Modular architecture confers robustness to damage and facilitates recovery in spiking neural networks modeling in vitro neurons

Takuma Sumi^1*Akke Mats Houben^2,3Hideaki Yamamoto⁴Hideyuki Kato⁵Yuichi Katori^6,7Jordi Soriano^2,3Ayumi Hirano-Iwata^1,4

• ¹Advanced Institute for Materials Research (WPI-AIMR), Tohoku University, Sendai, Japan
• ²Departament de Física de la Matèria Condensada, Universitat de Barcelona, Barcelona, Spain
• ³University of Barcelona Institute of Complex Systems (UBICS), Barcelona, Catalonia, Spain
• ⁴Research Institute of Electrical Communication (RIEC), Tohoku University, Sendai, Japan
• ⁵Faculty of Science and Technology, Oita Univeristy, Oita, Oita, Japan
• ⁶Graduate School of System Information Science, Future University Hakodate, Hakodate, Japan
• ⁷International Research Center for Neurointelligence (WPI-IRCN), The University of Tokyo, Tokyo, Japan

Impaired brain function is restored following injury through dynamic processes that involve synaptic plasticity. This restoration is supported by the brain's inherent modular organization, which promotes functional separation and redundancy. However, it remains unclear how modular structure interacts with synaptic plasticity to define damage response and recovery efficiency. In this work, we numerically modeled the response and recovery to damage of a neuronal network in vitro bearing a modular structure. The simulations aimed at capturing experimental observations in cultured neurons with modular traits which were physically disconnected through a focal lesion. The damage reduced the frequency of spontaneous collective activity events in the cultures, which recovered to predamage levels within 24 hours. We rationalized this recovery in the numerical simulations by considering a plasticity mechanism based on spike-timing-dependent plasticity, a form of synaptic plasticity that modifies synaptic strength based on the relative timing of pre-and postsynaptic spikes. We observed that the in silico numerical model effectively captured the decline and subsequent recovery of spontaneous activity following the injury. The model supports that the combination of modularity and plasticity confers robustness to the damaged neuronal network by preventing the total loss of spontaneous network-wide activity and facilitating recovery. Additionally, by using our model within the reservoir computing framework, we show that information representation in the neuronal network improves with the recovery of network-wide activity.