Some people go out of their way to help their peers, while others are more selfish. Some lend their trust easily, while others are more suspicious and distrustful. Some dive headlong into risky ventures; others shun risk like visiting in-laws. There’s every reason to believe that these differences in behaviour have biological roots, and some studies have suggested that they are influenced by sex hormones, like testosterone and oestrogen.
It’s an intriguing idea, not least because men and women have very different levels of these hormones. Could that explain differences in behaviour between the two sexes? Certainly, several studies have found links between people’s levels of sex hormones and their behaviour in psychological experiments. But to Niklas Zethraeus and colleagues from the Stockholm School of Economics, this evidence merely showed that the two things were connected in some way – they weren’t strong enough to show that sex hormones were directly influencing behaviour.
To get a clearer answer, Zethraeus set up a clinical trial. He recruited 200 women, between 50-65 years of age, and randomly split them into three groups – one took tablets of oestrogen, the second took testosterone tablets and the third took simply sugar pills.
After four weeks of tablets, the women took part in a suite of psychological games, where they had the chance to play for real money. The games were designed to test their selflessness, trust, trustworthiness, fairness and attitudes to risk. If sex hormones truly change these behaviours, the three groups of women would have played the games differently. They didn’t.
Their levels of hormones had changed appropriately. At the end of the four weeks, the group that dosed up on oestrogen had about 8 times more than they did at the start, but normal levels of testosterone. Likewise, the testosterone-takers had 4-6 time more testosterone and free testosterone (the “active” fraction that isn’t attached to any proteins) but normal levels of oestrogen. The sugar-takers weren’t any different. Despite these changes, the women didn’t play the four psychological games any differently.
In a “Dictator game”, they were given 200 Swedish Kronor and told to distribute it between themselves and a charity for the homeless – this was a measure of their selflessness. In an “Ultimatum game”, a partner offered them a share of a 400 Kronor pot. If they accepted the offer, both players took their respective shares, but if they rejected it, neither player got anything. The cut-off point where they would accept the offers, even if they were small ones, indicated their attitudes towards fair play.
In a “Trust game”, they had to give a share of a 150 Kronor pot to a “trustee”. That investment was tripled and the trustee had to decide how much to give back to the “trustor”. The women played both parts, with the trustor role measuring their level of trust in their peers, and the trustee role showing their own trustworthiness. Finally, the women had to choose between taking a 50/50 gamble on winning 400 Kronor and a certain but smaller payoff, from 80 to 300 Kronor. The point at which they took the chance determined their attitudes to risk.
Zethraeus found that the month-long regimen of hormones had absolutely no effect on the women’s decisions during these games (see graph below; T = testosterone; E = oestrogen; P = placebo). And when he looked at their results individually, he found that the extent to which their hormone levels went up had no bearing on their choices either. Based on these stark results, he safely concluded that “sex hormones have no significant impact on economic behaviour”.
That result may be surprising to some, especially in the light of earlier studies. According to some of these, testosterone-charged men are more likely to reject low, unfair offers in the Ultimatum game than those with lower levels, take bigger financial risks and make bigger profits. Women apparently take more risks during parts of their menstrual cycle when their oestrogen levels are highest.
But these results are mere correlations. This new trial is stronger, in that it actually manipulated hormone levels to see what would happen. It was randomised, which means that the decision about which tablet they received was made randomly. And it was “double-blind”, which means that neither the volunteers nor the researchers knew which tablets they were taking. That was only revealed at the end of the trial when all the results were in. These features are incredibly important. They remove the temptation and the ability of the researchers to skew or bias their results, even unconsciously.
Based on his negative results, Zethraeus believes that the links observed in other studies may reflect a third factor – something else that affects behaviour and varies with levels of sex hormones. Rather more harshly, he suggests that the previous results could be “spurious”. It’s possible that negative results, which didn’t find a link between behaviour and sex hormone levels, were simply never published, giving a rose-tinted and overly positive view of the evidence. This problem is called publication bias, and it means that some hypotheses wrongly appear to gain support, simply because negative results never see the light of day.
There are a few caveats to the current study. It only looked at women, although there’s little reason to believe that sex hormones would influence risk-taking, selflessness or trust in men alone. It only examined the effects of a short four-week course of hormone treatments, although the boosted levels of hormones after the four weeks can’t be ignored. Hormone treatments, which are commonly used to deal with menopausal symptoms and sexual disorders, produce changes of the size that Zethraeus saw. And finally, it only looked at changes in sex hormone levels later in life. Indeed, Zethraeus admits that his results don’t rule out the possibility that testosterone or oestrogen levels during childhood or puberty could affect decision-making behaviour later on.
Nonetheless, the study provides a solid blow to the idea that sex hormones affect our attitudes to trust or fairness, and it reminds us yet again to be cautious about relying too heavily on correlations.
Reference: Zethraeus, N., Kocoska-Maras, L., Ellingsen, T., von Schoultz, B., Hirschberg, A., & Johannesson, M. (2009). A randomized trial of the effect of estrogen and testosterone on economic behavior Proceedings of the National Academy of Sciences DOI: 10.1073/pnas.0812757106
More on hormones (although interesting in the light of this study): Testosterone-fuelled traders make higher profits
April 7th, 2009 at 9:17 am
Want to bet that this research will not get headlines in the major newspapers in Sweden or elswhere?
“Testosterone does NOT affect your risk taking!!”
Right or wrong, these null-results are never given anywhere near the attention that “positive” results do – especially when it comes to meale-female charactestics.
Thank you for highlighting this article.
April 7th, 2009 at 9:20 am
Which is *exactly* why I decided to write up this paper.
April 7th, 2009 at 9:47 am
Great choice of article and great analysis! It’s sad that null results so rarely get published. they’re just as important as the positive ones!
April 7th, 2009 at 10:14 am
If it’s not the current level of hormone, then there is an interesting question for more study.
Is it possibly the hormone level during development? Or does some genetic component affect both hormone level and behavioral style?
April 7th, 2009 at 11:05 am
This is a neat study. However, I fear that the very design of the study may lead to its misinterpretation.
We know that men & women have structural brain differences. By studying only women, the researchers eliminated the variable of sexual dimorphism in the brains of their subjects, important for clarity of results.
However, it’s likely that the general public will not appreciate this. People tend to think of “testosterone-charged men,” or women with hormonally induced mood swings; in other words, hormones at work in the brain of someone of a specific gender. I am concerned that the general press may report this as “Sex hormones have no effect on the brain.”
Just the $.02 of a humble clinical neurologist.
April 7th, 2009 at 12:53 pm
On the question of hormone levels in childhood and puberty, it could start even earlier than that–I know that the Autism Research Centre in Cambridge studies foetal testosterone, and they claim to have found correlations between that and behaviour in later life.
April 7th, 2009 at 3:27 pm
I’m so glad you did decide to write about this. “Nothing” is as important a result as “something.”
April 7th, 2009 at 3:31 pm
Re: Thomas K’s comment, I remember reading something about this as well. I believe a big part of the reason they started looking into this was the finding that autism is more common (or at least more often diagnosed) in boys than girls. I don’t remember which site had the info about their research, but I seem to recall a graph of their study statistics (the study was incomplete at the time, iirc) that was at that point decidedly inconclusive for their sample. Damn, I wish I could find that page, but a quick google doesn’t seem to give the right links.
April 7th, 2009 at 10:51 pm
On a digressory note, I wonder why women agreed to take testosterone at the age of about 60 years, or even estrogen for that matter. At that age, women are just recovering from the havoc created by fluctuating hormones and are probably feeling better after years of discomfort or worse.
April 8th, 2009 at 1:49 am
What thoe hoell is “oestrogen?”
April 8th, 2009 at 12:18 pm
It would be a sex steroid associated mainly with female sexual characteristics. I believe that Americans spell it “estrogen”, but the rest of the world thinks that they’re nuts. Who in their right mind would take spelling cues from a society that spells colour “color”. It’s just crazy.
April 8th, 2009 at 12:24 pm
Word. I’m keeping the ‘o’ in for the honour of the English language
April 8th, 2009 at 1:57 pm
I think it is possible there is still a small but cumulative effect that is reinforced into one’s personality. By the time you are 60, you have had almost a lifetime of reinforcing a certain level of fairness, trust, risk and altruism. I would think for the hormones to have a chance at countering this, its effect would have to be large enough that it should be immediately obvious in at least some cases (ie risk assessment changes as if going from sober to drunk within hours of ingestion)
April 8th, 2009 at 2:21 pm
Possibly – but note that these are biologically-relevant doses of hormones. As I said, similar regimens are used to treat conditions like erectile dysfunction and menopausal symptoms so they clearly produce physiological effects even in old age.
Also, you can compare the increase in hormone levels to the differences seen in other studies. The authors note that a previous piece of research suggested that doubling testosterone levels would increase risk-taking behaviour by 39%. Well in this study, testosterone levels were quadrupled to no effect.
April 8th, 2009 at 3:16 pm
Physiological and psychological effects are of course different. Unless there was a physiological component to the risk taking, such as how far they are willing to jump a gap in order to gain a reward, a physiological change isn’t necessarily going to lead to a psychological change.
I have not come to a conclusion either way. However, if we were to agree that the study shows what is claimed, then we would have to throw out the other correlation studies anyway. So, I suppose this at least you could say that this study refutes the correlation studies even if it does not show there is no difference at all over any time period.
April 8th, 2009 at 4:03 pm
Very interesting, thanks for the post!
What might be relevant is to thin about causality: it may also be that the more risky persons are more excited which in turn may cause them to have higher levels of steroids. This study thus could also indicate that high steroid levels and risk-taking behavior are simply correlated due to some underlying causality, but not affecting each other directly.
Causalities in endocrine systems are hard to examine and not always clear cut, although they sometimes may seem like they are.
April 8th, 2009 at 5:23 pm
“It’s sad that null results so rarely get published. ”
Hmmmmm. The Journal of Null Results.
That so needs to exist. Even as a blog. Anyone?
April 9th, 2009 at 9:58 pm
Rodney – it already does, Journal of Articles in Support of the Null Hypothesis: http://www.jasnh.com/
January 12th, 2010 at 3:16 am
This is an interesting study but I think if you believe that your level of testosterone and estrogen will affect your attitudes to fairness, trust, risk and altruism then it will because belief is a powerful agent.