Godfrey you are asking the question that leads me to my own conclusion: How are all three legs of the canon supposed to work and what makes them canonical?
I've always come to the conclusion that in order for the legs of the canon to serve as criteria of truth, they had to function "automatically" without the input of reason/opinion. Therefore I have always rejected the view that anticipations could be "concepts," because in my view that creates a feedback loop. If the opinion we form after experience becomes part of our standard of truth, then that just doesn't work if the main feature of the canon is that it is pre-rational.
Now in my mind there is a possibility that the anticipation faculty is some kind of "organizational" capacity that can be made sharper over time, just as perhaps our ear for music or our ability to pick out detail in sight might improve with experience. But that would just be improvement in the working of a non-rational faculty, and if you consider concepts like "ox" to be subjects of anticipations, then in my mind that's a non-starter. "Ox" is a human-developed category of living things summarized in a particular word "ox," and it's going to be a matter of opinion where the dividing line between an ox and a cow and a horse and sheep really lies.
So I think DeWitt is correct in ruling out the possibility of there being an "anticipation" of a concrete particular like an ox or Plato.
The process of deciding whether the thing headed toward us is an ox, or Plato, clearly does involve some kind of process in which the mind works to narrow down the possibilities and fit the data to a pattern we have developed over time, but at the point we're saying "that's an ox because it matches our definition of an ox," and at that point we are pretty far from what Epicurus was considering to be a faculty analogous to seeing and hearing.
Now being a lawyer I think I can take the other side of that argument. I can argue that, "Yes, since human experience isn't absolute and so much is relative to our perspective, then we should consider our previously-formed concepts and opinions to be a part of our canon of "truth." In saying that we would have to emphasize that "truth" is not absolute, so it's ok to incorporate own on reasoning conclusions as part of what we think is true.
And it seems that the "later" or "the Epicureans generally" did take that course, thereby creating a fourth leg of the standard of truth.
However DeWitt concludes that that was a big mistake, and I agree with DeWitt. Once you admit that the product of conceptual reasoning itself is a part of your measure of "truth," to me you are on the slippery slope to Platonic rationalism, because your holding the opinions of your own mind as equal in authority to the promptings of nature.
[Edit: I made some pretty significant revisions to clean up my poor typing and phrasing hopefully without changing the meaning.]
