Reverse Engineering Mechanisms of Information Processing in the Human Brain

Thursday, 2017, January 12 - 11:00
Max Planck Institute, Liepzig, Germany


Brain functions emerge from interactions at multiple spatial (from synapses to areas) and temporal (milliseconds to days) scales. Dynamical interactions in the healthy brain and their alterations in neurological disorders can be identified experimentally using different modalities (e.g. EEG: electroencephalography, fMRI: functional magnetic resonance imaging), but extracting general rules that apply across scales and support whole brain dynamics is difficult due to the size and complexity of empirical data sets.  A conceptual framework is needed that utilizes available complementary imaging data to infer the mechanisms that are occurring on finer scales than each imaging method on its own can resolve. We have developed The Virtual Brain (TVB,, an open-source neuroinformatics platform for the creation of large-scale brain network models. TVB uses connectome-based modeling to integrate empirical data with dynamical systems equations and generate simulated data across multiple spatiotemporal scales from local field potential and excitatory/inhibitory population firing, to EEG and fMRI. The unique feature of TVB is the capacity to directly link with empirical data.  For example, an individual person’s functional and structural MRI data can be used to set the initial conditions for a model, which provides insights into the unique characteristics of that individual’s brain dynamics. This feature places TVB in a unique position by providing the quantitative link between data sources that span scales and species within a single computational modeling framework. The ensuing models provide a deeper understanding of how electrical (e.g., firing rates, postsynaptic membrane potentials), chemical neurotransmission such as GABA, NMDA and microscopic structural (e.g., synaptic gain through STDP) activity in different populations of neurons are represented in macroscopic-level measurements of the human brain. The approach helps to better explain multiscale dynamics by constraining the targets for empirical studies to a smaller number of biophysical variations and will additionally serve to better integrate the work from animal models with human data. This is a unique avenue that can revolutionize our understanding of the biological activity underlying, and neuroinformatics content of, data collected using contemporary functional neuroimaging techniques.