Purchase this article with an account.
Jocelyn Sy, Andrew Tomarken, Vaama Patel, Randolph Blake; Distributional analyses of individual differences in binocular rivalry dynamics. Journal of Vision 2017;17(10):582. doi: https://doi.org/10.1167/17.10.582.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Binocular rivalry is the dissociation of physical stimulation and perceptual experience that occurs when conflicting monocular stimuli are imaged on corresponding areas of the eyes sparking alternations in perceptual dominance between the two eyes' views. There is increasing evidence for individual differences in rivalry dynamics with important correlates to neural inhibitory processes (e.g., GABAergic signaling) and function (e.g., autism and schizophrenia). While test-retest correlations between summary measures (e.g. means) have been used to demonstrate individual differences, this approach ignores potentially important distributional features of the data. We used a novel non-parametric approach to assess the within-subjects consistency of rivalry distributions. An intensive design was used in which 15 females viewed rivalry stimuli across 15 sessions consisting of 20 60-second blocks per session. We used kernel density estimation to generate smoothed, and minimally assumptive, distributions of percept durations for each individual for the odd and even sessions. We then computed the overlap coefficient (OC; potential range = 0 to 1) between odd and even distributions by numerical integration of the region of intersection. We found a remarkably high degree of within-subjects consistency: mean and median OCs = .95 (range = .91 to .97 ). Within-subject OCs were markedly higher than OCs with other subjects (mean OC=.63), thus indicating that the former reflected more than stereotype consistency. Permutation tests indicated that for 4 individuals we could not reject the null hypothesis that OCs=1. We classified with high accuracy the single-session performance of viewers from their aggregate scores across independent sessions. These results indicate (1) Marked consistency in the distributions of binocular dominance durations that are characterized by a unique subject-specific "signature"; and, (2) The utility of non-parametric density estimation and the overlap coefficient for quantification of similarity and individual differences in perception studies.
Meeting abstract presented at VSS 2017
This PDF is available to Subscribers Only