ABSTRACT
In the first part of the chapter, we introduce a general developmental model of intelligence and argue that the strict notion of optimization does not constitute a proper level of analysis for understanding human intelligence because a fixed, a priori construct (objective function) cannot account entirely for the interactive nature of intelligence. In the second part of the chapter we address the synthesis question by presenting simulations of a specialized architecture of the general model. The synthesis of this architecture was not based on the traditional optimization methodology in that the minimization of a fixed, predetermined global “cost function” was not the central design principle. No attempt was made to explicitly synthesize such a function. Rather we approached the problem in a more dynamical setting by integrating specialized architectures whose continuous-time interactions synthesize goals that change according to the interactions between the system and the environment. We present several simulations illustrating the interplay between system and environmental dynamics and show, for example, how the system can change its criterion (modify its “cost function”) according to prevailing environmental conditions.
