Purchase this article with an account.
Igor Utochkin, Anton Gura; Seeing variety: The determinants of visual representation of variance statistics. Journal of Vision 2014;14(10):878. doi: https://doi.org/10.1167/14.10.878.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Visual summary statistics (such as the average along a sensory dimension) are efficient to construct a compact description of multiple objects when access to their individual features is limited. A number of recent studies have reported that variance is an important statistical feature affecting averaging (Corbett et al., 2012; Fouriezos et al., 2008; Im & Halberda, 2013). In our study, we directly addressed the ability to extract summary variance statistics from visual sets. In all experiments, observers were briefly presented with sets of differently sized circles to the right and left from fixation and had to respond which one was more various. In Experiment 1 two factors – bandwidth and heterogeneity (the number of unique sizes per set) – were manipulated. We found in the result that bandwidth appears a predominant factor of variety discrimination. In Experiment 2, the smoothness of size distribution was manipulated within the constant bandwidth via additional intermediate sizes between smallest and largest items. We found that the sharpest two-peaks distributions consisting only of extremes are estimated as being more various than smoother ones. In Experiments 3-5, we used the same factors as in Experiment 1 and tested whether variety discrimination depends on additional factors such as mean size of circles, their spatial density, or numerosity. In Experiment 6, we precued either of sides from fixation drawing attention to one of the sets. Although mean size, density, numerosity, and attention manipulation elicited biases in judging variety of identical distributions, they did not affect overall discriminability of different distributions replicating the results of Experiment 1. So, the only found principal determinants of variety perception were those actually related to stimulus physical variance. The results of Experiments 3-5 also show that statistical representations of variety are rather abstract and and Experiment 6 shows that they are extracted in parallel.
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only