A number of studies have revealed that animals often show marked spontaneous color preferences when responding behaviorally to chromatic stimuli (lise, 1941;Tinbergen & Perdeck, 1950;Curtius, 1954;Hess, 1956;Muntz, 1962; Kear, 1964;Quine & Cullen, 1964; Hailman, 1967). The preference patterns in most cases cannot be related in any simple way to the spectral sensitivity of the receptors (Thompson, in press) and in a given species they appear to vary with the type of behavior the animal responds to the colored stimuli. The response and the color preference shown in such cases depends on the motivational state the animal is in when facing the stimuli (Impekoven, in press) and on the stimulus context in which coler is presented (G. Thompson, unpublished). The experiment reported here was aimed at establishing motivation-dependent color preferences in the pigeon, an animal which lends itself to further experimental manipulation including electrophysiological recording. Hunger and thirst were used as suitably controllable motivational variables. 2 APPARATUS The apparatus used was a brightly lit, octagonal Skinner box with eight transparent pigeon pecking keys, one on each wall. Exchangeable stimulus cards could be slid into these keys. Rewards were offered at the center of the box floor and were either grains delivered with a hopper for 4 sec, or water injected into a well by a pump and available for 4 sec. The box was programmed using conventional relay equipment so that pecks to each key were counted separately on eight counters, but a peck to any of the keys produced a reward. While the reward was being offered, however, no further pecks were counted, this ensuring that if the pigeons gave a burst of pecks, as they sometimes did, to a key before going to the center of the box for the reward, only the first response registered so that each response recorded was an independent key choice. Another counter shut down the trial after a preset total of 50 choices had been reached. METHODS Three pigeons (Columba IMa, var. domestica) of mixed breeds and from various sources, were used in the experiment. After having been gradually deprived to 80 per cent of their normal weight they were pretrained for 10 to 15 trials with neutral grey stimulus cards, by which tinle they pecked indiscriminately any of the keys for grain hopper access. The key cards were then replaced by a set of eight color cards: purple, red, yellow. orange, yellow-green, green, green-blue, and blue, and after four pretrials in which the animals settled down to responding in this situation they were run fo~ two blocks of eight trials each. Within each block the colored cards were randomly allocated one by one to each and all keys, the animals being maintained at 80 per cent body weight throughout the experiment. After eight days recovery in which they regained their full weight, they were deprived of water for 18 h before each trial, while having ad lib food, and after four pretrials run for two blocks of trials replicating the