Size-invariant but viewpoint-dependent representation of faces

The present study investigated the role of size and view on face discrimination, using a novel set of synthetic face stimuli. Face discrimination thresholds were measured using a 2AFC match-to-sample paradigm, where faces were discriminated from a mean face. In Experiment 1, which assessed the effect of size alone, subjects had to match faces that differed in size up to four-fold. In Experiment 2 where only viewpoint was manipulated, a target face was presented at one of four different views (0 degree front, 6.7 degrees, 13.3 degrees, and 20 degrees side) and subsequent matches appeared either at the same or different view. Experiment 3 investigated how face view interacts with size changes, and subjects matched faces differing both in size and view. The results were as follows: (1) size changes up to four-fold had no effect on face discrimination; (2) threshold for matching different face views increased with angular difference from frontal view; (3) size differences across different views had no effect on face discrimination. Additionally, the present study found a perceptual boundary between 6.7 degrees and 13.3 degrees side views, grouping 0 degrees front and 6.7 degrees side views together and 13.3 degrees and 20 degrees side views together. This suggests categorical perception of face view. The present study concludes that face view and size are processed by parallel mechanisms.