How does precise excitation-inhibition balance control the activity landscape in the brain.
The brain employs synaptic excitation (E) and inhibition (I) to control the excitability landscape of neurons. I am interested in understanding how using specific motif connectivity, different kinds of computations can be implemented using E and I connections. To do this, first, we need to understand the degree and kind of EI balance in different kinds of networks. Several theoretical positions have been put in literature, including detailed EI balance (Vogels and Abbott, 2009) where all combinations of presynaptic inputs are EI balanced and tight EI balance where E and I are highly correlated and balanced at fast timescales (Boerlin, Machens and Deneve, 2013). Using the mouse hippocampus as a model system, we projected random patterns of light to optogenetically activate CA3 neurons using a projector
Further, in the feedforward inhibitory network connectivity of the hippocampus, EI balanced inputs naturally arrive with a delay that varies with input size. Using model and experiments, we showed that this results in a novel computation called Subthreshold Divisive Normalization (SDN). As its name suggests, SDN is a subthreshold gain-control computation that divisively controls neuronal output. SDN also has the interesting property that it controls the fraction of information in spike-time vs spike counts.
We are currently looking at how these phenomena play a role in inhibitory gating, the formation of place cells and neural sequences in the hippocampus.