Reputation matters

We are going to give you a pot of money. You can give as much (or as little) as you like to us. Then we’re going to triple the amount you gave us and give it to someone else. They can give as much (or as little) from the pot back to us, and this time we’ll triple the amount and give it to you. What do you do?

This is the Trust Game, developed by accounting professor Joyce Berg. Giving away money advertises that you are a trustworthy person, but you take the risk that the other player in this game simply takes the money and runs. Because of this risk, classic economic theory predicts that you would keep the money, particularly if you are player two. So what actually happened?

Berg and others have found only about 11 per cent of player ones give no money. In fact, most player ones give more than half of what they have, and most player twos are even more generous. It seems more important for players to advertise they are trustworthy than to keep the swag. Why?

Craving a good reputation

Adam Smith, the Scottish enlightenment figure known as the father of Economics, claimed that nature has endowed us with a craving for a good reputation. He believed that humans value good reputation, even more than money. In the game, we want to advertise that we are generous. Some experiments use multiple players, mixing up the pairs each time. In this version, players learn who will be generous, and who won’t be. 

Of course, as two neuroscientists, we want to know what is going on in the brain during this game, in those little spaces in the interaction. Studies scanning the brains of players while they play have found two contrasting areas showing the most activity: the caudate nucleus, and the amygdala (Rilling & Sanfey, 2011). They are both part of the brain’s basic learning mechanism. After every iteration of the game, if your opposite number is more generous than you expected, the caudate nucleus activates. In other words, you get a signal telling you this is a good and trustworthy person. They earn a good reputation. Meanwhile, the amygdala activates when you encounter something you should avoid – such as a person who continually gives no money. They earn a bad reputation. 

Economists may be surprised that most people are instinctively nice. Psychologists aren’t. Many people are generous and altruistic, even in situations where they wouldn’t be punished if they were selfish. Learning to win games like this can reveal how our brains are building up, and testing each other’s, reputations. Keeping the money for yourself is not always the best policy. If you have a good reputation, you can get even more money. 

Separating truth from noise

But why should you believe what we are telling you about reputation? Many aspects of science have started to lose their own reputation in recent years, including brain imaging. Craig Bennett’s 2009 study on ‘neural correlates of interspecies perspective taking in the post-mortem Atlantic salmon’, comes to mind. Research by Deena Weisberg and others on ‘the seductive allure of neuroscience explanations’ suggests we believe studies more if they include pretty pictures of the brain, even if the image has nothing to do with the content of the paper (although see also Martha Farah and Cayce Hook’s 2013 paper on the ‘seductive allure of the seductive allure…). We often see articles showing a brain with a nice red splodge on it, indicating the location of some bit of thinking, but no live brain scan will show such an image. So, should we be wary? How do we separate the truth from the noise in our data?

Firstly, imaging one brain is not enough. You have to image several and overlay them, and this is a problem because brains are like people – all different shapes and sizes. To overcome this, all the images are distorted until they fit a standard brain (belonging to someone called Colin, from Montreal, since you ask). You slightly blur the image, so that if they don’t quite fit they will still overlap.

Magnetic Resonance Imaging, the technique we’re talking about, is very good for finding out where things are happening in space. In terms of time – when the activity is happening – it’s much better to use EEG. But again, one brain is not enough. 

Consider Benjamin Libet’s famous experiment, showing that the researcher is able to ‘see’ when a person is about to move their fingers… a wave of activity up to 600 milliseconds before someone reports having made the decision to do that. Libet didn’t actually combine the data from his five subjects, he presented them separately. He did, however, combine over time, because subjects had to lift their finger many times before you could see the signal. The line that showed a consistent, repeated change across all repeats must, surely, show the brainwaves that relate to the task at hand. It’s a little like trying to ignore a clutter of images in one’s mind to focus on just one thing. 

That one thing, one line on the EEG graph, is now called the signal – and this is where we start to see the truth. And Libet’s study is robust, it has been replicated many times with different paradigms (see Frith & Haggard, 2018). But what does it say about free will? Does it show that we could look into your brain, and predict when you’re about to turn the page? No, because averaging across trials means that although there is a robust signal, you cannot see it on a single trial so we cannot know your next move before you do.

So, by averaging across space and time we can extract our signal from the noise in the data, but is this enough to believe these pretty pictures of the brain in action? You must always be suspicious. It’s rather like tossing a coin. How many heads in a row before you decide this coin is special because it’s biased? Five heads in a row, 1 in 32 chance, that becomes suspicious? Our rule of thumb in science is that we get suspicious if we see activity that only has a 5 per cent chance. The only trouble is, when you are doing brain imaging, you are getting data from thousands of different brain areas or voxels, not just the ones we are interested in. That means you are going to reach this level of significance many times by chance. This is why you have to correct for multiple comparisons, and if you read a paper which doesn’t do that you should be particularly suspicious.

Tricks and guarding against them

All too often, this suspicion is lacking. Experimenters can blind themselves and others with data, especially if they’re looking for a very particular result. For example, they could just keep running the experiment on more subjects until enough of them conform to the hypothesis. Or they could use a variety of different statistical methods to analyse their data, until one of them produces an interesting result – so-called ‘data dredging’. Solutions to these problems have emerged, such as pre-registration of studies. 

More difficult is helping people to see that when data disproves a hypothesis, it can be just as valuable, if less fun. It’s up to editors to be bold enough to print those papers, again, even knowing that finding nothing tends to be a less dramatic story to read than finding something (see John et al., 2012). It’s also incumbent upon scientists to accept that science is slower than we imagine. It’s always tempting to give ‘the answer’ – it seems kind to give practical solutions, to improve lives – but often we simply don’t know. That doesn’t mean that we will never know, we just need more studies. It can be useful for scientists to say, for example, ‘this is a big problem’, rather than falling into the trap of seeking to provide definitive answers. 

Here’s an example of a study that has failed to replicate, contributing to Psychology’s reputation problem. In 1996, John Bargh and colleagues found that, if they primed a room full of students to think about old age, those students would then walk measurably slowly once they got up to leave the lecture hall. Like a lot of Psychology findings, it sounds so plausible that you may be tempted to shrug and say ‘sure, of course’. It turned out it couldn’t be replicated. Is the original study wrong, or the replication? Could it even be that society had changed in the interim, away from the stereotype of older people moving more slowly? We think this is unlikely. 

In 2012, a group in Belgium led by Axel Cleeremans added a twist to the priming story. They primed the experimenters before asking them to carry out the test. One set were told that the original 1996 study was definitely true. Another set were told the opposite. Both sets found exactly what they were primed to find. The irony, of high-level priming causing experimenter effects in a study on priming! This alludes to one of the main problems in Psychology: we have not taken enough account of what the subject actually thinks the experimenter is trying to do, and whether they want to help them or not. When we talk about top-down control, the top is not the frontal cortex but the experimenter.

Even if you get that strong signal, the ‘right’ result, it all appears true, we can still run into problems – such as the well-known fact that correlation is not causation. A nice example is from Mann and Labrosse, back in 1959. Patients diagnosed as schizophrenic all excreted high levels of substances called phenolic acids. Could this point the way to some brain or hormone-based cause? No, the differences were tracked to another factor – the patients drank more coffee than the controls.

If we know about these tricks that scientists can consciously or unconsciously use to mislead themselves about their data, we can guard against them. That’s a silver lining of the ‘replication crisis’: you should have become more suspicious about what you read, in scientific journals and in the wider world of the media. That’s a good thing: we must always be suspicious.

The audience effect

The science that we’re drawing on here is under the shadow of a dramatic loss of reputation, and that matters. It matters how others see us. On the title page of our book, the two of us are seen dancing. We’re only happy to be seen dancing in a drawing: it makes us wince to think that someone might actually watch us dancing (a privilege reserved for our grandchildren). The audience effect is a sign that reputation matters. 

If nature has endowed us with a craving for reputation, as Adam Smith said, do we even need to teach it to children, or do they spontaneously get it? Remarkably, children as young as 10 months can be seen adjusting their behaviour (for example, smiling differently: Collins & Hong, 1991) when they know there’s an audience watching. But, perhaps this has more to do with early attempts at communication. There are certainly marked changes as children grow up.

As they grow up, they learn to value different audiences differently, but there are not actually many studies of this. In one, ‘theory of mind and children’s desire to perform’ (Chaplin & Norton, 2014), children were given the choice of two activities: singing and dancing, or sitting quietly and doing some colouring. About half of the 3-4 year-old children chose to sing and dance, but none of the 11-12 year olds made that choice. They were afraid to be judged by some audience. They cared about their reputation. It seems this is quite a spontaneous change in their development, it does not have to be taught.

How does the audience effect work when we can’t see our audience, for example when we use the internet? We actually published a study ten years ago with Claudio Tennie in 2010, on reputation management in the age of the world-wide web. It seems we don’t know much more than we knew then. Subjective experience has shown us that Twitter – as with most forms of social media – has a weirdly addictive quality, and it is based on reputation management. It has both rewarding and punishing effects on us, but punishment does not seem to diminish our dependency.

The power of gossip

Of course, Twitter and other forms of social media can be a time-saver and powerful source of information, but this information is only as reliable as the people posting it. This makes social media very similar to old-fashioned gossip. Gossip has always been a key part of human interaction – something we constantly indulge in, and that we might also be a bit addicted to. We often feel guilty about this and hear the advice that we should ignore gossip, but that would be a mistake. Instead, a case can be made in praise of gossip (Spacks, 1982). We heard earlier that we can learn through our direct experience whether to trust a stranger, and therefore choose him as a co-operator, and we don’t even have to think about it. We can do this through our evolved ability to learn through reward and punishment, depending on the accuracy of our assessment of others’ trustworthiness. But we only rarely have time to learn by direct observation. Gossip offers a shortcut. It is more efficient to learn through other people: they have already filtered the information you need. Gossip tends to come from several sources (Sommerfield et al., 2008), and that’s more reliable than information from a single source (even if it’s yourself). You can learn from other people’s mistakes. 

At least we should respect gossip, and in the lab it has been taken quite seriously. We’re even learning, from behaviour and brain imaging studies, that gossip is a more powerful factor than our own observations…

A deep influence

We come back to the Trust Game. Remember that the caudate nucleus was automatically tracking guesses about the trustworthiness of your partner, and it responded particularly when you were pleasantly surprised. A 2005 experiment tried a variation of the game, adding an element of gossip (Delgado et al., 2005). They told participants about the moral character of their partner beforehand: for example, either they were someone volunteered at a food bank and had once saved a child from drowning, or they had once been accused of plagiarism and read comics in science classes. People gave more money to partners who had been presented as more trustworthy.

What was more interesting was that the brain itself behaved differently. The caudate nucleus was practically asleep – it seems people were relying on the gossip, rather than on building up their own picture during the game.

In that study, though, information was supplied by the experimenter. What about more real, everyday gossip? Would you pay attention to subjectively gathered information, or go against it if your own observation went against it? One study [Sommerfield et al., 2008)] involved multiple groups of people, who started by watching each other playing a trust game. After watching several rounds, they gossiped with each other about the way the players behaved during the game. The experimenters then observed the group as a whole playing rounds of the trust game with each other. They took careful note of how each person made decisions, based on what the gossip had said about their opponents (these were actually statements that had been written down by the observers). It turned out they were deeply influenced by the gossip. They made quick and firm decisions in rounds played with people who had been the subject of ‘positive’ gossip, and they did not override these when the actual behaviour of the partners contradicted it.

This finding may not strike you as surprising, and this is of course a problem with psychological science: when it confirms what we already know from this great experiment called ‘life’, we forget all the other things that we know science has not confirmed. 

To sum up, gossip should be respected, at least to protect us from trusting an untrustworthy person. It can quickly allow us to accept a trustworthy person as a partner. In this way, gossip is an aid to our human ability to cooperate. We may, of course, all have to abandon that wonderful convenience of pooling information from many other people because, in the age of the internet, we have far more uncertainty and reason to worry about gossip. Anyone can make up anonymous accounts, give higher ratings for goods and services to themselves and their friends, and spread misinformation. Anonymity should be taken as a danger signal.

In our graphic biography we abandon anonymity and reveal some of the more personal side of ourselves. We hope you will take it as a signal to trust us. We have laid our reputation on the line by appearing so prominently in this book. We often worried about this, but our son, who actually produced the book and wrote every line of text, assured us that we would often be portrayed in a ridiculous light, because after all, you would expect ridiculous figures in a comic. 

There is another reason that we agreed to act as leads. We hope to demonstrate that scientists are not special, but ordinary people. Equally, we hope that revealing a more personal side to ourselves will be taken as a signal to trust us as scientists. 

A fine line

As ordinary people and as scientists we have personally experienced the ‘craving for a good reputation’ that Adam Smith identified long ago. The question that haunts us is whether the craving matches reality. This question is not for us to answer, but for the reader. 

The reader, we hope, will remember that we think a genuinely good reputation is about doing good things, that are noticed and talked about. And in our case, this is not just a matter of doing good science, but also about encouraging our colleagues and, especially, encouraging and helping our students. It’s also about being consistently good. Perhaps only saints can achieve this. 

But there is a final twist. Those who are doing good deeds must be careful not to advertise their goodness too blatantly. This is to guard against the anger of those who are not doing good deeds. They don’t like to be shamed, and that’s why good deeds are sometimes punished (Raihani & Power, 2021). 

We are always treading that fine line. Reputation management – in personal and professional lives – is hard, because we are not entirely in control. But we are all dependent on each other, and we can do better together.

Box text: Non-human reputations 

Gossip is just the high-end level of a universal truth: that most of what we know about the world, we learn from others. This applies to fruit flies as well as people: they learn from watching other fruit flies. 

In the book, we talk about meerkats; and about cleaner wrasse.

These fish guard their reputation very carefully. They bite the fish they are cleaning less often when they are being watched by potential ‘clients’ (see Nichola Raihani’s book The Social Instinct). All the audience effects are happening. The difference with us is that they have hundreds if not thousands of these interactions every day, which is probably sufficient for basic learning processes to achieve that reputation, but we can sometimes do it much more quickly. The basic evolutionary drive is the same, just with a different mechanism for achieving the same result.

- Dame Uta Frith is Emeritus Professor of Cognitive Development at University College London.

- Chris Frith is Emeritus Professor of Neuropsychology at University College London.

Their book, Two Heads: Where Two Neuroscientists Explore How Our Brains Work, with Alex Frith and illustrator Daniel Locke, is out now with Bloomsbury Publishing. For your chance to win a signed copy, see our Twitter feed.

Key sources

Bargh, J.A., Chen, M. & Burrows, L. (1996). Automaticity of social behavior. Journal of Personality and Social Psychology, 71, 230-44.

Bennett, C.M., Baird, A.A., Miller, M.B., & Wolford, G.L. (2009). Neural correlates of interspecies perspective taking in the post-mortem Atlantic salmon: an argument for proper multiple comparisons correction. J Serendipitous Unexpected Results, 1, 1-5.

Berg, J., Dickhaut, J. & McCabe, K. (1995). Trust, reciprocity, and social history. Games and Economic Behaviour, 10, 122-42.

Chaplin, L.N. & Norton, M.J. (2014). Why we think we can’t dance: Theory of mind and children’s desire to perform. Child Development, 86, 651-58.

Collins, K. & Hong, H. (1991). An audience effect on smile production in 10-month-old infants. Psychological Science, 2, 45-9.

Delgado, M.R., Frank, R.H. & Phelps, E.A. (2005). Perceptions of moral character modulate the neural systems of reward during the trust game. Nature Neuroscience, 8, 1611-18.

Doyen, S., Klein, O., Pichon, C. L., & Cleeremans, A. (2012). Behavioral Priming It’s All in the Mind, but Whose Mind PLoS ONE, 7, e29081

Farah, M.J. & Hook, C.J. (2013). The seductive allure of ‘seductive allure’. Perspectives on Psychological Science, 8, 88-90.

Frith, C.D. & Haggard, P. (2018). Volition and the brain: Revisiting a classic experimental study. Trends in Neurosciences, 41, 405-7.

John, L.K., Loewenstein, G. & Prelec, D. (2012). Measuring the prevalence of questionable research practices with incentives for truth telling. Psychological Science, 23(5), 524-32.

Mann, J.D. & Labrosse, E.H. (1959). Urinary excretion of phenolic acids by normal and schizophrenic male patients. AMA Archives of General Psychiatry, 1, 547-51.

Raihani, N. (2021). The social instinct. How cooperation shaped the world. London: Random House.

Raihani, N.J. & Power, E.A. (2021). No good deed goes unpunished: the social costs of prosocial behaviour. Evolutionary Human Sciences 3, e40.

Rilling, J.K. & Sanfey, A.G. (2010). The Neuroscience of Social Decision-Making. Annual Review of Psychology, 62, 23-48.

Sommerfield, R.D., Krambeck, H.J. & Millinski, M. (2008). Multiple gossip statements and their effect on reputation and trustworthiness. Proc Biol Sci, 275, 2529-36.

Spacks, P.M. (1982). In praise of gossip. Hudson Review, 35, 19-38.

Tennie, C., Frith, U. & Frith, C.D. (2010). Reputation management in the age of the world-wide web. Trends in Cognitive Science, 14, 482-88.