Abstract
The effect of size and contrast on motion direction detection thresholds has been successfully explained by center-surround interaction. The effect of the spatial extent of attention on the thresholds, however, has not been systematically studied before. Here we studied this via a novel experimental design and developed a model to explain the behavioral results. The stimulus consisted of drifting sinusoidal gratings with 98% and 2% Michelson contrast. A central grating (diameter: 1.5°) was presented either alone or surrounded by an annular grating (width:2.5° or 9.2°) with a 1.3° gap between them. Drift direction in the central and annular gratings could be the same or opposite. There were two attention conditions. In the narrow attention (NA) condition, participants (N=10) were asked to attend the central grating and report its motion direction. In the wide attention (WA) condition, participants first reported the drift direction of the central grating, then reported whether the central and annular gratings drifted in the same direction. This was done to encourage the participants to widen their attentional fields. In each trial, the duration of the presentation was adjusted following an adaptive 1-up 3-down staircase procedure, based on the participants’ judgments of the central grating drift direction. Results showed that surround suppression was significantly stronger in the WA condition compared to the NA condition. The magnitude of this effect increased as the stimulus size increased in 2% contrast condition, but did not change in 98% contrast condition. Next, we developed a model by incorporating the spatial extent of the attentional field in the normalization model (Reynolds and Heeger, 2009). The model could successfully predict the observed behavioral outcomes. These findings unveil the critical role of spatial attention on surround suppression and show that behavioral results and the effect of attention can be explained by a generalized normalization model.