Browsing by Subject "STDP model"

Effective population coding is dependent on connectivity, active and passive postsynaptic membrane parameters but how it relates to information transfer and information representation in the brain is still poorly understood. Recently, Brendel et al. (2020) showed how spiking neuronal networks can efficiently represent a noise input signal. This "D_Model” successfully showed that spiking neural networks can recreate input signal representations and how these networks can be resilient to the loss of neurons. However, this model has multiple unphysiological characteristics, such as instantaneous firing and the lack of units related to physical values. The aim of the present study is to build upon the D_Model to add biological accuracy to it and study how information transfer is affected by biophysical parameters. We first modified the D_Model in the MATLAB environment to allow for the simultaneous firing of the neurons. Using our CxSystem2 simulator in a Python environment (Andalibi et al. 2019), we built a network replicating the one used in the D_Model. We quantified the information transfer of Leaky Integrate-and-Fire units that had identical physiological values for both inhibitory and excitatory units (Comrade class) as well as more biologically accurate physiological values (Bacon class). We used various information transfer metrics such as granger causality, transfer entropy, and reconstruction error to quantify the information transfer of the network. We examined the behaviour of the network while altering the values of the capacitance, synaptic delay, equilibrium potential, leak conductance, reset potential, and voltage threshold. Broad parameter searches showed that no single set of biophysical parameters maximised all information transfer metrics, but some ranges fully blocked information transfer by either saturating or stopping neuronal firing. This suggests theoretical boundaries on the possible electrophysiological values neurons can have. From narrow searches within electrophysiological ranges, we conclude that there is no single optimal set of physiological values for information transfer. We hypothesise that different neuronal types may specialise in transferring different aspects of information such as accuracy, efficiency, or to act as frequency filters.