Recognition of faces in the presence of two-dimensional sinusoidal masks

The spatial frequencies most relied upon by subjects in a recall task for face recognition were found to lie in the midfrequency range. A linear systems analysis model cannot account for these masking data in terms of retinocortical processing limitations alone. In order to account for the greater disruption of the face recognition task by masks in the range of 2.2 cycles/deg, the existence of unequal filtering of spatial frequency components must be recognized. This unequal filtering may occur either during memory deposition or retrieval of the input stimulus in the recall task or at any time in between.