Abstract
To measure the acuity of visual attention, a search target is placed in one of several to-be-atttended areas surrounded by unattended areas filled with false targets. Shaping the attended areas into gratings of different spatial frequencies enables a Fourier Analysis derivation of the spread function of spatial attention. The assumptions are: all stimulus elements are represented in perception with a certain amount of random error, targets and false targets have an incrementally greater mean perceptual strength s than distracters, s is attenuated according to the spatial frequencies of the attended-unattended area arrangement. A Fourier-derived attention spread function accurately accounts for the performance of observers in the 20 such grating displays used to derive it. To test the theory, observers were trained to detect targets in new displays in which to-be-attended areas of were randomly generated. In moderately complex random displays, the Fourier theory, with one new "difficulty" parameter predicted performance in the 72 (of 144) attended locations. Ultimately, in still more complex displays, observers partially or completely neglected certain to-be-areas thereby indicating capacity limits of attention that depend both on the number of areas to-be-attended and the shape-complexity of the to-be-attended areas.
Meeting abstract presented at VSS 2014