Max Planck Institute for Dynamics and Self-Organization -- Department for Nonlinear Dynamics and Network Dynamics Group
Personal tools
Log in

Modelling the role of neural oscillations in information routing

It has been hypothesized that dynamic patterns of oscillatory coherence modulate the direction and the efficiency of communication between different brain regions. We have directly confirmed this hypothesis, studying information routing in simple multi-areal circuits and verifying that changes of phase-locking induce rerouting of information transfer [1]. However, in this early study, the simulated oscillations were unrealistically persistent. In contrast, oscillations in vivo are transient and their frequency varies stochastically [2]. We have recently generalized our model of inter-areal information routing to operate in a regime, which we call the “edge of synchrony”. In such a regime, collective oscillations are meta-stable, but self-organize to give rise to transient episodes of phase-locking, which still regulate information transfer. We thus prove that stochasticity of oscillations in vivo is not incompatible with their use for flexible routing [3].


Figure 1: State-resolved Transfer Entropy analyses reveal that the direction and efficiency of information transfer is still regulated by oscillatory phase coherence, even at the edge of synchrony where it is highly transient.

Recent massively parallel electrophysiological recordings suggest that communication-through-coherence in different directions exploits different frequency bands. Fast oscillations would mediate bottom-up communication, while beta or even slower oscillations top-down influences from higher order cortical areas [4]. We have studied a mean field model of two model regions wired according to the so called canonic local and long-range cortical connectivity [5]. Such a model automatically causes different cortical layers to oscillate at different frequencies as an effect of chaotic inter-layer entrainment. As a result, the emergent frequency-dependent patterns of inter-areal phase-locking naturally reproduce experimental findings. We show that this effect would not occur in random circuit, thus proving that the experimentally observed connectivity affects dynamics in a highly specific manner. However we also predict that other functionally-equivalent connectomes may exist, by constructing suitable examples.


Figure 2: When multiple brain regions are wired according to canonic cortical connectivity, our model reproduces the experimental finding that top-down and bottom-up inter-areal communication operate in different frequency bands.



[1]    D. Battaglia, A. Witt, F. Wolf, and T. Geisel, PLoS Comp Biol 8, e1002438 (2012).

[2]    D. Xing, Y. Shen, S. Burns, C.-I. Yeh, R. Shapley, and W. Li, Journal of Neuroscience 32, 13873 (2012).

[3]    A. Palmigiano, T. Geisel, F. Wolf, D. Battaglia, under review

[4]    A. M. Bastos, J. Vezoli, C. A. Bosman, J.-M. Schoffelen, R. Oostenveld, J. R. Dowdall, P. De Weerd, H. Kennedy, and P. Fries, Neuron 85, 390 (2015).

[5]    M. Helmer, X. J. Chen, W. Wei, F. Wolf, and D. Battaglia, bioRxiv (2015).

Members working within this Project:

 Theo Geisel 
 Fred Wolf 

Former Members:

 Markus Helmer 
 Demian Battaglia 
 Agostina Palmigiano 

Selected Publications:

C. Kirst, M. Timme, and D. Battaglia (2016).
Dynamic information routing in complex networks
Nature Comm. 7:11061. download file