Exploring how different frequency oscillations interact and influence the routing of information in the brain

Abstract

Cognitive tasks require the flexible communication between multiple brain regions, which may be facilitated by neural oscillations. Several studies have emphasized the role of transient oscillatory coherence in a specific frequency band (e.g. gamma) to mediate directed communication. However, oscillations in the brain occur simultaneously in multiple frequency bands, which dynamically cooperate to perform the cognitive tasks. In this study, we systematically investigate multi-frequency oscillation patterns and their effect on inter-region communication by constructing computational spiking networks models of structurally coupled local regions oscillating at different frequencies. We consider simulated network activity at different synchrony levels, ranging from asynchronous to edge-of-synchrony and more strongly synchronous regimes. We firstly identify multi-frequency oscillations patterns (MFOPs) that can transiently emerge during spontaneous dynamics. These MFOPs are characterized by the joint occurrence of oscillatory bursts across multiple frequencies and regions. Remarkably, we find that as an effect of inter-regional excitatory or inhibitory interactions, oscillatory burst at fast or slow frequencies can arise in all populations, irrespectively of their natural local resonance frequency. Secondly, we use information theoretical analyses to extract the Information Routing Patterns (IRPs) associated to each type of MFOP. We reveal that the joint occurrence of oscillatory bursts at different frequencies leads to a boosting of directed information transfer, particularly enhanced at one or more specific latencies within the slower oscillation cycles. Each MFOP map to a different spatio-temporal motif of transfer boosting. Therefore, the variety of possible MFOPs gives rise to a rich dictionary of emergent IRPs, in which regions can exchange information multiple times in alternating directions, on time-scales different from the ones initially hardwired at the neuronal resonance level. Overall, our computational analyses predict that interacting faster and slower frequency oscillations can impact information routing in much more complex ways than postulated by current theories of frequency-multiplexing of directed communication or gating via cross-frequency coupling. They also emphasize that inter-regional information transfer is an active computation, not limited to the passive reception of information from a source region but also encompassing the active integration within the target region of multiple information chunks received at different times.