Abstract
On many vision tasks, people improve with practice, and the generality of transfer of training can be quite revealing with respect to underlying neural mechanisms. Using dynamic arrays composed of hundreds of small grating patches, we trained one group of observers to discriminate spatial forms defined by dynamic fluctuations in luminance of a subset of gratings and another group to discriminate shape defined solely by common temporal structure (TS) among a subset of the gratings (Lee & Blake, 1999). For both types of dynamic display, grating orientations were random throughout the array, and each grating moved in a direction orthogonal to its contour orientation and changed direction of motion irregularly over time. The task itself involved 2AFC shape discrimination. Performance thresholds for both TS and luminance-defined shape were measured before and immediately following training. For some observers, training and testing was monocular. All observers improved at their respective task with practice. Those trained using a luminance cue showed strong, immediate transfer when tested with TS-defined forms, which is not so surprising since the luminance display contained dynamic, TS information. Observers trained on TS, in contrast, did not show transfer to the luminance based discrimination. Moreover, effects of practice on the luminance task transferred from the trained to the untrained eye, but practice did not transfer interocularly on the TS task. These preliminary results suggest an early locus for the underlying neural mechanisms of this form of temporal structure.
Supported by EY07760 and EY07135.