But, rationally, the gap between the first three beliefs, and the last one, is relatively small. Purely on the basis of rationality, you ought to expect a smaller dropoff than we in fact see. Hence, most people in the first group are avoiding the repugnant conclusion for non-rational reasons. (Or believing in the first three, for non-rational reasons.)
Well, my comment from http://lesswrong.com/lw/1ww/undiscriminating_skepticism/1sek would probably be better here. I still dispute that argument, as I think this drop-off is justified, even for rationalists.