Abstract
The gist of a natural scene can be extracted very rapidly (e.g., Rousselet, et al., 2005), without focal attention (Li et al., 2002) and even by pigeons (Kirkpatrick et al., 2014). These findings seem to indicate that the extraction of scene gist is a prioritized process. Consistent with this assumption, one study showed that the gist of a natural scene could be largely extracted when it was crowded in the visual periphery (Gong et al., VSS 2015). However, no study has directly investigated whether the gist of scenes is more readily to extract than other information. Here we employed a visual crowding task to compare the extraction of scene gist (to categorize scene pictures into building, forest, highway and mountain) with the extraction of facial expression (to categorize faces into angry, fear, happy and neutral) and orientation (to discriminate the facing direction of letter "E" into left, right, up and down). In all three tasks, the target, either appeared alone (uncrowded condition) or crowded by two flankers (crowded condition), presented at three eccentricities (9°, 13°, 20°) to the left or right of the fixation for 100 ms. Results showed that the accuracy in the scene gist categorization task was always significantly higher than the accuracies in the other two tasks in the crowded condition, though this was not true in the uncrowded condition. When using accuracy in the uncrowded condition as the baseline to calculate crowding strength, we found that the crowding strength was significantly weaker in the scene gist categorization task than that in the other two tasks at all three eccentricities. These findings indicate that scene gist can better get through the bottleneck of visual crowding and may suggest that the extraction of scene gist is a prioritized process.
Meeting abstract presented at VSS 2018