Abstract
According to the Selective-Tuning model (Tsotsos, 1990), convergence of neural input and selection of a attended feature will result in surround suppression within that feature domain: the nearby feature values in the same feature dimension will be inhibited, but not the feature values farther away in the dimension. We present three experiments that support this hypothesis. The first experiment used a feature-cue paradigm. The subjects' attention was first attracted to an orientation cue, and then subjects made a perceptual judgment about a stimulus with same or different orientation as the cue. We found that orientation attention actively suppressed the nearby orientations (5 ∼ 10 degree from the cue), but did not influence far away orientations (20 ∼ 90 degree from the cue), replicating the results of Tombu & Tsotsos (2008). In the second experiment we used the same paradigm but added a distractor to the cue, and the stimulus sequence became cue -[[gt]] distractor -[[gt]] probe. This time the subjects must actively ignore the distractor to focus their attention on the cue. By increasing the difficulty of the subjects pay attention to the cue, we found an even stronger Mexican-hat profile of attentional suppression. In the third experiment, we extended our findings from orientation to color. Again, we acquired a Mexican-hat profile of the feature attention in color dimension. These results further extend the evidence supporting the Selective Tuning explanation of spatial and feature attention and suggest a more general mechanism of signal-to-noise enhancement applicable to any feature dimension.