August 2012
Volume 12, Issue 9
Vision Sciences Society Annual Meeting Abstract  |   August 2012
Temporally specific visual working memory representations revealed by multivoxel pattern analysis
Author Affiliations
  • Timothy Vickery
    Department of Psychology, Yale University
  • Brice Kuhl
    Department of Psychology, Yale University
  • Marvin Chun
    Department of Psychology, Yale University
Journal of Vision August 2012, Vol.12, 1272. doi:
  • Views
  • Share
  • Tools
    • Alerts
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Timothy Vickery, Brice Kuhl, Marvin Chun; Temporally specific visual working memory representations revealed by multivoxel pattern analysis. Journal of Vision 2012;12(9):1272.

      Download citation file:

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

  • Supplements

Visual working memory often requires remembering not only what was seen, but when it was seen. We asked whether temporal features are encoded in visual representations, using fMRI pattern classification and an N-back memory task. While being scanned, participants (N=14) viewed a slow stream of images from three classes (faces, objects, and scenes). Separate runs required either 1-back or 2-back memory tasks. Classifiers were trained to discriminate either the 1-back category or the 2-back category from activity following each trial, and tested using a leave-one-run-out cross-validation procedure. Two broad regions were examined - ventral temporal cortex (VTC: inferior temporal, fusiform, and parahippocampus) and occipital cortex (OCC: all occipital, including lateral occipital cortex). We found that the 1-back item could be significantly decoded in both regions and both tasks. The 2-back item could also be decoded above chance in both regions, but only in the 2-back task. We then examined confusion errors to determine whether this memory was temporally specific. If the decoded features are not temporally specific, then when the classifier made errors during 2-back decoding it should have tended to select the 1-back category, rather than the 3rd (unrepresented) category. Indeed, when the VTC classifier made errors it tended to classify 2-back items according to the 1-back category, suggesting that memory in these regions was not temporally specific. However, when the OCC classifier made errors it did not show a bias to select the 1-back category; in fact, it was more likely to choose the unrepresented category. Thus, despite the fact that 1-back items were robustly represented in OCC, this information was clearly distinct from representations of 2-back items. These results suggest that 2-back tasks induce temporally-specific visual working memory representations in occipital visual regions.

Meeting abstract presented at VSS 2012


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.