Individuals, groups, and decisions

Imagine buying something from a friend: do you think you’d give him a better or worse offer than you’d give a stranger? Would you buy something you might not normally want if pressured into it by a friend?  The thing is, our preferences and decisions aren’t consistent from moment to moment, they’re always changing.  One pet interest of mine has long been how decisions change when in groups than when alone.  Do we make the same decisions?

It turns out that we don’t.  People in groups are more akin to ruthless machines, making the economically “rational” self-interested decision even when the socially-optimal decision is something else.  Take, for example, the Trust and Ultimatum games.  Whereas people – westerners, at least – tend to make more pro-social offers, sharing and avoiding pure self-interest, when a group of individuals make a decision, they’re more likely to make the ‘rational’ decision and screw the other group.  Groups of people favor themselves, even when the individuals in the groups may have been willing to share if on their own.  When you consider that a lot of our economy is built on how much we trust each other, this could be a bad thing.

But it’s not all bad!  Groups of individuals may choose the economically optimal decision – but sometimes this decision is good for everyone.  Take games that have multiple equilibria, especially ones that require coordination.  When individuals are playing in groups for a coordination game, they are more likely to be able to coordinate with the other group in order to help everybody.  Even an individual that is simply on their own, but identified as part of some group, will be able to coordinate better.

Humans strongly identify with in- and against out-groups, and it shouldn’t be surprising to anyone with a knowledge of psychology that people in groups may be more self-interested.  But what is interesting is how often the self-interest helps everyone.  In general, in conditions where there is a unique pure-strategy equilibrium, the group will find this equilibrium and hurt everybody.  In conditions where there are multiple equilibria that require coordination, groups will be more efficient and help everybody.  Groups are better at strategizing, better at taking into account (so far) unchosen strategies, and better at anticipating the actions of other groups.

There are a few different possible explanations for this.  It could be – and is extremely likely – that there are strong in-group versus out-group effects.  It could be that being in a group increases the motivation of individuals.  But then there’s a new study in Nature that suggests a possible alternative: that the individuals are being given more time to think about their decision, and people who are given more time to think are more self-interested.  Not that this should be a surprise to those with beer-goggles.

If you just strap a single person down, and ask them to play a social dilemma game, GO NOW FAST!  They will make a choice and it will, more often than not, be somewhat fair.  If you strap that person down, and ask them to play, but hey, take your time?  They’ll think about it, maybe, who knows, nurse some grievances, and decide that they are more important and be less fair.  In fact – and I think this is way cool – if you look at their data, how much people are willing to give is almost perfectly negatively linearly correlated with the logarithm of the decision time.  I mean, look at this beauty:

And maybe this can partly explain what is happening with the groups: they’re simply given more time to rationalize their decisions.  I’m sure that’s not everything – as I said, in-group effects are strong in people – but it is a lot.  It turns out that the two great rationalizers are taking your time, and worrying about your group.

G Charness & M Sutter (2012). Groups make better self-interested decisions Journal of Economic Perspectives DOI: 10.1257/jep.26.3.157
Rand DG, Greene JD, & Nowak MA (2012). Spontaneous giving and calculated greed. Nature, 489 (7416), 427-30 PMID: 22996558

Testosterone: cooperation or competition?

In my last post, I gave an introduction into a couple aspects of testosterone: how it rises and falls, and how it affects decision-making.  I forgot to mention that, neurally, it appears to act substantially through three areas of the brain: the nucleus accumbens, amygdala, and orbitofrontal cortex (OFC).  The nucleus accumbens is a major dopaminergic center, the molecule generally seen as responsible for decision-making and action selection.  Amygdala, as we all know, mediates fear and emotional responses (generally…).  The more interesting area is OFC, which is typically thought to be an area that is involved in self-control.  I couldn’t find many papers that I really wanted to talk about on this aspect of testosterone, so I’ll wait for another day to delve into it.

So let’s look at how testosterone affects social behavior.  In what must have been the Most Fun Study To Participate In Ever, Oxford et al. asked subjects to play Unreal Tournament in teams.  When men are playing the game against other teams, the players that contribute the most show increases in testosterone.  However, when men are forced to play against their own teammates, they have decreased testosterone!  And the subjects who contributed the most to a win showed the largest change.

But if testosterone is increasing during competition against other groups, what affect might it have on social behavior?  Eisenegger et al. used the ultimatum game, where a pair of subjects are given a small amount of money.  One of the subjects then makes an offer of part of the money to the other subject, who can either accept the money or reject it; when the second subject rejects it, neither subject gets any money.  It is well-known that people will generally reject unfair offers.  Following the framework of past studies, female subjects were given either testosterone or placebo and asked to play the game.  They found that subjects who were given testosterone made larger offers than placebo subjects.  Although the authors try to make the claim that this is because being turned down is a ‘status concern’, it could just be because they think that they will make more money that way?  Maybe this is risk-aversion?  I should also note that different study found that subjects given testosterone and asked to be the second subject will also reject more unfair offers.  But the most interesting part about the study is that the subjects who thought that they had received testosterone made much smaller offers – presumably because they already thought they knew what testosterone should do, even though they were wrong!

In a response, van Honk et al. tried using a different game.  He used the ‘public good game’ which is where all players receive 3 moneys, and can contribute some to the public good.  When at least two players contribute to the public good, all players receive 6 moneys.  Note that in this version of the game, with the contribution rate of other players, the expected value is highest when you contribute to the public good.  And subject who have testosterone administered to them give more often to the public good!  So it’s not clear whether they are being more pro-social or just smarter…

The interesting thing about this paper, though is that they also measured the ratio of ring to index finger.  This is a measure of prenatal testosterone exposure, although it doesn’t predict adult levels of testosterone.  Those with a high 2D:4D ratio (ie, those with low maternal testosterone, figure left) are most likely to contribute to the common good, and the less prenatal testosterone, the more of an effect the testosterone given to subjects has.  van Honk et al. suggest that prenatal exposure may change something physically to make subjects more receptive to testosterone, whether it is metabolism or receptor level.  They had found a similar result in a previous study which showed that suspicious individuals didn’t become any more suspicious from testosterone, but the most trusting individuals became much more suspicious when given testosterone (figure right).

The data is a bit hard to interpret, but the general feeling now is that testosterone can act as either a pro-social hormone, or one that makes you more concerned about your social status (egocentrism?).  Although I’d love to give a good clean explanation here, I cannot come up with – and have not yet found – a good unifying framework that unites all the social effects of testosterone .


Eisenegger, C., Naef, M., Snozzi, R., Heinrichs, M., & Fehr, E. (2010). Prejudice and truth about the effect of testosterone on human bargaining behaviour Nature, 463 (7279), 356-359 DOI: 10.1038/nature08711

van Honk, J., Montoya, E., Bos, P., van Vugt, M., & Terburg, D. (2012). New evidence on testosterone and cooperation Nature, 485 (7399) DOI: 10.1038/nature11136

Oxford, J., Ponzi, D., & Geary, D. (2010). Hormonal responses differ when playing violent video games against an ingroup and outgroup Evolution and Human Behavior, 31 (3), 201-209 DOI: 10.1016/j.evolhumbehav.2009.07.002