Anderson proposes that human cognitive architectures will have adapted optimally to the problems posed in their environment. Therefore, discovering the optimal solution to the problem posed by the environment, independant of the architecture, is equivalent to discovering the mechanism used by the architecture. A 'Rational Analysis', as it is called, takes into account the available information in the environmnet, the goals of the agent, some basic assumptions about computational cost (in terms of a 'general' architecture mechanism),and produces the optimal behavioral function. This function, then of course can be tested empirically and assumptions modified if it proves inaccurrate. A contrasting point of view to this is espoused by Simon, and is centered around the claim that, in a rational analysis, the assumptions about the architecture actually do most of the work.