fMRI (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, adaptation of the fMRI signal showed that the same neural population responds to faces falling along single identity axes within this space.

Humans are remarkably sensitive in detecting small deviations from circularity. In tasks involving discrimination between closed contours, either circular in shape or defined by sinusoidal modulations of the circle radius, human performance has been shown to be limited by global processing. We assessed the amount of global pooling for different pattern shapes (different radial modulation frequencies, RF) when circular deformation was restricted to a fraction of the contour. The results show that the improvement in performance depends on the modulation frequency (the pattern shape) when increasing the number of cycles of an RF pattern. Global processing only extends up to modulation frequencies between 5 and 10. For higher frequencies, performance can be predicted by probability summation. Position uncertainty cannot explain these effects. In a circumstance where global pooling exceeds probability summation (RF=5), we split the pattern up into five identical segments conserving the total amount of information presented. Thresholds are significantly affected by different global arrangements of these segments: (a) Occluding small parts of the pattern shows a significant effect on the position of occluders with performance lowest when gaps are placed at the points of maximum curvature. (b) Shifting segments away from the pattern centre (exploded condition) or displaying them out of concentric context (spiral condition) shuts down global processing. (c) Jittering segments radially disrupts both global and local processing. We conclude that RF patterns in the global processing range are analysed by detecting the points of maximum curvature and that, in this range, the visual system can only reliably process up to about 5 local curvature extrema.

To simplify the study of visual face processing, we introduce a novel class of synthetic face stimuli based upon 37 measurements (head shape, feature locations, etc.) extracted from individual face photographs in both frontal and 20 degrees side views. Synthetic faces are bandpass filtered optimally for face perception and include both line and edge information. Pilot experiments establish that subjects are extremely accurate in matching a synthetic face with the original grayscale photograph, even across views. To determine the perceptual metric of face space, we introduce face cubes in which the geometric differences between any faces in a four-dimensional face subspace can be precisely determined. Experiments on face discrimination using face cubes establish the metric of synthetic face space as locally Euclidean, with discrimination thresholds representing 4-6% total geometric variation (as a percent of mean head radius) between faces. Discrimination thresholds are lowest for face cubes constructed around the average face, thus indicating that the mean face for each gender represents a natural origin for face space. Finally, synthetic faces exhibit a pronounced inversion effect for 20 degrees side views and a characteristic "Thatcher effect" for inverted front views. Synthetic faces and face cubes thus provide a useful new quantitative approach to the study of face perception and face space.

This review focuses on low and intermediate stages of contour shape processing. It is split into two main sections, 'Contour Detection' and 'Shape Discrimination and Representation'. The first section examines contrast detection of elements within a contour ("collinear facilitation") and the detection of contours in noise ("contour integration"). The second section deals with the discrimination and representation of simple and complex shapes. Perceptual effects on contour detection have been linked to low-level, long-range lateral interactions between neighbouring neurons in V1. Experimental results suggest a complex network of interactions that are context dependent, with collinearity being the dominant factor. While lateral connections are an obvious candidate for linking contour elements into spatially extended contours, the long-range interactions are insufficient to account for human performance in a variety of tasks. Data suggest the existence of global mechanisms that integrate information beyond that of neighbouring cells and are influenced by the overall features of a stimulus. Evidence from psychophysics and physiology is converging towards the identification of an intermediate level of shape processing, where sensitivity to such global attributes emerge.