Neuronal Networks

Neurons in the cerebral cortex form a network in which information flows in the form of short pulses called `spikes’. Recent modelling studies have shown that repeated exposure of a neural network to various stimuli will result in the formation of ‘memory traces’ – characteristic patterns of spikes associated with each stimulus (Izhikevich 2006. Neural Computation).

During my PhD, I developed a computational model of the mechanism underpinning this process in biologically plausible networks of neurons, and have tested it against a set of fundamental criteria. For example, a key requirement for any memory encoding is that incomplete cues should suffice to recall familiar stimuli, much like our ability to infer entire words despite missing lettrs. We have shown that the randomly connected network models normally used often fail in this regard. However, investigating a range of different network topologies, we found that so-called `small-world’ architectures enable reliable discrimination between inputs even when prompted by 80% incomplete recall cues. Furthermore, we have shown that small-world architectures operate at significantly reduced energetic cost and that they introduce biologically realistic constraints on the optimal input stimuli. Interestingly, recent studies have found real cortical networks to display these small-world properties. Taken together, our results suggest that the architecture of cortical networks is optimal and perhaps necessary for this type of neuronal encoding mechanism.

In collaboration with: Tom Duke, Danielle Bassett



In my transition from physics to computational neuroscience, there are a number of resources which I found invaluable:

Perlewitz’s Computational Neuroscience on the Web
contains amongst other things:
-   a list of conferences and workshops on computational neuroscience
-   a list of people working in the field
-   a list of relevant journals and books

is an online peer reviewed encyclopedia written by scholars from around the world. It looks like wikipedia, but it focusses on specific fields:
-   Encyclopedia of Computational Neuroscience
-   Encyclopedia of Dynamical Systems
-   Encyclopedia of Computational Intelligence

Seminars and talks online
-   Seminars at the Redwood Center for Theoretical Neuroscience
-   Heller Lecture Series in Computational Neuroscience
-   Talks at the Kavli Institute’s ‘Dynamics of Neural Networks’ conference in 2001 and talks from the corresponding pedagogical program