Purchase this article with an account.
Patrick Monnier; The disciminability of a target defined along multiple dimensions can be accounted for by probability summation. Journal of Vision 2004;4(8):684. doi: https://doi.org/10.1167/4.8.684.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
BACKGROUND & PURPOSE The majority of visual search models have an initial parallel stage where basic features (e.g., color, spatial frequency, orientation) are encoded by neural mechanisms. Signals from these mechanisms are subsequently combined in a limited capacity attention stage to ultimately form an object representation. Recent research has elaborated the limited capacity attention stage of these models by, among other things, emphasizing the importance of “top-down” processing. In the present study, benefits of coding a single target along multiple dimensions were assessed. Multidimensional targets may offer an advantage if signals from low level mechanisms can be combined. METHODS Difference thresholds were measured for briefly flashed stimuli using a yes-no task. A target differed from homogeneous distractors along a single dimension (e.g., chromaticity OR orientation) and along two dimensions (e.g., chromaticity AND orientation). Exp. 1: Elements were blobs and the target differed from the distractors either in chromaticity, luminance, and both. Exp. 2: Elements were elongated blobs and the target differed from the distractors either in chromaticity, orientation, and both. Exp. 3: Elements were Gabors and the target differed from the distractors either in chromaticity, spatial frequency, and both. Search performance was assessed at set sizes 2 and 8. RESULTS The multidimensional spaces were normalized using the single dimension thresholds. Thresholds were subsequently measured along diagonal directions (35, 45, 55 deg) in the normalized spaces. Thresholds for targets that differed from distractors along 2 dimensions were generally consistent with probability summation, regardless of the dimensions and set size. The threshold increases with set size were consistent with an unlimited capacity model of attention. CONCLUSIONS The results are consistent with a model where the parallel stage consists of independent signals that, in these conditions, were not attentionally combined to facilitate search.
This PDF is available to Subscribers Only