Purchase this article with an account.
Deborah Apthorp, Jason Bell; Symmetry: Less than meets the eye. Journal of Vision 2014;14(10):391. doi: https://doi.org/10.1167/14.10.391.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Introduction: Symmetry is a ubiquitous feature in visual scenes and human observers are highly sensitive to it. Does the presence of symmetry bias our estimate of scene content? We consider this question in the context of numerosity judgments. Recently, it has been suggested that a simple model using the relative response of mechanisms tuned to low and high spatial frequencies can predict systematic errors in both number and density estimations by human observers (Dakin et al., PNAS 2011). Here we ask whether these estimations can also be biased by the higher-order statistics of the displays. Method: We asked observers to estimate the relative number of elements in symmetrical compared to asymmetrical dot displays, using a 2-interval, forced-choice paradigm with adaptive staircases (QUEST). Reference displays contained 50, 100 or 200 dots, and comparison displays were adjusted to obtain the point of subjective equality (PSE). To assess whether the effect was due to perceived global structure within the symmetric display, we also tested with concentric Glass patterns (which had global structure but not symmetry) compared to randomly-oriented Glass patterns. Results: Symmetrical displays were consistently judged as less numerous than asymmetrical displays, and this effect persisted across different dot numbers, dot densities and axes of symmetry. Symmetrical displays required approximately 10% more elements to appear as numerous as asymmetrical displays, although the effect was smaller at higher dot numbers and at oblique axes of symmetry. For Glass patterns, the effect was greatly reduced, and was abolished at higher element numbers, suggesting the bias could not be attributed solely to the perceived structure in a symmetric pattern. We discuss the results in terms of current models of number and density judgments.
Meeting abstract presented at VSS 2014
This PDF is available to Subscribers Only