**Larry Yaeger**

(Indiana University)

Wednesday 9th November 2011

16.00pm

** **

B10 Seminar Room,

Alexandra House, 17 Queen Square, London, WC1N 3AR

**Evolutionary Selection of Neural Network Structure and Function**

I will examine the relationship between evolved neural network structure and function, applying graph theoretical tools to the analysis of the topology of evolved artificial neural networks and information theoretical tools to their dynamical neural complexity.

Results suggest a synergistic convergence between network structures emerging due to physical constraints, such as wiring length and brain volume, and optimal network topologies evolved purely for function in the absence of physical constraints. Increases in clustering coefficient are observed in concert with decreases in path length, together producing a driven evolutionary bias towards small-world networks relative to comparable networks drawn from a passive null model. These small-world biases are exhibited during the same period that evolution actively selects for increasing neural complexity, and during which the model's agents are behaviorally adapting to their environment, thus strengthening the association between small-world network structures and complex neural dynamics. I will also introduce a new measure of path length in graphs, "normalized path length", that is better behaved than existing metrics for networks comprised of disjoint subgraphs and disconnected nodes, a novel method of quantifying the degree of evolutionary selection for small world networks, "small-world bias", and an open source suite of graph theoretical tools, "bct-cpp", which is a C++ version of Sporns, Kötter, and Rubinov's MATLAB "Brain Connectivity Toolbox".

#