Imagine that your boss wants you to work, but you'd rather watch cat videos. Your best strategy is to learn your boss's patterns so that you'll appear hard at work only when she's watching. New research shows that chimpanzees are better than humans at making this kind of Machiavellian deduction.
Caltech behavioral economist Colin Camerer wanted to understand the evolution of optimal decision-making. Working with a team including Kyoto University's Tetsuro Matsuzawa and PhD student Christopher Flynn Martin, he designed a game that he taught to six chimpanzees from Kyoto University Primate Research Institute. The "Inspection Game" is no more complicated than hide-and-seek. Two chimpanzees sit back to back, each with a computerized touch screen in front of him or her. Two blue boxes appear on the left and right sides of the screen. Each chimpanzee simply has to choose one of them, but the two chimps have different jobs. One of them is the "hider" – he wins if he chooses the opposite of his partner's choice. The other is the "matcher" – she wins if she chooses the same as her partner.
Let's say Ayumu is the hider and Ai is the seeker. Ayumu wins if he chooses "right," but his mom chooses "left" (or vice versa), while Ai wins whenever she and her son pick the same side. Each pair played 200 rounds, and winners received chunks of apple for their efforts.
One way to go about the game is to play randomly, but that isn't a particularly fruitful strategy. Not if your goal is to get as many pieces of apple as possible. Instead, the chimpanzees should try to accurately predict what their partners will do next, and play the game accordingly, without being too predictable themselves.
That's essentially the same as in the scenario I asked you to imagine above. Your goal should be to anticipate your boss's behavior, so that you can anticipate when she'll be watching you work and when you can get away with watching cat videos. At the same time, your boss needs to anticipate your strategy, perhaps by becoming more unpredictable.
Still, if both players behave most strategically, game theory predicts that there's a limit to how often either player can win, and it's called the "Nash equilibrium," named for mathematician John Nash Jr., who won the 1994 Nobel Memorial Prize in Economics.
In experiment after experiment, no matter the particular details of the game being played, humans tend to deviate widely from the Nash equilibrium, suggesting that there's something about human reasoning that prevents us from constantly making the most strategic decisions. That's why Camerer and his colleagues were so surprised by the performance of the chimpanzees in the Inspection Game. They performed almost at the Nash equilibrium!
Even when the researchers tried to confuse the chimpanzees by changing the reward distributions, or by arbitrarily switching who was the "hider" and who was the "seeker," the chimpanzees rapidly adjusted, maintaining their play at the most optimal of levels.
And not only that, but the chimpanzees' performance "match[ed] game theory predictions as closely as in any species and comparable learning setting ever observed," in any experiment!
Since Camerer and his team were so surprised at how competent the chimpanzees were at the game, they decided to test two groups of human players: Japanese human students from Kyoto University, and adult men from the village of Bossou, in Guinea, West Africa.
The two human groups couldn't have been more different: the students were enrolled in universities in a large Japanese city, and if they won they received a handful of 1 Yen coins (1 Yen is less than one US cent). The villagers from Bossou are poor, and education levels there are low. If they won, they were given a handful of 100 Guinea franc coins. 100 Guinea francs are also roughly equivalent to one US cent, but a handful of such coins amounts to a full day's wage. While the Japanese students played with computer screens, as the chimpanzees had, the Guinean villagers played a version of the game using bottle caps.
Despite the differences in education level and motivation to win, the two groups of humans performed similarly, and both were dramatically less successful at behaving optimally than were the chimpanzees. Thus, the chimpanzees' success rates can't be attributed to motivation, or to familiarity with the computer screens.
So why are chimpanzees so much better at this sort of reasoning that humans, with all our fancy language and tools and so on?
The researchers offer at least two possible explanations. This first is Matsuzawa's "cognitive tradeoff hypothesis." According to Matsuzawa, as the human brain developed specialized regions for uniquely human capacities like language, other cognitive abilities suffered. Particularly abilities related to short term memory. Camerer and Matsuzawa point out that no human has ever been able to match chimpanzees (like Kyoto University's Ai and Ayumu) when it comes to memory tasks.
That tradeoff may have placed the humans at a relative disadvantage, compared to the chimpanzees, who may be more adept at reasoning their way through competitive situations like the Inspection Game.
That leads to the second possible explanation, which is that chimpanzees evolved in an environment that rewarded competition more than cooperation, and that a fundamentally competitive spirit is necessary to ever approach the Nash equilibrium. Chimpanzee social life is dominated by competitive strategic interactions, like predatory stalking, raiding crops from human farms, play chasing, and border patrolling (which is itself very much like the Inspection Game). That's not to mention the competitive nature of chimpanzee social hierarchies themselves. "Because competitive payoff games are common in chimpanzee life, evolutionary theory predicts that chimpanzees would have developed cognitive adaptations to detect patterns in opponent behavior and to create predictability in their own behavior," the researchers write. "Some have argued that the capacity to randomize effectively evolved because primate predatory behavior and routine social interaction selects for unpredictability in counter-strategies. Experiments also show that chimpanzees are better at competitive tasks than at comparable cooperative ones," at least when dealing with non-kin.
It isn't necessarily clear just where humans fall on the cooperation-competitive continuum, but it seems reasonable to say that human society is in general more prosocial and cooperative than is chimpanzee society.
Scientists often worry whether tests they give to animals are "fair," or if they're too based upon human behavior. For example, some animals fail spectacularly at recognizing their own mirror images, but can convincingly display evidence of self-awareness if allowed to use senses other than sight. Could it be that these sorts of game theory models are similarly "unfair" to humans, because we evolved in an environment that generally rewarded cooperation and altruism over competition and selfishness? Or because the human players were placed in a situation that robbed them of the chance to use language?
Camerer thinks so. "It is notable that in our protocol humans are deliberately deprived of an extraordinary cognitive ability – language. In competitive games language cannot increase group rewards. But in games requiring coordination and cooperation, where language is particularly useful in our ecology, the [evolutionary] prediction is that humans will outperform other species."