Abstract
INTRODUCTION: The direction of social attention from groups can provide stronger cueing than from an individual (Gallup et al., 2012). It has previously been shown that basic visual features such as size or orientation can be averaged (Dakin 2001), however recently it has also been shown that observers can average gaze direction in a group of 4 faces (Sweeny & Whitney, 2014) . We investigated averaging of both gaze direction and head rotation using an equivalent noise (EN) paradigm. METHODS: Participants judged either (a) the overall direction of gaze, or (b) overall head rotation for an array of faces presented for 300ms. The direction of gaze (or head rotation) for each element in the array was drawn from a normal distribution. External noise was added to the stimulus by increasing the standard deviation of the gaze (or head) distribution (0.5° to 32°). Direction discrimination thresholds were collected at 7 levels of external noise, and an EN function was fit to the data to extract estimates of internal noise (accuracy of direction estimate for a single face) and sampling efficiency (number of faces averaged). Faces were either distributed in a large (16 faces) random array with no fixation or a small (6) circular array around fixation (ecc= 4°). RESULTS: Observers were able to average the direction of a crowds’ gaze and head rotation across all conditions, though not very efficiently when judging gaze in the large arrays (sample estimates between 1.5 and 2.5). Averaging of head rotation had less internal noise and was more efficient than for gaze direction (4-6 samples). Internal noise was greatest when faces were presented away from fixation. CONCLUSIONS: Head rotation and gaze direction can be averaged from groups of faces to reliably judge information about a crowd, although this is more reliable when judging head direction.
Meeting abstract presented at VSS 2015