Accession Number : AD0684609

Title :   A PATTERN-RECOGNITION MODEL APPLIED TO THE PERCEPTION OF PITCH.

Descriptive Note : Technical rept.,

Corporate Author : NEW YORK RESEARCH GROUP INC N Y

Personal Author(s) : Rothenberg,David

Report Date : JAN 1969

Pagination or Media Count : 111

Abstract : A mathematical model for pattern recognition is applied to the perception of complex tones. A possibly infinite and continuous space (over which sensory stimuli range) is mapped into a finite space of discrete points (the classifications of such stimuli). A simple ordering of points in the former space together with a feedback-determined preordering of pairs of such points is assumed. The number of classifications required is also given and is assumed to derive from feedback. The model chooses a 'characteristic' finite subset of the stimulus space which defines a function and a 'range' about each element of the subset such that the union of all such ranges is maximal and the classification of pairs of points contained in such union is unambiguous. Restrictions on the choice of such subset derives from the limitations in information carrying capacity of the resulting classification system. A 'Gestalt' description of perception results in which extremely complex and varied phenomena can be perceived without proportionately large human memory. Specific distortions of perception in specified contexts are predicted. Musical practice is shown to conform to boundaries deduced from the model. Experiments for testing of the model are described. (Author)

Descriptors :   (*PITCH DISCRIMINATION, PATTERN RECOGNITION), MUSIC, ADAPTIVE SYSTEMS, LEARNING, SET THEORY, CLASSIFICATION, MATHEMATICAL MODELS, DISTORTION

Subject Categories : Psychology

Distribution Statement : APPROVED FOR PUBLIC RELEASE