September 2017
Volume 17, Issue 10
Open Access
Vision Sciences Society Annual Meeting Abstract  |   August 2017
The role of uncertainty in perceptual organization
Author Affiliations
  • Yanli Zhou
    Center for Neural Science, New York University
    Department of Psychology, New York University
  • Luigi Acerbi
    Center for Neural Science, New York University
  • Wei Ji Ma
    Center for Neural Science, New York University
    Department of Psychology, New York University
Journal of Vision August 2017, Vol.17, 746. doi:10.1167/17.10.746
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Yanli Zhou, Luigi Acerbi, Wei Ji Ma; The role of uncertainty in perceptual organization. Journal of Vision 2017;17(10):746. doi: 10.1167/17.10.746.

      Download citation file:


      © ARVO (1962-2015); The Authors (2016-present)

      ×
  • Supplements
Abstract

Perceptual organization is the process of grouping scene elements into whole entities, for example line segments into contours. Previous studies have reframed classic Gestalt principles of perceptual organization in terms of Bayesian models, in which the observer computes the probability that a whole entity is present in the scene. These studies, however, leave open the possibility that people apply a fixed, learned stimulus-response mapping that only mimics Bayesian inference, instead of actually computing with probability distributions. Proper probabilistic computation requires people to flexibly take sensory uncertainty into account even when it varies from trial to trial. Here, we vary uncertainty from trial to trial to distinguish between probabilistic and non-probabilistic inference in a simple form of perceptual organization. Subjects (n = 8) judged whether two line segments separated by an occluder were collinear. In this task, an optimal observer would be probabilistic and utilize knowledge of uncertainty when deciding whether a measured offset between the line segments is due to non-collinearity or to sensory noise. We compare this model against an alternative model that applies a fixed, uncertainty-independent decision boundary. Finally, motivated by a nonparametric examination of the data, we also test a probabilistic heuristic model whose decision boundary is linear in eccentricity. The fixed-boundary model fits by far the worst (leave-one-out cross-validation score, fixed - optimal = -25.5 ± 13.6, fixed - heuristic = -82.8 ± 15.2; mean ± SEM across subjects), providing evidence for probabilistic computation. Moreover, we find that the heuristic model performs better than the optimal model (leave-one-out score, heuristic - optimal = 57.3 ± 10.5), suggesting that people take uncertainty into account in a suboptimal way. The model comparison did not change qualitatively when we estimated parameters from an independent discrimination task. Our work opens the door to investigating the role of uncertainty in more natural forms of perceptual organization.

Meeting abstract presented at VSS 2017

×
×

This PDF is available to Subscribers Only

Sign in or purchase a subscription to access this content. ×

You must be signed into an individual account to use this feature.

×