Cortical oscillations support sampling-based computations in spiking neural networks

Author summaryActivity oscillations are a ubiquitous and well-studied phenomenon throughout the cortex. At the same time, mounting evidence suggests that brain networks perform sampling-based probabilistic inference through their dynamics. In this work, we present a theoretical and a computational analysis that establish a rigorous link between these two phenomena: background oscillations enhance sampling-based computations by helping networks of spiking neurons to quickly reach different high-probability network states, i.e., computational results.Such an acceleration of sampling is required for efficient learning and inference in neural networks. Our results show that oscillations provide this acceleration robustly over different frequency bands and in different network conditions. This suggests a similar functional role of oscillations throughout the cortex. As unspecific background input is enough to evoke this acceleration, our proposed mechanism has a very general scope. We show how such a view on oscillations ties in with a multitude of experimental observations and discuss various opportunities for constraining our model with new experimental data.Overall, the mechanism we put forward is general and robust and leads to a new understanding of oscillations in the context of sampling-based computations. Our model offers a computational explanation for many related experimental observations that are linked to cortical oscillations. Being permanently confronted with an uncertain world, brains have faced evolutionary pressure to represent this uncertainty in order to respond appropriately. Often, this requires visiting multiple interpretations of the available information or multiple solutions to an encountered problem. This gives rise to the so-called mixing problem: since all of these "valid" states represent powerful attractors, but between themselves can be very dissimilar, switching between such states can be difficult. We propose that cortical oscillations can be effectively used to overcome this challenge. By acting as an effective temperature, background spiking activity modulates exploration. Rhythmic changes induced by cortical oscillations can then be interpreted as a form of simulated tempering. We provide a rigorous mathematical discussion of this link and study some of its phenomenological implications in computer simulations. This identifies a new computational role of cortical oscillations and connects them to various phenomena in the brain, such as sampling-based probabilistic inference, memory replay, multisensory cue combination, and place cell flickering.