Attention, the presolution period, and choice accuracy in pigeons.

Six pigeons were trained first on a color then on a form discrimination; four other pigeons were trained first on form and then on color. One of two colors or one of two forms (sample stimuli) appeared in the center of a touch sensitive monitor for 5 pigeons and in the center and in 16 other locations for 5 other pigeons. A peck anywhere within the region in which the sample stimuli appeared produced two white disks (comparison stimuli), one on the left and one on the lower right corners of the screen. Correct left-right choices provided food. Although of no consequence, the location of pecks in presence of the sample was predictive of the pigeon's subsequent choice. Accuracy, choice of the correct comparison stimulus, was greater when the sample stimuli appeared in the center as well as 16 other locations than when it appeared only in the center. The presolution period, the period of chance accuracy prior to evidence of discrimination learning, was decreased on each task following training on the other task. This evidence of facilitation following an extra-dimensional shift was attributed to continued relevance of the conditions under which the first task was learned. The duration of the presolution period was inversely related to asymptotic accuracy-data accounted for by Heinemann's (1983) theory of information processing during the presolution period.

Processing of conflicting and redundant stimulus information by pigeons.

Selective attention in processing of visual information by pigeons, trained on alternating sessions with two colors (red and green) and two forms (a diamond and an X shape) differentially associated with a left-right key choice task, was examined. A color and a form were presented together on probe trials during sessions in which, on other trials, only one of the dimensions, color or form, was shown. The dimension in effect on the surrounding trials had no influence on choice when the information provided by the two dimensions on probe trials was in conflict--color correct for one choice and form for the other. When both color and form redundantly cued the correct choice, there was no increase in accuracy in comparison with that associated with one dimension. Following separate training on the color and form discriminations, pigeons appeared to base their choices on color on some trials, on form on other trials, but not on both simultaneously. These findings are discussed in terms of an exemplar model of information processing.

Effects of identical context on visual pattern recognition by pigeons.

The effects of identical context on pattern recognition by pigeons for outline drawings of faces were investigated by training pigeons to identify (Experiment 1) and categorize (Experiment 2) these stimuli according to the orientation of the mouth-an upright U shape representing a smiling mouth or an inverted U shape representing a sad mouth. These target stimuli were presented alone (Pair 1), with three dots in a triangular orientation to represent a nose and eyes (Pair 2), and with the face pattern surrounded by an oval (Pair 3). In Experiment 1, the pigeons trained with Pair 1 were most accurate, those trained with Pair 2 were less so, and those trained with Pair 3 failed to acquire the discrimination within eighty 100-trial sessions. The same ordering was found in Experiment 2 for pigeons trained on the three pairs simultaneously. The authors suggest that a contrasting finding in humans, the face superiority effect, might be due to a gain in discriminability resulting from recognition of the pattern as a face. An exemplar model of information processing that excludes linguistic coding accounts for the present results.