Costs and benefits of using rhythmic rate codes

Neural oscillations are observed ubiquitously in the mammalian nervous system, and the benefits of oscillatory coding have been the topic of frequent analysis. Many prior studies focused on communication between populations which were already oscillating and sought to understand how these rhythms and overall communication interact. We take a contrary view here. In this paper, we focus on measuring the costs of translating from an aperiodic code to a rhythmic one. We study two models. The first is simulated independent populations of neurons subjected to a theta-band (6 Hz) pacemaker using Linear-Nonlinear Poisson (LNP) sampling. The second is a model of beta-gamma oscillations using biophysical neurons with self-organized dynamics. We measure benefits and costs in both models using information theory. In both models oscillations can only benefit communications by increasing spiking by specific amounts, in effect, correcting for "undersampling" of the stimulus. This is mechanistically similar to theories for how deep brain stimulation can enhance cognition and is consistent with older studies of gamma entrainment. Yet this trend was not universal. No one guiding principle of dynamics determines the cost of a translation in the models we studied. In our models to predict the benefits or costs of an oscillatory translation we need to understand the exacting physical details of the intrinsic connections, the population size, and the external drive.