Item

Abstract

Everyday life requires us to identify different faces in many different poses and views, despite this complexity, we are capable of recognizing familiar faces reasonably well. In this
study, we focus on the question of what kind of information human observers use to recognize
faces across variations in viewpoint; specically, whether they use only holistic information,
or whether they encode and store the local information contained in facial parts (featural information)
as well as their spatial relationships (congural information). Three experiments were
conducted, modelled after the inter-extra-ortho experiments by B¨ulthoff Edelman, 1992,
who used novel objects as stimuli (wire- and amoebae-like 3D objects). Experiment 1 was
designed as a base line for the subsequent experiments. Ten face-stimuli were presented in
frontal view and 45
r
side view. At test they had to be recognized among ten distractor faces
at different viewpoints. We found systematic effects of viewpoint (recognition performance:
inter = extra
s
ortho) similar to results in the original study. Experiments 2 and 3 were designed
analogously to Experiment 1 except for the fact that in the testing phase the faces were
presented scrambled or blurred. The results showed that human observers are capable of recognizing
faces across different viewpoints on the sole basis of isolated featural information
and of isolated congural information. Furthermore, we found systematic effects of viewpoint
for both isolated congural information and isolated featural information. The results provide
further support for two routes in face recognition and clearly show that part-based processing
is at least as viewpoint dependent as congural information. This effectively demonstrates
how view-based recognition pertains even to different processing routes. In addition to the psychophysical
experiments, systematic differences between the effects of viewpoint are discussed
in a computational framework based on key frames.