Purchase this article with an account.
Thiago L Costa, Andrey R Nikolaev, Cees van Leeuwen, Johan Wagemans; Attenuated brain responses to Gestalts at threshold: differential predictive processing behind Gestalt phenomena?. Journal of Vision 2019;19(10):36d. doi: https://doi.org/10.1167/19.10.36d.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The aim of this work was to assess bias and expectation effects towards Gestalts and their corresponding neural correlates, in a task with minimal visual search or salience confounders. We presented stimuli at threshold to equate task difficulty and salience between different stimulus types. By making target discrimination more demanding, we make room for stronger expectation effects. We used a change detection task to measure responses to the emergent features of proximity and orientation. The base stimulus was a pair of dots. The dots could change in their configuration (orientation / proximity, i.e. Gestalt conditions) or change location without changing configuration (two control conditions, moving horizontally or vertically). To assess expectations, we had two types of blocks: blocks where only one stimulus type was presented (blocked condition) and blocks where all stimuli were presented with equal likelihood (mixed condition). We performed a QUEST psychophysical procedure (manipulating the amount of displacement of the dots) prior to the EEG recording to ensure that all stimulus types were presented at 81% correct performance levels for each participant. In summary, the task measures implicit expectations, without substantial target selection or distractor suppression demands and requires no classification of stimulus type: responses are based on change of location. As planned, we found no relevant significant differences between stimulus types at accuracy or RTs (except for proximity being different from control in one of the blocks). Analyses of the P1, N1, P2 an P3 event-related potentials showed that Gestalt conditions generally led to smaller responses than control conditions (significant from N1 to P3). This effect was larger in the mixed condition and also larger for orientation. Analyses of prestimulus activity will be performed, but these results already provide initial support for the hypotheses derived from reverse hierarchy theory and predictive coding models of visual processing.
This PDF is available to Subscribers Only