A Note on the Logic of Transcendental Claims
A conversation yesterday (ostensibly about Heidegger) prompted me to clarify my thoughts about the logic of transcendental claims. Take, for example, the claim that the unity of consciousness requires objective experience. Let us pretend for the moment that we know what both ‘unity of consciousness’ and ‘objective experience’ mean. One natural way to gloss this claim is as follows: It is a necessary truth that any subject that enjoys unity of consciousness also enjoys objective experience. We might then formalise this
□Ax (Ux→Ox) {forgive my quantifiers, I haven't worked them out for this font yet}
Now a transcendental claim of this form has the following features: It is a necessary truth, it is universally quantified, it is synthetic and a priori (!), and just to top things off it has a conditional in the middle.
Ignoring standard qualms about quantified modal logic, the synthetic a priori, and the demand that we explain how to interpret the conditional, there are at least two distinct ways in which such claims can be challenged:
It might be suggested that we are not justified in accepting the claim as necessary.
It might be suggested that we are not justified in accepting the claim as universal.
To see this, suppose we were to become convinced that it was possible that someone could enjoy unity of consciousness without objective experience
◊ Ex(Ux & ~Ox)
How might we respond? Well, we could just dump the whole transcendental claim and give up. But that might seem overly drastic.
Another move is to drop the necessity operator, either entirely or replace it with ‘It is physically necessary that...’. The first option turns the transcendental claim into a contingently true generalisation. The second into a physical/nomic necessity. The second option is (i) only available if we think that the putative counterexample involves an infraction of the laws of nature, and (ii) we can plausibly maintain that this truth really does follow from the laws – for we have an idea of what an epistemology of nomic necessities looks like (science). So on either option, it seems that the a priority of the transcendental claim would have to go. [Then again, might transcendental claims be understood as contingent a priori truths? One would have to look for similarities with the best Kripke-style examples of such things. I need to think about this].
However, an alternative move is to restrict the quantifiers in such a way as to disqualify the putative counterexample. So, we might say that the domain is restricted to beings like us in respect R. In this case, we would still have a necessary truth and knowledge of it might still be a priori. The difficulty with this move would be that we would need to specify what respect R is in such a way to prevent the transcendental claim looking like a trivial consequence of an ad hoc move. Still, that might be preferable to the earlier option of turning transcendental claims into contingent generalisations or physical necessities.
What if we were to discover an actual counterexample [Ex (Ux & ~Ox)]? Well, the first option (dumping/changing the necessity operator) is not an option, but the second remains a possibility. I’m thinking of Beatrice’s interpretation of Foucault as putting forward transcendental claims limited to contemporary westerners. Of course this would have nothing like the strength of the original claims, but might still be of some interest.
posted by Joel at 1:15 pm
4 Comments:
Just a quick comment. I think obviously the way to go would be to restrict the quantifiers in transcendental claims. My own favoured way of doing this would be to restrict them to all x's of whom certain 'conceptual truths' hold. An example of a conceptual truth would be that a discursive subject can only differentiate properties if she plots them onto a unified heterogeneous domain of presentation.(This conceptual truth is due to Marks Sacks, see Chapter 7 of Objectivity and Insight, although I have modified the details a little. It is worth pointing out that Mark does not refer to conceptual truths.) Of course we would have to justify why this is a conceptual truth. But if we could do that, it would be a priori true that property differentiation depended on the maintenance of such a domain. We would then make the claim that a unified heterogeneous domain can only be maintained if a discursive subject has a perception of permanence. This would be a further conceptual truth. Finally, we would come to the 'ampliative' or synthetic a priori move. This is obviously where the threat of contingency enters the proof. The claim would be that only a perception of a unified domain of spatio-temporal objects would provide the perception of permanence which is necessitated by the conceptual truths just mentioned. So the quantifiers in a transcendental claim would be restricted to x's of whom the relevant conceptual truths held, and to possible worlds in which the synthetic fact upon which the obtaining of the former depends, is the only plausible candidate.
I don't see this as overly problematic. Restricting quantification to conceptual truths about discursive subjects is not too harmful to the universality of transcendental claims, viz. any future human being will minimally at least be discursive. And restricting quantification to certain synthetic facts does not seem ad hoc, i.e. possible worlds where physical objects are non-spatio-temporal are extremely distant if intelligible at all.
One final thought. Joel mentioned we would have to explain the relation of implication appealed to in transcendental arguments. How about doing that in terms of supervenience? The latter is after all a relation that naturalists appeal to all the time, and we are, after all, trying to convince naturalists of the cogency of transcendental claims.
Just a comment on my comment: obviously we would have to show that the perception of permanence necessary for the maintenance of a unified heterogeneous domain of presentation could only be provided by external objects. This part of the proof might prove quite tricky.
"An example of a conceptual truth would be that a discursive subject can only differentiate properties if she plots them onto a unified heterogeneous domain of presentation"
Of course, this claim itself has the same logical form as the original claim that unity of consciousness requires objective experience. Now, limiting the domain of entities to discursive subjects seems fine, and it is pretty clear that a Kantian should do at least this. But limiting the domain to discursive subjects for whom it is true that they "can only differentiate properties if plotted onto a unified heterogeneous domain of presentation", seems to put the cart before the horse as that is one of the things that needs to be shown. That is, to limit our investigation to just those subjects is to assume too much and make the claim very weak. For it is consistent with the suggestion that a subject could be discursive and yet not satisfy such a condition.
So, it seems to me, we are left with the task of having to make good a claim about all (meaning ALL) discursive subjects. I don't mean to express here any scepticism with regards to this possibility. All that I wanted to point out in the original post was that, supposing we are faced with a compelling case that the notion of discursive subjectivity doesn't necessitate X (our favourite transcendental condition), we don't necessarily need to abandon our condition. Rather we can limit our claim, by restricting the domain, whilst keeping it (potentially) necessary and a priori. And, as you point out, this is what we should do, since the other options (moving to contingent generalisations or physically necessary truths) look profoundly un-transcendental.
The problem shifts, then, to justifying our limitation of the domain. Building too much into this limitation seems to give too much to the oponent, and can look ad hoc.
Yes, you're right, the claim that a discursive subject "can only differentiate properties if she plots them onto a unified heterogeneous domain of presentation," is itself a universal claim. But if the very idea of a discursive subject could be said to entail such a truth (which is why I called it a 'conceptual truth'), then I wouldn't see relativizing a transcendental claim to the domain of subjects so conceived as either unnecessarily restrictive nor ad hoc. But obviously work would have to be done to clarify why the idea of discursive subjectivity entails such a conceptual truth (and what exactly 'entail' means here).
I saw the point at which the proof would be in danger of making ad hoc moves to be when it asserts that certain synthetic facts make possible the alleged conceptual truths. Hopefully the moves up until that point would only involve a priori reasoning about what it means to be a discursive subject.
Post a Comment
<< Home