[27] Add that when an animal or primitive man has recognized that a certain object possesses a particular attribute, he often finds it difficult to recognize that simply analogous objects possess the same attribute. I was one day making a kitten run after a wooden ball as a dog would do; the ball struck it and hurt it; it cried out and I petted it and then wanted to begin playing once more; it would run willingly even after large stones when I threw them, but obstinately refused to run after the ball. So that it evidently conceived that the ball alone possessed the attribute of power to injure it; the kitten looked upon the ball, no doubt, with an evil eye, regarded it perhaps as an evil being who was unwilling to play; by a fault of generalization the kitten created for itself a sort of fetich which it did not adore indeed, but which it feared, and fear is a step toward adoration. Mr. Spencer himself admits in savages a certain inaptitude for generalization. This opinion, paradoxical as it may seem, is perhaps an important truth. If primitive intelligences, as M. Taine among others remarked, are especially prompt at noticing the superficial resemblances of things, that fact is not always a mark of genuine perspicacity, for the resemblance perceived between two sensations may be explicable less as an intelligent generalization than as a sort of confusion of the sensations themselves; if sensations are analogous or indistinct, they may naturally be mistaken for each other without any exercise whatsoever of intelligence. Thence the comparative insignificance of many examples taken from the case of language. True generalization seems to consist, more than anything else, in the reduction of facts to law; that is to say, in a conscious abstraction of differences, in a conscious recognition of the fundamental determinism which binds things up together and which precisely eludes both savages and animals.