<body class="blog" onload="">

Ratiocination

The Transcendental Deduction


I here articulate an interpretation of Kant’s famous Transcendental Deduction. My goal shall be to sketch a minimally plausible argument that can be attributed to Kant (with some minimum degree of plausibility). There are two dangers when engaging in this project: on the one hand, I could give a textually faithful rendition of Kant's argument on which the argument turns out to be radically implausible or obviously invalid. On the other hand, I could state an argument which just might be in the text and that's also valid and minimally plausible. I've erred in the direction of this latter danger (as I always try to do when engaging in the history of philosophy--why study arguments that aren't valid or at least minimally plausible?)

I shall first state Kant’s theories of analyticity and transcendental idealism. On my reading, the Transcendental Deduction is a modal argument for the latter.

The Synthetic and Analytic

For Kant, it is acts of judgment which are analytic or synthetic (not the abstract propositions of contemporary metaphysics nor the sentence types or tokens that express them). The things eligible for analyticity and syntheticity are event-like objects--things that happen or occur. It remains unclear whether Kant has in mind event types (so-called `Chisholm events’) or event tokens (so-called `Kim events’).

An analytic judgment occurs iff a connection of subject and predicate is `thought through identity,’ while a synthetic judgment occurs iff a connection of subject and predicate is thought, but not through identity.

(I shall not discuss Kant’s anachronistic theory of a prioricity here, but simply note that he thinks synthetic a priori knowledge is possible. His pet example of this is the judgment that `everything that happens has a cause.’)

Transcendental Idealism

On Kant’s taxonomy, empirical idealism is the conjunction of two theses:

a) The mind is directly acquainted only with objects as they appear to us (roughly: Lockean ideas or medieval phantasms).

b) There are no objects in themselves (roughly: Lockean substrata).

Berkeley accepts both (a) and (b); Hume accepts (a) and (sometimes!) seems to accept (b). Locke accepts (a) and rejects (b); Aristotle, as an empirical realist rejects both (a) and (b).

Transcendental idealists, on the other hand, reject (a) and (b), but not in the manner of Aristotle.

Transcendental Idealism is the thesis that there are subjective conditions grounding the possibility of intelligible experience (and more specifically, the intelligible experience of synthetic a priori knowledge)--and that these conditions obtain. An experience is intelligible iff (roughly) we can make sense of it, which entails among other things that it has a discernible ordering. Subjective conditions here are features about a subject that make possible such experience. It’s not just that these features are necessary conditions of such experience; rather, such experience is possible in virtue of these features. The relation is one of both dependence and entailment.

The Transcendental Deduction

Kant’s Transcendental Deduction is both one and many. One: it’s an argument schema or general strategy of argument. Many: Kant advances many instances of the schema, letting the variables in the schema take as their values different varieties of intelligible experience and different varieties of subjective conditions.

And in fact, one can read Kant as employing two schemata in his Transcendental Deduction. The first is aimed at empirical idealists and empirical realists who accept that there are such things as intelligible experiences. The second is aimed at the skeptic who grants that such intelligible experiences are possible, but withholds judgment as to their actuality. I shall concentrate on the first schema in the sequel.

The general strategy of the Transcendental Deduction is to identify some experience and then show some necessary conditions that ground this experience. Given that experience happens, it follows then that the necessary conditions obtain. QED.

Finally, the Transcendental Deduction is a modal argument, employing the notions of possibility and necessity.

In the sequel, let x be a kind of intelligible experience, let p be transcendental idealism with respect to x, and let q be the thesis that someone (a thinker like us) has an intelligible experience of kind x.

Schema the First:

1. Necessarily, if not-p, then not-q.
2. q
3. p (from 1 and 2, modus tollens/double negation)

Note that this argument has a strong premise; it assumes that intelligible experience of a particular sort is actual. The skeptic, thus, won’t be convinced by any argument of this form, since she withholds judgment about its second premise. Kant has another argument schema in mind. This one will employ premise one of the first schema but won’t assume that intelligible experience actually occurs. More on that later.

Kant has an argument for the first premise. And it goes something like this: either the necessary grounds of intelligible experience are (i) only in objects themselves, (ii) only in objects as they appear to us, or (iii) in us. But not (i), since we can’t know objects in themselves (one might worry whether Kant begs the question here against the empirical realist). And not (ii), since the thought of appearances without objects involves contradiction (there wouldn’t be any objects to appear to us). Therefore (iii). Kant thinks it follows (from the definition of `grounds’ and the above trilemma) that premise (1) is true. That is: necessarily, we’ve got intelligible experience only if the grounds for those experiences are in us. Equivalently: if these grounds don’t obtain, we can’t have intelligible experience.

I shall now present (informally) an instance of the above argument schema. Let x takes as its value experience of cause and effect. So we get something like this:

If minds like ours have intelligible experiences of ordered appearances where one seems to us to inevitably produce the other, there are grounds making possible such experience. These grounds couldn’t merely be in the appearances of objects to us; for as Hume taught us, we cannot infer from a mere series of appearances that there are necessary connections between them. Nor could these grounds be in objects themselves, since we’ve got no access to those. So these grounds must be in us.

What are these grounds? Call them `categories.' They’re structures (roughly: the interpretive schema of contemporary evolutionary psychology) that filter and shape the manifold of intuition and experience. It’s only because our minds are structured in this way that intelligible experience of cause and effect happens.

We can repeat this procedure as often as we like, replacing experience of cause and effect with substance, time, space, and the like. I take it that the conjunction of the conclusions of all these mini-arguments will itself be the totality of Transcendental Idealism.

Now tie this strategy into the overall Critical Project. Kant is particularly interested in synthetic a priori knowledge since such knowledge grounds the theoretical sciences (especially physics). So when x takes as its value some kind of synthetic a priori knowledge, the Transcendental Deduction will help us see the grounds of such knowledge. And in giving us this, Kant has given us a sure foundation for the theoretical sciences. The foundation is not only in appearances, and it’s not only in objects themselves. It’s in us. This is the Copernican Revolution Kant is fond of referring to; to show that what matters is not whether our knowledge conforms to objects, but rather whether objects conform to our knowledge (better: to the faculties whereby we know).

I have presented one iteration of the Transcendental Deduction. Here is the second I have alluded to, though I won’t comment much on it:

Schema the Second:

1. Necessarily, if not-p then not-q.
2. Possibly, q.
3. Possibly, p (from 1 and 2)
4. If possibly p then p.
5. Therefore, p (from 3 and 4, modus ponens)

The second schema employs a weaker second premise than the first schema did. To maintain validity, however, Kant must add another premise to the argument--premise four. Kant’s argument for premise four is something like this (I leave it to my reader to judge whether this argument commits any modal fallacies):

If it’s possible that some minds (like ours, at least) exhibit a certain structure making possible intelligible experience, then any minds (like ours) must exhibit this structure. There’s just one way minds could be that could make minds able to have intelligible experiences. And if it must be the case that minds exhibit this structure, then our minds exhibit it actually.

3 Comments:

Add your own


Noumena at 8:24 AM  said... On my reading, the Transcendental Deduction is a modal argument for the latter.

Oh dear. I don't think Kant would accept any kind of (metaphysical) modal argument for anything.

His pet example of ... [a synthetic a priori judgement] is the judgment that `everything that happens has a cause.’

This looks more like a sentence expressing a proposition than an event or an object like an event. Maybe an easy fix here would be to say that a judgement can be expressed as a sentence with an implied `I think [that]'. That would go along with one of the key steps in the Deduction, that an `I think [that]' can always be attached to any judgement.

It’s not just that these features are necessary conditions of such experience; rather, such experience is possible in virtue of these features. The relation is one of both dependence and entailment.

This sounds like it's on the right track. But which directions do the (one?) relation of dependence and entailment go?

And I'm afraid you might be off track about the purpose of the Deduction. Kant's goal is to show that, for all beings like us, the Categories apply to sensation -- that our empirical experiences are structured by the Categories, and not just Space and Time. He doesn't need to show that they're structured by Space and Time (Transcendental Idealism for Space and Time) because he accomplished that (or at least he think he did) in the Aesthetic.

Also, you seem to be treating the Categories as metaphysical principles, like the causal principle above. And you call them concepts. And then I winced as you related them to evolutionary psychology. But they're none of these things. They're forms of judgements -- the rule-patterns of valid syllogisms in ordinary logic. In the case of causation, the form of judgement is the hypothetical syllogism (modus ponens). Kant's goal, then, is to show that our empirical experiences are structured by the ordinary laws of logic. Roughly, that our empirical experiences obey the ordinary laws of logic. It's not so much a metaphysical claim as a logical one. Metaphysical principles like universal causation don't show up until the Categories have been `schematized', which presupposes the conclusion of the Transcendental Deduction.

Here's a way to get at the problem: Imagine you're a Humean being. You're constantly having all these sense experiences, and you'd like to make predictions about future experiences. How do you know that you'll never get things wrong using modus ponens to make these predictions? Since modus ponens is just an empirical generalisation, you don't, not in any strong sense of `know'.

Kant's trying to solve the same problem by slightly modifying the setup. Now you don't just have the sense experiences -- you also have these cognitive structures in place. Modus ponens isn't just an empirical generalisation. It's part of the way your mind came wired. But it only automatically applies when you're reasoning, using your mind discursively. It doesn't automatically apply -- at least, not without an argument -- when you're using your mind to process sense experience.

Finally, the structure of the arguments of the Deduction aren't modal. They're proto-phenomenological, concerning the way in which we cognitively collect and arrange experiences into objects. In particular, neither of your argument schema mention the Transcendental Unity of Apperception, which is the pivot around which the whole B Deduction turns. Unfortunately it's been too long for me to recall any details here. Fortunately, what you have is probably completely sufficient to get a passing grade in the extraordinarily unlikely event that you're asked to explain the Deduction.



Andrew Bailey at 8:39 AM  said... Dan, thanks for the comments. They'll be helpful if I get a Kant question on my comps next week. =)

And incidentally, I think the arguments have *got* to be interpreted as modal ones if we take seriously language like `possibility of experience.' Whether this modality is a thick metaphysical one (or something else) is another matter.



Anonymous at 7:23 PM  said... "Possible experience" is not a modal notion. The notion of possibility at work here is "transcendental" possibility rather than logical possibility. That is, transcendental possibility concerns experience which is able to occur for us as human beings (equipped with the categories and the pure forms of intuition: space and time). This same transcendental/logical distinction occurs in the context of "necessity" as well in Kant. Sometimes when Kant uses "necessary," he means logically necessary. For example, analytic statements are logically necessary. But usually when he uses "necessary" he means "transcendentally necessary," that is, necessary for us as human beings (equipped with the categories and the pure forms of intuition: space and time). For example, synthetic a priori judgements are transcendentally necessary, but not logically necessary. They can't be logically necessary, or else they would be analytic.