Arnold, G. & Sieroff, E.
Institut de Psychologie, Université Paris Descartes, CNRS, FRANCE
When faces are learned from rotating view sequences, novel views may be recognized by matching with multiple discrete views, or with an integrated representation of the sequence. A view-matching process should benefit from long view durations, allowing the attention to sequentially focus on each view during the encoding of the sequence. Conversely, an integrated-representation process should benefit from short view durations, allowing the distribution of attention over the entire sequence in a short temporal window. In a sequential comparison task, we tested the recognition of novel interpolated and extrapolated views after learning faces from rapid (240 ms for each view) and slow sequences (960 ms for each view). In a first experiment, recognition was tested with internal views (learned and interpolated). In a second experiment, recognition was tested with internal and extrapolated views. Results showed a global superiority of rapid over slow sequences, in favour of the integrated-representation hypothesis. In addition, the recognition pattern for the different viewpoints in the sequence depended on the absence (Experiment 1) or presence (Experiment 2) of extrapolated test views. The presence of extrapolated views affects the global representation of the face, modifying the “centre of gravity” of the representation.