Abstract
This experiment examined the effects of adding five different kinds of prominent monocular features to a large-disparity random-dot stereogram. It was found that features which enclosed the disparate area produced the shortest initial perception times for fusion. The longer initial perception times for stimuli containing features without this enclosing property are explained in terms of less-helpful guidance of saccadic eye movements prior to the establishment of fusion. Subsequent reductions in perception times for these latter stimuli could be due to perceptual learning within the eye movement control system.