Everybody is familiar with the sense of shock and betrayal at having been lied to. At the same time, people are familiar with the temptation to lie to benefit themselves. Many will have done so.
And that raises an interesting question—given the chance to lie for their own benefit, which people will take the opportunity? What percentage always tell the truth regardless of how much is at stake? And what percentage always lie to maximize their gain?
Psychologists also want to know whether these behaviors are intuitive. Are we hard-wired to lie or tell the truth? The answer has important implications for our understanding of human nature and our efforts to encourage or discourage certain behaviors in society.
Today we get some insight into these questions thanks to the work of Hélène Barcelo at the Mathematical Sciences Research Institute in Berkeley and Valerio Capraro at the Middlesex University Business School in London. They have devised a clever way to test our inherent veracity and say their data suggest that humans fall into three categories: the good, the bad, and the angry.
At the simplest level, deciding whether to lie is a binary problem—either we tell the truth or we don’t. Psychologists have devised numerous experiments that explore this scenario.
But in the real world, the decision is usually more complex. It involves a calculation to determine the benefit we can gain from lying but also the punishment it might engender, and whether the potential benefit outweighs the potential loss.
Psychologists have devised ways to test this, too, by offering people lots of different ways to lie so that they must calculate which one benefits them the most.
But no experiment tests all these factors at the same time. Until now. Barcelo and Capraro have devised an ingenious experiment that tempts people to lie and also gives them the opportunity lie in different ways that change the benefit to them.
The experiment is an online test for workers on Amazon’s Mechanical Turk crowdsourcing website, available to many different people. “By looking at the distribution of choices, we can divide people in types according to the strategy they implement,” say Barcelo and Capraro.
Participants are told that they will be shown a randomly generated list of numbers between 1 and 90. At the same time, they are also shown a single number that indicates a position on the list.
The task for the Turkers is to work out which number appears at this position. Their pay will be the value of this number in cents.
Unknown to the Turkers, all are shown the same list of numbers. The question that Barcelo and Capraro investigate is whether participants identify the correct number or choose one of the others on the list.
The list of numbers is as follows: 25, 3, 63, 54, 28, 70, 37, 36, 26, 31, 43, 15, 30, 60, 33, 37, 15, 63, 16, 50, 4, 71, 79, 2, 85, 48.
Half the Turkers are asked to identify the number in position 19 and the other half to identify the number in position 22. The respective answers are 16 and 71.
These answers represent very different outcomes. The unlucky Turkers asked to identify 16 know they will receive just 16 cents but can easily improve the outcome by choosing the number on either side: 63 or 50. That’s a seemingly easy mistake to make and could be interpreted as stretching the truth.
Indeed, Turkers can improve their payout even further by choosing other numbers. The quickest option is to choose one of the first high numbers on the list, such as 63 in position 3 or 70 in position 6. Or they could choose the highest payout of 85 cents at position 25.
By contrast, the Turkers asked to identify 71 are much luckier. Their payout is close to the maximum of 85 cents, so they have less incentive to lie.
Barcelo and Capraro asked some 800 Turkers to complete the experiment. These people had an average age of 37; 52 percent were male, 48 percent female. In half the cases, the researchers put a time limit on the decisions to study the effect of time pressure.
The results make for interesting reading. It turns out that most people are honest—some 84 percent of them. But a higher percentage of people in the unlucky group lie than in the lucky group.
This suggests that people calculate the outcome and lie only if the truthful payoff is bad for them. In other words, they are conditionally honest.
By contrast, Some 50 percent of people tell the truth regardless of how it affects them: they are unconditionally honest.
Curiously, nobody stretches the truth. Not a single unlucky Turker chose the positions next to the unlucky one.
Instead, liars all maximized their payoff by choosing higher numbers elsewhere in the list. These liars divide into conditional and unconditional liars. Some always lie, and some decide to lie only after assessing the outcome for themselves.
A smallish percentage of unconditional liars choose the maximum payoff regardless of the truthful outcome. The data don’t allow Barcelo and Capraro to work out the exact proportion, only that it is more than 0 percent and less than 16 percent of people.
None of the Turkers rely on other strategies, such as choosing randomly.
The researchers summarize their results by pointing out that all Turkers fall into three categories. In the first are people who always tell the truth regardless of whether they are lucky or unlucky in their payoff. These are the good people and make up about 50 percent of us.
In the second are conditional liars who assess the payoff and lie if they are upset with the outcome. These are angry people and make up perhaps 35 percent.
In the final group are bad people. These are those who always lie to get the biggest payout for themselves. This group is made up of no more than about 15 percent of the total.
Interestingly, the researchers say that women are consistently more honest than men, a result that is in line with many previous studies. “Men lie more than women in self-serving situations,” say Barcelo and Capraro.
The experiments show that being dishonest generally takes more time, which could suggest that this is a reflective process rather than an intuitive one. However, the experimental setup plays an important role in determining this kind of response, and further work is needed here.
The heartwarming result is that so many of us are unconditionally honest—at least in this experiment. That suggests that in many situations we ought to be able to rely on people’s truthfulness. But it’s also news that dishonest actors can exploit for their own benefit. Such is the nature of humanity!
Ref: arxiv.org/abs/1712.10229 : The Good, the Bad, and the Angry: An Experimental Study on the Heterogeneity of People’s (Dis)Honest Behavior
What’s next for AI regulation in 2024?
The coming year is going to see the first sweeping AI laws enter into force, with global efforts to hold tech companies accountable.
Meet the economist who wants the field to account for nature
Gretchen Daily is working to make the environment more of an element in economic decision-making.
Three technology trends shaping 2024’s elections
The biggest story of this year will be elections in the US and all around the globe
Four lessons from 2023 that tell us where AI regulation is going
What we should expect in the coming 12 months in AI policy
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.