Echoic functional class formation by a bottlenose dolphin

Abstract

A bottlenose dolphin was trained to perform separate two-choice simple discrimination tasks using echolocation and vision. The targets were air-filled shapes constructed from PVC pipe. Targets varied in length and width but not depth, and were separated into classes of letters and numbers. Target echoes were measured, and classes were set up so that they could not be grouped on the basis of perceptual similarities between class members. The subject showed successful learning set formation (win-stay, lose-shift) in visual and echoic discrimination reversals consisting of specific target pairings. The subject also successfully grouped targets into classes as shown by performance in a discrimination reversal procedure involving sets of targets rather than individual pairings. The subject was unable to group the same targets using vision instead of echolocation. No evidence of bimodal performance enhancement or immediate cross-modal transfer was shown. The results suggest that dolphins can arbitrarily classify dissimilar targets using echolocation, but that performance can be independent of visual information. Successful cross-modal class formation likely results from high levels of experience associating echoic and visual stimuli, therefore is more likely to be shown in the context of a conditional discrimination rather than a simple discrimination task. [Work supported by ONR.]