Abstract
A widely accepted view in vision science is that humans exhibit spatiotopic adaptation to perceived number. This conclusion is supported by over thirty published studies and is bolstered by a variety of phenomenologically compelling demonstrations. Moreover, number adaptation is theoretically significant: It is offered as one of the strongest pieces of evidence that number is a primary visual attribute (akin to other features like color, size, or motion; Burr & Ross, 2008). Here, we report a series of three pre-registered experiments that challenge the existence of genuine number adaptation. In Experiment 1, we show that apparent reductions in perceived number (that ostensibly result from visual adaptation to a large number) are significantly affected by the physical overlap of items in the adaptor and test displays. This suggests that classic demonstrations of number adaptation may simply result from observers adapting to items and their features rather than number per se. What about supposed cases of reverse adaptation, where adapting to a small number of items increases perceived number in a test display? In Experiment 2, we replicate reverse adaptation effects in a double-adaptor experiment (wherein observers adapt to a small number of dots on one side of space and a middling number on the other). But in Experiment 3, we take each trial from Experiment 2 and ‘split it in half’; observers now only see one adaptor at a time. Here, we show that perceived number is *not increased* when observers adapt to low numbers but instead that it is *decreased* when observers adaptor to middling numbers — the opposite of what a number adaptation account should predict. Cumulatively, these results provide strong evidence against the idea of genuine number adaptation and suggest a novel alternative: that humans perceptually adapt to perceived items, but do not adapt to their numerical quantity.