Optimizing Measures of Information Encoding in Astrocytic Calcium Signals

While most models of brain information encoding focus on neurons, recent studies have shown that calcium dynamics of astrocytes, the major class of non-neural cells in the brain, can add information about key cognitive variables that is not found in the activity of nearby neurons. This raises the question of what could be the contribution of astrocytes in information processing, and calls for analysis tools to characterize this contribution. Here we construct simulations with realistic dependencies of astrocytic activity on external variables and we use these simulations to understand how to optimally set parameters of information theoretic analysis of astrocytic activities. Applications of our techniques to simulated and real astrocytic data show how to set parameters of information analyses that provide conservative, yet reliable, estimates of astrocytic calcium dynamics contribution to circuit-level brain information processing.