Several commentators have suggested that the attractiveness of average facial configurations could be due solely to associated changes in symmetry. If this symmetry hypothesis is correct, then averageness should not account for significant variance in attractiveness ratings when the effect of symmetry is partialed out. Furthermore, changes in attractiveness produced by manipulating the averageness of individual faces should disappear when all the images are made perfectly symmetric. The experiments reported support neither prediction. Symmetry and averageness (or distinctiveness, the converse of averageness) made independent contributions to attractiveness (Experiments 1 and 2), and changes in attractiveness resulting from changes in averageness remained when the images were made perfectly symmetric (Experiment 2). These results allow us to reject the symmetry hypothesis, and strengthen the evidence that facial averageness is attractive.

According to the multidimensional space framework, faces are represented as locations in a psychological face-space. Our aim was to test whether the locations of veridical, caricatured, and anticaricatured face stimuli in face-space, derived from multidimensional scaling analyses, could account for identification accuracy and distinctiveness ratings for these stimuli. Caricatures were identified more accurately and rated as more distinctive than veridical images, whereas anticaricatures were identified less accurately but were not significantly less distinctive than veridical images. In the face-space derived from multidimensional scaling, caricatures were further from the origin and less densely clustered than veridical images, whereas anticaricatures were closer to the origin and located in denser regions of the space. A quantitative model successfully predicted identification performance from the spatial locations of the stimuli. In general, the physically transformed faces were systematically mapped in the psychological space.

Several brain imaging studies have identified a region of fusiform gyrus (FG) that responds more strongly to faces than common objects. The precise functional role of this fusiform face area (FFA) is, however, a matter of dispute. We sought to distinguish among three hypotheses concerning FFA function: face specificity, individuation, and expert individuation. According to the face-specificity hypothesis, the FFA is specialized for face processing. Alternatively, the FFA may be specialized for individuating visually similar items within a category (the individuation hypothesis) or for individuating within categories with which a person has expertise (the expert-individuation hypothesis). Our results from two experiments supported the face-specificity hypothesis. Greater FFA activation to faces than Lepidoptera, another homogeneous object class, occurred during both free viewing and individuation, with similar FFA activation to Lepidoptera and common objects (Experiment 1). Furthermore, during individuation of Lepidoptera, 83% of activated FG voxels were outside the face FG region and only 15% of face FG voxels were activated. This pattern of results suggests that distinct areas may individuate faces and Lepidoptera. In Experiment 2, we tested Lepidoptera experts using the same experimental design. Again, the results supported the face-specificity hypothesis. Activation to faces in the FFA was greater than to both Lepidoptera and objects with little overlap between FG areas activated by faces and Lepidoptera. Our results suggest that distinct populations of neurons in human FG may be tuned to the features needed to individuate the members of different object classes, as has been reported in monkey inferotemporal cortex, and that the FFA contains neurons tuned for individuating faces.

Disruption to the anterior thalamus (AT) may be an important factor in diencephalic amnesia. Rats with small lesions of the anteromedial (AM) or anteroventral (AV) nucleus showed persistent working-memory and reference-memory deficits in a 12-arm radial maze, although they were comparable to controls during the early part of training. The only activity difference in the maze was that lesioned rats failed to run more slowly when revisiting a baited arm. For all groups, both working and reference memory were impaired after extramaze cues were removed; removal of intramaze cues further impaired performance relative to the original conditions. These findings suggest the AT makes a distinct contribution to mnemonic functions, probably as part of an integrated system involving limbic cortex and the hippocampal formation, and that AT lesions produce a general rather than a specific deficit in spatial or working memory.

Valentine's (Valentine T. Q J Exp Psychol 1991;43A:161-204) face recognition framework supports both a norm-based coding (NBC) and an exemplar-only, absolute coding, model (ABC). According to NBC; (1) faces are represented in terms of deviations from a prototype or norm; (2) caricatures are effective because they exaggerate this norm deviation information; and (3) other-race faces are coded relative to the (only available) own-race norm. Therefore NBC predicts that, for European subjects, caricatures of Chinese faces made by distorting differences from the European norm would be more effective than caricatures made relative to the Chinese norm. According to ABC; (1) faces are encoded as absolute values on a set of shared dimensions with the norm playing no role in recognition; (2) caricatures are effective because they minimise exemplar density and (3) the dimensions of face-space are inappropriate for other-race faces leaving them relatively densely clustered. ABC predicts that all faces would be recognised more accurately when caricatured against their own-race norm. We tested European subjects' identification of European and Chinese faces, caricatured against both race norms. The ABC model's prediction was supported. European faces were also rated as more distinctive and recognised more easily than Chinese faces. However, the own-race recognition bias held even when the races were equated for distinctiveness which suggests that the ABC model may not provide a complete account of race effects in recognition.