The only rational way to play Prisoner’s Dilemma, as a selfish myopic utility maximizer (Economic Man) is to play D (defect). Yet this way one ends up with less than the maximum pay-off, which is possible only if the two players agree to cooperate and keep that agreement.

How do real humans play? In numerous PD game experi-ments, the degree of cooperation starts at about 50%, then dips down, and eventually comes up again to above 50%. (The latter is the average between CC and DD lock-ins – more CC than DD.) Yet, by rational selfish myopic logic, people should play DD all the time. Why are real humans so much better than strict logicians?

Partly because they expect (and often get) reciprocity from their partners (opponents?), especially if they expect to meet them again. (I.e. in long protocols or in computer tournaments.) Axelrod showed that tit-for-tat reciprocity can persists in spite of DD attacks, and even successfully invade a society of DD players. (Cf. “Survival of Cooperation”.) Partly they learn as they play. And partly they know how to communicate with their partner (who is not really an opponent) by non-verbal signals, i.e. their mode of play. In sum, real humans are much smarter than strict logicians. Or, put another way, rationality is not the apex of intellectual capability. Experience far outruns strict reasoning-by-the-rules when it comes to winning playful games, and even dead-serious games of survival in the real world. Humans have evolved to play these dead-serious games, not to follow syllogisms or true/false tables.

Another instructive example of this propensity is so-called prospect theory, which has to do with decision-making. The rational way to make decisions is expected utility calculation: one outlines all the options, assigns to each outcome its utility (how much one values it) and its probability of being realized in that option, multiplies each utility by its probability of attainment (to obtain the so-called “expected utility”), and then chooses the largest expected utility.

However, again real people deviate from such rational behaviour. The deviations are systematized in so-called “prospect theory”. They are of several kinds.

1. People tend to over-estimate losses and under-estimate gains.
2. People tend to judge losses and gains from the present state (status quo) rather than from an absolute base.
3. People tend to prefer a small but sure gain over a probable larger gain even if expected utilities are equal or reversed.
4. People tend to prefer a sure security against losses to probable security against larger losses even if the expected utility is the same.
5. People react differently when risks are stated in percentages rather than in actual numbers.
6. People tend to react differently to very small probabilities regarding gains and losses: e.g. gamble to win in lotteries but disregard dangers of road accidents.

Looking over this list, it seems that a brain attuned to survival rather than abstract thought could be expected to have exactly this kind of deviations from rationality. For example, it would prefer absolute security against large losses (especially death) to anything else. Gains are nice, but relatively less important.

To conclude: rationality in the sense of cost-benefit calculations plays a part in human decision-making, but not the predominant part. Experiments show that real people are much more flexible in their reasoning and interactions with others, and much more oriented toward survival (as Darwinian theory would predict) than the austere but arid logician.