A party game, a lab task
“Two truths and a lie” is known as an icebreaker: everyone says three things about themselves and the others try to spot the lie. In this experiment we don’t talk about ourselves but about the world. Each round has three surprising facts; two are true and backed by a reliable source, and one is a plausible lie we wrote.
Over eight rounds you pick the lie and say how sure you are. Someone guessing at random catches 2.7 lies in 8 rounds on average. On the results screen you can see your score, the crowd’s average and which true fact was most often mistaken for a lie.
Why are we bad at catching lies?
Charles Bond and Bella DePaulo (2006) combined hundreds of lie-detection studies in a single meta-analysis: 206 documents and 24,483 judges in total. When people watched someone speak and decided whether they were lying, average accuracy was 54%; tossing a coin would give 50%.
Even more interesting was the direction of the errors. People recognised 61% of truth-tellers as honest but caught only 47% of liars. In other words, judges mostly said “they’re telling the truth”, even when they should have been suspicious. This tendency is called truth bias.
The truth default
Timothy Levine (2014) interpreted truth bias not as an error but as a default. According to truth-default theory, people automatically believe what they hear until they see a reason for suspicion. Because most everyday communication is honest, this default usually leads us to the right conclusion and makes communication possible.
According to Levine, lies are mostly caught not through behavioural cues (like avoiding eye contact or fidgeting) but through content and context: what is said contradicting what we know, or evidence that turns up later. That is all you can do in this experiment too: compare the fact with what you know.
Surprising truths, familiar lies
There is another trap in this game of facts: the world is often stranger than we expect. Wombats leaving cube-shaped droppings, or Cleopatra living closer to the Moon landing than to the pyramids, sound made up. The most surprising truths may be the ones most often taken for lies.
A good lie, on the other hand, feels familiar. Hasher, Goldstein and Toppino (1977) showed that people rate statements they have heard before as truer, regardless of whether they are true; this is called the illusory truth effect. That is why much-repeated myths, like the goldfish’s few-second memory or Napoleon being very short, are among our lies.
Being sure isn’t being right
In each round we also ask how sure you are. The meta-analysis by DePaulo and colleagues (1997) found that the relation between confidence and accuracy in lie detection is close to zero (an average r of .04 across 18 samples). Confidence was related more to a person’s tendency to say “they’re telling the truth”.
On the results screen we compare how accurate the crowd was in rounds where they said “very sure” with rounds where they said “not sure”. In a trivia game general knowledge helps, so a link between confidence and accuracy here wouldn’t be surprising. What we are curious about is how weak or strong that link is.