I said earlier that the difficulty for the capacity reading was explaining how the regress in DA 3.2 got started, and the difficulty for the activity reading was explaining how it stopped. Let me explain the latter. It has to do with whether the activity reading can account for Aristotle's "we ought to posit this in the first instance" ( e0pi\ th~j prw&thj tou~to poihte/on).
Consider the following argument.
Everything chosen, is chosen for the sake of something. Something is chosen either for the sake of something else that is chosen, or for the sake of itself. Suppose an action is chosen for the sake of something else that is chosen. Either this proceeds to infinity, or some action is chosen for its own sake. We ought to posit, then, that each action is chosen for its own sake.The conclusion doesn't follow. It doesn't follow, from the fact that 'chains' involving choices must terminate in something chosen for its own sake, that all such chains are just one link long.
A similar difficulty affects the activity reading, in its construal of the regress argument. I'll state the activity reading using a distinction between, as I shall call it, perceptions and perceivings. A perception is a particular act; a perceiving is a relation involved in that act. According to Caston, Aristotle's view is that each perception of the sense of sight involves two perceivings: a perceiving of a color, and perceiving of itself (of that perception itself). Using this language, we can state the activity reading's construal of the regress argument in the following way:
For each perception, there is a perceiving of that perception. That perceiving is accomplished either by some other perception, or by that perception itself. Suppose the perceiving is accomplished by some other perception. Then either this proceeds to infinity, or there is some perception, the perceiving of which is accomplished by that perception itself. We ought to posit, then, that the perceiving of each perception is accomplished by that perception itself.The conclusion no more follows in this case than in the analogous argument given above. It doesn't follow, from the fact that any 'chain' involving perceiving that one perceives, must terminate in a perception which is a perceiving of itself, that all such chains involve no more than one perception. On the activity reading, then, when Aristotle says "we ought to posit this in the first instance" ( e0pi\ th~j prw&thj tou~to poihte/on), he's committing a fairly gross fallacy.
This difficulty is effectively conceded by Caston. Near the end of his article, he considers the objection of whether a regress doesn't break out among perceivings ('types', he calls them), if not among perceptions ('tokens', he calls them). Caston replies that:
the premiss used for the earlier regress, namely,
B. Whenever we have a perception, we have a perception of that perception.
is satisfied even if every perception were to instantiate only two contents, at the first- and second-orders. (797)
That is to say (using the language of 'perceptions' and 'perceivings', as above) the premise that There is a perceiving of every perception (as we put it) is not contradicted (it's 'satisfied') if it's true that The perceiving of each perception is accomplished by that perception itself. But that's not to say that the latter is implied, given the former. Caston's fudge term is 'satisfied'. He needs to say that the view he favors is implied by the regress consideration, not simply that it 'satisfies' it.
A defender of the activity reading might say that Aristotle from the start has been presuming that there is no more than one level of reflection built into each perception, and that it's obvious that our experience contains no chains of perceptions of perceptions, involving more than one link. This is something clear from instrospection. But if this were obvious, no regress argument would be necessary to establish it. Nor could the regress argument establish it, since it would have no more force than that initial introspective observation.
On the activity reading, Aristotle is investigating the 'structure of consciousness'. He's not entitled, then, to say "we ought to posit this in the first instance", unless that's how things are. And if his reason for saying "we ought to posit this in the first instance" is in the end simply a conviction that that's how things are, then the regress argument becomes otiose.
Note that this difficulty does not affect the capacity reading. On that reading, as we have seen, Aristotle is investigating whether additional faculties need to be posited to account for the ability of a sense to monitor whether it is operative or not. Aristotle's "we ought to posit this in the first instance" is a reasonable appeal to parsimony, exactly as one would expect.
I (provisionally) take this to be a decisive consideration against the activity reading. Or does anyone disagree? Have I missed something (entirely possible, given the difficulty of the subject matter)?