Mammalian brains exhibit population oscillations, the structures of which vary in time and space according to behavioural state. A proposed function of these oscillations is to control the flow of signals among anatomically connected networks. However, the nature of neural coding that may support selective communication that depends on oscillations has received relatively little attention. Here, we consider the role of multiplexing, whereby multiple information streams share a common neural substrate. We suggest that multiplexing implemented through periodic modulation of firing-rate population codes enables flexible reconfiguration of effective connectivity among brain areas.

SummaryThe mammalian brain exhibits profuse interregional connectivity. How information flow is rapidly and flexibly switched among connected areas remains poorly understood. Task-dependent changes in the power and interregion coherence of network oscillations suggest that such oscillations play a role in signal routing. We show that switching one of several convergent pathways from an asynchronous to an oscillatory state allows accurate selective transmission of population-coded information, which can be extracted even when other convergent pathways fire asynchronously at comparable rates. We further show that the band-pass filtering required to perform this information extraction can be implemented in a simple spiking network model with a single feed-forward interneuron layer. This constitutes a mechanism for flexible signal routing in neural circuits, which exploits sparsely synchronized network oscillations and temporal filtering by feed-forward inhibition.Video Abstract

The recently developed ‘two-step’ behavioural task promises to differentiate model-based from model-free reinforcement learning, while generating neurophysiologically-friendly decision datasets with parametric variation of decision variables. These desirable features have prompted its widespread adoption. Here, we analyse the interactions between a range of different strategies and the structure of transitions and outcomes in order to examine constraints on what can be learned from behavioural performance. The task involves a trade-off between the need for stochasticity, to allow strategies to be discriminated, and a need for determinism, so that it is worth subjects’ investment of effort to exploit the contingencies optimally. We show through simulation that under certain conditions model-free strategies can masquerade as being model-based. We first show that seemingly innocuous modifications to the task structure can induce correlations between action values at the start of the trial and the subsequent trial events in such a way that analysis based on comparing successive trials can lead to erroneous conclusions. We confirm the power of a suggested correction to the analysis that can alleviate this problem. We then consider model-free reinforcement learning strategies that exploit correlations between where rewards are obtained and which actions have high expected value. These generate behaviour that appears model-based under these, and also more sophisticated, analyses. Exploiting the full potential of the two-step task as a tool for behavioural neuroscience requires an understanding of these issues.

Gamma oscillations in the dentate gyrus and hippocampal CA3 show variable coherence in vivo, but the mechanisms and relevance for information flow are unknown. We found that carbacholinduced oscillations in rat CA3 have biphasic phase-response curves, consistent with the ability to couple with oscillations in afferent projections. Differences in response to stimulation of either the intrinsic feedback circuit or the dentate gyrus were well described by varying an impulse vector in a two-dimensional dynamical system, representing the relative input to excitatory and inhibitory neurons. Responses to sinusoidally modulated optogenetic stimulation confirmed that the CA3 network oscillation can entrain to periodic inputs, with a steep dependence of entrainment phase on input frequency. CA3 oscillations are therefore suited to coupling with oscillations in the dentate gyrus over a broad range of frequencies. Many brain areas, including the dentate gyrus and hippocampal CA3, show prominent gamma oscillations whose inter-region coherence varies during awake activity 1,2 . Structures of gamma-oscillatory activity in the temporal lobe are modulated by both brain state 3 and task demands 4,5 . Such changing patterns of coherent oscillatory activity may be involved in controlling information flow among connected networks 6-8 . The dynamical mechanisms underlying the emergence of coherence among anatomically coupled networks are, however, poorly understood.During gamma oscillations, individual neurons spike irregularly 1,2,9-11 , but the collective dynamics of the local network are oscillatory. Thus, the emergence of phase coherence between regions is a process of synchronization between local network oscillators. Important determinants of the synchronization properties of systems of coupled oscillators are the phase-response curves (PRCs) of the constituents 12-14 . Synchronization through excitatory connections is strongly promoted by biphasic PRCs, in which the same input can either advance or delay the phase of oscillation depending on when the perturbation occurs during the oscillatory cycle. Such PRCs allow oscillators to entrain to periodic inputs at both higher and lower frequencies than the unperturbed oscillation frequency. Although the mechanisms generating gamma oscillations in local networks have been extensively studied 9,11,15,16 , the phase response dynamics of the resulting network oscillators have not been measured, nor have their entrainment properties in response to periodic inputs been