September 2015
Volume 15, Issue 12
Free
Vision Sciences Society Annual Meeting Abstract  |   September 2015
A rapid whole-brain neural portrait of scene category inference
Author Affiliations
  • Pavan Ramkumar
    Sensorimotor Performance Program, Rehabilitation Institute of Chicago and Departments of Physical Medicine & Rehabilitation and Neurobiology, Northwestern University, Chicago IL, USA
  • Bruce Hansen
    Department of Psychology and Neuroscience Program, Colgate University, Hamilton NY, USA
  • Sebastian Pannasch
    Department of Psychology, Technische Universität Dresden, Dresden, Germany
  • Lester Loschky
    Department of Psychological Sciences, Kansas State University, Manhattan, KS, USA
Journal of Vision September 2015, Vol.15, 351. doi:10.1167/15.12.351
  • Views
  • Share
  • Tools
    • Alerts
      ×
      This feature is available to authenticated users only.
      Sign In or Create an Account ×
    • Get Citation

      Pavan Ramkumar, Bruce Hansen, Sebastian Pannasch, Lester Loschky; A rapid whole-brain neural portrait of scene category inference. Journal of Vision 2015;15(12):351. doi: 10.1167/15.12.351.

      Download citation file:


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

      ×
  • Supplements
Abstract

Perceiving the world around us is a process of active inference from incoming visual information. One opportunity to study brain processes underlying perceptual inference is when perception deviates from reality. Here, we focus on errors in rapid scene categorization. How do humans accurately categorize natural scenes after extremely brief presentations (< 20 ms)? To elucidate the role of brain areas involved in visual encoding and perceptual inference, we measured cortical activity using whole-scalp magnetoencephalography (MEG). Scenes were flashed for 33 ms and subjects responded with one of six scene categories. We localized single-trial sensor-level data to the cortical surface reconstructed from individual MRIs. Next, we computed categorization confusion matrices (CMs) using support vector machines based on (1) cortical activity, (2) spatial envelope image features, and (3) behavioral responses. We then used these CMs to examine the functions of different cortical areas. Behavioral categorization confusions can result from either visual representation errors or perceptual inference errors. Thus, if confusions in neural decoders (neural CMs) are driven by errors in image feature-based decoders (image-feature CMs), then any associated cortical activity is attributable to the visual representations. Conversely, if confusions in the perceived category (behavioral CMs) can explain errors in neural CMs, then any associated cortical activation could be attributed to errors in perceptual inference. Using multiple linear regression at each cortical vertex and each millisecond time bin to explain neural CMs as a function of image-feature CMs and behavioral CMs, we found that neural CMs within early visual cortices were explained primarily by image-feature CMs from 90–110 ms, whereas neural CMs within regions such as PRC, PHC, RSC, and OFC were explained primarily by behavioral CMs during 120–200 ms. Our results suggest that medial temporal areas and OFC actively infer visual percepts rather than passively representing categorical information.

Meeting abstract presented at VSS 2015

×
×

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.

×