ABSTRACT
Bounded rationality has been embodied in models of individual and organizational behaviour mainly – if not exclusively – in terms of limitations in the (algorithmic) computational capabilities of agents. Much less attention has been given to the bounds agents encounter in their capabilities of representing the complex problems they face. Most of economic theory, even when allowing for informational and computational limitations, assumes that agents correctly represent the problems they face. In this chapter I present a simple abstract model of a class of problems (puzzles) which require the coordination of a large number of interdependent elements and discuss how different representations of the problem induce different levels of difficulty. In particular, it is shown that by acting on the representation of the problem agents can radically change its degree of decomposability into independent or quasi-independent subproblems and thus show that changes in the representation can be more powerful a problem-solving heuristic than devising more sophisticated search algorithms. Moreover, I argue that there exists a trade-off between the level of detail of the representation and the level of difficulty it implies. Finally, I discuss implications of the model for individual and collective problem-solving and, in particular, for the theory of markets and organizations, viewed as institutions which socially construct problem representations.
