Humans are notorious for overestimating themselves. We think of ourselves as more beautiful, more popular, and better at estimating risk than we really are. Oh, and we probably overestimate our capacity for logic, too.

Brian Gallagher at Nautilus takes on the research of the late psychologist Peter Watson in a new article, exploring the “Watson selection task,” a famous and oft-repeated experimental method that tests subjects’ logical reasoning processes.

In the test, Watson showed a volunteer four cards, two showing numbers, and two colored cards without numbers. He then asked them how they would go about, in the least number of steps, proving whether or not even numbered cards always have a blue face on the back. Though he thought the test “deceptively easy,” as he wrote in a paper on it, 90 percent of his subjects got it wrong. Subjects even admitted that if they had to do it over, they’d probably still choose wrong—highlighting the irrationality of humans.

Years later, psychologist Daniel Kahneman hypothesized that the difficulty of the task has to do with a battle between two cognitive systems triggered by the word choices of the question. One system tends to take mental shortcuts, because it’s faster and easier, while abstract reasoning, the second system, is harder.

We won't spoil just how those mental shortcuts function in this case. Play the game through the YouTube video below, then head over to Nautilus for a deeper explanation of why you probably got it wrong.

[h/t Nautilus]