ENERGY RESTRICTION EFFECTS IN A SMALL COMPUTATIONAL MODEL

Biological neurons exhibit remarkable energy efficiency, executing intricate computations with minimal energy expenditure. An understanding of these mechanisms could inform the design of efficient computational models and energy-aware learning systems. In this study, we developed a simulated environment within NEST wherein a small network of biophysically plausible artificial neurons was configured. Specifically, the simulations incorporated energy-dependent neuron and synaptic plasticity models: the Energy-Dependent Leaky Integrate-and-Fire (EDLIF) model and Energy-Dependent Spike-Timing-Dependent Plasticity (EDSTDP). This network, driven by synaptic plasticity, is currently under investigation to determine the optimal configuration capable of accurately modeling a neural culture. We have observed that manipulation of energy parameters within this model directly influences the network's response. So far, our results suggest that more available energy (or a greater energy regeneration ratio) and the size of the network can positively or negatively affect its performance in terms of synaptic activity. Furthermore, while constant stimulation with a static configuration does not reduce plasticity over time, it does lead to the formation of synaptic weight clusters. These findings offer an insight on the sensibility of the simulation on the parameters , important for future experiments, such as replicating the dynamics of cultured neurons (MEAs).