We present a quantitative framework for interpreting the results of multidimensional stimulus generalization experiments in animals using concepts derived from the geometrical approach to human cognition. We apply the model to the analysis of stimulus generalization data obtained from pigeons trained with different sets of stimuli varying along two orthogonal dimensions. Separable pigeons were trained with stimuli varying along the dimensions of circle size and line tilt, dimensions found to be separable in previous human research; Integral pigeons were trained with stimuli varying along two dimensions of rotation in depth, dimensions which are intuitively integral and which hold special interest for theories of object recognition. The model accurately described the stimulus generalization data, with best fits to the City-Block metric for Separable pigeons and to the Euclidean metric for Integral pigeons.
KeywordsMultidimensional generalization; integrality; separability; metric; pigeon
Integrality/Separability of Stimulus Dimensions and Multidimensional Generalization in PigeonsThe objects and scenes that organisms encounter in their natural environments are often extraordinarily complex and they vary along multiple dimensions. For this reason, a key question in the study of animal visual cognition is how different dimensions of a stimulus interact with each other. One way to investigate this dimensional interaction is through studies of multidimensional stimulus generalization.Consider the simple case in which a training stimulus is changed along two dimensions. If it is possible to find a combination rule to predict generalization after a bidimensional change in the stimulus from generalization after unidimensional changes along each dimension alone, then this rule might provide important insights into how these dimensions are processed, represented, and combined.In human research, the distinction between separable and integral stimulus dimensions has been of considerable value in the study of dimensional interaction (Garner, 1974;Shepard, 1991). Separable dimensions are those that can be attended to and processed independently of each other; here, the dimensional structure itself determines the similarity between stimuli. Integral dimensions are those that cannot be attended to and processed independently of each other; here, the similarity between stimuli is directly perceived and the notion of multiple