Abstract
A model of context effects on pattern vision must account for results from several paradigms. We measured the detection threshold for a Gabor pattern as a function of the contrast of flanking Gabor patterns, and the contrast discrimination function (threshold vs. pedestal contrast) with and without flanking Gabor patterns of different contrasts. All the patterns had the same spatial frequency (4 c/deg), space constant (0.25 deg), and orientation (vertical), and when flankers were present they were aligned with the target and presented simultaneously (90 msec). The target was always 4 degrees to the left of the fixation point. The detection threshold is a smooth “dipper-shaped” function of flanker contrast and it is well fitted by the nonlinear excitation divisive inhibition model (Foley & Chen, 1999). In the absence of the flankers, the contrast discrimination function is sometimes a smooth dipper-shaped function and sometimes has a second minimum at high pedestal contrast. This second minimum is accounted for by the intrusion of a second mechanism sensitive to the target. In the presence of the flankers, the contrast discrimination function changes in a complex way. When the flanker contrast is low, thresholds are reduced at low pedestal contrast; when flanker contrast is high, thresholds are increased at all except the highest pedestal contrasts. In addition there is a local maximum at a pedestal contrast about 4 dB below the flanker contrast and a local minimum near the pedestal contrast equal to the flanker contrast. All of these results can be accounted for quantitatively by an elaboration of the model that adds two processes to account for the flanker effects: 1) the flanker not only adds to the divisive inhibitory signal, but also increases or decreases the effect of the target and the pedestal, 2) the flanker produces a local decrease in the contrast response function around the contrast of the flanker.