Nonlinear dynamics in computational neuroscience. Contributions of the international workshop `Nonlinear dynamics in computational neuroscience: from physics and biology to ICT', September 2015 (Q721125): Difference between revisions

This book contains eight chapters, all written by different sets of authors, resulting from a workshop held in Italy in 2015. The unifying theme is nonlinear dynamics in neuroscience and the contributions vary from mathematical modelling to purely experimental. The chapters are now summarised. Chapter 1 discusses ``next generation'' neural mass models. These are evolution equations for macroscopic quantities derived from large groups of coupled hetereogeneous point neurons, but unlike traditional neural mass (or field) models, the ones presented here are rigorously derived from networks of theta neurons, using the Ott/Antonsen ansatz. Numerical bifurcation analysis of the presented models shows that they are capable of producing macroscopic oscillations, and also event related synchronisation and desynchronisation. Chapter 2 considers a network of excitatory and inhibitory leaky integrate-and-fire (LIF) neurons. Depending on coupling strengths, the network can be bistable, between a low- and a high-frequency macroscopic steady state. Including ``fatigue'' in the form of a slow, activity-driven inhibitory current results in the network showing bursts, i.e., slow, almost periodic alternation between an active state and a quiescent state. This behaviour is well reproduced by a reduced mean-field model. Chapter 3 discusses nonlinear time series analysis of local field potential recordings from mice. The authors introduce power spectra, and linear and nonlinear measures of coupling strength and coupling directionality between two signals. They also discuss power law correlations and fractal dimensions. These quantities are then measured for normal mice and mice having a model of focal epilepsy, mice which have had a stroke induced, and mice reared with environmental enrichment. There are some significant differences in the values of the quantities measured in the different sets of mice. In Chapter 4, the authors consider a sparse random network of excitatory LIF neurons with short term synaptic depression. They also include correlations between a neuron's in-degree and out-degree, and between the excitability of a neuron and the sum of its in- and out-degrees. They find that the network shows approximately periodic ``population bursts'', in which the neurons fire quasi-synchronously, separated by longer periods of asynchronous firing. They find a small number of neurons which, when either stimulated with a DC current or deleted from the network, cause the bursting to cease. These neurons are also found to fire in a particular order and timing before each burst and thus form a ``functional clique''. Chapter 5 considers a symmetric pair of coupled populations of LIF neurons. The coupling strengths within and between the two populations are different. For full connectivity the system is known to produce a variety of stable states including those for which the level of synchrony is different within the two populations, sometimes referred to as a ``chimera'' state. The authors consider the effects of ``dilution'' on some of the dynamics, by only connecting neurons with a fixed probability, thus forming an Erdös-Rényi network. They lastly consider adding noise to a slightly diluted network by randomly choosing the reset value for each neuron after each firing, and consider the effect of this on some dynamics. Chapter 6 is entitled ``Nanotechnologies for neurosciences'' and briefly discusses recent research regarding micro- and nano-technological approaches in cell biology and physiology such as high-resolution imaging and fabrication, and smart materials and molecules. The last two chapters discuss memristors. In Chapter 7, a memristor is characterised by a set of Volterra kernels, so that the output for any input is given by the sum of integrals of the input multiplied by the relevant kernels. A method for analytically deriving low-order kernels is given, and the results of this approach are compared with the standard approach of directly solving the equations governing the memristor and other circuit elements. The approach is then repeated for circuits containing two memristors. Chapter 8 considers a two-dimensional array of diffusively coupled memristors and shows that depending on parameters, the system can support travelling solitary waves or Turing patterns.