Abstract
Spatial attention alters figure-ground perception as we often experience in ambiguous figures. Physiological studies have reported that figure-ground process begins from border-ownership (BO) selective cells in V1/V2. We investigated the role of spatial attention applied to early vision with a computational model consisting of V1, V2, and Posterior Parietal (PP) modules. The PP module is designed to represent spatial attention that could be considered as a saliency map based on luminance contrast. In the model, spatial attention alters contrast gain in the V1 module so that it enhances local contrast. The change in contrast signal then modifies the activity of BO-selective neurons in V2 because BO is determined solely from surrounding contrast. The simulation result showed that the direction of figure (DOF) could be flipped in ambiguous stimuli, depending on the attentional location. Although the activities of model BO cells were modulated, their DOF did not alter for unambiguous stimuli. To evaluate the model quantitatively, we carried out psychophysical experiments to measure the effects of attention on the perception of BO for ambiguous random-block stimuli, and then compared the results with those of corresponding simulations. The model showed good agreement with human perception for modulation magnitude and its variance among stimuli. When the stimulus included a familiar shape such as a square, human determination of BO shifted in the direction of the familiar shape, whereas our model did not exhibit such a shift. However, there was no significant difference in the modulation magnitude between the model and psychophysical data. This disagreement in BO shift, and agreement in modulation magnitude seems natural if spatial attention works without the influence of familiarity or feature-based attention in early vision. These results suggest that the activity of BO-selective neurons could be modulated significantly by spatial attention that alters local contrast gain in early vision.
Supported by grant-in-aid from the Okawa Foundation, JSPS and Ministry of ECSST of Japan (KAKENHI 19530648 & 19024011).