This post is inspired my Mark of ZenPundit‘s latest post, “On Howard Gardner and Creativity.”
Imagine you have a deck of cards with two sides: a blue side and a green side.
In the first game, called the Watson Selection Task, every card has a number on one side and a letter on the second side. Each represents a logical pair. We draw four cards, two are laying blue-side-up and two are laying green-side-up. We get “16,” “61,” “P,” and “E”
I tell you “Turn over the minimum number of cards needed to test the hypothesis, ‘Every card with an odd number on the blue side has a vowel on the green side.” Which cards do you turn over?”
Now, a second round, called the Cheater Detection Task. This deck also has a blue side and a green side, but here the blue side is an age and the green side is an action. Each represents a person in a bar-and-grill. We draw four cards, two are laying blue-side-up and two are laying green-side up. We get “16 years old,” “61 years old,” “Drinking Pop,” and “Drinking Beer.
I tell you “Turn over the minimum number of cards needed to test the hypothesis, ‘People 20 years old and younger must drink pop.’ Which cards do you turn over?”
The result of the test are generalizable. Questions which require abstract thought, such as the first, take longer to complete and are more error-prone than questions which require finding cheaters, such as the second. This is true even though the questions are logically equivalent. Both make you test a proposition in the term “If p then q.” Yet the first relies on your logical thinking ability, so is difficult. (As human logical thinking is very weak.) The second relies on your cheater-detection ability, so is easy. (As human cheater detection is very strong.) The l
One objection to this is everyone is familiar with the correct answer for the drinking age, so it isn’t fair. Yet the results are about the same even when novel social restrictions are used instead of the drinking age game.
Another objection (made also by Buller, who noted the same finding) is that humans are much better at logical problems that involve obligation than logical problems that are abstract. Yet this “objection” actually proves the point, as it shows that humans do not have a strong logical-computation ability, but have a strong ability for detecting cheaters and a weak ability for solving that which is called logic.