Speaker
Description
Understanding how individuals recover function after stress is a central question in resilience research. From a neuroscience perspective, identifying neural circuits that promote resilience could reveal whether resilience is a universal principle across scales.
We address this question by studying spiking neural networks subject to disturbances. Prior work in similar settings has focused on robustness (the preservation of function after attack) and its correlation with structural graph metrics. However, resilience, defined as the recovery of network function over time, remains less explored. Here, we quantify network function by tracking neuronal activity with information-theoretic measures, such as entropy and mutual information (MI), computed before and after disturbances. We evaluate functional metrics including (i) the return of entropy to its pre-attack baseline and (ii) the MI between pre-attack and relaxed post-attack activity patterns.
Preliminary analyses show a trade-off between the two recovery readouts; MI-based recovery is negatively associated with entropy recovery. Across generated network structures, recovering pre-attack information patterns often coincides with reduced population variability, thereby lowering information capacity. Structurally, MI recovery increases with global efficiency and transitivity but decreases with small-worldness; entropy recovery shows the opposite pattern.
Both social and physical interactions with the environment shape neural population dynamics and their impact has been characterised by information-theoretic measures. We view stressors, injuries, or clinical disorders, as perturbations to the underlying neural network. Our resilience assessment aims to link network structure and mechanisms to functional outcomes that impact behaviour.
