Number blindness in human vision

There is an ongoing controversy over whether human vision first estimates area and number, deriving our sense of density via division, or if it first estimates area and density, deriving our sense of number via multiplication. If number and area are both primary independent dimensions of visual perception then we should observe cross-magnitude influence between them in a simple choice task, especially if that influence would improve performance and this is explicitly explained to the participants. In contrast, here we show that human vision exhibits a specific kind of number blindness: performance on an area-choice task (which of these rectangles is larger?) is not improved by the addition of a perfectly correlated number signal (the larger one always has more dots on it) that creates equivalent density - even when explanations, reminders, and accurate feedback are given to the participants. This replicated across two experiments (N = 82, 122) with slightly different stimuli. Control analyses with brightness in Experiment 1 indicate that this is not a general resistance to the predicted cross-magnitude influence. This indicates that density, not number, is the primary independent perceptual dimension in human vision.