The paper first argues that under conditions of domestication animals are necessarily selected (incidentally or otherwise) for (1) responsitivity to a broad band of stimuli and (2) behavioral plasticity. The consequent tractability of domestic animals is contrasted with the stimulus-boundness and response stereotypy of the fixed action patterns observed in wild animal behavior. It is suggested that this difference accounts for the differential trainability of wolves and dogs. The second section of the paper presents observational evidence that although the wolf is not very amenable to instrumental conditioning, it possesses a highly developed capacity for observational learning. It is then noted that since observational learning requires recognition of means-ends relationships this conclusion is inconsistent with the claim that wolf behavior is largely instinct-bound. Finally, these conclusions are reconciled by hypothesizing that the wolf possesses a “duplex” information-processing system, a primitive “instinctual” system that mediates basic survival responses and a more recently acquired “cognitive” system that evolved as the wolt became a group hunter. Neurobehavioral and developmental comparisons of wolf and dog suggest that these two systems have become integrated into a single scheme in the course of the dog's domestication.