Teleosemantical pseudoproblem?

Apparently a speaker at the phil. graduate conference last night wanted to try to dissolve a difficulty that arises from teleosemantical theory. This is the supposed problem of resolving an apparent conflict between a description at the teleosemantic level with a description at another (lower) level. The description would be of a function or process in an organism, e.g. a from detecting an object with certain properties and nabbing it with its tongue. The conflict seems to arise when the description is given as e.g. "detects small dark object, sticks out tongue" and "sees frog chow, tries to eat it"; and the situation is problematic when a non-frog-chow item is so nabbed. So for instance the frog detects a BB instead of a fly, its object detector functions properly, but it doesn't try to eat frog chow (because no frog chow is present). The speaker attempted to show that this is a pseudoproblem by using some ideas from R. Millikan, among others. I don't buy that account (maybe more on it later) but I do thing this is nothing but a confusion resulting in a "problem" for the teleosemanticist... But then I don't necessarily think teleosemantics has much going for it either...