Purchase this article with an account.
Steven Thurman, Javier Garcia, Emily Grossman; Determining the feature sensitivity of visual areas to biological motion using brain-based reverse correlation. Journal of Vision 2011;11(11):688. doi: 10.1167/11.11.688.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Introduction. In recent work we have employed the spatio-temporal “bubbles” paradigm to investigate the diagnostic space-time features used by human observers and a computational model to discriminate biological motion patterns (Thurman, Giese, & Grossman, 2010, Journal of Vision). Here we applied this paradigm to functional brain-imaging data to determine the space-time features of biological motion that drive high-amplitude BOLD responses in various visual regions of interest (ROIs). Methods. The bubble method samples the stimulus space randomly on each trial with a space-time mask, revealing portions of the stimulus through small Gaussian apertures. Reverse correlating brain responses with the bubbles masks across many trials reveals the space-time features that drive the BOLD responses in those regions. In a targeted region of interest (ROI) analysis, we independently localized the following ROIs in each subject: posterior STS (biological vs. scrambled motion), EBA and FBA (static bodies vs. objects; see Downing et al., 2001, Science), and human MT+ (dot flow fields vs. static dots). Each subject then completed 20 “bubbles” scans in which 30 trials of masked point-light walking patterns were discriminated in a slow, event-related experimental design. We computed classification movies by reverse correlating the random “bubbles” masks with peak BOLD responses across all 600 trials for each subject (n = 8) and each ROI individually. Maps were then combined across subjects and hemispheres to create group classification movies. Results. Despite the noise inherent to fMRI measurements, this method effectively yielded classification movies for each ROI. The classification movies across our targeted ROIs were highly correlated and revealed key features that were also apparent in the behavioral classification movie. In particular, dynamic features in the lower body that are critical for discriminating walking patterns behaviorally also appear to drive brain responses in this network.
This PDF is available to Subscribers Only