September 2021
Volume 21, Issue 9
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2021
Without adaptive stimulus sampling, comparison-of-comparison tasks skew perceptual bias estimates
Author Affiliations
  • Man-Ling Ho
    UCL
  • D. Samuel Schwarzkopf
    UCL
    University of Auckland
Journal of Vision September 2021, Vol.21, 1994. doi:https://doi.org/10.1167/jov.21.9.1994
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Man-Ling Ho, D. Samuel Schwarzkopf; Without adaptive stimulus sampling, comparison-of-comparison tasks skew perceptual bias estimates. Journal of Vision 2021;21(9):1994. https://doi.org/10.1167/jov.21.9.1994.

      Download citation file:


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

      ×
  • Supplements
Abstract

Measuring perceptual bias is method-dependent. When judging perceived stimulus size, we previously found a traditional two alternative forced choice (2AFC) task produced bias estimates that were significantly greater than a novel perceptual matching (PM) task, a variant of a comparison-of-comparisons (CoC) task (e.g. Finlayson, Papageorgiou, & Schwarzkopf, 2017). CoC tasks are thought to better control for decisional criterion issues than 2AFC because it constrains stimuli to differ only in the stimulus parameter of interest (Morgan, Melmoth, & Solomon, 2013; Jogan & Stocker, 2014; Patten & Clifford, 2015). Using the Ebbinghaus illusion, here we aimed to test if decisional bias could be driving the difference in bias estimation between 2AFC and PM. In Experiment 1 (n=7), we collected confidence reports alongside each 2AFC judgment and estimated perceptual bias as the point of maximum uncertainty, because metacognitive bias estimates have also been suggested to control for criterion shift (Gallagher, Suddendorf, & Arnold, 2019). We found no difference between 2AFC and metacognitive estimates, while PM estimates were significantly lower than both. In Experiment 2 (n=22), we tested if PM or 2AFC tasks were better at capturing true perceptual bias. We varied test target sizes using idiosyncratic bias estimates from PM and 2AFC and found that 2AFC estimates were superior in nulling the Ebbinghaus illusion. Subsequent simulations revealed that CoC tasks skew bias estimates because 2AFC requires judgment against the position of the true bias, while CoC involves judging the distances from the true bias. In Experiment 3 (n=32), we showed that this dependency can be corrected via adaptive stimulus sampling. On average, adaptive PM produced comparable bias estimates to 2AFC and both methods were equally good at nulling the Ebbinghaus illusion.

×
×

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.

×