Effect of selective and distributed training on visual identification of orientation

An experiment contrasted the effect of four training schedules in a visual orientation reproduction task. Two selective schedules involved repeated presentation of a single target orientation. Two non-selective schedules involved targets covering the first quadrant either at fixed, equispaced orientations, or distributed randomly. In pre-training sessions, we observed the classical oblique effect (precision for vertical and horizontal stimuli higher than for oblique ones). Practice improved precision with both distributed schedules, but was ineffectual for non-selective schedules. However, a significant oblique effect persisted under all conditions. We argue that the pattern of results is compatible with the hypothesis that the oblique effect reflects both the intrinsic neuronal properties of the primary visual system, and the structure of the visual space imposed by higher, more cognitive processes. The results challenge the thesis that only attentional and post-perceptual factors are able to affect the working of the early visual system.