Portal de Periódicos da CAPES

Social Intelligence in Games

2003; Mohr Siebeck Verlag; Volume: 159; Issue: 1 Linguagem: Inglês

10.1628/0932456032975050

1614-0559

Gerd Gigerenzer, Richard McElreath,

Economic theories and models

In his essay “An anthropologist on Mars,” the neurologist Oliver Sacks describes his encounter with Temple Grandin, a highly remarkable autistic person who holds a Ph.D. and teaches at Colorado State University. Like other autists, Temple Grandin seems to be largely devoid of social intelligence. When Temple watches the social games normal people play, the social rituals they follow, and their powerful emotions that can change from love today to hate tomorrow, she feels, as she put it, like an anthropologist on Mars. As SACKS [1995, p. 270] describes her, “Lacking it [social intelligence], she has instead to ‘compute’ others’ intentions and states of mind, to try to make algorithmic, explicit, what for the rest of us is second nature.” Oliver Sacks studies people who have neurological disorders. Economists and cognitive psychologists usually do not; they study normal people. Nevertheless, there is a similarity between the autistic personality and the rational models that many economists and some psychologists embrace. If by “rational” we mean conforming to the classical expected-utility model, or backward induction, then we have an “autistic” conception of human rationality. Just like Temple Grandin, homo economicus – defined in that way – lacks social intelligence and is puzzled by the strange behavior of normal people. Experimental economists rely on simple games to elicit behavior that is blatantly at odds with the assumption that players are attempting to maximize their own expected utility. Consensus is growing that the rational-choicemodel is descriptively wrong. This message has been hammered home by two groups of researchers: (i) experimental economists studying social games such as the ultimatum game, and (ii) psychologists demonstrating that individual people’s judgments and decisions violate the axioms or consequences of the expected-utility calculus, the laws of probability, or logic. Real humans are capable of acting and punishing altruistically and of valuing fairness, and are guided by emotions, obligations, feelings of guilt, and other moral sentiments. What to do with a model of rational manwho is socially unintelligent, that is, a model that is descriptively wrong? One reaction is just to go on, close one’s eyes, and ignore this research. A second reaction is to engage in a “repair program”: One sticks with the expected-utility