My interests are in information processing in both natural and artificial systems. Principally my work develops statistical techniques to deal with large, stochastic systems.

In natural systems, my viewpoint is that biological systems encode relevant functional
characteristics with the molecular and physical tools they have available. Our task is to primarily understand what this functionality is, the mechanisms behind this being somewhat secondary to this. The rationale behind this viewpoint is that the functionalities are key to applying insights from neurobiology to different areas -- for example, to make an artificial visual or auditory system that goes beyond low-level signal processing. In doing so, I hope to shed light on observed experimental behaviour demonstrating that the observed complexity in the neural system may be explicable by simple rules which, nevertheless, lead to complex effects.

Whilst progress in understanding individual neural behaviour is clear, relatively little has been learned about the interactions and properties of neural assemblies.

My most relevant work in theoretical neurobiology concerns learning in networks of spiking neurons. I showed that effects such as Spike Time Dependent Plasticity are natural consequences of a collection of coupled neurons wishing to optimally store patterns, and given the fact that refractoriness is an unavoidable biological constraint. These results are significant in that they are able to explain STDP as a consequence of a simple objective (functional) which is implemented with the biological processing units at hand.

I'm particularly interested in auditory processing. My work on signal processing concerns learning and extracting fundamental sound units based on complex non-linear processing. My goal is to show how the cochlea mimics the objective of recognising core sub-units of sound that in-composition make up the rich acoustic events of everyday life. In doing so, we hope to derive improved artificial speech and acoustic processing systems.