Wednesday, August 15, 2012

Prospects for an infinitist theory of knowledge

To: Christian M.
From: Geoffrey Klempner
Subject: Prospects for an infinitist theory of knowledge
Date: 26 February 2008 12:47

Dear Chris,

Thank you for your email of 17 February, with your essay/ paper, 'Prospects for a Positive Motivation for Infinitism'.

Despite your intentions, you have succeeded in persuading me that Klein's theory is a strong prospect for a third alternative to foundationalist and coherentist theories of justification.

What I mean is that infinitism doesn't just appear as the only remaining alternative -- however implausible -- after foundationalism and coherentism have been rejected. There is a strong (for me) intuition at the basis of infinitism which connects with something which many would accept at a pre-philosophical level: namely, that the reasons for a truth are plentiful and inexhaustible, whereas the reasons for a falsehood require increasingly contrived and artificial 'justifications' which finally give out when one reaches the point of total absurdity.

As with any theory in the context of philosophical analysis, intuitions are not always easy to articulate and pin down, as the objections which Klein considers and the two objections which you look at in your essay demonstrate.

However, I am going to have a go at meeting your two objections, the objection of 'adequate truth theory' and the objection of 'convergence behaviour'.

Adequate truth theory

As you point out, coherentism naturally goes with a coherence theory of truth, although it is possible to be a coherentist who rejects the coherence theory of truth. Indeed, a coherence theory of truth is much harder to defend than a coherentist theory of justification.

Similarly, foundationalism naturally goes with a realist/ correspondence theory of truth, although once again it is perfectly possible to be hold a foundationalist theory of justification and yet be an anti-realist about truth.

So what about infinitism? If we view the infinitist as not holding that there 'exists' in some timeless sense an actually infinite chain of justifications but rather holding that justifications are inexhaustible, then the appropriate model for 'truth' would seem to bear a remarkable resemblance to what Popper says about corroboration.

Although Popper holds a correspondence theory of truth (Popper's 'third world') we can describe a 'falsificationist' theory of truth and meaning analogous to Dummett's 'anti-realism', which defines 'truth' negatively as a proposition which is falsifiable but resists every attempt at falsification. This is not the same as the possibility you consider, of an 'ideal observer' theory which defines truth as ideal convergence at the limit of inquiry. The 'truth' -- i.e. that at which judgements aim -- is that which has the potential to withstand every objection, that is to say, for every attempt to cast doubt on the proposition a suitable defence can be found.

This is a radical move, because in a sense it does away with truth in the traditional sense.

Having said that, I see no reason why the infinitist about justification should not simply help themself to a correspondence theory of truth, just as in Popper's own account of the methodology of scientific research.

Getting back to the quote from Ginet, 'inference cannot originate justification, it can only transfer it from premises to conclusions', the correct response should be that justification is nothing other than the process of weeding out beliefs which rest on false assumptions. The more thorough one's attempt at weeding out, the more confidence one is justified in having that the proposition in question is in fact true.

Convergence behaviour

I don't see that each proposition in the inferential chain must necessarily 'add to the total credibility' of the proposition in question. It must pass the test of being a recognizable reason for believing the proposition, and this assumes that in the background we already have a conception of what is or is not a 'good reason' for the case in question.

This is an important concession in itself, because it is not as if everything is up for grabs. We have rationality, our background beliefs about the kind of thing that looks like a reason or explanation for something.

There was a famous case in the last century of children who 'discovered' that they had fairies living at the bottom of their garden. The claim was never conclusively refuted. Well, suppose I made this claim. You ask me why, and I have to tell you something which has the capacity to function as a relevant reason. I can't say, 'Because the tomatoes are growing much better in the poor soil than I expected.' Any number of more plausible explanations can be given for the health of the tomato bush.

On the other hand, if I said, 'There are fairies at the bottom of my garden because there are pixies there, and where there are pixies you will always find fairies,' all I have succeeded in doing is create a bigger target to shoot down. Now I have to justify fairies and pixies.

A photograph of a 'fairy sighting' would be a possible justification. Now we have to look at reasons for accepting that the photograph is genuine and not just another example of the kind of thing UFO hoaxers concoct in the darkroom.

You can see where this is heading. There are three broad categories of potential reason:

1. Those which would be an adequate explanandum, like the tomatoes, but where we have any number of more plausible explanations.

2. Those where we merely succeed in increasing the size of the target, hastening the point at which 'total absurdity' is reached.

3. Those which offer genuine evidence which, however, rests on assumptions which themselves require further justification.

Realistically, therefore, I do not accept the objection that one could 'spin out a long enough chain of inference' for any belief. There are practical limits to what one can spin, and when those limits are reached, the falsehood is seen for what it is.

All the best,