Improving Self-Confidence: An Advise-Based Evolutionary Model

Abstract

A main concern in the simulation of adaptive behaviors is the characterization of organizational principles or architectures of adaptive behaviors in animals and synthetic animats. These characterizations are made through experimentation and analysis over well-defined models. In this work, we present a model to study the Vertical Cultural Transmission (VCT), as a catalytic factor over the interaction between individual learning and population evolution. Our model is based on the incorporation of advice, in the form of suggestions, as a way of information transmission between agents. An evolutionary reinforcement learning technique called classifier systems is used as the mechanism for individual learning. In particular, a recent variant based on the accuracy of the predictions made by the rules, called XCS, is applied. In order to control the environmental complexity, the Latent Energy Environments model, LEE, is adopted.

Finally, we present experimental results comparing three populations that are differentiated by their adaptation mechanisms: VCT learning, individual learning and evolution. For analise our results, we used self-organizing maps to clusterize two phenotypical traits that characterize our advise VCT model, belief and self-confidence.