On the nature of non-responding in discrimination learning with and without errors

In human subjects, discrimination learning with errors results in active responding incompatible with the reinforced response. The direction of such incompatible behavior is opposite to that of the reinforced response. Responding occurs only during the stimulus correlated with extinction. The frequency of active non-responding is maximal shortly after the start of discrimination training (the time at which the frequency of errors decreases most rapidly) and approaches zero as discrimination training continues. The magnitude of behavioral contrast is not related systematically to the number of errors. Instead it is related directly to the frequency of active non-responding. Active non-responding appears to be motivated by the aversiveness of self-produced frustration, in the sense that active non-responding allows the subject to avoid the aversiveness of non-reinforced responding.     

The dimensional nature of same-different discrimination behavior in pigeons

We studied the dimensional nature of same-different discrimination behavior in pigeons. Birds first learned to discriminate between simultaneously presented displays of 16 identical items (Same arrays) and 16 nonidentical items (Different arrays), conditional on the color of the background. After discrimination mastery, we tested the birds with Mixture arrays comprising both identical and nonidentical items. Accuracy increased and reaction time decreased as the disparity in entropy (a measure of variability) between the arrays increased. As well, within each entropy disparity level, lower entropy values were more discriminable than higher entropy values. These results accord with a logarithmic relation between entropy and discriminative behavior and, thus, with the idea that the discrimination of Same from Different arrays follows Weber's Law.     

Conditional discrimination learning in the pigeon

Four pigeons received conditional discrimination training in which reinforcement contingencies were related to specific combinations of color and form, but were unrelated to either color or form considered separately. During discrete-trial training, each response in the presence of two of four color-form displays produced reinforcement and terminated the trial; responding to the other two displays was never reinforced, and each such response prolonged the particular trial on which it occurred. Subsequently, the subjects received multiple-schedule training in which responding to either of the displays previously associated with reinforcement was now reinforced on a variable-interval schedule, and extinction was the schedule again correlated with the other two displays. After differential responding to the stimuli was clearly evident, intensity of the combination displays was changed in subsequent training sessions. Complex stimulus control was generally maintained across variation in intensity, although there were temporary disruptions in performance associated with onset of some of the intensity changes. Finally, a component-stimulus test revealed considerably more responding to the forms than to the colors.     

The multifaceted nature of unsupervised category learning

A substantial portion of category-learning research has focused on one learning mode--namely, classification learning (a supervised learning mode). Subsequently, theories of category learning have focused on how the abstract structure of categories (i.e., the co-occurrence patterns of feature values) affects acquisition. Recent work in supervised learning has shown that a learner's interactions with the stimulus set also plays an important role in acquisition. The present study extends this work to unsupervised learning situations involving simple one-dimensional stimuli. The results suggest that categorization performance is a function of both learning mode (i.e., study conditions) and learning problem (i.e., category structure). Unsupervised learning, like supervised learning, appears to be multifaceted, with different learning modes best paired with certain learning problems.