Poster D131, Monday, March 27, 8:00 – 10:00 am, Pacific Concourse
Assessing hierarchical self-similarity processing with univariate and multivariate analysis approaches
Florian Ph.S Fischmeister1,2, Georg Langs3, Mauricio Martins4,5,6, W. Tecumseh Fitch4, Roland Beisteiner2; 1High Field Magnetic Resonance Centre, Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna, Austria, 2Department of Neurology, Medical University of Vienna, Vienna, Austria, 3Computational Imaging Research Lab, Department of Biomedical Imaging and Image-guided Therapy, Medical University of Vienna, Vienna, Austria, 4Department of Cognitive Biology, University of Vienna, Vienna, Austria, 5Berlin School of Mind and Brain, Humboldt Universität zu Berlin, Berlin, Germany, 6Max Planck Institute for Human Cognitive and Brain Sciences, Leipzig, Germany
The ability to understand and generate complex hierarchical structures is a crucial component of human cognition. The investigation of the underlying neural bases is thus essential to understand the foundations of human cognitive architecture. Pattern analysis approaches such as Gini contrast represent a powerful new tool to probe functional imaging data for distributed multivariate patterns and link them to cognitive processes. Such an approach is of particular interest when studying subtle correlates across distributed networks serving complex cognition. Here we compared three approaches, classic GLM modeling, functional connectivity, and Gini contrast, with respect to their ability to describe and differentiate between recursive and iterative processing during the encoding and decoding of visual stimuli. On the group level all three methods showed a considerable overlap of brain regions commonly associated with hierarchical processing. Gini contrast revealed several network specific brain areas that were not visible when using standard GLM analysis. Brain areas more specific for recursive processing include the posterior cingulate cortex and lateral temporal cortices. Evidence from functional connectivity analysis suggest these additional areas to be part of the default mode network. Additional areas linked to non-recursive processing were detected e.g. in the fronto-parietal control network (e.g., inferior precuneus). In conclusion, each method reflects different components and aspects of the underlying neuronal process. Gini contrast, due to its multivariate nature, seems to represent a viable tool to describe complex cognitive functioning as processes utilizing distinct neuronal resources in a shared neuronal network.
Topic Area: THINKING: Other