Yaacov pulls a tu-quoque on the Classical Theorist.  O&S charge that Prototype Theory cannot handle conceptual combination--STRIPED APPLE, PET FISH, etc.--and they're right.  But they imply that the CT is better off in that regard, and Yaacov argues that it's (emphatically) not.  The intersection model doesn't work well for CT either: LUNCH BOX, DESK CHAIR, KITCHEN CABINET, BUS STATION, COFFEE TABLE, LOGIC BOOK,..., and my favorite, ELECTRICAL ENGINEER.
     Yaacov's complaint is presaged by O&S themselves, in fns 8 (p. 267) and 12 (p. 272); they say there are kinds of problems they "will not consider."  Their tone suggests that they're just being nice to PT, in that they could have gone on and bashed PT with those further kinds of counterexamples as well.  Their examples are GOOD COUNTERFEIT DOLLAR and SMALL GALAXY, for which, they point out, PT would give the wrong values.  O&S fail to notice that COUNTERFEIT DOLLAR alone is a counterexample, since a counterfeit dollar is not a dollar at all, not just not a good one.  Cf. FAKE FUR, IMITATION PEARLS.
     What these examples have in common with Yaacov's is that they are all nonintersective or nonconjunctive.  (I'll say a combination FG is intersective/conjunctive iff to be FG is just to be F and be G.)  There are many different types of nonconjunctive concept.  One type is generated by the nullifying modifiers such as "counterfeit," "fake," et al.  Another type involves so-called "attributive adjectives" such as "small."  A small G is not something that is small and is a G, but rather something that is a G and is small for a G --hence "small galaxy," "small moose," etc.  A third type involves functional modification:  A lunch box is a box used for [carrying] lunch, not something that is both a lunch and a box.  There are many more types here.  (A Paul Ziff example: Contrast OIL CAN, OIL LAMP, OIL PAINTING.)  Analogical predication, a favorite Yaacov topic, plays a role.
     Yaacov is right in contending that CT does no better by nonconjunctive combinations than does PT.  But I think he does not give O&S full credit.  For they're right to maintain that for conjunctive concepts, CT does better than PT.  And their examples are conjunctive: STRIPED APPLE, PET FISH.  This is still a victory for CT.
     It was a tiny bit self-serving of O&S to rule nonconjunctive combinations out of consideration, because as Yaacov has shown, CT does no better with them than does PT.  Also, Yaacov might be right to say that once we see how many conceptual combinations are nonconjunctive in the first place, O&S's victory over PT is small to the point of insignificance.   It remains to be seen, since maybe classical semantics will go on to do better by the nonconjunctive combinations than PT can, once it addresses them.  (For a survey of some of the nonconjunctive combinations and an excellent classical-semantic approach to them, see Romane Clark, "Concerning the Logic of Predicate Modifiers," Noûs 4 (1970).  Of course, that is about linguistic meanings, not directly about anything psychological.)