Abstract
Contour erasure describes the phenomenon that objects completely disappear or are filled in by the background after being adapted to flickering contour outlines that match the objects edges. Such pre-adaptation of the contours drastically speeds up the filling-in process. Here, we quantitatively examined the mechanisms underlying such a contour erasure effect by measuring target contrast discrimination threshold following contour adaptation. The target and pedestal were homogenous disks of 1-degree radius of various luminance contrast levels with respect to the background. The contour adapters were high-contrast rings of 0.2-degree thickness that superimpose spatially on the edges of the target and pedestal. In a two-alternative forced-choice task, on each trial, two contour adapters positioned on the upper-right and lower-left quadrants first flickered at 3Hz for 3s. After an 83.3ms inter-stimulus interval, the pedestals were presented at both positions for 83.3ms, whereas the target was superimposed randomly on one of the two pedestals for the same period. The observers were instructed to indicate the target location by pressing the corresponding key. A Bayesian adaptive staircase algorithm determined the target contrast for the next trial. We defined the target threshold as the contrast level at 86% accuracy. Results showed that the target threshold first decreased then increased as the pedestal contrast. The target threshold was raised after contour adaptation regardless of the pedestal contrast. Such an adaptation effect increased with eccentricity (from 3 to 11 degrees). We fitted the psychophysics data with a contrast gain control model, in which the pattern response is determined by an excitatory component raised by a power then divided by an inhibitory component plus a normalizing constant. The adaptation effect can be captured by changes in the sensitivity parameters and a normalizing constant. We are the first to investigate the impact of contour erasure as a function of contrast.