December 2022
Volume 22, Issue 14
Open Access
Vision Sciences Society Annual Meeting Abstract  |   December 2022
Drift-like dynamics of working memory representations in human cortex
Author Affiliations & Notes
  • Hsin-Hung Li
    Department of Psychology, New York University
    Center for Neural Science, New York University
  • Wei Ji Ma
    Department of Psychology, New York University
    Center for Neural Science, New York University
  • Clayton Curtis
    Department of Psychology, New York University
    Center for Neural Science, New York University
  • Footnotes
    Acknowledgements  This research was supported by the National Eye Institute (NEI) (R01 EY- 016407 to C.C., R01 EY-027925 to C.C. and W.J.M. H.-H.L. is supported by a Swartz Foundation Postdoctoral Fellowships.
Journal of Vision December 2022, Vol.22, 3620. doi:https://doi.org/10.1167/jov.22.14.3620
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Hsin-Hung Li, Wei Ji Ma, Clayton Curtis; Drift-like dynamics of working memory representations in human cortex. Journal of Vision 2022;22(14):3620. https://doi.org/10.1167/jov.22.14.3620.

      Download citation file:


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

      ×
  • Supplements
Abstract

Working memory (WM) enables people to hold information in mind temporarily for actions and decision-making. The neural processes underlying WM are noisy and memory error increases with time. Whereas many neurophysiological studies on animals and computational models have investigated the dynamics of WM, how the neural representations of WM evolves over time in the human brain remains largely unknown. We studied the formation of WM error in the human cortex when human participants performed a memory-guided saccade task in an MRI scanner: A target dot (12° eccentricity) appeared at a randomly chosen location (polar angle) in each trial. After a delay, observers made a saccade to the remembered target location, and then reported their uncertainty by a post-decision wager. We used a Bayesian decoder (van Bergen and Jehee, 2021) to decode the memorized locations from fMRI BOLD response, for each time point during the memory delay. We found that even though the location of the target can be decoded as early as 2 to 3 seconds from the delay onset, a positive correlation between decoding error and (behavioral) memory error only emerged later at about 6 seconds from the delay onset, ramped up over time and reached a peak at the time of memory report. This correlation appeared the earliest in the intraparietal sulcus (IPS) and the dorsal exstrastriate cortex, and was only significant at the end of the delay in the striate cortex. Moreover, by binning the trials based on the magnitude and the direction (clockwise vs. counterclockwise) of memory error, we found that the decoding error accumulated over time in the direction predictive of the direction of behavioral memory error. Overall, these patterns are consistent with the drift-like dynamics proposed in the previous bump attractor network models of WM and the neurophysiological studies on nonhuman primates.

×
×

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.

×