Bounded rationality and constrained maximisation

Research output: Contribution to conferencePaperpeer-review


David Gauthier’s (1986) theory of rational morality is heavily based on economic theory premises. Morals by Agreement proposes a moral theory based on economic rationality and suggests an account of constrained maximisation which relies on mainstream economic rationality. However, constrained maximisation has more in common with bounded rationality and heterodox economics than it does with neoclassical assumptions of rational agency.

A constrained maximiser is conditionally disposed to act “in ways that, if followed by all, would yield outcomes that she would find beneficial” (Gauthier, 1986: 167). So for Gauthier, a rational agent is disposed to act cooperatively in interactions represented by the prisoner’s dilemma game and therefore by economic theory standards, irrationally. The argument relies on two conditions: First, that the constrained maximiser acts on her disposition and second that there are other similarly disposed agents.

Constrained maximisation is only rational in a society where others are also disposed to constrain their maximisation and therefore, its rationality depends on others’ disposition. As such, one’s social environment is central to the argument for constrained maximisation. Rationality is then examined in a social context.

Repetitiveness is implied by the need for joint strategy and translucency, the capacity of rational agents to be able to predict others disposition. Repeated interactions ensure that these two assumptions are met, assuming that they entail non-random interactions within a social group, not just pair-wise interactions. One may successively interact with more than one interlocutors and as such, participate in repeated interactions. Agent A may interact with agent B once and then interact with agent C; this, for agent A, can be seen as a repeated interaction allowing her to form a joint strategy with her interlocutor of each interaction, and use her history to predict others’ behaviour.

Rationality as constrained maximisation is similar to the account proposed by heterodox economics and theories of bounded rationality. Boundedly rational agents are not hyper-rational as the agents in neoclassical economics. They are constrained by information availability, incomplete memory, reasoning capabilities, time and social environment. Thus, boundedly rational agents “look around them, they gather information and they act fairly sensibly on the basis of their information most of the time” (Young, 1998).

Information that affects behaviour depends on interactions and interlocutors. An agent whose interaction history is made up mostly of cooperative/non-cooperative interactions is more likely to expect similar future interactions, and act accordingly. For this agent the information available points to the fact that cooperation/defection is rational. In this context, the social environment affects and frames rationality and vice versa, rational strategies formulate the social environment. Whether it is rational to cooperate depends on whether one’s neighbours are disposed to cooperate (Skyrms, 2004). In a word of nasties, it pays to be a nasty (Sugden, 2004), and in a world of Fooles, it would be irrational to be a constrained maximiser (Gauthier 1986).

Therefore, although constrained maximisation is supposed to be supported exclusively by the premises of neoclassical rationality, it ends up being closer to bounded rationality and heterodox economics.
Original languageEnglish
Publication statusPublished - 2016


Dive into the research topics of 'Bounded rationality and constrained maximisation'. Together they form a unique fingerprint.

Cite this