Abstract
It is widely believed that faces are processed differently by the visual system than most non-face stimuli. Despite substantial evidence in support of this belief, many questions remain as to how specialized face processing is achieved, and if it affects the raw visual input as a means of operating upon incoming face stimuli. We have previously shown that face inversion effects depend on stimulus duration, emerging 52–75ms after stimulus onset. This result suggested that specialized face processing is not continuously active but is deployed deliberately when a face-like stimulus is detected (Schwartz, VSS07). In the current study, we leverage this finding in order to measure the effect of specialized face processing on low-level contrast detection. Using a backward masking paradigm, we measured contrast sensitivity in subjects who were asked to detect a Gabor stimulus that was preceded by an upright face, an inverted face, or by random Gaussian noise that was structurally dissimilar but spectrally similar to face images. Mask and Gabor stimuli were displayed for 82ms, separated by a 12ms ISI. Gabor stimuli were presented randomly at 8 spatial frequencies ranging from 0.25 to 10cpd, and 3 orientations (horizontal, vertical, oblique). Detection thresholds were measured using the QUEST procedure (Watson and Pelli, 1983) augmented by 0% and 30%-contrast catch trials. Compared to a non-masked control condition, we found that sensitivity decreased by an average of 0.3 log units for all SFs masked by pink noise. In both face conditions, similar decreases were observed for all SFs except 2–4cpd (16–32cpf) which remained equal to control levels despite the mask. These data suggest that specialized face processing involves the selective enhancement of SFs between 2–4cpd. While sensitivity was preserved for both upright and inverted face masks, additional research is necessary to determine how these frequency bands operate in subsequent specialized processes.