Discovering and maintaining behaviours inaccessible to incremental genetic evolution through transcription errors and cultural transmission

In this work the question of whether the introduction of both transcription errors and cultural transmission, in the form of learning by imitation, can enable the evolution of behaviours inaccessible to incremental genetic evolution alone is assessed. To answer this a neural network model using a hybrid of two different networks was implemented: one capable of demonstrating reactive qualities, the other controlling deliberative goal selecting behaviours. Animats using this model were evolved in an adaptation of the environment proposed by Robinson et al. (2007) to solve increasingly difficult tasks. Simulations were run on populations with and without learning by imitation to assess the relative success of each strategy, leading to the conclusion that populations with learning by imitation can successfully demonstrate the most complex behaviour, which was empirically found to be inaccessible to non-learning populations.