Abstract

We consider key assumptions and data underlying the Gabor function (GF): Gaussian-modulated cosine descriptions of cortical simple cell receptive fields (RFs). We show that the implied metric of spatial and spatial-frequency localization has not been justified by others. In fact, alternate metrics yield plausible receptive fields. We review the interdependence of such metric considerations with information theoretical explanations of the RF function. We review comparisons of fits to neurophysiological and psychophysical data which show that other RF functions are superior to GFs.

© 1988 Optical Society of America