The concept of spatial frequency channels cannot explain some visual masking effects.

Visual masking is assumed to be predictable by the degree to which mask and signal spatial frequency amplitudes are similar. In this masking experiment we have clearly shown this not to be the case by studying the effects of factorial combinations of signal-mask amplitude and phase spectra. Results show that the amplitude spectra characteristics do not predict masking, and a better predictor of these results appears to be the correlations between the mask and signal's luminance profiles. These results, therefore, show that the spatial similarity between two images, as may be processed by "spatial frequency channels" cannot be determined by the similarities between these channel outputs as defined by the (output) modulated amplitudes.