A blindsight conundrum: how to respond when there is no correct response.

Whereas research on blindsight customarily defines the correct responses to all visual stimuli presented to the cortically blind field, we here introduced a small number of unexpected 'no stimulus' trials in a localization task, to discover whether they would elicit the same responses as blind field targets. As no correct responses existed for the blank stimuli, our subjects, three hemianopic and one normal monkey, and one human hemianope who was aware of many blind-field targets, could either respond to these catch trials as to a target or refrain from responding. Visual stimuli were presented singly at four possible positions, two in the blind field of the hemianopes, and all subjects correctly localized the vast majority of targets in either hemifield. On blank trials, the monkeys, but not the human, often failed to respond, and when they did respond, all hemianopes almost invariably touched a target position in the blind field. Analysis of reaction times showed that necessarily false responses to blank stimuli took longer than responses to blind field targets. However, apart from one hemianopic monkey, incorrect target responses took as long as responses to blank stimuli. The human hemianope showed the same pattern of reaction times as the hemianopic monkeys unless he had to report on stimulus awareness and confidence. Then, his confidence reports and response times mirrored his awareness of the stimuli, but neither differed for correct versus false responses once these were separated for 'aware' versus 'unaware' trials. The hemianopic monkeys' response probability and reaction time data indicate that they, implicitly or explicitly, registered differences between target and blank stimuli and, in one case, even between false responses to blind-field and blank stimuli.