Purchase this article with an account.
Omer Daglar Tanrikulu, Andrey Chetverikov, Arni Kristjánsson; Variance modulates temporal weighting during integration of sequentially presented visual ensembles. Journal of Vision 2019;19(10):193. doi: https://doi.org/10.1167/19.10.193.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
While most studies on visual ensembles focus on spatial integration of simultaneously presented groups of items, recent studies have shown that observers can also estimate summary statistics from sequentially presented items. Very little is known, however, about how these spatial and temporal summaries are combined and whether such combined information is used during perception. We examined how observers temporally integrate two different orientation distributions that were presented sequentially in a visual search task. We manipulated the variance of the two distributions to investigate the influence of variance on their temporal integration. Participants performed streaks of sequential odd-one-out visual search for an oddly oriented line among distractors. In a streak of sequential learning trials, the distractor orientations were sampled from two different Gaussian distributions on alternating trials. After a certain number of trials, observers were given a test trial where the orientations of target and distractors were switched which resulted in slowed search due to role reversal effects. The reaction times from test trials revealed observer’s internal model of distractor distributions. The variance of distractor distributions and the number of learning trials were manipulated. Our results revealed that summaries of orientation ensembles were largely biased by the orientation distribution presented in the last trial of a streak. However, this bias interacted with the variance of the two alternating distributions. When the variances of the distributions were relatively large, observers took into account the mean of the distractors from the earlier trial, which, in turn, weakened the influence of the last trial. These results indicate that complex weighting of information takes place in temporally encoded ensembles, where the recency effect depends on the variance of the integrated distributions.
This PDF is available to Subscribers Only