Abstract
Perception represents the world probabilistically. If memories are approximate markers, where will an observer end an action intended to identify a remembered location? We investigated this question straightforwardly. Participants viewed a target dot, followed by a mask. At test, a mouse cursor appeared, and participants moved the cursor to the remembered center of the initial target dot. So that we could evaluate the location where a response was made relative to the starting point of an action, we manipulated the starting position of the mouse cursor. We found a clear effect of whether the mouse began above or below the true target location: when it began above, the mouse was moved, on average, too far (expansion). And when the mouse began below, it was moved too little (compression). Helping to explain these effects, response distributions were always centered below a target’s true position. Thus, observers ‘aimed’ for the same point —below a target— leading to observed compression and expansion depending on the mouse starting position. We hypothesize that this occurs because observers fixate the remembered location while executing the mouse movement, and because attentional resolution is better below compared to above fixation. Eye tracking experiments explore this hypothesis, as well as experiments with alternative response methods. But readers can confirm for themselves: where do you fixate as you move a mouse to an icon position? Practically, these results are important because many studies use mouse responses to investigate the spatial structure of perception without accounting for the mouse starting point, attentional resolution, or fixation locations. Theoretically, these results identify an interaction between attention, perception and fixation when translating perceptions into actions. Specifically, higher resolution locations, with respect to fixation, act as attractors to an executed action. As a result response distributions become distorted relative to the distributions stored in memory.
Meeting abstract presented at VSS 2013