[21] _Psychology_, II., 348 et seq. James, however, recalls the case of another dog accustomed to find and carry wedges for splitting wood. One day he did not return. After half an hour they looked for him; he was biting and tearing at the handle of a hatchet stuck in a block (the wedge was not forthcoming). Had this animal clear perception of the common character of the two instruments used for splitting? “This interpretation is possible, but it seems to me far to transcend the limits of ordinary canine abstraction.” (_Loc. cit._, p. 352.) James attempts another explanation. It is singular that he does not invoke training, and association with man: that this is an influential factor in the intellectual development of animals cannot be doubted. It is advisable to adduce exclusively their spontaneous inventions, with no possible suggestion: such facts alone are clear and convincing. [22] Lloyd Morgan, whose tendencies have already been indicated, distinguishes three sorts of inferences: (1) unconscious inference on immediate construction (perceptual); (2) intelligent inference (conceded to animals), dealing with constructs and reconstructs (perceptual); and (3) rational inference, implying analysis and isolation (conceptual). (_Op. cit._, p. 362.) [23] _Nouveaux Essais_, Book III., Chapter I. [24] Cf. Taine, _L’Intelligence_, Vol. I., Book I., Chapter II., Part 2, Note 1. Preyer, _Die Seele des Kindes_, Chapter XVI. [25] Romanes, _Mental Evolution in Man_, p. 283. [26] B. Pérez, _op. cit._, 210. [27] Houzeau, _op. cit._, II., 202. [28] Cf. _Revue Philosophique_, July, 1890.