Cooperation in the lab vs. real world

A recent outpouring of naturalistic studies on cooperation brought 3 interesting pieces of research with somewhat counter-intuitive results.

1. Prisoners without dilemmas

The Prisoner’s Dilemma game has been with us since 1950, but only now has someone bothered to check whether actual prisoners behave as predicted by the Nash equilibrium. They don’t. Cooperation flourishes in prison and this may be due to a number of reasons (parochialism to name one). Inmates cooperated 56% of the time whereas students only 37%.

Khadjavi, M. & Lange, A. (2013) Prisoners and their dilemma. Journal of Economic Behavior & Organization, 92:163–175.

2. Selfish Dictators

The idea that there is a universal willingness to share resources with strangers has been alive since Henrich et al.’s cross-cultural study with Dictator games. A caveat: participants in these games knew that they were participating in an experiment.

In their recent study, Winking and Mizer questioned the external validity of  such Dictator Game experiments. The researchers approached people at bus stops close to casinos in Las Vegas offering them free casino chips that could be exchanged for money. The researchers also made sure they had a believable excuse for why they themselves didn’t need the chips. In one of the conditions the experimenter suggested that the person should share the chips with a stranger (“an unmenacing 35-year old male confederate”). In contrast to results obtained in the traditional experimental setting, NO participants gave ANY chips to the stranger.

This is a bit shocking and flips the currently popular view that humans are cooperative by nature. Or, could it be that the money-oriented location had some effect on these findings? Also, the authors mention that “individuals were excluded if they were visibly intoxicated, mentally unstable or homeless” which tells us a bit about the specific decadent aura of Vegas bus stops.

Winking, J., & Mizer, N. (in press). Natural-field dictator game shows no altruistic giving. Evolution and Human Behavior

Henrich, J. et al. (2004) Foundations of Human Sociality: Economic Experiments and Ethnographic Evidence from Fifteen Small-Scale Societies. Oxford University Press.

3.  A/C martyrs

Having recently visited Florida, I know how important A/C is to Americans. It is very important. Given A/C is something people care so much about in hot US states, using  it as a currency for measuring cooperation is a spot-on choice.

Overexploitation of A/C on particulary hot days causes blackouts, making it a nice example of the tragedy of the commons. Some utility companies provide voluntary schemes through which residents can choose to have their A/C adjusted when there is a blackout danger. Of course such an adjustment prevents them from getting their desired temperature.

Yoeli and colleagues looked at how people subscribe to this scheme under various experimental conditions. In line with other studies on participation in public dilemmas they showed that when participation was non-anonymous it was three times higher than when the subscribers could not be identified by other residents.

Yoeli, E., Hoffman,M., Rand, D. G. & Nowak M. A. (2013) Powering up with indirect reciprocity in a large-scale field experiment PNAS 110:10424-10429

Cooperate, quick!

Within about one year of each other, two groups of researchers produced contradictory findings about the intuitiveness of cooperative behaviour, both published in Nature.

In 2012, Rand, Greene and Nowak provided exciting evidence that people are more cooperative when they take less time to make a decision about whether to cooperate or not, sugessting that we are predisposed to cooperate. However, a few days ago, a group from Sweden published a paper in which they describe their failure to replicate Rand et al.’s results despite using a sample of 2,500 people from three countries.

Rand et al. come back with a sharp riposte: Tinghög et al.’s study is not a faithful replication of theirs and a meta-analysis of 15 studies investigating the effect of time pressure on cooperation indicates that the effect is real. It would be interesting to see whether this effect is present in a more naturalistic setting, for example, looking at charity donations under pressure.

Rand, Greene & Nowak (2012) Spontaneous giving and calculated greed. Nature 489, 427–430

Tinghög,  Andersson,  Bonn,  Böttiger, Josephson,  Lundgren, Västfjäll, Kirchler, Johannesson (2013) Intuition and cooperation reconsidered. Nature, 498


Can people  accurately predict the behaviour of others when it comes to cooperation? We finally got an answer to this question that has long been perturbing scientists. Vogt, Efferson and Fehr conducted a study to test how well people can guess behaviour of participants in a Prisoner’s Dilemma (cooperation) and Stag Hunt (coordination) games. They proposed an evolutionarily-rooted theory, according to which, participants should not be able to predict cooperation, but should be able to predict coordination.

This logic comes from the design of the Prisoner’s Dilemma (PD) and Stag Hunt (SH) games. In PD, the highest payoff goes to a player who defects while the other player cooperates. Hence, the defectors should try not to give away any predictive cues to their behaviour. In SH, the most beneficial result for both players happens when both choose the same option. So, if they can communicate their intentions, both will benefit.

Two caveats:

  1. Both PD and SH games, to some extent, measure risk aversion, which may confound the results.
  2. In some environments, the optimal strategy may be not to maximise the total payoff but the difference between one’s own and other’s payoff.

Anyway, the long-awaited answer to the question whether we can accurately predict intentions in inter-dependent economic games is…: maybe, but probably not. Contrary to hypotheses, participants did not exceed the chance level when predicting behaviour in SH. Also, although participants seemingly had higher accuracy level when guessing behaviour in PD, computer simulations indicated that this result isn’t robust.

The inconclusiveness of this study is disappointing, yet it is comforting to know that, as long as you produce a coherent piece of research, the fact that you failed to support your hypotheses will not preclude you from getting published in Evolution and Human Behavior.

Vogt, S., Efferson, C. & Fehr, E. (in press) Can we see inside? Predicting strategic behavior given limited information. Evolution and Human Behavior

Carrot-and-stick works in the US but not in Romania

Recently, an anonymous peer reviewer of one of my papers brought to my attention an exciting cross-cultural study on cooperation, rewards, and punishment. The study was conducted with US and Romanian students who were asked to repeatedly play:

  • public good games (PGGs)
  • versions of Prisoner’s Dilemma (PD)
  • PGGs followed by PDs
  • PGGs followed by punishment opportunity.

The main findings are:

  1. US and Romanian students did not differ in their cooperation measured by repeated PGGs and PDs.
  2. However, when given an opportunity to reward (through PD) or punish the behaviour of others, students from the two countries differed considerably:
    • American students frequently rewarded cooperation and withheld rewards towards non-cooperators.
    • Romanian students tended to reward those who rewarded them previously, but rewards were not linked to group-benefitting cooperation. In other words, Romanian students valued the prosperity and success within pairs more than within group (This finding suggests that reputation-building may not be a cooperation-enhancing mechanism in all environments/cultures.)
  3. When presented with an opportunity to punish, American students used punishment mostly to enforce cooperation, whereas Romanian students displayed high levels of antisocial punishment. Punishment meted out by Romanian students was often indiscriminate: targeted against both non-cooperators and cooperators.

I liked the study a lot but was not convinced by the high-level interpretation of the findings. The authors say: “Broadly speaking, we suggest that some societies prosper because they are good at building shared understanding.” and “The ultimate reason that an engaged group member is able to foster cooperation is that the other group members understand what is going on and adjust their behavior both with respect to the group at large and with respect to the engaged group member.” How do the above results address the problem of shared understanding? Didn’t Romanian students understand what was “going on” and adjust their behaviour in response to how others behaved? Although they didn’t foster cooperation, the strategy they adopted might have been better suited to the socio-environmental conditions they live in.

Ellingsen, Tore, Herrmann, Benedikt, Nowak, Martin A., Rand, David G. and Tarnita, Corina E., Civic Capital in Two Cultures: The Nature of Cooperation in Romania and USA (November 1, 2012). Available at SSRN: or

The academic fitness of HBE

Nettle. et al. (2013) summarise the condition of Human Behavioural Ecology. It’s doing alright. Evolution and Human Behavior and Human Nature seem to be the most fertile publishers of research in HBE between 2000 and late 2011. They are followed by American Journal of Human BiologyCurrent Anthropology and Proceedings Royal Society B.

As strong points of the field, the authors identify:

  • vitality – increasing number of published papers
  • broad scope and conceptual coherence
  • high ecological validity in comparison to evolutionary psychology and behavioural economics

HBE fares less well when it comes to:

  • integration with (Animal) Behavioural Ecology
  • the variety of topics:  there a very few, if any, papers on diet, resource extraction, resource storage, navigation, spatial patterns of habitat use, hygiene and social coordination

From my perspective, an exciting development of HBE would be research on  how reputations are built and spread under varied ecological pressures.

Nettle, D., Gibson, M. A., Lawson, D. W. & Sear, R. (in press) Human behavioral ecology: current research and future prospects. Behavioral Ecology


Patience and punishment

The use of altruistic and antisocial punishment appears to be related to temporal discounting. In a study conducted in Spain, present-oriented participants tended to mete out antisocial punishment, while future-oriented participants were inclined to impose altruistic punishment. This fits well with the finding by Herrmann et al. (2008), in which they showed a negative correlation between the expression of antisocial punishment and GDP per capita. Discounting the future and focusing on present competition might bring high payoffs in unpredictable environments with scarce resources. In contrast, enforcing cooperation with an expectation of future benefits is likely to be a successful strategy in more stable and wealthier places.