December 2022
Volume 22, Issue 14
Open Access
Vision Sciences Society Annual Meeting Abstract  |   December 2022
A Novel approach for estimating spatiotemporal population receptive fields in human visual cortex
Author Affiliations & Notes
  • Insub Kim
    Stanford University
  • Eline Kupers
    Stanford University
  • Garikoitz Lerma-Usabiaga
    Basque Center on Cognition, Brain and Language (BCBL)
  • Won Mok Shim
    Center for Neuroscience Imaging Research, Institute for Basic Science (IBS)
    Sungkyunkwan University
  • Kalanit Grill-Spector
    Stanford University
    Wu Tsai Neurosciences Institute, Stanford University
  • Footnotes
    Acknowledgements  Funding: NEI R01 EY023915
Journal of Vision December 2022, Vol.22, 4060. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Insub Kim, Eline Kupers, Garikoitz Lerma-Usabiaga, Won Mok Shim, Kalanit Grill-Spector; A Novel approach for estimating spatiotemporal population receptive fields in human visual cortex. Journal of Vision 2022;22(14):4060.

      Download citation file:

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

  • Supplements

The visual system integrates sensory input across space and time. To investigate how spatial and temporal information are processed in the human visual system, we developed and tested a spatiotemporal population receptive field (pRF) model. The spatiotemporal pRF models spatial responses with 2D Gaussians (Dumoulin & Wandell, 2008) and temporal responses with nonlinear neural impulse functions in millisecond resolution (Stigliani et al., 2­­019; Zhou et al., 2019). First, we developed a software that synthesizes fMRI responses given a spatiotemporal pRF, a spatio-temporal visual stimulus, and a noise level. Additionally, given an fMRI time series and stimulus, the software solves the spatiotemporal pRF parameters. This allowed us to compare model solutions to ground truth. Our model recovered spatiotemporal pRF parameters from noiseless synthetic fMRI time series with 99% accuracy. Second, we evaluated how well different spatiotemporal pRF models (conventional spatial pRF, spatiotemporal pRF with compressive temporal summation, and 2-channel spatiotemporal pRF) predict empirical fMRI data. In the experiment, ten observers viewed, while fixating, a bar containing colored cartoon stimuli presented at varying temporal rates (33ms to 5s duration, 33ms to 200ms interstimulus intervals) that swept the visual field (12° radius, 4 directions, 9 steps). Across ventral and lateral visual streams (V1, V2, V3, hV4, LO1, LO2, and TO1), the spatiotemporal pRF model showed progressively higher accuracy for predicting single voxel responses than the spatial pRF model. This indicates that the incorporation of temporal nonlinearities to the spatial pRF is especially important for higher visual areas. Additionally, estimated pRF sizes were larger for the spatial than spatiotemporal models, suggesting that the standard spatial pRF method may overestimate pRF size to account for extended temporal responses. Together, we provide a new framework and computational model to synthesize and predict responses in the visual system to any dynamic visual stimulus.


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.