September 2018
Volume 18, Issue 10
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2018
Optimal change detection without ensemble statistics
Author Affiliations
  • William Harrison
    Department of Psychology, University of CambridgeQueensland Brain Institute, The University of Queensland
  • Paul Bays
    Department of Psychology, University of Cambridge
Journal of Vision September 2018, Vol.18, 190. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      William Harrison, Paul Bays; Optimal change detection without ensemble statistics. Journal of Vision 2018;18(10):190.

      Download citation file:

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

  • Supplements

Despite experiencing a richly detailed visual world, our ability to remember the appearance of objects even for a fraction of a second is greatly limited. Recent studies have suggested that this limitation may be circumvented by storing higher-order regularities of a scene (i.e. ensemble or summary statistics) that typically aren't captured in lab-based experiments. Here we investigated the influence of ensemble statistics on working memory using a modified change detection task. An observer's task was to remember the colors of items in a sample display, and report whether a test display, presented one second later, was the same or different. Memoranda were two (Experiment 1) or four (Experiments 2 and 3) colored disks, randomly chosen from a circular color space. On change trials in Experiment 1 and 2, changes to individual colors were chosen to keep the mean the same, while changing the variance, or vice versa. In Experiment 3, we included a condition in which both the mean and variance changed. In Experiments 1 and 2, we found consistent evidence that sensitivity to a change in the mean color is the same as to sensitivity to a change in color variance. In Experiment 3 we found that sensitivity is not improved when both the mean and variance change, revealing these statistics do not additively influence performance. Indeed, rather than finding these ensemble statistics influenced memory performance, our results instead were precisely predicted by an ideal observer model that optimally summates evidence from each individual item independently of the group mean or variance. These results present a challenge to the claim that ensemble statistics are automatically extracted and stored in visual working memory.

Meeting abstract presented at VSS 2018


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.