Abstract
The present study examines the influence of structured form information on human heading perception. Random-dot kinematograms (100 light-increment dots) were used to generate expanding optic flow (70.1°H × 70.1°V) with a focus of expansion (FOE) at −15°, 0° and 15° from the display center. Form signals were introduced into the stimulus by assigning each dot in the display a partner dot to form a dipole of a particular local orientation. Dipoles were configured to produce a radial “Glass” pattern implying a “form” FOE at −20°, −10°, 0°, 10° and 20° from the display center. The stimulus thus affords that for certain configurations form and motion information each indicates a different FOE. Observers were asked to fixate on a cross in the center of the display and clicked a mouse button to start a trial. The stimulus was displayed for 1.5 sec and observers were required to indicate their perceived heading direction along a horizontal line in the display center using a mouse-controlled probe. For 10 observers (8 naïve), for flow patterns with a motion FOE at the display center (0°), the judged heading is shifted towards the implied form FOE from the Glass pattern with a bias equivalent to a weighting of approximately 0.3. For flow patterns with a motion FOE at 15° to the left or right of the display center, a systematic effect was evident only when motion and form FOE were in same direction. No such effect was observed with anti-Glass patterns, possibly due to a weaker association between opposite-polarity dots forming dipoles. Form information plays an important role in human heading perception from optic flow. The findings are consistent with a computational procedure that weights and averages form and motion estimates of heading.
Supported by: Hong Kong Research Grant Council, HKU 7471//06H.