NETWORK TOPOLOGY AND CELL-TYPE COMPOSITION SHAPE SPIKING DYNAMICS AND COMPUTATIONAL CAPACITY

Understanding how network topology and cellular composition shape neural dynamics is central to linking brain structure to function across scales. Here, we present a flexible spiking neural network framework designed to systematically probe how distinct connectivity motifs, cell types, and synaptic polarities influence spiking activity, population dynamics, and computational properties. The model comprises spatially embedded, three-layer networks of excitatory and inhibitory Izhikevich neurons, with configurable input, hidden, and output populations. Connectivity can be instantiated using a range of topological motifs, including feedforward, partially recurrent, and modular architectures inspired by microscale circuit patterns observed in biological neural networks.

Using a custom Python-based simulation and analysis pipeline, we introduce controlled input pulses and background activity to examine emergent firing rates, burst structure, synchrony, and functional connectivity derived from spike trains. By varying topological motifs, excitation–inhibition balance, and cell-type composition, we aim to characterise how structural constraints shape dynamical regimes and information processing capacity. Although analyses are ongoing, we hypothesise that biologically inspired motifs—particularly partially recurrent and modular topologies—promote richer dynamics, improved efficiency, and enhanced computational capacity compared to random or purely feedforward networks. This framework provides a general tool for testing mechanistic links between network structure, dynamics, and computation in spiking systems.

Image: Example network topology inputted to network simulation (simplified depiction of lollipop shape graph for fun!). 1000 neurons simulated over 10 s. Excitatory neurons in the cluster, excitatory and inhibitory neurons in the tail are shown at the bottom, middle and top of the raster, respectively.