Three groups of pigeons (n = 8) learned a successive discrimination, 538-nm S+ versus 576-nm S-, in one context, say , houselight and tone, and the reversal in another context, say , dark and noise. In Group 1, the single reversal paradigm was used ; for Group 2, the problems alternated daily; and for Group 3, they alternated minute by minute. Wavelength generalization gradients obtained in both contexts revealed conditional control by context, which was weakest in Group 3.Conditional discrimination learning refers to a situation in which the (positive or negative) value of a particular stimulus depends on the other stimulus or stimuli with which it is combined. Pigeons typically learn conditional discriminations in operant free-response situations only with great difficulty and after extensive training (Boneau & Honig, 1964;Richards, 1979). In a study by Thomas, McKelvie, Ranney , and Moye (1981), however, such learning was achieved very rapidly. Pigeons first acquired a successive wavelength discrimination and then its reversal, each in a different physical context (e.g. , houselight and tone , HL T, dark and noise, HL N). The subjects were then tested for generalization in each context, and every gradient peaked at the appropriate S + value, that is, that of the S + trained in that context.Both the procedure and the conditional stimuli used by Thomas et al. (1981) differ from those previously used , and which of these factors accounts for the results that were obtained needs to be determined. Thomas, McKelvie, and Mah (1985) found that the auditory component of HL T played no role in the effectiveness of the compound . They also showed that striped versus plain walls were effective conditional cues, whereas background key color did not gain conditional control over a line angle discrimination in the single reversal paradigm. They suggested that pigeons may use global visual cues to identify the place in which food is obtained and may be predisposed to associate such place cues with the learning that occurred in those locations.The present study investigates the temporal factors that may affect operant free-response conditional discrimination learning. With regard to the alternation of training conditions, the single reversal paradigm is at one end of a continuum. At the other end is the typical procedure in which the cues alternate on a minute-by-minute basis (cf. Boneau & Honig, 1964;Richards , 1979). Intermediate between these extremes is conditional serial discrimination reversal learning (CSDR ) as used by Thomas and McKelvie (1982). In their study, a wavelength discrimiRequests for reprintsshouldbe addressed to David R. Thomas, Department of Psychology, Campus Box 345, University of Colorado. Boulder. CO 80309. nation and its reversal were presented on alternate days, with a different line angle accompanying each problem. The birds learned the conditional relationship, and after approximately 30 days of training, they began each session by responding appropriately even before the first reinforcer had bee...