Entropia na evolução simbólica através de algoritmo genético

Abstract

This work proposes a computational model using genetic algorithms, due to its correspondence to naturally found evolutive pressures, to study the dynamics of attributing meaning to initially arbitrary expressions. The model was characterized and its behavior’s physical properties were studied, especially regarding entropy. Entropy in this modeled system decays exponentially from initially elevated values to minimum ones, where it stabilizes. It also was capable of distinguishing between different groups of conceptual repertoires. Systems that behave like this model are able to quickly achieve high levels of efficiency, despite having as a starting point an ensemble of individuals with few shared elements. This work also studies the effects of imposing a grammar structure on message formulation, which did not show significant impacts on its dynamics. However, within this work context and its limitations, this absence of effect corroborates certain theories of language (human) as fortunate byproducts.