Abstract
Background. A short delay between the presentation of a reference and a test stimulus in a two-alternative hue estimation task biases estimates toward the average stimulus value (Olkkonen & Allred, VSS 2013). Here we characterize the conditions for the bias by 1) manipulating the delay between reference and test ("internal noise" ) and 2) manipulating the chromatic variability in the reference ("external noise"). Methods. Two-alternative hue judgments were collected in two experiments for stimuli varying on the blue-yellow continuum. In Experiment 1, observers made judgments between two patches displayed on left/right of fixation across delays of 0.4, 2, and 4 seconds, interleaved. In Experiment 2, a new group of observers made judgments across a fixed 2 second delay with three interleaved levels of chromatic noise in the reference stimulus. In both experiments, the reference was always presented first; three interleaved references were employed. Test hue was varied to measure psychometric functions (PMF), from which bias and precision were estimated. Bias was defined as the difference between each point of subjective equality (PSE) and veridical reference hue; precision was the reciprocal of discrimination threshold (75th -50th percentile of the PMF). Results. Experiment 1. Hue estimates across the shortest delay were not biased, but a central tendency bias emerged with the longer delays. Thresholds increased slightly overall with delay, but there was no significant relationship to bias. Experiment 2. Hue bias increased monotonically with increasing chromatic noise in the reference. Thresholds increased slightly overall, and were moderately correlated with bias magnitude across observers and noise level (r=0.45, p<0.001). Conclusion. Biased hue percepts can be elicited both with an internal (delay) and an external (chromatic variability) noise manipulation. The difference in relationship between bias and thresholds for the two manipulations suggests that different mechanisms might underlie hue appearance bias with external vs. internal noise.
Meeting abstract presented at VSS 2014