Abstract
Several theories propose that the phenomenon of visual crowding - the propensity for neighbouring objects to interfere with target recognition - is caused by a reduction in the spatial resolution of visual attention. We tested this hypothesis by examining the relative effects of visual crowding and attentional load on orientation discrimination. The task involved identifying the mean orientation (clockwise vs. anti-clockwise relative to vertical) of six spatially distributed orientation-filtered noise targets. These were arranged iso-eccentrically, either in isolation or in the presence of distractor objects (randomly oriented filtered noise), located at a constant distance between fixation and the targets. By measuring orientation-discrimination performance at various levels of target-orientation variability, equivalent noise analysis yields (a) the precision with which subjects can estimate the orientation of any one patch (local noise) and (b) the number of samples over which they are averaging (global sampling). In some conditions subjects also performed an attentionally demanding foveal “dual-task”. This was to report the orientation of a white ‘T’ amongst a dynamic stream of randomly oriented black ‘T’s embedded in noise, the proportion of which was varied to maintain 75% correct foveal performance. We report a dissociation between the effects of crowding and attention: whereas crowding induces increases in local noise, attention produces global undersampling. A dual-pass experiment (ie. Two runs using identical stimuli) reveals that crowding does not disrupt the relationship between stimulus-specific-agreement (across runs) and percent-correct performance, indicating that crowding is stimulus-driven rather than an effect of attentional localisation. This psychophysical dissociation between crowding and attention suggests that they are mediated by distinct neural mechanisms.
Funded by the Wellcome Trust