• Citer

Compact hybrid type electronic neuron and computational model of its dynamics

par Shlyonsky, Vadim ;Dupuis, Freddy ;de Prelle de la Nieppe, Bertrand ;Erneux, Thomas ;Osee, Michel ;Nonclercq, Antoine ;Gall, David
Référence Nonlinear dynamics, 112, 16, page (14343-14362)
Publication Publié, 2024-08

Article révisé par les pairs

• Accès en ligne
• Détails
• Contenu
• Statistiques

Résumé :

We propose a Neurosimilator, novel analog neuron circuit and mathematical model of its dynamics that both simulate with high precision the spiking behavior of known types of excitable cells. The analog circuit is compact and can be built using off-the-shelf components. This could facilitate its use in teaching neuroscience and biophysics. The circuit is scalable down to the pF-valued capacitors, presenting an advantage in research on the analog nerve fiber networks. The equations of circuit dynamics contain exponential non-linearities and Heaviside functions, so that the model combines features from the generalized adapting exponential integrate-and-fire neuron model and from the intermittent feedback Hindmarsh-Rose model, but it is not directly related to them. Our four-dimensional system (4D-Neurosimilator) simulates most of excitable cells’ spiking, bursting and chaotic behavior depending on only fewer predefined parameters. In bursting and chaotic oscillatory patterns, the model demonstrates self-adaptive energy flow redistribution. The energy expenditure amounts to ≈36 µJ per one spiking event in original model and to ≈0.63 pJ in its down-scaled version. The model has computational cost comparable to that of the Hodgkin–Huxley model, but it tends to handle noisy input stimulations more efficiently. Our work provides novel insights to the simulation of neuron’s non-linear dynamics and may constitute another choice of available model in computational neuroscience research that expands the limits of a tradeoff between accuracy, biological explainability, noise-resistance and computing time.