Item

Abstract

Some years ago a quantitative analysis of the optomotor responses in the beetle Chlorophanus was worked out. This analysis has led to the formulation of a model for the functional structure of perception describing stimulus response relations. The model consists of linear filters with different transfer characteristics and correlator units. The transfer characteristics of the filters was determined on the basis of experiments with rotating patterns of sinusoidally changing contrasts. The model enabled us to predict responses to moving patterns of various contrast configurations. In one of the predictions the response curve to the movements of a “white noise“ contrast pattern was calculated and compared with reactions of the beetle which were elicited by a rotating cylinder carrying randomly distributed (vertically oriented) striped shades from white through black 1. The comparison showed rough agreement between theory and experiment. Recently THORSON has pointed out that Equ. (38) in our 1959 paper 1, which gives the analytic expression for the response to a moving “white noise“ pattern, contains an algebraic error. Prompted by this finding, we have again taken up this part of the investigation in the present paper. If the algebraic error is corrected and the calculated response curve plotted, a misfit between the theoretically derived response and the experimental data is found. If however, the actual Fourier spectrum of the pattern is used in the calculations instead of the hypothetical “white noise“, an acceptable fit is achieved. The approximation can be improved if the time constants τF and τH of the low pass filters of the model are slightly changed from the formerly obtained values. Furthermore it was investigated whether a reduced model, containing only one or two different filters lead to better approximations. The reported results favour the model originally suggested containing three different types of filters.