Transformation-invariant cues in the recognition of simple visual patterns

Abstract

Two alternative hypotheses about the discriminative cues in visual patterns were tested by comparing the speed with which Ss could apply two different classification rules for identifying a set of 16 simple dot patterns, consisting of four groups of four transformations. One rule required discrimination between groups of transformations, and another rule required discrimination between transfonnations within groups. The patterns within a transfonnation group were less similar in their positioning of dots than were patterns of the same transformation in different groups. Speed of identification, however, was more rapid for the discrimination between groups than for the discrimination of transformations within groups and was also invariant with respect to the specific transformations included in the same category under the between-groups classification rule. The discriminative cues in these patterns were thus indicated to be relationships among dots that remained invariant under the group of transfonnations.

On the relation between time and space in the visual discrimination of velocity.

“…In the case of moving objects, the invariance or coherence is defined on the relation between the spatial and temporal changes rather than on the discrete spatial and temporal positions. In most previous demonstrations of nonlinear perceptual relations among component stimulus variables (e.g., Lappin, 1971;Pomerantz & Garner, 1973;Weintraub, 1971), the relationships involved a single physical variable, which was usually spatial. It is noteworthy that the perceptually effective relationship in the present study was defined on two different physical dimensions.…”