Abstract
The thesis investigates principles of self-organization that may account for the observed structure and behaviour of neural networks that generate locomotor behaviour and complex spatiotemporal patterns such as spiral waves, metastable states and persistent activity. This relates to the general neuroscience problem of finding the correspondence between the structure of neural networks and their function. This question is both extremely important and difficult to answer because the structure of a neural network defines a specific type of neural dynamics which underpins some function of the neural system and also influences the structure and parameters of the network including connection strengths. This loop of influences results in a stable and reliable neural dynamics that realises a neural function. In order to study the relationship between neural network structure and spatiotemporal dynamics, several computational models of plastic neural networks with different architectures are developed. Plasticity includes both modification of synaptic connection strengths and adaptation of neuronal thresholds. This approach is based on a consideration of general modelling concepts and focuses on a relatively simple neural network which is still complex enough to generate a broad spectrum of spatio-temporal patterns of neural activity such as spiral waves, persistent activity, metastability and phase transitions. Having considered the dynamics of networks with fixed architectures, we go on to consider the question of how a neural circuit which realizes some particular function establishes its architecture of connections. The approach adopted here is to model the developmental process which results in a particular neural network structure which is relevant to some particular functionality; specifically we develop a biologically realistic model of the tadpole spinal cord. This model describes the self-organized process through which the anatomical structure of the full spinal cord of the tadpole develops. Electrophysiological modelling shows that this architecture can generate electrical activity corresponding to the experimentally observed swimming behaviour.
Document Type
Thesis
Publication Date
2008
Recommended Citation
Cooke, T. (2008) Locomotor patterns and persistent activity in self-organizing neural models. Thesis. University of Plymouth. Retrieved from https://pearl.plymouth.ac.uk/fose-theses-other/16
