August 2016
Volume 16, Issue 12
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2016
Rapid category learning: Naturalized images to abstract categories
Author Affiliations
  • Alison Campbell
    University of Victoria, British Columbia
  • James Tanaka
    University of Victoria, British Columbia
Journal of Vision September 2016, Vol.16, 400. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Alison Campbell, James Tanaka; Rapid category learning: Naturalized images to abstract categories. Journal of Vision 2016;16(12):400.

      Download citation file:

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

  • Supplements

Object categories are the perceptual glue that holds our visual world together. They allow us to recognize familiar instances and extend recognition to novel ones. Although object categorization has been studied using supervised learning techniques, less is known about how they are spontaneously acquired through unsupervised learning. In this study, we examined how temporal contiguity contributes to this spontaneous abstraction of object categories during passive viewing. We hypothesized that viewing exemplars of the same category closer in time would support better abstraction of the visual properties that distinguish one object category from another, and facilitate better category formation. Participants passively viewed a continuous sequence of 160 natural images of four warbler species (40 images per species). Images were presented serially for 500 ms per image with no visual masking. In a blocked condition, participants viewed images grouped by species (e.g., 40 images of Cape May warblers, followed by 40 images of Magnolia warbler, etc.). In a mixed condition, participants viewed images presented in random order. Participants then completed a "same/different" test using novel warbler images. A study image was presented for 500 ms, and then a test image was presented for 500 ms. Participants responded "same" if the images depicted warblers of the same species or "different" if they depicted different species. Participants in the blocked presentation condition performed reliably better on the same/different task (d' = 1.96) than participants in the mixed presentation condition (d' = 1.30, p < .05) and participants in a control condition who received no presentations prior to test (d' = 1.19, p < .01). Performance in the mixed presentation and control conditions did not reliably differ, p > .10. These results suggest that temporal contiguity may enhance the visual system's ability to rapidly extract statistical regularities involved in category learning.

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.