This episode of You Are Not So Smart explores research in social psychology that suggests, well, maybe you are. Tom Stafford, a University of Sheffield psychologist, delves into his research showing people have more intelligence than they get credit for – it just takes the right context to make their smarts operational. Stafford explains to host David McRaney exactly what that context is, and how computer scientists are working to bring it to online interactions.
- Humans are bad at thinking individually.
- When people solve problems in groups, they outperform individuals thinking solo.
- Online platforms favor individual arguments and disfavor group problem-solving.
- Scientists are working to create online environments that facilitate social problem-solving.
Humans are bad at thinking individually.
Cognitive reflection testing shows most people are lazy, irrational thinkers who often leap to the obvious – but wrong – answer. The most famous cognitive reflection test question goes as follows: If a bat and ball together cost $1.10, and the price of the bat is a dollar more than the ball, what’s the price of the ball? Most people give 10 cents as the answer, and they’re wrong. (It’s five cents.) For the overall standard cognitive reflection test, only 17% of people answer all the questions correctly, and fully one-third of test-takers answer them all incorrectly.
In another “gold standard” test for human reasoning, the Wason Selection Task, people are shown four cards, showing the letters and numbers E, K, 2 and 7. The tester tells the subject each card has a number or letter printed on its reverse, and that the following rule might or might not be true: Cards with a vowel on one side will always have an even number on the reverse. To prove or disprove the rule, the subject can turn over two of the four cards. Which two should the person turn over? Only 10% of people correctly choose E and 7. Indeed, doctors, lawyers and scientists don’t exceed a 10% success rate.
When people solve problems in groups, they outperform individuals thinking solo.
New studies indicate people’s poor cognitive performance might depend heavily on whether they’re solving problems alone or in a group. When groups of three or more work on the Wason Selection Task together, they get the right answer 80% of the time – and in a full half of those cases, all the individuals in the group had gone into the group discussion believing a wrong answer.
“If you assumed human reasoning was a tool for individual cognition, it would seem really flawed. But if you look at it as a tool for group-based reasoning…it all makes sense.” (David McRaney)
Researchers at Cambridge University wondered if the success of groups would carry over to online collaboration. In a study, people solved the Wason Selection Task by working in groups of two to five, communicating anonymously on a text-based chat platform. One-third of these groups obtained the correct answer – more than tripling the rate of individuals working alone.
Online platforms favor individual arguments and disfavor group problem-solving.
Evolution has led to a division of cognitive labor in which individuals propose arguments – often biased and not well-thought-out – and then groups evaluate those arguments, discarding poorly reasoned arguments and combining well-reasoned ideas into a plan.
“Online discussion is not famous for generating consensus or being productive.” (Tom Stafford)
Unfortunately, although a large portion of discourse has moved to the internet, most online environments facilitate the production of individualistic arguments over group-based evaluation of those arguments. And certain features of online communications hamper productive group collaboration – such as anonymity and the ease of presenting information out of context.
Scientists are working to create online environments that facilitate social problem-solving.
The Cambridge study of online group collaboration on the Wason Selection Task yielded a data set that psychologists and computer scientists are studying to learn more about how group deliberation happens, and how online environments could facilitate social problem-solving.
“Like trust, good deliberation is slow to build up and easy to destroy.” (Tom Stafford)
That same data set could serve as a corpus to train “dialogue agents” – bots – to participate in online discussions and insert messages, such as asking questions or probing for people’s reasoning, that enhance deliberations.
About the Podcast
Cognitive scientist Tom Stafford studies learning and decision-making at the University of Sheffield. He is the author of Mind Hacks and For Argument’s Sake. The You Are Not So Smart podcast explores humans’ unwarranted confidence in their own reasoning, perception and motivation, and is inspired by host David McRaney’s best-selling book of the same name. McRaney is also the best-selling author of How Minds Change.