August 2014
Volume 14, Issue 10
Vision Sciences Society Annual Meeting Abstract  |   August 2014
Plinko: A spatial probability task to measure learning and updating.
Author Affiliations
  • Alex Filipowicz
    Department of Psychology, University of Waterloo
  • Derick Valadao
    Department of Psychology, University of Waterloo
  • Britt Anderson
    Department of Psychology, University of Waterloo
  • James Danckert
    Department of Psychology, University of Waterloo
Journal of Vision August 2014, Vol.14, 881. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Alex Filipowicz, Derick Valadao, Britt Anderson, James Danckert; Plinko: A spatial probability task to measure learning and updating.. Journal of Vision 2014;14(10):881.

      Download citation file:

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

  • Supplements

Research has demonstrated that humans efficiently learn the statistics of their visual environment (e.g., Fiser & Aslin, 2001). Typical studies present participants with series of events and ask them to predict which event will occur on specific trials. Responses are then aggregated over bins of trials to represent a probability distribution of participant predictions. Although informative, these tasks provide limited information about how participant expectations evolve over the course of a task. We present a novel spatial probability task that attempts to overcome this limitation. Based on the game Plinko (the modern incarnation of Galtons Bean Machine), participants view balls that drop through pegs and land in slots. On every trial, participants are asked to estimate how likely a ball will fall in each slot. Participants adjust a cup or bars under the slots to represent their likelihood estimations. We exposed participants to four distinct distributions of ball drops and measured how accurately they could represent each distribution (Experiment 1) and shift from one distribution to the next (Experiment 2). Rather than representing participant expectations by building probability distributions over multiple trials, our measures provide a probability distribution on each trial of the task. Participants managed to use the cup to accurately track the mean and variance of each distribution by adjusting the cups center position and width throughout the task. Participants were also efficient at using the bars, matching the computers distributions with an average accuracy of 80%. Participants also managed to effectively shift from one distribution to the next using either the cup or bars, and this without being made explicitly aware that any changes would occur. These results suggest that our task provides an effective measure of spatial probability learning while also providing a rich representation of changes in participant predictions over the course of the task.

Meeting abstract presented at VSS 2014


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.