Event Abstract

Back to Event

An information theoretic framework for neuroimaging data analysis: stimulus modulations, representational interactions and causal communication of specific information content

Robin A. Ince1*, Nicola J. Van Rijsbergen1, Guillaume A. Rousselet1, Joachim Gross1, Stefano Panzeri2 and Philippe G. Schyns1
  • 1 University of Glasgow, Institute of Neuroscience and Psychology, United Kingdom
  • 2 Istituto Italiano di Tecnologia, Laboratory of Neural Computation, Italy

Information theory provides a principled and unified statistical framework for neuroimaging data analysis. A major factor hindering wider adoption of this framework is the difficulty of estimating information theoretic quantities in practice. We present a novel estimation technique that combines the statistical theory of copulas with the closed form solution for the entropy of Gaussian variables (Ince et al., 2016a). This results in a general, computationally efficient, flexible, and robust multivariate statistical framework that provides effect sizes on a common meaningful scale, allows for unified treatment of discrete, continuous, uni- and multi-dimensional variables, and enables direct comparisons of representations from behavioral and brain responses across any recording modality. Open-source Matlab and Python code implementing the new methods is available at: https://github.com/robince/gcmi We validate the use of this estimate as a statistical test within a neuroimaging context, considering both discrete stimulus classes and continuous stimulus features. We demonstrate the benefits of a multivariate statistical approach with examples such as MEG vector magnetic fields (including separate quantification of stimulus modulations of amplitude and direction) as well as considering the instantaneous temporal derivative together with the raw values of M/EEG signals as a two-dimensional response. A particular advantage of the information theoretic framework is its ability to quantify representational interactions between different neuroimaging responses: for example, different cortical regions, frequency bands, time periods or recording modalities (i.e. simultaneous recorded EEG and fMRI). Such interactions are quantified as redundancy (overlapping information content) and synergy (joint information not available in the individual responses). Conceptually, this is similar to methods such as Representational Similarity Analysis or the cross-classification decoding technique, which quantify the similarity of stimulus representation in different neural responses (but cannot address synergistic effects). However, the information theoretic approach can be applied with a wider range of experimental designs, and to univariate responses, which allows mapping of representational interactions with the full spatial and temporal resolution of the recorded signals. By combining the Gaussian copula mutual information estimator and the information theoretic approach to quantifying representational interactions within the Wiener-Granger framework for causal inference, we have developed an approach to measuring functional connectivity that is grounded in specific stimulus information content. We have demonstrated within-subject cross-hemisphere communication of information about eye visibility in a face detection task (Ince et al., 2016b) using sensor space EEG data. We have also developed Directed Feature Information (DFI), a novel measure of directed functional connectivity which quantifies communication about the specific stimulus features subtending perceptual decisions (Ince et al., 2015). These developments allow for network level analyses of functional neuroimaging data that are directly grounded in the representation, processing and communication of specific stimulus features, and so provide the promise of a new perspective on the algorithmic basis of many cognitive functions.

References

Ince, R.A.A., Giordano, B.L., Kayser, C., Rousselet, G.A., Gross, J., Schyns, P.G., 2016a. A statistical framework for neuroimaging data analysis based on mutual information estimated via a Gaussian copula. bioRxiv 043745. doi:10.1101/043745

Ince, R.A.A., Jaworska, K., Gross, J., Panzeri, S., Rijsbergen, N.J. van, Rousselet, G.A., Schyns, P.G., 2016b. The Deceptively Simple N170 Reflects Network Information Processing Mechanisms Involving Visual Feature Coding and Transfer Across Hemispheres. bioRxiv 044065. doi:10.1101/044065

Ince, R.A.A., van Rijsbergen, N., Thut, G., Rousselet, G.A., Gross, J., Panzeri, S., Schyns, P.G., 2015. Tracing the Flow of Perceptual Features in an Algorithmic Brain Network. Sci. Rep. 5, 17681. doi:10.1038/srep17681

Keywords: EEG, MEG, fMRI, Information Theory, mutual information, representational interactions, redundancy, Synergy, Gradient, functional connectivity, transfer entropy, statistics, multivariate statistics, multivariate analysis

Conference: Neuroinformatics 2016, Reading, United Kingdom, 3 Sep - 4 Sep, 2016.

Presentation Type: Investigator presentations

Topic: Neuroimaging

Citation: Ince RA, Van Rijsbergen NJ, Rousselet GA, Gross J, Panzeri S and Schyns PG (2016). An information theoretic framework for neuroimaging data analysis: stimulus modulations, representational interactions and causal communication of specific information content. Front. Neuroinform. Conference Abstract: Neuroinformatics 2016. doi: 10.3389/conf.fninf.2016.20.00009

Copyright: The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers. They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.

The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.

Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.

For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.

Received: 29 Apr 2016; Published Online: 18 Jul 2016.

* Correspondence: Dr. Robin A Ince, University of Glasgow, Institute of Neuroscience and Psychology, Glasgow, United Kingdom, robin.ince@glasgow.ac.uk