September 2019
Volume 19, Issue 10
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2019
Building color-concept association distributions from statistical learning
Author Affiliations & Notes
  • Melissa A Schoenlein
    Department of Psychology, University of Wisconsin-Madison
    Wisconsin Institute for Discovery, University of Wisconsin-Madison
  • Karen B Schloss
    Department of Psychology, University of Wisconsin-Madison
    Wisconsin Institute for Discovery, University of Wisconsin-Madison
Journal of Vision September 2019, Vol.19, 299a. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Melissa A Schoenlein, Karen B Schloss; Building color-concept association distributions from statistical learning. Journal of Vision 2019;19(10):299a. doi:

      Download citation file:

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

  • Supplements

Knowledge about the properties of objects is continually updated with experience. The Color Inference Framework proposes that this is the process by which color-concept associations are formed--people update color-concept association distributions with every co-occurrence between colors and concepts (Schloss, 2018). We tested whether these distributions can be formed through statistical learning using two tasks (1) category learning, and (2) color-concept association ratings. In the category task, participants saw aliens from two species: Filks and Slubs. Body shape (pointy vs. round) was perfectly diagnostic of species, whereas color distribution (warm vs. cool) was partially diagnostic. The aliens appeared in eight saturated hues with different color distributions between species. One species’ color distribution was warm-biased (20-red, 25-orange, 20-yellow, 15-chartreuse, 10-green, 5-cyan, 10-blue, and 15-purple aliens). The other species was cool-biased (opposite frequencies). On each trial, an alien appeared below the species’ names and participants indicated the species to which the alien belonged. There was immediate feedback. In the association task, participants rated the association strength between each species name with each of 32 colors: the eight saturated hues, plus light, muted, and dark versions of each hue. After, participants reported their categorization strategy. Most reported only using shape, with only one-third of participants noticing color patterns between species. Results indicate that participants learned color distributions, regardless of noticing patterns. Color frequencies during category learning predicted color-concept associations for the saturated colors (mixed-effect linear regression: p < .001), with strong correlations between exposure frequencies and average color-concept associations for the Filk-warm/Slub-cool group (r = 0.78, p < .001) and Filk-cool/Slub-warm group (r = 0.57, p < .05). Learned associations tended to generalize to the same hues with different saturation/lightness levels for each group (r = 0.40, p < .01; r = 0.26, p = .07). This study demonstrates that people can form new color-concept associations through statistical learning.

Acknowledgement: Wisconsin Alumni Research Foundation 

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.