Abstract
Using the method of fMR-adaptation (also termed repetition suppression) we have recently demonstrated that the neuronal tuning to faces is very narrow, and follows the perceptual discrimination of subjects (Gilaie-Dotan and Malach, 2007). However, it is not clear whether this relationship between neuronal selectivity and perception is unique to faces or is a more general property of object representations. In order to examine this issue, we have compared our initial finding with an identical experiment in which the faces were inverted upside-down. This manipulation disrupts familiarity and expertise effects while preserving all low-level feature compositions. In the upright and inverted experiments, subjects were exposed to faces under four conditions: different faces, repeated presentation of an identical face and two levels of morphing which introduced small or large differences between faces. Regions of interest were localized using a separate face localizer experiment containing both upright and inverted faces. Behaviorally, we found that there was a significant broadening of the selectivity for inverted compared to upright faces. Examining the fMRI response amplitudes revealed only slight reduction in activity for inverted faces in the fusiform face area (FFA). Yet importantly, the tuning of fMR-adaptation in the FFA as well as in the object-related Lateral Occipital Complex (LOC) showed significant broadening for inverted faces. This broadening nicely correlated with the perceptual discrimination behavior for upright and inverted faces. Our results support the notion that the ability of human observers to discriminate objects and faces can be attributed to the tuning properties of single neurons underlying these representations rather than being a consequence of some vector representation of broadly tuned neurons. This mode of coding is apparently not unique to faces but appears to be a property of object representations in general, regardless of their familiarity or behavioral relevance.
This work was supported by ISF, Minnerva and Clore grants to R. Malach.