that there can be truths of the form, 'This is a tree', or 'It is day time'. Xenophanes also believed (at least on my interpretation) that there are truths about the nature of physical reality which only God knows. Because these latter truths can be known only by God and not by man, Xenophanes concluded that there can be no such thing as theoretical knowledge. The only knowledge human beings can have is perceptual knowledge, or knowledge derived from perception without the aid 'theory'.

Now, one point that could be raised here is whether, in fact, any perceptual claim is completely free from theory. According to Descartes two thousand years later, even a simple claim such as, 'I am seated by the fire' implies the theory, 'There exists an external world because God is not a deceiver'. Lacking proof of the existence of God, the belief in an external world becomes as theory laden as any obtuse speculation about the nature of physical reality. It seems this problem never occurred to Xenophanes. However, it does indicate a potential incoherence in the idea of separating two distinct realms of inquiry, the realm of simple perceptual truth, and the realm of doubtful theory. (Later, Plato brilliantly reversed this distinction, arguing that only the non-physical world of Forms can be an object of knowledge, while the world of everyday, changeable things remains a world of doubtful opinion.)

The crux of the matter is how to save the title 'knowledge' for scientific theory. You and I are both agreed on that.

Your recommendation, if I understand it correctly, is that scientific theories can be knowledge, provided that we give up talk of scientific 'truth'. 'I think we should define theoretical knowledge simply as justified belief... It means... that knowledge can turn out to be false. But after it's falseness is discovered it is no longer knowledge.'

In other words, in order to save scientific knowledge, we must get rid of Xenophanes' God.

Or, putting the same point in non-religious terms, in order to save scientific knowledge, we must give up the correspondence theory of truth.

If belief in a scientific theory is not belief in its 'truth', what is it? The standard account of belief is that belief aims at truth. To believe that P is to believe that 'P' is true. Here are some familiar alternatives:

A. 'To believe in theory P is to choose P as your instrument for predicting experimental results.'

B. 'To believe in theory P is to judge that P coheres best with your other theoretical beliefs.'

C. 'To believe in theory P is to accept that P is good for us to believe.'

To which the correspondence theorist replies that a 'good instrument for predicting results' can still be false (because it is still not the best instrument). A theory that coheres best with other theories can still be false, if they are all false too. The 'truth' of the belief that theory P is good to believe (for example the Politburo's belief that it is good for us to believe what they say) cannot itself be decided on the basis of 'what it is good to believe' without entering on a vicious regress.

Depending on which of the three alternatives you choose, I assume you will say, 'Yes, that's OK, because I am not offering an alternative theory of truth. I don't accept that there is such a thing as theoretical truth.'

My own personal view, however, is that this is not a coherent position. The very fact that we are aware of the provisional nature of theory (a better 'instrument' can come along, 'coherence' is not proof, etc.) shows that we *do* operate with at least a minimal correspondence notion of truth, namely, that:

'P' is true if and only if P.

It doesn't matter whether 'P' is a report of immediate sense experience or an obtuse physical or metaphysical theory, the effect is the same. Any statement you make aims at truth, because it means what it says. 'Rocks are made of water' is true, is a correct statement, if and only if rocks are made of water. You were justified in asserting that rocks are made of water (because that's the best available explanation) even if you later accept you were wrong to believe that rocks are made of water, because a better explanation comes along. Now, your judgement is that 'Rocks are made of water' is false.

All we need for theoretical knowledge is recognition of the real possibility that rocks are made of water. We do not need to prove that rocks are made of water, so long as there can be sufficiently good reasons for putting forward the theory that rocks are made of water.

The result, according to the standard analysis of knowledge is that:

'Rocks are made of water' is knowledge if and only if:

1. We believe that rocks are made of water.

2. We have sufficiently good theoretical grounds for believing that rocks are made of water.

3. Rocks *are* made of water.

The added emphasis here makes no difference to the meaning. But nor does saying, '"Rocks are made of water" is true.'

As you say, if it turns out we were wrong, then it wasn't 'knowledge' after all.

- One last point. Since the famous article by Paul Gettier ('Is Knowledge Justified True Belief?'), it is generally agreed that justified true belief is not sufficient for knowledge, because a statement can be justified, and true, and yet its truth can turn out to have been a lucky accident in relation to the justification that was put forward for its truth. It is an interesting question whether that affects the defence I have offered above of theoretical 'truth'.

All the best,

Geoffrey