Abstract
PURPOSE. Although color memory has been extensively studied, its relationship to color perception is controversial. We investigated this relationship for achromatic color by characterizing the bias and precision for both perception and memory for lightness within the same experimental paradigm.
METHODS. The task was lightness discrimination. On each trial, a reference and a test stimulus were presented horizontally on either side of fixation for 500 ms, either simultaneously or separated by 2.5 seconds. The observer indicated which one appeared lighter. Reference stimuli were either decrements or increments relative to their background. The test stimulus background matched the reference background (symmetric condition) or was lighter (asymmetric condition). Interleaved staircases controlled the intensity of the test stimulus. Background and delay conditions were blocked. Points of subjective equality (PSE) and discrimination thresholds were determined from a cumulative Gaussian fit to the proportion lighter data. We defined bias as the shift of the PSE relative to the reference stimulus, and precision as the reciprocal of the threshold.
RESULTS. 1. Bias: The asymmetric background caused a shift in the PSEs in the simultaneous comparison as expected. For both increments and decrements, the background bias was largest for low-contrast reference stimuli (~20%) and decreased with increasing reference contrast. Interestingly, there was a bias of up to 15% in the symmetric delayed condition, but the pattern across reference stimuli was more complex: in some cases the bias had an opposite sign for decrements and increments. Finally, the delay and background biases approximately added in the asymmetric delayed condition. 2. Precision: Background did not have a measurable effect on thresholds in the simultaneous condition. The delay caused an increase in thresholds for both decrements and increments on both backgrounds.
CONCLUSION. Under these experimental conditions, memory for lightness appears to be both biased and less precise relative to perception.
Meeting abstract presented at VSS 2012