4 Towards dealing with the complexity of modelling agents - modelling modelling

4.4 Processes of model development

Despite the fact that there are a multitude of candidate mechanisms to choose from in AI and cognitive science two mechanisms have dominated economic models of learning: those of optimization and evolution (GA/GP/EP etc.). Of these approaches those based on the GP paradigm [13] appear the most appropriate, due to the potentially expressive format of the genome. Using such an approach one can represent an agent as a population of competing mental models [8]; a possible architecture for such an agent is shown in figure 3.

Figure 3: Using a genetic population to model an economic agent

The evolutionary approach has the weakness that the development of variation is blind to the process of selection. In other words it avoids the intentionality inherent in human learning. There are, however many other credible models for learning [3].

Generated with CERN WebMaker