Pigeons were trained on four matching-to-sample tasks with various schedule requirements in effect on the sample key. Differential sample-schedule requirements (a differentialreinforcement-of-low-rates of 3 sec in the presence of one sample and a fixed-ratio 16 in the presence of the other) produced rapid rates of acquisition that did not differ across tasks. Nondifferential sample-schedule requirements (fixed-ratio 1, fixed-ratio 16 or a differential-reinforcement-of-low-rates of 3 sec in the presence of both samples) produced slower rates of acquisition, which depended on the difficulty of the discriminations between samples and between comparisons. Patterns of stimulus and position preferences were influenced both by the comparison stimuli in each task and by the sample-schedule requirements. Detailed analyses of acquisition revealed frequent instances of complete differential sample control of comparison responding at intermediate levels of overall "accuracy". RespondingIn a conditional discrimination, the relationship between the discriminative stimuli and the reinforcement contingencies depends on the stimulus context in which they appear . The matching-to-sample procedure is a conditional discrimination procedure in which the subject is first presented with a sample stimulus, and then required to select the correct stimulus from a set (usually two) of comparisons. In identity matching, the correct choice is the comparison that is most like the sample. In nonidentity matching, the relation between (Ginsburg, 1957, referred to this as "amatching"; Cumming and Berryman, 1965, used the term "symbolic matching".) The terms "matching" and "matching to sample" in the present paper refer to the procedures described above and not to the subject's performance on those tasks.Carter and Eckerman (1975) compared the acquisition of identity and nonidentity matching in pigeons. Using two stimulus dimensions, hues and lines, they examined acquisition of hue-hue and line-line identity tasks and hueline and line-hue nonidentity tasks (where the first term designates the sample stimulus dimension and the second term the comparisons).

The matching-to-sample experimental procedure was altered by reinforcing the selection of the non-matching comparison hue rather than the matching stimulus. Six birds were trained with red, green, and blue alternatives and a simultaneous presentation of stimuli in which the sample was present at the same time as the choice stimuli. The acquisition functions began well above the chance level but displayed a very slow improvement thereafter, which was different from that shown under matching conditions. Transfer of the oddity performance was tested by substituting a yellow light whenever a blue stimulus had previously been programmed. The results from the transfer test are considered in terms of both a “coding hypothesis” and the stimulus rules which appear to govern the performance of the “oddity” task.

Six birds were trained to match-to-sample with red, green, and blue stimuli on a zero-delay procedure in which the sample stimulus is presented and then removed at the same time the choice stimuli are presented. The acquisition functions for zero-celay matching show it to be a more difficult task for pigeons than simultaneous matching where the choice response can be made with the sample present. After 42 sessions with red, green, and blue stimuli, yellow stimuli were substituted wherever the blue stimuli had appeared to test the transfer of the matching performance. The results from the transfer tests are considered in terms of a “coding hypothesis.”

Three groups of pigeons were given conditional discrimination training in which the number of standard stimuli was varied across groups. In the presence of each standard, a pigeon adjusted the comparison stimulus on a second key until the two keys matched. A report of this match (response on the first key) was reinforced. Transfer of the matching performance was investigated by adding new standards to the ones already available. All pigeons were exposed to two extinction sessions after 155 sessions of training. Rapidity of acquisition was inversely related to the number of standards presented. Generalization gradients derived from the several comparison stimuli showed that all pigeons reached a high level of accuracy in the presence of at least one standard, and some pigeons did so in the presence of as many as four of the six standards. There was no evidence of a systematic effect of extinction upon overall accuracy, or the individual generalization gradients. When a new standard was added, a given pigeon's performance (in terms of responding to the comparisons) was similar to performance in the presence of one of the old standards. However, the pigeons did not show evidence of confusion among the comparisons.

Pigeons were trained on a differential autoshaping procedure in which both components of two-stimulus sequences predicted delivery or nondelivery of food. All birds acquired the conditional discrimination. When the subjects were exposed to an extinction procedure, the stimuli maintained conditional control as long as the birds continued to peck the key. When a delay interval was imposed between the two components of a stimulus sequence using a titration procedure, the stimuli maintained conditional control up to delay values of 7 to 10 sec. These data are consistent with the view that the controlling stimuli in conditional discrimination situations are compounds of stimulus elements.Key words: conditional discriminations, autoshaping procedure, delay performance, pigeonsThe term "conditional discrimination" has been used to refer to situations in which the behavior exhibited in the presence of each of several discriminative stimuli varies with the value of an additional "superordinate" stimulus (Heinemann and Chase, 1970). As the term "discriminative stimulus" implies, this behavior has been studied predominantly within the instrumental paradigm. In Lashley's classic experiments (1938), for example, rats were rewarded for jumping to an erect, rather than an inverted triangle, when these triangles appeared on a black background, but were rewarded for jumping to the inverted triangle, rather than the erect one, when they appeared on a striped background. In the three-key matching-to-sample situation (e.g., Cumming and Berryman, 1965) the sample or standard stimulus is the "superordinate" stimulus that determines the relation between discriminative stimuli (the comparisons) and the reinforcement contingencies. A variety of organisms, including rats (Lashley, 1938), pigeons (Ginsburg, 1957); and monkeys (Nissen, Blum, and Blum, 1948) have all learned conditional discriminations when exposed to these operant procedures. Systematic studies of conditional performance in Pavlovian, response-independent procedures are less common, but a few clearcut cases exist. For example, Asratyan (1961) reported an experiment by Struchkov in which dogs were presented with pairings of both a buzzer with food and tactile stimulation with shock in one room, and the reverse pairings in another room. The dogs learned to salivate or lift their paws to the buzzer or the tactile stimulation depending on the room in which the stimuli were presented. Saavedra (1975), employing an eyelid conditioning procedure, found that rabbits can successfully acquire a conditional discrimination when exposed to stimulus pairs AC+, AD-, BC-, BD+. In both of these studies, as in the response-dependent procedures described above, no single stimulus in isolation was uniquely associated with reinforcement or with nonreinforcement; discriminative performance must arise from joint control by "subordinate" and "superordinate"