Computational Neuroscience

Two questions particularly intrigue us: how do populations of neurons learn?  And is the joint activity of many neurons really a simple dynamical system? As neurons across the brain must work together to make bodies function in the world, we’ve attempted to make sense of the neural activity in the basal ganglia, brainstem, sensory and prefrontal cortex, and the sea-slug locomotion system. Irritatingly, dopamine keeps popping up everywhere – even in the sea slugs. I also find the simple pleasures of network theory to be a welcome distraction from the unfathomable complexities of the brain. 

You can find more about our work at  Humphries Lab.

I write about (systems) neuroscience for a broad audience at The Spike.

In collaboration with international institutions (e.g. UCLA Medical Center, King’s College Hospital), we study the simultaneous activity of single neurons recorded from microelectrodes implanted in the brain of epileptic patients for possible curative surgery. This approach led us to show for the first time that individual neurons in the human brain change their firing to link associations when a new memory is formed. We have recently started to implement neural network models, including short-term plasticity and STDP, to further our understanding of memory formation.

To uncover the neuronal underpinnings of visual search we are developing novel data analysis techniques to bridge the gap between eye movements and non-invasive (EEG/MEG) brain recordings. This research has recently led us to propose a novel data-driven framework, providing the link between local processing (from individual fixations) and the underlying mental program.

I also specialise in characterising inter-individual differences in brain morphology, particularly with respect to aging, dementia, and cognitive abilities.

I conduct research across a variety of topics, including emotional memory, risky decision-making, and embodied cognition. I study these topics using behavioural paradigms, as well as fMRI, EEG, and structural MRI. Additionally, some studies involve computational modelling--either in the form of advanced statistical methods and machine learning, or through the development of specific models designed to distinguish between particular theoretical hypotheses.

In addition, I have developed a number of data analysis methods for neuroscience, including the 'Van Rossum' spike distance measure. I have taught and organised various courses in the field of computational neuroscience. I have directed multiple PhD training programmes in Edinburgh, and I was nominated for a FENS supervision award. I am very excited about the new programmes that I have set up in Nottingham, as for the first time in the UK these will directly integrate computational neuroscience with psychology and mathematics.

To support students who have less exposure to computational neuroscience research, I have lectured many times in so-called summer schools and tutorials. My lecture notes have been used worldwide. For my research I mainly use computer simulations and theoretical analysis, and of course extensive conversations with fellow scientists. I am always interested in setting up new collaborations. Current research includes the study of schema and rule extraction from memory events, and the role of metabolic constraints on neural computation. The most up-to-date list of publications can be found on Google Scholar.

I am interested in the computational principles underlying simple cognitive processes such as sensation, action and decision making. I believe that computational and statistical methods can provide step-changes in our ability to understand the brain - in particular the automation of data processing tasks with machine learning. I also believe in studying simple systems with as much experimental precision as possible. For these reasons my current research is focussed on understanding the rodent whisker system with a range of computational tools (in collaboration with Rasmus Petersen, Manchester, UK).

My PhD and early postdoc years were earned with Prof. Tony Prescott in Sheffield, where I developed models of whisker based tactile perception, and robot technologies based on the rodent whisker system. See  https://mathewzilla.github.io/ for more information.

With this approach, I hope it is achievable to obtain important insights into the biology in the brain. Currently, my primary research focus is on the impacts of metabolic cost on plasticity and network circuitry. Besides that, I am building a novel neural network model that describes the temporal dependence of reconsolidation and extinction of fear memory.

Learning from experience requires the combined integration of sensory stimuli and past events to drive future decision. This adaptive behaviour relies on complex dynamics of large-scale brain circuitry.  Understanding of these complex processes represents an important goal towards the comprehension of network abnormalities observed in learning disabilities and other psychiatric disorders.  My approach to address this challenge is to analyse large neural recordings of animals learning new tasks.  By using network theory and machine learning techniques I aim to understand cortico-hippocampal contribution when learning a spatial-navigation task.