Thursday, April 21, 2011

The transcendental deduction

To: Tony B.
From: Geoffrey Klempner
Subject: The transcendental deduction
Date: 7 February 2002 14:43

Dear Tony,

Thank you for your e-mail of 28 February, with your latest essay towards the Associate Award, 'Line of advance: the Transcendental Deduction.'

It is a good plan to concentrate on a critical exposition of Kant's refutation of scepticism/ idealism.

When I did my term's supervision with Strawson for my Oxford B.Phil, the first essay I wrote for him was on the 'Metaphysical deduction of the categories'. This relatively brief section is often overlooked by students of Kant, but I saw it as important because this is the place where Kant has the chance to say what he means by an 'objective judgement'. My admittedly contentious gloss on the argument (once we have stripped the argument of its excessive reliance on the Aristotelian forms of syllogistic reasoning) is that we can deduce from the existence of logic that the world is a place where it is possible to gain knowledge by deductive inference. Deductive inference has real utility only in a world where one is able to put forward a theory of how objects are arranged in space and time, on the basis of a sequence of purely temporal data.

In your essay, you show yourself to be well aware of the charge that the so-called 'transcendental deduction' is merely a piece of analysis.


(1) X is the necessary condition of Y
(2) Y

Therefore, (3) X.

All 'X is the necessary condition of Y' says is that X is necessary for Y. Or, in other words, if Y then X. The argument now becomes a simple example of Modus Ponens, nothing 'transcendental' about it.

One suggestion is that if there is something special about a 'transcendental' argument that distinguishes it from a simple case of modus ponens, it has something to do with the role of possibility: X is the necessary condition for the *possibility* of Y.

But if that is the case, then how does the argument proceed?

(1') If Y is possible, then X
(2') Y

Therefore, (3') X

is an example of a logically invalid argument.

In that case we need to insert at least one extra step:

(1'') If Y is possible, then X
(2'') Y
(3'') Y is possible (from 2'')

Therefore, (4'') X.

With the aid of an axiom from modal logic (If P, then it is possible that P) the argument is now formally valid.

The question now becomes, what makes the third argument interestingly different from the first argument. What *is* so special about a transcendental argument?

Your example of music suggests that what is doing the work is something to do with experience, rather than the mere logical form of the argument. What to do with experience? Might there be transcendental arguments that concern, not experience, but some other concept? Meaning, perhaps? (I am thinking of Wittgenstein's later philosophy, in particular the private language argument.) What is so special about the concepts of experience or meaning that makes them suitable for conducting transcendental arguments?

Suppose someone said, 'All you are doing is analysing the concept of experience (analysing the concept of meaning).' In that case, we are not dealing, as Kant thought, with a synthetic a priori judgement but with a mere analytic judgement. 'If we have experience (we use words with meaning), then Y. We have experience (we use words with meaning). Therefore Y.'

I find your response inadequate here. What you write can be read as saying that we are not dealing any particular experience (any particular piece of music) but experience as such. So the critic will reply that what Kant's transcendental argument does is analyse the concept of experience as such. That's all a transcendental argument amounts to.

- Maybe we shouldn't after all be too bothered about the difference between analytic and synthetic. Didn't Quine once write a rather famous paper to that effect?

The important thing is that we learn something very special from investigating the concept of experience.

It should hardly be necessary to add that a deduction about experience as such is different from an empirical deduction, for example of the bee and the hive.

However, the critique of Locke does suggest that what is at stake here is a special kind of concept that in some way yet to be explained relates to experience but is not derived from experience: that's what the transcendental deduction is really about.

- This looks like a promising strategy. I'll be interested to see how the argument proceeds from this point.

All the best,