Abstract

Relative spatial localization of widely separated objects (>30–50 min of arc) cannot be completely accounted for by a neural representation consisting of spatial frequency channels; another, more proximal, mechanism is also required.¹ Is this localization mechanism selective for orientation? Relative localization accuracy was measured as a function of orientation using two dot stimuli with a mean separation of 100 min of arc. The observer’s task was to compare the inter-dot separation on a given trial to the average separation. Accuracy varied systematically with orientation, being poorest at 0°, 45°, and 90° (vertical). This suggests that the large-scale localization mechanisms are tuned for orientation. To explore further this orientation selectivity, localization accuracy was measured again, but now one dot was displaced orthogonal to the direction of measurement by a distance that varied randomly from trial to trial.

© 1986 Optical Society of America