September 2021
Volume 21, Issue 9
Open Access
Vision Sciences Society Annual Meeting Abstract  |   September 2021
Visual attention maximizes expected information gain in goal inference
Author Affiliations & Notes
  • Bohao Shi
    Zhejiang University
  • Zhen Li
    Zhejiang University
  • Yazhen Peng
    Zhejiang University
  • Zhuoxuan Liu
    Zhejiang University
  • Jifan Zhou
    Zhejiang University
  • Mowei Shen
    Zhejiang University
  • Footnotes
    Acknowledgements  This research was supported by National Natural Science Foundation of China [Grant 31871096, 31600881].
Journal of Vision September 2021, Vol.21, 2187. doi:https://doi.org/10.1167/jov.21.9.2187
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Bohao Shi, Zhen Li, Yazhen Peng, Zhuoxuan Liu, Jifan Zhou, Mowei Shen; Visual attention maximizes expected information gain in goal inference. Journal of Vision 2021;21(9):2187. doi: https://doi.org/10.1167/jov.21.9.2187.

      Download citation file:


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

      ×
  • Supplements
Abstract

Visual attention is a mechanism of information-seeking and uncertainty reduction. However, the fundamental question of how attention helps people reducing uncertainty is still poorly understood. This study aims to explore this issue by relating attention to the concept of Expected Information Gain (EIG) in information theory, which describes the expected value of uncertainty reduction. In three experiments, we designed a task to probe the relationship between EIG and attention: participants were asked to determine which of two targets was the goal of a moving agent. When the agent was at some locations in its path, observing the next move of the agent would have higher EIG (i.e., the observations were more likely to reduce uncertainty in participants' inference), while at other locations, the observations (with lower EIG) were less likely (even impossible) to reduce uncertainty. We used eye trackers to record participants' eye movements and pupil sizes to measure their attentional levels during the task. Experiment 1 showed that participants had greater pupil dilation as the levels of EIG increased. Moreover, participants' positions of fixations were less frequent to move away from the locations of the agent when EIG levels were higher. These results illustrate that the attentional levels increase with EIG levels. Experiment 2 added obstacles to agents’ paths to break the distribution of EIG relative to spatial locations, finding participants' attentional levels still changed in parallel with EIG levels. In Experiment 3, the agent’s paths were exactly the same as those in Experiment 1, but difference levels of EIG was made to vanish by manipulating the order that targets appeared. Consequently, the effects in Experiment 1 were greatly diminished. Combined, these results demonstrate that visual attention might be the process of resolving uncertainty as much as possible by maximizing the EIG.

×
×

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.

×