April 2014
Volume 55, Issue 13
ARVO Annual Meeting Abstract  |   April 2014
Top-Down Effects on Categorizing Incomplete Complex Scenes
Author Affiliations & Notes
  • Beliz Hazan
    Psychology Program, Graduate Center of CUNY, New York, NY
  • Daniel D. Kurylo
    Psychology Program, Graduate Center of CUNY, New York, NY
    Psychology, Brooklyn College, New York, NY
  • Zeynel Baran
    Experimental Psychology, Hacettepe University, Ankara, Turkey
  • Naomi Bowens
    Psychology, Brooklyn College, New York, NY
  • Footnotes
    Commercial Relationships Beliz Hazan, None; Daniel D. Kurylo, None; Zeynel Baran, None; Naomi Bowens, None
  • Footnotes
    Support None
Investigative Ophthalmology & Visual Science April 2014, Vol.55, 786. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Beliz Hazan, Daniel D. Kurylo, Zeynel Baran, Naomi Bowens; Top-Down Effects on Categorizing Incomplete Complex Scenes. Invest. Ophthalmol. Vis. Sci. 2014;55(13):786.

      Download citation file:

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

  • Supplements

Purpose: Superordinate level of complex scenes (natural vs. man-made) has been previously examined in terms of spatial envelope properties (including degree of naturalness, openness, perspective). In addition to these stimulus metrics, top-down factors, such as familiarity and contextual information, contribute to lower-level processing as well as cognitive components of categorization. It was hypothesized that scene inversion, which disrupts familiarity, should interfere with categorization similarly within, and between, superordinate levels.

Methods: Participants categorized briefly presented complex scenes as either forest, mountain, or ocean coast (categories of natural scenes) or houses, highway, or city skyline (man-made scene categories). Perceptual capacity was indexed in terms of scene completion, in which each picture was initially presented with a high level of occlusion, which was progressively reduced across trials. The highest occlusion level at which scenes were correctly categorized served as the threshold for categorization.

Results: Results indicated that for upright images performance across categorize was similar, and that performance did not predict superordinate levels. However, inverted images reduced performance for man-made categories, but had little effect on natural scenes.

Conclusions: These results may reflect fundamental differences in stimulus components between superordinate levels, such as the occurrence and extent of linear borders. These results suggest that cues mediating categorization differ in relative importance depending on component features of complex images.

Keywords: 641 perception • 640 pattern vision  

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.