Falsification: How to Destroy your Best Ideas

“The human mind is a lot like the human egg,
and the human egg has a shut-off device.
When one sperm gets in, it shuts down so the next one can’t get in.”

— Charlie Munger


Sir Karl Popper wrote that the nature of scientific thought is that we could never be sure of anything. The only way to test the validity of any theory was to prove it wrong, a process he labeled falsification. And it turns out we're quite bad at falsification.

When it comes to testing a theory we don't instinctively try to find evidence we're wrong. It's much easier and more mentally satisfying to find information that proves our intuition. This is known as the confirmation bias.

In Paul Tough's book, How Children Succeed: Grit, Curiosity, and the Hidden Power of Character, he tells the story of an English psychologist Peter Cathcart Wason, who came up with an “ingenious experiment to demonstrate our natural tendency to confirm rather than disprove our own ideas.”

Subjects were told that they would be given a series of three numbers that followed a certain rule known only to the experimenter. Their assignment was to figure out what the rule was, which they could do by offering the experimenter other strings of three numbers and asking him whether or not these new strings met the rule.

The string of numbers the subjects were given was quite simple:


Try it: What’s your first instinct about the rule governing these numbers? And what’s another string you might test with the experimenter in order to find out if your guess is right? If you’re like most people, your first instinct is that the rule is “ascending even numbers” or “numbers increasing by two.” And so you guess something like:


And the experimenter says, “Yes! That string of numbers also meets the rule.” And your confidence rises. To confirm your brilliance, you test one more possibility, just as due diligence, something like:


“Yes!” says the experimenter. Another surge of dopamine. And you proudly make your guess: “The rule is: even numbers, ascending in twos.” “No!” says the experimenter. It turns out that the rule is “any ascending numbers.” So 8-10-12 does fit the rule, it’s true, but so does 1-2-3. Or 4-23-512. The only way to win the game is to guess strings of numbers that would prove your beloved hypothesis wrong—and that is something each of us is constitutionally driven to avoid.

In the study, only 1 in five people was able to guess the correct rule.

And the reason we’re all so bad at games like this is the tendency toward confirmation bias: It feels much better to find evidence that confirms what you believe to be true than to find evidence that falsifies what you believe to be true. Why go out in search of disappointment?

There is also a video explaining Wason's work.