Neuroscience
Learning and memory in biological neural networks
Biological neural networks are composed of large, complex dynamical systems. They exhibit a rich spatial structure, as well as many dynamical ingredients spanning over vastly different timescales. While many individual ingredients are now well studied, how they fit into the bigger picture is still an open question, for instance:
- How do local, micro-circuits communicate with each other?
- How do local plasticity mechanisms give rise to memory and learning?
- How does the interplay of complex topologies and different timescales lead to robust, global behaviour?
Our interest lies in defining modelling, simulation and analysis techniques to address these issues in an efficient way.
This requires:
- To be able to model hierarchical systems based on well-defined atomic elements.
- To handle efficiently the modelling and simulation of possibly large models.
- To take into account various relationships between inter-dependent parameters.
- To collect, visualize and analyse the simulation results.
By using an efficient and precise event-based modelling and simulation strategy, under the control of state of the art modelling formalisms, we are able to manipulate these large, complex artificial spiking neural systems.
Biologically constrained neural network models
I am a research fellow and a member of the SECSE multi-disciplinary collaborative project investigating properties of spatially embedded networks. My work focuses on biologically constrained neural network models and I specialize in investigating the phenomenological properties of the unitary parts which provide functionality in the network as a whole. In particular I consider how adaptation, regulation and plasticity may be combined to generate robust networks capable of carrying out cognitive tasks. I am particularly interested in the processing and learning of sequential information (Womble, 2004), and have a passing acquaintanceship with the evolution of language faculty in the brain (Womble and Wermter, 2002a, 2002b). I have also studied connectionist models of language production (Womble and Wermter, 2001). I am committed to the use of quantitative measures of system performance in particular studying population decoding schemes (Womble, Wermter and Treves 2002). More recently I have tested the results of computational studies in cognitive tasks, predicting the effect of caffeine on word retrieval failures (Lesk and Womble 2004). The results of this work have been presented to a general audience (see for instance the BBC article here) The development of more detailed network models which can capture the full range of effects found in the cognitive testing (of human volunteers and in an aphasic patient) is a current research direction.
The C. elegans locomotion nervous system
C. elegans is one of the simplest creatures of the animal kingdom. With a mapped genome and the only mapped neural circuitry, this organism offers a first tangible opportunity to understand an entire living, behaving and learning system bottom-up and top-down. As such, it offers great promise to systems biologists, neuroscientists and roboticists alike. Despite its relative simplicity, C. elegans possesses many of the functions that are attributed to higher level organisms, including feeding, mating, complex sensory abilities, memory and learning. Can we understand the underlying engineering designs that allow this tiny nematode to survive and flourish? What insight can we gain into universal principles that give rise to adaptive and robust life-forms or to the unique architecture of its nervous system? Meeting this challenge requires a large multi-disciplinary effort, combining insight and expertise from biology, physics, engineering and computer science. Our group focuses on the C. elegans locomotion system and its neural control. This is a combined theoretical and experimental effort, spanning behavioural experiments and imaging studies as well as mathematical and simulation modelling.