Effect of spike-defective neurons in a nonlinear neuronal cellular automata model

Abstract

In our previous works, we investigated a cellular automata model of neuronal network with a nonlinear response based on the activation function of the Hodgkin-Huxley model. The observed steady-state dynamics of the spiking activity are classified into Class 0: a) slow-decay, and b) fast-decay; and Class 1: a) low activation, and b) high activation steady-state. Our investigation shows that varying neighborhood and boundary conditions imposed on the lattice does not alter any of the aforementioned steady-state dynamics. Meanwhile, this work focuses on the implications of adding spike-defective neurons that approximates neurons damaged due to injuries or aging. A population of N = 2500 neurons are arranged in a 50×50 lattice with random initial activity a. A fraction of the population D are assigned randomly as spike-defective neurons that neither takes input nor produces output activity. We found that Class 0b systems are the most sensitive to the fraction of defective neurons. Further investigation shows that the peak activity of Class 0b systems occurs at T ≈ 2.16 in arbitrary units (arb. unit) with a dynamic range of T ∈ [1.15, 16.16] (arb. unit). This suggests that defective neurons are necessary in the network to produce a more dynamical activity analogous to what we observed in real neuronal networks.