Understanding network dynamics of compact assemblies of neurons in zebrafish larvae optic tectum during spontaneous activation

How does intrinsic network structure shape spontaneous coordinated neural dynamics? Previous work has shown that overlapping tectal neuronal assemblies can be circuit mechanisms for robust visual detection [1]. To better understand how these assemblies give rise to emergent dynamics, we use single-photon calcium imaging of triple transgenic zebrafish larvae optic tectum (OT), letting us simultaneously record spontaneous activity of 1000s of neurons in the absence of stimulation. What is the composition of cell types within the assemblies? What role do the inhibitory and excitatory neurons play in modulating assembly dynamics? We look at the E-I ratio of neurons within the assemblies, as well as the ratio of E-I activity during and outside of spontaneous assembly activation. We observe that assemblies maintain a consistent ratio of E-I activity during and outside of activation, even though total neural activity goes up during acti- vation. How are neurons interacting within the assemblies? Next we investigate the structure of the neural correlations using topological approaches invariant under nonlinear monotonic translations of the data, such as those introduced when measuring neural activity via calcium signal strength. We observe that topological signatures of neural correlation differ during versus outside of activation, in particular suggesting low dimensional dynamics during activation. Modeling assembly dynamics with threshold-linear networks lets us further relate these observed topological signatures to features of the network’s structure influential on its dynamics.