Holistic and subordinate-level face processing in monkeys

Face perception in humans differs from perception of most objects: faces are recognized at the individual or subordinate level (e.g. Madonna, Collie), whereas objects are recognized at the basic level (e.g. face, dog). Additionally, faces are perceived holistically, i.e. features are not functionally independent. To date, these criteria have yet to be tested in monkeys.

Using a dishabituation paradigm borrowed from developmental psychology, rhesus macaques elicited an image or blank square, alternately, by directing gaze towards the monitor, and terminated a stimulus by looking away. After 20 seconds of cumulative stimulus display time, a new image was presented, constituting the beginning of a dishabituation trial. Image preference was measured as the time spent looking at the new image versus the cumulative display time. In experiment 1, conspecific faces preceded by either another animal or face showed higher preference than those preceded by the mirror-reversed image, demonstrating individuation among monkey faces. In contrast, animals (dogs and birds) did not elicit higher preference when preceded by another same-category exemplar, than when preceded by the mirror-reversed image, demonstrating only basic level differentiation. In experiment 2, composite monkey faces consisting of top and bottom halves were either aligned or misaligned. In dishabituation trials, only the bottom half was replaced. If monkeys perceive faces holistically, this would cause perception of a new face only in the aligned condition. Indeed, aligned faces elicited greater preference than misaligned faces in dishabituation trials. Thus monkeys, like humans, show individuation and holistic processing of conspecific faces.