Johan Wagemans¹ (), Claudia Damiano², Eline Van Geert¹; ¹KU Leuven, Department of Brain & Cognition, ²University of Toronto, Department of Psychology

Context effects and sequential effects are ubiquitous and well-known in vision science. Lesser-known but also important are set effects: Perception of stimulus attributes is affected by the set within which the stimuli are embedded. This phenomenon has been studied as “Einstellung” in the early days of psychology but has been largely forgotten or ignored by contemporary vision scientists. Here we demonstrate the role of set effects on a wide range of ratings of various kinds of images. 1839 participants rated 810 stimuli (90 stimuli x 9 categories: fractals, patterns, textures, scenes, art photographs, paintings–some figurative, some abstract) on 5 dimensions with 7-point bipolar scales (complexity, order, interest, pleasure, beauty). The critical manipulation was to have 9 homogeneous blocks (1 block of 90 stimuli from 1 category only) and 9 heterogeneous blocks (1 block of 10 stimuli from each of the 9 categories), assigned to different groups of participants. Clear differences were obtained between these conditions: the variance of the ratings was always higher for the heterogeneous blocks than for the homogeneous blocks. For all rating scales, categories with lower ratings on average were scored lower in heterogeneous than in homogeneous blocks, and categories with higher ratings on average were scored higher in heterogeneous than in homogeneous blocks. Heterogeneous blocks thus allowed participants to differentiate the stimuli better, while homogeneous blocks induced regression to the mean. This effect was more pronounced for ratings of more subjective attributes (beauty, pleasure, interest) but still present for more objective attributes (order, complexity). Intercorrelations between the scales were systematically higher for heterogeneous blocks. Subjective complexity ratings correlated more strongly with objective complexity measures for more abstract stimulus categories, and these correlations were generally higher for homogeneous than for heterogeneous blocks. This study clearly demonstrates the presence of set effects on judgments of stimulus attributes.

Acknowledgements: This research has been funded by an ERC Advanced Grant (no. 101053925, GRAPPA) awarded to JW and an FWO postdoctoral fellowship (no. 1244825N) awarded to EVG.