Abstract
Detection of low-contrast luminance-defined stimuli can involve spatial summation over a large portion of the visual field. For example, contrast thresholds for grating detection decrease as a function of the width of the grating, up to a width of about 8–10 cycles. There is evidence, however, that high levels of stimulus noise may shut down long-range spatial pooling. Kersten (1984), for example, found that contrast thresholds for grating detection in noise bottomed out for gratings only one cycle in width. Here we use a classification image technique to directly test for variations in the extent of spatial summation as a function of noise contrast. Stimuli were large (26 deg) vertical gratings. Classification images for grating detection were estimated at grating frequencies of 0.5 and 1.7 cpd and noise contrasts ranging from 4%–50%. Classification images were fit with a 4-parameter Gabor model tuned to the frequency and phase of the signal. In all conditions, the estimated summation fields extended over many cycles of the stimulus. No systematic variation in the extent of summation as a function of noise contrast was observed. Linear and nonlinear pooling models were evaluated in terms of trial-by-trial consistency with the human data. In general, probability summation over localized responses of broadband (1.7 octave) detectors was found to better predict human judgements than a purely linear model. Our results suggest that the extent of spatial summation for grating detection is relatively unaffected by high contrast stimulus noise.
We speculate that Kersten's previous results may be explained by an increase in spatial uncertainty with stimulus size.
This work was supported by grants from NSERC, GEOIDE, CRESTech and the PREA.