Discrimination and association processes for faces and non-faces: The effect of rotation

Most current theories of face perception claim that inversion leaves the coding of non-face stimuli largely unaffected, while causing a qualitative change in the coding of faces. Empirical support for this hypothesis mainly stems from recognition studies which typically show a larger inversion decrement for faces than for other stimuli. Several recent studies using experimental paradigms that do not contain a substantial memory component have however yielded contradicting results. This observation suggests that the disproportionate effect of inversion for faces might be related to the presense, or absence, of a memory component in the experimental task. In order to explore this hypothesis we investigated the effect of inversion within a discrimination learning paradigm, which contains a memory component comparable to that included in a recognition paradigm. We compared the effect of rotation on discrimination and association processes for faces and cars. Subjects learned to discriminate pairs of similar faces and similar cars and to associate them with neutral responses. The stimulus pairs were presented upright, inverted, and additionally in two intermediate orientations. We found that discrimination performance was generally better for faces than for cars and that associations were learned faster for faces than for cars. However, we did not find any evidence that rotation affected discrimination and association processes for faces differently than for cars. In this sense, our results provide no evidence for the hypothesis that memory processes are responsible for the disproportionate effect of inversion which is found in recognition experiments.