Eye movements do not preferentially test inferences in the blind spot

Humans make eye movements to regions of high uncertainty to maximize the information gain in their visual search. Along with veridical sensory information, there is also perceptually inferred information that arises at the gaps caused by anatomical or environmental factors. It is unclear how those inferences are treated in comparison to veridical information during search behavior. Here, in two experiments, we tested if eye movements are preferentially directed towards the blind spot as an area of high uncertainty and high information gain in a monocular visual search task. The results show that the first saccade was not directed primarily to the blind spot when "invisible" targets in the left and right blind spot occured interleaved. Only when viewing conditions were blocked, such that "invisible" targets occured always in the same blind spot side, participants learned to look preferentially at the blind spot. These results show that perceptual inferences in the blind spot are not preferentially tested by eye movements in general, but that they can be optimized by using contextual information.