The problem for a 'moderate capacity reading' is whether it can explain why the regress gets going at all. The problem for other readings is whether they can explain why it should stop.
First, let's look at the text. Here is the argument once again, and the two translations:
GREEK TEXTSmall points on the translations:
e1ti d' ei0 kai\ e9te/ra ei1h h( th~j (15) o1yewj ai1sqhsij, h2 ei0j a1peiron ei]sin h2 au)th& tij e1stai au(th~j: w3st' e0pi\ th~j prw&thj tou~to poihte/on.
And furthermore, if indeed the sense of sight were different [b15], then either it will go on to infinity or some sense will be of itself, [b16] so that we should do this in the case of the first [sense].
Further, if the perception of seeing is a different [perception], either this will proceed to infinity or some [perception] will be of itself; so that we ought to posit this in the first instance.
1. Presumably Caston should really want to have 'if the perception of the seeing'.
2. Also, Johansen's taking e0pi\ th~j prw&thj to refer back to 'sense' seems correct, so Caston might help himself to 'we ought to posit this in the case of the first [perception]'.
Now can the moderate capacity reading explain why a regress gets started? At the BACAP seminar, Kosman raised the worry that it couldn't. If we suppose--Kosman said-- that, if we actually perceive something, then we must actually perceive that we perceive, then an obvious regress results of actual perceptions. But why, he wondered, must there be an infinite regress of mere faculties or capacities?
Johansen, I thought, had an elegant response. A faculty or capacity, he said, is an explanatory construct in Aristotle's psychology. Roughly, if an activity of a certain sort is performed, or even if in principle we may perform it, then a capacity needs to be postulated to account for it (that is, assuming that considerations of economy cannot account for that activity in some other way, which is precisely at issue in DA 3.2). So then, assume that, for any perception, we are capable of perceiving that perception. Assume that this capacity always needs to be explained by a faculty distinct from that which is responsible for the original perception. (Assume also no 'loops'.) Then an infinite series of faculties needs to be postulated. As Johansen put it:
Note that Aristotle is not committed here to saying that any act of perception, pn, must itself be perceived; only that any act of perception may be perceived, and if it is to be so perceived, then ... it must be explained by the stipulation of a further sense.This seemed to me an entirely adequate way of handling the argument and Kosman's concerns.
I would simply add that the argument seems even more elegant if one adopts Osborne's interpretation. Suppose that when Aristotle begins the passage with, 'since we perceive that we see and hear', what he means is (as I think): that we are seeing or hearing (rather than not) is something that we sense as well. What he is supposing then is: no faculty of sense, without the capacity to monitor whether that sense is operative. His concern in the passage is then over whether one can intelligibly attribute this capacity to the sense itself. Suppose one cannot. Then another sense needs to be postulated. But there is no faculty of sense, without the capacity to monitor whether that sense is operative, and then, using the same idea that this capacity must be attributed to another sense, we are off on an infinite regress.
In either case, it seems clear that a 'moderate capacity' reading can give a good account of the regress argument, its motivation and form.