Computational Systems Neuroscience

Computational Systems NeuroscienceThe Computational Systems Neuroscience group on one hand supports the experimental efforts in Systems Neuroscience with theoretical and mathematical analyses, and on the other hand inspires new experiments and technical approaches by advancing hypotheses and making theory-based predictions. We have particular strengths in signal processing, data analysis / assimilation, model building and (massively parallel) computer simulation. While we are interested both in applied and fundamental research questions, our modelling and analysis is practically always focused on actual experimental data and/or realistic anatomy and physiology.

Neural population models

Computational Systems NeuroscienceThe brain has intricate anatomical structure at many different size scales. This includes strong enervation at mesoscopic scales of about 0.1-3 mm, binding together tens of thousands to millions of neurons. Such large groups can show collective states of coherent activity, allowing a simplified mathematical description of their “mass action”. It turns out that the signals of brain activity recorded with non-invasive neuroimaging modalities (EEG, MEG, fMRI,...), and even some invasive ones (ECoG, VSDi, …), derive primarily from such collective states. Furthermore, since these models describe coherent states of actual neurons, one can directly relate some features of such models to the experimentally accessible anatomy and physiology of individual cells. We use neural populations models to describe simultaneous EEG / fMRI recordings, predict the effect of psychoactive drugs like general anaesthetic agents, understand processing in the retina and primary visual cortex, analyse epileptic seizures, etc. Currently we are also working on applying these models to motor control, and using them in data assimilation.

Structure, function and closed loops

Computational Systems NeuroscienceOur modelling and analytic efforts also focus on the relationships between structure and function, and in particular on the role of closed loops in neural and cognitive information processing. We are asking such questions across different scales. At a microscopic (cellular) level we work on methods for the reconstruction of neuron morphology from microscope images; and combine detailed morphology with Hodgkin-Huxley type of models of excitable membranes in order to elucidate how morphology shapes neural activity. At the meso- and macroscopic level, we employ tools to capture and model the evolving complex networks of synchronous activity between different neuronal populations and brain regions, respectively, and the role such dynamics plays in information processing. We follow a data-driven approach in which we study Empirical Mode Decomposition methods for extracting meaningful oscillations from neural activity; and we characterise statistics of phase locking in order to tease out neural correlates of cognitive processing.

Balance of excitation and inhibition

Non-invasive neuroimaging techniques (EEG, fMRI, …) have been widely used in human brain research, and are increasingly used for the medical diagnosis of disease. However their utility is limited by the fact that the interpretation of their signals in terms of the underlying neural mechanisms remains elusive. The focus of our research is to investigate the neural origin of these signals through mathematical modelling of in vivo experimental data. Recent research in intracellular recordings using whole-cell patch clamp techniques provided new insights into the interaction between excitatory and inhibitory postsynaptic activities of pyramidal neurons. They were shown to be proportionally balanced, and co-tuned during normal brain function both in response to stimulation and at rest. Guided by these findings, we design and conduct electrophysiological experiments to manipulate the balance between neural excitation and inhibition and measure concurrently neural and haemodynamic responses of the local neural population. These experiments provide a rich data set for dynamic modelling and system identification techniques to establish a heuristic model of neural activity and of neurovascular coupling.

Group members

Things to do now

  • Find out about our postgraduate research
  • PhD opportunities
  • Meet our staff

Follow us

Read our brochure

Contact us

Professor Slawomir Nasuto
s.j.nasuto@reading.ac.uk

+44 (0)118 378 6701

Page navigation

 

Search Form

A-Z lists