This research suggests a new model of decision under uncertainty in a multi-period framework where the decision maker (an individual or an organization) is assumed to be boundedly rational. The model differs from existing ones in two key assumptions: (1) the process is assumed to have an infinite history, i.e., to be in a "steady state," and (2Õ) the bounded rationality notion is modeled by a new computational model--a Turing machine with memory. The research objectives are: to define formally a Turing machine with memory; to study the various models of machines with memory, compare their computational ability and analyze the trade-offs among their complexity, size of memory and length of recall; to study the effects of interaction, assuming more than one decision maker is involved, and; to generalize the model in order to allow for the appearance and disappearance of decision makers (or agents) in the context of cooperative and competitive models.
