Abstract
When a sensory feature itself is too ambiguous to judge its magnitude, we can interpret it by relying on prior knowledge; co-occurrence relation with other sensory features. For example, if one must judge the brightness of outdoor illumination (bright versus dark), the subjective brightness of illumination is very ambiguous. Depending upon co-occurring features such as time of day (daytime or night), one might be biased to judge the illumination as either bright or dark. This study investigated whether biasing effects of several co-occurring features on participants' judgments of a target feature would disappear in situations where the target feature and co-occurring features had no correlation. Subjects had to judge whether the visual contrast of a Gabor patch was high or low while ignoring the visual orientation of this patch and ignoring the concurrent sound of a pure tone. The contrast of patches was mostly high or low, sometimes medium. Over trials, the probabilities of co-occurrence between visual contrast and visual orientation (rightward or leftward) and that between contrast and auditory loudness (loud or soft) were both controlled. We created three conditions: uncorrelated contrast-orientation, uncorrelated contrast-loudness, and uncorrelated contrast-orientation/contrast-loudness conditions. In each condition we examined whether subjects' judgment of visual contrast in high and low alternatives to medium contrast changed. If subjects used co-occurring features in judging visual contrasts, then judgments for mid-contrast should not be chance level; they should show systematic biases responding for uncorrelated relationships between target contrast and patch orientation and/or those between target contrast and tone loudness. Results indicated that participants' mid-contrast judgments were biased both by visual orientation and by tone loudness initially. However, on later trial, judgments of mid-contrast targets did not differ significantly from chance levels. These findings suggest that the influence of co-occurring features on visual contrast percepts could disappear.
Meeting abstract presented at VSS 2016