Complex Computations by Simple Brains – The Selfish Herd Model Revisited

This thesis used two computational frameworks -- agent based modelling and reservoir computing -- to provide insight into a
renowned explanation for the evolution of gregarious species, namely the selfish herd. The anchoring assumption of the theory
was validated with a computationally complex movement rule in an agent-based model illustrating that the selfish avoidance of
a predator does result in aggregation. These rules were then shown to be learnable by some of the most basic neural networks,
namely a reservoir computer, which suggests that more intelligent animal brains may also evolve such behaviour.