Abstract

The global geometrical arrangement of face parts (eyes over nose over mouth), commonly referred to as first order configural (FOC) properties, is believed to constitute a fundamental aspect of face detection. Indeed, several brain regions in the face processing network have been shown to display increased activity in response to abstract stimuli for face-like as compared to random spatial arrangements. However, the absence of a mathematical formalization of FOC properties makes it difficult to systematically study their contribution to face perception. To fill this gap, we used analytical geometry to define a set of 19 FOC features. In two psychophysical studies, a 2-alternative forced choice task assessing difference in perceived faceness between two stimuli and a Likert test assessing measuring perceived faceness of individual stimuli, we evaluated the contribution of each feature to the perceived faceness of abstract stimuli consisting of four black rectangular shapes reflecting the two eyes, the nose and the mouth. From the two experiments two subsets of features relevant for face detection were selected, including 10 and 11 features, respectively. Interestingly, 7 features are shared between the two sets. The difference between the subsets reflects processing of individual FOC features in the first study and holistic processing in the second study. The partial superimposition, on the other hand, is reflective of a shared basic mechanism in the perception of first order configural properties.