Abstract
Complex patterns, such as faces, can be described by the combination of their Fourier frequency and phase components. Whereas the role of frequency information has received the majority of research attention, the importance of phase information in face perception has been largely neglected. In the experiments reported here, we sought to investigate the role of phase information on face perception using a discrimination task on arrays of face morphs. In the first experiment, we varied the amount of aligned Fourier phase in different regions of the face frequency spectrum in order to determine whether the information in some regions was more important than others and whether the properties of the underlying neural processes are best understood in terms of frequency bandwidth or number of phase-alignments. In the second experiment, linear filtering was implemented to estimate the information content in different face frequency bands and to determine whether it is the number of phase-alignments or the signal-to-noise ratio of phase-alignments that matter. In the third experiment, we varied the distribution of phase-aligned frequencies to ascertain whether it is the number of contiguous phase-aligned frequencies or the global signal-to-noise ratio that matters. We conclude that there are underlying processes that depend on a certain signal-to-noise ratio of phase-alignments within a contiguous range of face frequencies (we termed these critical band of phase alignments) which operate with equal efficiency throughout the face frequency spectrum.
Canadian Institutes of Health Research (CIHR) grant: MT 108-18 to RFH, and an FRSQ Doctoral Fellowship to RF.