Michael Cohen^1,2, Helen Feibes³, Mabel Shanahan¹; ¹Amherst College, ²MIT, ³NIH

How much do observers believe they can see across the visual world, and how accurate are these intuitions. Here, we asked participants to make a series of predictions about their perceptual abilities across the visual field and then tested the accuracy of those predictions. First, we asked observers to predict how far out into the periphery they could tell whether 1) a face was Taylor Swift or Margot Robbie, 2) an object was big or small, 3) a scene was entirely colorful or desaturated in the periphery, and 4) a scene was entirely preserved or scrambled in the periphery. In this case, we used a custom-built mechanical device that allowed observers to move their arms into the periphery to “show” how far out they believed they could identify faces, objects, colors, and scenes. After extensively showing participants the stimuli at fixation, they predicted how far they could identify these stimuli at their 50% threshold. Next, we placed observers in a custom-built display dome that used a projection system to create a 180° field of view across the horizontal axis and used a staircase procedure to identify the point at which observers would were at chance with these tasks. In this case, the results were unambiguous: Observers drastically overestimated their perceptual abilities. For example, participants believed they could tell Taylor Swift from Margo Robbie 53° into the periphery, when they could only do it 17° into the periphery. Taken together, these results demonstrate that observers do not see nearly as much of the world across their visual field as they believe they do. The fact that these effects hold for numerous stimulus types and to such a large degree shows that it is a ubiquitous aspect of visual perception: Observers do not perceive nearly as much as they think they do.