Abstract
To process the abundance of visual stimuli available to us on a daily basis, our visual system often handles redundancy by compressing similar items into an ensemble representation (Alvarez, 2011). This can be something as comparatively simple as the size of a set of circles (Ariely, 2001), or as complex as the expressions of multiple faces (Haberman & Whitney, 2009). We investigated whether this compression also occurs for the attractiveness of faces. Attractiveness of a single face can be accurately judged from incredibly short exposures (Olson, 2005). We tested whether an accurate estimate of attractiveness can also be summarised from multiple faces. Participants saw sets of nine faces for varying display durations (250, 500, 1000ms, and an unlimited condition) and were asked to indicate whether it contained more attractive or more unattractive faces. We found performance was higher than chance in all but the most difficult conditions, and improved with a clearer majority, with generally better performance when this majority was unattractive. Longer exposures also increased performance, but 1000ms proved as sufficient as the unlimited condition. During this task we recorded eye movements to establish whether attractive or unattractive faces were drawing visual attention. The order of fixations to spatial locations was unaffected by the attractiveness of either the set or the face in the location. Common scan paths in the two shortest displays did not vary between accurate and inaccurate trials, remaining centrally focussed. For 1000ms, accurate trials showed slightly more systematic patterns than inaccurate trials, and the unlimited condition showed a left-to-right top-to-bottom reading pattern for accurate trials that was absent from inaccurate trials. Longer exposures appear to improve performance (often with more faces being fixated), suggesting that in brief exposures limited sampling may lead to an ensemble that is incomplete, but still allows some accurate performance.
Meeting abstract presented at VSS 2013