October 2020
Volume 20, Issue 11
Open Access
Vision Sciences Society Annual Meeting Abstract  |   October 2020
That just doesn’t add up: Continuous visual features fail to determine the number of items, and number behavior, in visual scenes
Author Affiliations & Notes
  • Emily Sanford
    Johns Hopkins University
  • Justin Halberda
    Johns Hopkins University
  • Footnotes
    Acknowledgements  National Science Foundation Graduate Research Fellowship Program
Journal of Vision October 2020, Vol.20, 751. doi:https://doi.org/10.1167/jov.20.11.751
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Emily Sanford, Justin Halberda; That just doesn’t add up: Continuous visual features fail to determine the number of items, and number behavior, in visual scenes. Journal of Vision 2020;20(11):751. https://doi.org/10.1167/jov.20.11.751.

      Download citation file:


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

      ×
  • Supplements
Abstract

How do we determine the number of things in a scene? Previous research has suggested that seemingly number sensitive behavior may not reflect the representation of number per se but may instead result from a summation of information across low-level, non-numerical features (e.g., Dakin et al., 2011; Durgin, 2008; Gebuis et al., 2016). We took two approaches in evaluating the validity of this claim. First, we investigated whether number could be predicted from continuous features in children’s counting books and real-world images, supposing that humans’ number extraction abilities could have been acquired through the learning of this relationship. In both datasets, the number of items in the scene was not reliably predicted by any individual continuous feature nor by their linear combination. Further, these features failed to predict signatures of human number perception such as scalar variability. Next, we evaluated the extent to which peoples’ representations of non-numerical visual features predict their numerical responses. Subjects estimated the convex hull, average area and number of objects in children’s book illustrations and real-world photographs. Subjects had more precise representations for number than continuous features (Counting books: F(2,39) = 16.77, p < .001; Real-world photographs: F(2,21) = 14.64, p < .001). Additionally, subjects’ number responses were much better accounted for by the actual number of objects in the image than by their non-numerical judgments on the same images (Counting books: R2Continuous Features = .28, R2Number = .92; Real-world photographs: R2Continuous Features = .03, R2Number = .41). This indicates that subjects were not computing a combination of their internal representations of continuous features to derive a number response but instead responded directly to the number of items in the scene. We conclude that number representations do not reduce to simple combinations of non-numerical features and that the content of number representations is truly numerical.

×
×

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.

×