Here's the best way to tell if someone's lying, according to science

It's not always this easy to spot a lie.
Humans are notoriously poor lie detectors. In scientific experiments, our accuracy levels are only slightly greater than chance — and sometimes lower.
The problem is that most of these experiments focus on people's ability to spot untruths on their own. What would happen if people pooled their cognitive resources and worked together to suss out dishonesty?
That's the premise of new research led by Nadav Klein, a doctoral student at the University of Chicago, and Nicholas Epley, Ph.D., a professor there. The study found that groups are significantly better at detecting white and high-stakes lies than individuals are.
The researchers conducted a series of experiments designed to test how well people could predict lies working individually and in groups.
In one experiment, participants watched video recordings of people who had been told to either tell the truth or fib about things like the best vacation they'd ever taken. Some participants reported individually whether they thought the people in the video were lying, while others worked in groups of three to come to a consensus.
In another experiment, participants watched video clips of "Golden Balls," a British game show in which people try to convince their partners that they're willing to split a sum of money — even though they might be trying to steal all the money for themselves. Lies in these cases were considered "high stakes" because they involved significant financial risk.
Results showed that, in the white-lies experiment, groups were about 8% better at detecting lies than individuals. In the high-stakes-lies experiment, groups were about 4% more accurate.
The key to groups' superior performance seems to be the discussion that goes on before they reach their verdict. In a subsequent yet unpublished study, in which groups didn't have to come to a consensus, Klein found that it didn't even matter if everyone in the group agreed— as long as they had a chance to discuss their opinions, their individual accuracy increased.
Previous research suggests that crowds of people can produce more precise judgments than individuals simply because they'll come up with a relatively accurate average. In one famous experiment, for example, people tried to guess the weight of an ox and their average guess was only a pound off.
Yet this strategy probably works better for estimating numbers than for making social judgments, as participants were asked to do in this experiment.
Groups of three were able to discuss their opinions — and that's probably why they were better lie detectors.
This research has some important practical implications.
For one, Klein says, "All the money that's spent on training individuals to be better lie detectors might be better spent just getting untrained individuals to do it in groups. It's a potentially inexpensive way of increasing lie detection."
There's also the opportunity for people to apply these findings in their professional lives. When a manager is trying to figure out if an employee is lying, the best bet could be to bring in a coworker or two and discuss the evidence. The same goes for insurance agents who are trying to detect fraud — it helps to have another person around.
These findings are relatively straightforward, but potentially revolutionary. If we could give up the idea of trying to become human polygraphs and accept that we need some outside help, we could drastically improve the accuracy of our social judgments.