fMRI evidence for the neural representation of faces

ARTICLESfMRI evidence for the neural representation of faces.com/natureneurosciencehttp://www.nature2005 Nature Publishing Group Gunter Loffler1, Grigori Yourganov2, Frances Wilkinson2 & Hugh R Wilson2fMRI (functional magnetic resonance imaging) studies on humans have shown a cortical area, the fusiform face area, that is

specialized for face processing. An important question is how faces are represented within this area. This study provides direct

evidence for a representation in which individual faces are encoded by their direction (facial identity) and distance

(distinctiveness) from a prototypical (mean) face. When facial geometry (head shape, hair line, internal feature size and

placement) was varied, the fMRI signal increased with increasing distance from the mean face. Furthermore, adaptationof the fMRI signal showed that the same neural population responds to faces falling along single identity axes withinthis space.At the highest levels of the primate form vision pathway in inferior

temporal cortex (TE), neurons respond selectively to highly complex

object forms such as faces1,2. fMRI has been used to identify an

analogous area in the human fusiform gyrus, the fusiform face

area (FFA), that is specialized for face processing3,4. It has been

hypothesized that faces are represented in a multi-dimensional space,

with individual faces encoded by their direction and distance from the

mean face5, and psychophysical data support this view6. If faces are

indeed represented in face space as a function of distance and direction

from an appropriate mean face5 (Fig. 1a), two possible neural

encoding schemes suggest themselves. The first is a Cartesian tiling

of the multidimensional space with different neural populations

tuned to different ranges of distance and direction (Fig. 1b). Alternatively, distinct neural populations might each respond at all distances from the mean face within a restricted range of directions, akin

to a generalized polar coordinate representation (Fig. 1c). In both

cases, response regions are represented as growing in size with distance

from the mean, reflecting data that indicate decreased face discriminability far from the mean7,8. The polar coordinate representation

predicts that responses will increase with distance from the mean,

whereas the Cartesian representation predicts roughly equal responses

at all distances.Here, we report evidence in favor of the polar representation by

recording fMRI signals from FFA as a function of varying facial

geometry (head shape, hair line, internal feature size and placement).

We found...