I AM not sure whether it is a badge of honour or a mark of shame, but a paper I published a few years ago is now ranked No. 8 on a list of studies that other psychologists would most like to see replicated. Good news: People find the research interesting. Bad news: They don't believe it.

The paper in question, written with my former student Margo Gardner, appeared in the journal Developmental Psychology in July 2005. It described a study in which we randomly assigned subjects to play a driving video game, either alone or with two same-age friends watching them. The mere presence of peers made teenagers take more risks and crash more often, but no such effect was observed among adults.

I find my colleagues' scepticism surprising. Most people recall that as teenagers, they did far more reckless things when with their friends than when alone. Data from the Federal Bureau of Investigation indicates that many more juvenile crimes than adult crimes are committed in groups. And driving statistics conclusively show that having same-age passengers in the car substantially increases the risk of a teenage driver crashing but has no similar impact when it comes to adults.

Then again, I am aware that our study challenged many psychologists' beliefs about peer pressure, for it showed that the influence of peers on adolescent risk-taking does not rely solely on explicit encouragement to behave recklessly. Our findings also undercut the popular idea that the higher rate of real-world risk-taking in adolescent peer groups is a result of reckless teenagers' being more likely to surround themselves with like-minded others.

My colleagues and I have replicated our original study of peer influences on adolescent risk-taking several times since 2005. We have also shown that teenagers take more chances when their peers are around partly because of the impact of peers on the adolescent brain's sensitivity to rewards.

In a study of people playing our driving game, my colleague Jason Chein and I found that when teens were with people their own age, their brains' reward centres became hyperactivated, which made them more easily aroused by the prospect of a potentially pleasurable experience. This, in turn, inclined teenagers to pay more attention to the possible benefits of a risky choice than to the likely costs, and to make risky decisions rather than play it safe. Peers had no such effect on adults' reward centres, though.

In other studies, we have shown that being around peers not only made adolescents more reward-sensitive but also drew them to immediate, rather than longer-term, rewards. Using an experimental set-up in which individuals were asked to choose between smaller immediate rewards (US$200 now) and larger, delayed ones (US$1,000 in six months), we found that university students were significantly more likely to pick the immediate reward when their decision-making was being observed by people their own age.

On average, as we mature through adolescence, we become more willing to delay gratification in order to obtain a bigger prize. Indeed, when university students are alone, their ability to delay gratification resembles that of people in their late 20s. But when they are being watched by their peers, they display the myopia of 14-year-olds.

Our studies have important implications for psychologists who study risky decision-making, as most other research in this area has tested individuals when they were by themselves. Many such studies have found no differences between teenagers and adults, but this may be a spurious result of testing people when they were alone, rather than when they were with others, which is frequently the context in which risky choices are made.

Perhaps the most intriguing of our studies of peer influences on adolescent behaviour is one that we published earlier this year in Developmental Science.

In this paper, we replicated our earlier studies but, this time, using mice rather than humans. We created "peer groups" of mice by raising them in triads composed of animals from three different litters. We then tested whether, if given unfettered access to alcohol, they would drink more when they were with their peers than when they were alone. Mice tested when they were fully grown drank equally in both contexts. But adolescent mice - tested shortly after puberty - drank significantly more in the presence of their peers than when they were by themselves.

The propensity for teenagers to do more risky things when they are with their peers - which understandably worries their parents and which should concern those who supervise teenagers in groups - is not only real, but it may also be hard-wired.

NEW YORK TIMES

The writer is a professor of psychology at Temple University.