| Abstract: | The learning efficiency of an Atlantic bottlenose dolphin was evaluated using auditory discrimination learning-set tasks. Efficiency, as measured by the probability of a correct response on Trial 2 of a new discrete-trial, two-choice auditory discrimination problem, reached levels comparable to those attained by advanced species of nonhuman primates. Runs of errorless problems in some cases rivaled those reported for individual rhesus monkeys in visual discrimination learning-set tasks. This level of stimulus control of responses to new auditory discriminanda was attained through (a) the development of a sequential within-trial method for presentation of a pair of auditory discriminanda; (b) the extensive use of fading methods to train initial discriminations, followed by the fadeout of the use of fading; (c) the development of listening behavior through control of the animal's responses during projection of the auditory discriminanda; and (d) the use of highly discriminable auditory stimuli, by applying results of a parametric evaluation of discriminability of selected acoustic variables. Learning efficiency was tested using a cueing method on Trial 1 of each new discrimination, to allow the animal to identify the positive stimulus before its response. Efficiency was also tested with the more common blind baiting method, in which the Trial 1 response was reinforced on only a random half of the problems. Efficiency was high for both methods. The overall results were generally in keeping with exceptations of learning capacity based on the large size and high degree of cortical complexity of the brain of the bottlenose dolphin. |
