October 2020
Volume 20, Issue 11
Open Access
Vision Sciences Society Annual Meeting Abstract  |   October 2020
Neurocomputational Mechanisms of Action-Outcome Prediction in V1
Author Affiliations & Notes
  • Clare Press
    Birkbeck, University of London
  • Emily Thomas
    Birkbeck, University of London
  • Sam Gilbert
    Institute of Cognitive Neuroscience, University College London
  • Floris de Lange
    Donders Institute for Brain, Cognition and Behaviour, Radboud University
  • Peter Kok
    Wellcome Centre for Human Neuroimaging, University College London
  • Daniel Yon
    Birkbeck, University of London
    Goldsmiths, University of London
  • Footnotes
    Acknowledgements  Leverhulme Trust; Wellcome Trust
Journal of Vision October 2020, Vol.20, 712. doi:https://doi.org/10.1167/jov.20.11.712
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Clare Press, Emily Thomas, Sam Gilbert, Floris de Lange, Peter Kok, Daniel Yon; Neurocomputational Mechanisms of Action-Outcome Prediction in V1. Journal of Vision 2020;20(11):712. https://doi.org/10.1167/jov.20.11.712.

      Download citation file:

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

  • Supplements

Goal-directed action depends on our ability to anticipate the outcomes of our movements. Recent accounts have suggested that the predictive mechanisms deployed during action operate according to general principles of perceptual prediction – with observers using top-down knowledge about likely action consequences to bias perception of expected outcomes and to ‘sharpen’ representations in the sensory brain. However, it remains unclear what kind of mechanism generates these effects, and there is continuing controversy surrounding the relationship between predictive effects on the sensory brain and behavior. Here we present a new experiment addressing this controversy by combining multivariate fMRI with computational modelling of participant choices in an action and perception task. In a behavioral acquisition phase, participants acquired perfect associations between manual actions and gratings with particular orientations. In a subsequent MRI test session, participants produced manual actions either with no visual effect (33%), or to generate gratings with an orientation that was expected (33%) or unexpected (33%) on the basis of the preceding training. When expectations were valid, decisions about the grating orientation were faster than on unexpected trials and modelling indicated that this effect was explained by biases in sensory evidence. Representations of outcomes in primary visual cortex (V1) were also ‘sharpened’ relative to unexpected trials, such that linear support vector machines classified the identity of the gratings from patterns of V1 activity with superior accuracy. Moreover, we performed a number of analyses allowing us to examine the relationship between the V1 processing and behavioral decisions, revealing how effects of action prediction in primary visual cortex are related to what agents see and what they decide.


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.