The rationality of a system simply is a measure of to what extent the system's actions are always consistent with it's goals and knowledge. There may be many reasons why a system is not absolutley rational, but the main reason is that as knowledge increases, so does the time it takes to compute the completely rational answer. Therefore many systems trade off 'rationality' or 'quality' of a solution for a timely response. The following categorization breaks the architectures we are disussing in to three categories. First, those systems that are always purely rational. Second, those systems that would compute the rational solution if given enough time, and thirdly, systems which always compute the answer which is 'most probably' correct.