中華心理學刊 民 79，32 卷，7-30
Chinese Journal of Psychology 1990, Vol.32, 7-30
Chao-Ming Cheng（Department of Psychology, National Taiwan University）；Su-Jane Chen （Department of Psychology, National Taiwan University）
Abstract
The issue of similarity is of fundamental importance in psychology, yet the mechanism underlying similarity judgment has long eluded research. In this paper, we propose a model that the proximity between two two-dimensional patterns, f(x, y) and g(x, y), is rated by first computing the autocorrelation of the two patterns, As(2,y) and Ag(x, y), which is then followed by computing the D1 statistic given by
D1=1-(2∫∫ As( x, y) dx dy/∫∫ (As(x,y)+Ag(x,y)dx dy)
where As(x, y) represents the autocorrelation of the pattern, s(x, y), consisting of all component features shared by both f(2,y) and g(x,y). Predictions of this model and of other correlation models were tested against judged-similarity data. Results show a better prediction by the present model than by the others. The generality and limitations of the model are discussed.
Keywords: Pattern similarity, Autocorrelation
