we can consider the purest model of how abstract reasoning might undermine the temptations of violence, the Prisoner’s Dilemma. In his popular Scientific American column, the computer scientist Douglas Hofstadter once agonized over the fact that the seemingly rational response in a one-shot Prisoner’s Dilemma was to defect. You cannot trust the other player to cooperate, because he has no grounds for trusting you, and cooperating while he defects will bring you the worst outcome. Hofstadter’s agony came from the observation that if both sides looked down on their dilemma from a single Olympian vantage point, stepping out of their parochial stations, they should both deduce that the best outcome is for both to cooperate. If each has confidence that the other realizes that, and that the other realizes that he or she realizes it, ad infinitum, both should cooperate and reap the benefits. Hofstadter envisioned a “superrationality” in which both sides were certain of the other’s rationality, and certain that the other was certain of theirs, and so on, though he wistfully acknowledged that it was not easy to see how to get people to be superrational.
Can higher intelligence at least nudge people in the direction of superrationality? That is, are better reasoners likely to reflect on the fact that mutual cooperation leads to the best joint outcome, assume that the other guy is reflecting on it as well, and profit from the resulting simultaneous leap of trust? No one has given people of different levels of intelligence a true one-shot Prisoner’s Dilemma, but a recent study came close by using a sequential one-shot Prisoner’s Dilemma, in which the second player acts only after seeing the first player’s move. The economist Stephen Burks and his collaborators gave a thousand trainee truck drivers a Matrices IQ test and a Prisoner’s Dilemma, using money for the offers and payoffs. The smarter truckers were more likely to cooperate on the first move, even after controlling for age, race, gender, schooling, and income. The investigators also looked at the response of the second player to the first player’s move. This response has nothing to do with superrationality, but it does reflect a willingness to cooperate in response to the other player’s cooperation in such a way that both players would benefit if the game were iterated. Smarter truckers, it turned out, were more likely to respond to cooperation with cooperation, and to defection with defection.
The economist Garrett Jones connected intelligence to the Prisoner’s Dilemma by a different route. He scoured the literature for all the Iterated Prisoner’s Dilemma experiments that had been conducted in colleges and universities from 1959 to 2003. Across thirty-six experiments involving thousands of participants, he found that the higher a school’s mean SAT score (which is strongly correlated with mean IQ), the more its students cooperated. Two very different studies, then, agree that intelligence enhances mutual cooperation in the quintessential situation in which its benefits can be foreseen. A society that gets smarter, then, may be a society that becomes more cooperative.
More (#2) from Better Angels of Our Nature: