Abstract
Our previous experiments showed that the strength of facilitation to the detection of a target bar by collinear, bilateral, flanking bars first decreases and then increases with the contrast of the contextual bars, making an U-shaped curve of detection threshold vs. contextual contrast (Jingling and Zhaoping, 2005, Applied Vision Association Meeting, Aston, UK). Here, we further explore the effect of the spatial configuration in this contrast-dependent modulation by contextual bars. The detection threshold of a vertical target bar is measured with a staircase method by a two temporal-interval alternative-force-choice procedure. The target bar is always at the center of the screen, and the contextual bars are of the same shape and orientation as the target. Four contrast values, 0%, 1%, 5%, and 20%, of the contextual bars are used. We found that the detection of the target could be facilitated by the contextual bars when they form a regular pattern with the target (e.g., a collinear or parallel array of bars, or two arrays crossing into a X shape, or a regular texture of bars). This is not due to uncertainty reduction of the target location, since replacing the context by four dots around the target does not facilitate detection. In addition, when contextual bars in the regular configurations are not collinear to the target, the detection threshold varies monotonously (rather than U-shaped) with the contrast of the contextual bars. We suspect that the spatial configuration-dependent modulation by the context is beyond the known mechanism in the primary visual cortex.
Gatsby Charitable Foundation