Rapid and Coarse Face Detection in Cortex

Humans have structures dedicated to the processing of faces, which include cortical components (e.g. areas in occipital and temporal lobes) and subcortical components (e.g. superior colliculus and amygdala). Although faces are processed more quickly than stimuli from other categories, there is a lack of consensus regarding whether cortical or subcortical structures are responsible for rapid face processing. In order to probe this, we exploited the asymmetry in the strength of projections to subcortical structures between the nasal and temporal hemiretina. Participants detected faces from unrecognizable control stimuli and performed the same task for houses. In Experiments 1 and 3, at the fastest reaction times, participants detected faces more accurately than houses. However, there was no benefit of presenting to the subcortical pathway. In Experiment 2, we probed the coarseness of the rapid pathway, making the foil stimuli more similar to faces and houses. This eliminated the rapid detection advantage, suggesting that rapid face processing is limited to coarse representations. In Experiment 4, we sought to determine whether the natural difference between spatial frequencies of faces and houses were driving the effects seen in Experiments 1 and 3. We spatially filtered the faces and houses so that they were matched. Better rapid detection was again found for faces relative to houses, but we found no benefit of preferentially presenting to the subcortical pathway. Taken together, the results of our experiments suggest a cortical, coarse rapid detection mechanism, which was not dependent on spatial frequency.