An Experimental Analysis of the Stimulus-Reinforcer Relations in Matching-to-Sample

This study investigated the stimulus-reinforcer relations possible in matching-to-sample, and determined the extent that each relation contributed in the transfer to subsequent matching-to-sample. Fifty three homing pigeons were autoshaped on a single key, with two stimuli, red and green. Forty eight subjects were then divided equally in 12 groups receiving the experimental treatments, and five subjects served as no-treatment controls. Group assignments were first based on autoshaping to either an identity, non-identity, or identity and non-identity stimulus relation. Each stimulus relation consisted of a center stimulus followed by an outer key stimulus. For identity, the center and outer stimuli were the same. For non-identity the center and outer stimuli were different, The second aspect determining group divisions was the correlation of reinforcement to the stimulus relation; explicitly paired, explicitly unpaired, truly random, or omission (identity only). The subjects received 10 sessions of autoshaping where each session consisted of 50 exposures to the stimulus sequence correlated with the reinforcement. Subjects were then transferred to an operant task--two color zero-delay matching-to-sample. In matching-to-sample training, the 6-second sample was followed by the presentation of 2 outer key comparison stimuli. The comparisons remained illuminated for 6 seconds or until a response occurred. A response to the matching comparison was reinforced and a response to the non-matching comparison resulted in the inter-trial interval. Sessions consisted of 50 trials. The results from the single stimulus autoshaping sessions revealed that all subjects acquired reliable autoshaping within 150 stimulus-food pairings. The results from the autoshaping sessions revealed that stimulus sequences explicitly paired with food produced reliable responding to both the center and outer stimuli.. No significant differences were observed between center and outer key responding for subjects trained with an explicitly paired food arrangement. Subjects receiving identity stimuli explicitly paired with food and non-identity stimuli explicitly unpaired with food responded differentially to the outer key stimuli, showing control by the identity and non-identity stimulus relations. The truly random, explicitly paired/omission, or explicitly unpaired conditions resulted in center and outer key responses decreasing across sessions. The results from the matching sessions revealed that only autoshaping discrimination training (identity explicitly paired and non-identity explicitly unpaired with food) facilitated transfer to the operant task. Autoshaping training on a single stimulus relation in matching-to-sample (identity or non-identity) did not facilitate matching transfer. The results from the omission training suggest that the key pecking was very susceptible to operant controls. These data were taken to indicate that exposure to and differential responding in the presence of the conditional discriminative stimuli in matching-to-sample may be necessary for matching performance. Exposure and behaving in the presence of components of matching did not produce matching. Therefore any explanation of matching in terms of learning a single stimulus relation rule may be questioned