August 2016
Volume 16, Issue 12
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2016
The power of populations: How the brain represents features and summary statistics
Author Affiliations
  • Shaul Hochstein
    ELSC Safra Brain Center, Neurobiology Department, Life Sciences Institute, Hebrew University, Jerusalem
Journal of Vision September 2016, Vol.16, 1117. doi:https://doi.org/10.1167/16.12.1117
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Shaul Hochstein; The power of populations: How the brain represents features and summary statistics . Journal of Vision 2016;16(12):1117. https://doi.org/10.1167/16.12.1117.

      Download citation file:


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

      ×
  • Supplements
Abstract

Introduction. Much recent interest has been directed at perception of summary statistics. As discussed in a variety of presentations at this and previous VSS meetings, intensive research has uncovered diverse dimensions of summary statistic perception, including simple and complex dimensions (from circle size and Gabor orientation to face emotion and attractiveness), the type of statistics acquired (mean, variance, range), and our ability to summarize elements presented simultaneously or sequentially and to divide displays into separate groups, detecting statistics of each. Methods. I now tackle a central question that remains unanswered: How does the brain compute scene summary statistics without first attaining knowledge of each scene element? One possible solution is that the brain uses implicit individual element information to compute summary statistics, which become consciously accessible first. I show that this added step is superfluous. On the contrary, direct acquisition of summary statistics is not surprising and no novel computational principle is required for summary perception. Results. A simple population code representation, as found for single element parameters, may be scaled up to compute mean values for element groups. The range of active neurons is broader, but the computation is the same for sets of elements as for a single element. Using a population code adds tremendous power, as it allows direct determination of which elements to include in the set, which elements are outliers – to be excluded and trigger pop out attention – and how to divide between simultaneously presented sets. Conclusion. Population coding provides a direct and efficient representation of set summary statistics. As suggested by Reverse Hierarchy Theory, conscious perception may begin with summary statistics, seeing many similar elements as a group, and only later focus attention to individual elements. Interestingly, a similar population code representation may underlie categorization, including both category prototype and its boundaries.

Meeting abstract presented at VSS 2016

×
×

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.

×