Abstract
Blindsight refers to the ability of patients with visual cortex damage to detect, localise or discriminate visual stimuli that they deny “seeing”. In this study we sought to create an analog of blindsight in normal observers using binocular masking via continuous flash suppression. We had participants reach toward and pretend to grasp an imaginary object represented by a pair of parallel lines, oriented horizontally or vertically, made “invisible” via masking. The dependent variable was the orientation of the hand, horizontal or vertical, at grasp. Over a period of training (six 100 trial sessions) during which trial-by-trial feedback was provided, participants showed a significant improvement in their ability to orient their hand to grasp the “invisible” target. No such improvement was evident when other participants responded using only verbal responses, even with a matched period of training. Trial-by-trial reports of subjective visibility and confidence in performance revealed that, while participants never reported “seeing” the target, performance was highly correlated with confidence. These data have important implications for the treatment of cortically blind patients, and highlight the indistinct boundaries between what we consider visible and invisible inputs.