ABSTRACT
Everyone will agree that an economic agent is, before all, a rational agent, but disagreements appear when it comes to characterizing the rationality involved.1 Among economists, it is common enough to refer to an ideal agent, of which homo oeconomicus is supposed to be the paradigm insofar as this agent is presumed to be capable, thanks to appropriate calculations, of maximizing whatever the value that is at stake and, consequently, of making optimal decisions. However, more and more objections have been raised against this way of conceiving of an economic agent, since it does not account for important traits of actual rational decisions. Therefore, various models of rational agents that avoid this kind of idealization have been proposed. Among them, the model developed by Herbert Simon in his life-long work on ‘bounded rationality’ is no doubt the most well known example. It is interesting to note that Simon’s conception places an emphasis on previously almost unexplored aspects of the rationality of agents. It is with the help of concepts such as ‘satisfactory pay-offs’ (Simon 1957: 245-252) and ‘aspiration level’ (ibid.: 254) that Simon manages to account, in a novel fashion, for subtle relations between preferences, beliefs and inferences. Other models of agents with limited rationality have been proposed, for example, by the psychologist Daniel Kahneman whose model P (as designated in Kahneman 1996b: 252-254) reduces in an even more radical fashion the traditional features attributed to the rational agent by economists. Indeed, in this model, rules and attitudes count among the components of rational behaviour. Kahneman’s explicit aim in devising this model was to explain any behaviour, whether strictly rational or ‘less rational’, with the help of the same mechanism, which can operate differently in different circumstances. It is also suitable in this context to consider the view of the philosopher Christopher Cherniak, for whom the rationality of the agent can be perfectly well described as ‘minimal’. According to Cherniak, the fundamental condition of minimal rationality can be formulated in the following way: ‘If A has a particular belief-desire set, A would undertake some, but not necessarily all, of those actions that are apparently appropriate’ (Cherniak 1986: 9; my emphasis). Note that the word ‘apparently’ is important here, because as minimal rationality does not imply that the beliefs are true, the actions suggested by an inference correctly drawn from false beliefs could be revealed as inappropriate while being
minimally rational. According to Cherniak, minimal rationality requires a (minimal) condition of consistency as well: ‘If A has a particular belief-desire set, then if any inconsistencies arose in the belief set, A would sometimes eliminate some of them’ (ibid.: 16). Even though they might appear somewhat restrictive and vague, these propositions are sufficient to highlight the fact that we are dealing with a form of rationality, since beliefs and desires are assumed to be related to the actions undertaken in an appropriate fashion, and since the few eliminated inconsistencies are eliminated precisely because their presence within the ensemble of desires and beliefs ends up being intolerable. Furthermore, as Cherniak suggests (ibid.: 19), the rather vague character of laws and principles of action that rest on such a minimal rationality would in no way deprive them of their predictive power. It is also a kind of minimal rationality that I am referring to when I claim that it is sufficient and less deceiving to simply assume that economic agents are not stupid and that they usually make decisions that are normally guided by reasons that they judge suitable for themselves. For sure, many economists will not like this somewhat informal way of characterizing rationality and agency and will claim that an economic science must be based on a much more precise definition of homo oeconomicus. Therefore, in order to vindicate the notion of minimal rationality, which risks being contested on this ground, I would like first to challenge an argument that seems to be a central tool in the arsenal of the defenders of the idealized model of the economic agent. Even though the rejection of this argument will not guarantee that resorting to a relatively minimal model of rationality in economics is well founded, it should contribute to assuring the degree of credibility that was denied to this approach by the argument in question. But first let us establish what this argument consists in.
