Economics as psychology of rational behaviour

Whatever is used as a criterion to classify behaviour as “rational”, it must be distin­guished between (1) substantive rationality of (the results of) choice making as observed from an outside observer’s (“external”) point of view; and (2) procedural rationality from the (“internal”) point of view of the choice-making entity.

1. The first point of view broadly corresponds to that of a behaviourist psychologist who observes how rats in a maze respond to various stimuli. Along these lines, economists have studied adaptive behaviour of animals in broadly speaking micro­economic settings (see Kagel 1995). In such studies, the utility functions can, under certain minimum conditions of consistency, represent the choices that individuals are assumed to make from choice sets. Actors - rats, men, corporate actors and so on - frequently behave as if maximizing.

For the reasons indicated above, it is a mistake to drop the “as if” condition and to transfer the external description to motives in the internal mental model of the actor without a nomological theory justifying this move. Motives can be ascribed on the basis of overt behaviour only as theoretical terms of some underlying empirical theory of cognitive processes. The theory must link behaviour and motives by nomological hypotheses. The form of the utility function representing observed behaviour must be explained by the theory as well. Unless such a theory is presented the common practice of economists to put all sorts of concerns into the utility functions - “to make them more realistic” - remains “ad hoc”. Without systematically testing theories justifying such inclusions it is but another variant of “revelations” that are supported merely by plausibility. A possible remedy emerges if we leave the “mindless economy” and go from substantive to procedural accounts of rational choice making.

2. In the second, procedural, sense (minimally) rational actors choose on the basis of some mental model of the action situation. In forming such a model, rational actors will try to distinguish between what is and what is not subject to the causal influence they can exert through choices they make (see Kliemt 2009). Against the background of the distinction between what is and what is not subject to the causal influence of the acting entity, there are two ways to come to terms with stronger rationality requirements.

(a) Choice makers can rank states of the world along several value dimensions (according to desires and so on). Taken separately, these dimensions rank the states of the world to be influenced by choices. If this kind of evaluation allows for aggregation (corresponding to the formation of a decathlon table on the basis of its ten performance measures), an overall function comprising all dimensions of value may possibly be formed (see MacKay 1980). Then the choice maker might be seen as a person who seeks to make choices that will maximize the value of that aggregate function intentionally (see Broome 1991). This will come very close to the classical utilitarian model that assumed that humans would not merely behave as if maximizing a function but would indeed be guided by the intention to act such as to maximally fulfil their desires and some value criterion like, say, “aggregate happiness”.

(b) One might want to avoid the assumption that there is a maximand guiding intentional action. Individuals who do not intend to maximize nevertheless evaluate the expected consequences of possible choices and have certain aspi­rations that they want to fulfil (see Simon 1957). They intentionally satisfy aspirations rather than intentionally maximizing value. A theory of boundedly rational economic behaviour that does away with the necessity to describe behaviour as if maximizing could conceivably emerge from this move.

To explicate what the attribute “rational” - as an evaluative term - in the context of the two competing forms of procedural rationality - “2(a)” intentional maximiza­tion and “2(b)” intentional satisficing - means, it will be unavoidable to evaluate the results of the procedures.

More concretely, if actors guided by, say, a decision support procedure conform­ing with alternative “2(a)” would “on average” be substantively better off compared with actors who decide according to some decision support procedure correspond­ing to “2(b)”, then this would count as an argument in favour of procedure “2(a)”. However, even though this comes rather close to providing a tentative answer to Aumann’s question it does not explicate rationality fully in substantive terms. Substantive criteria on the ultimate level do not translate into substantive criteria on the proximate level of behaviour. On the contrary, behavioural procedures that are proximately neither maximizing nor satisficing may ultimately be the most success­ful and in that sense substantively rational.