Thursday, January 7, 2010

The Blunders of Belief: Part I - Confirmation Bias

This is the first of a series of posts that will examine beliefs from a psychological perspective. There are numerous studies demonstrating that (1) belief is a fundamental part of human experience (even for notorious skeptics) and (2) humans are hard-wired to cling to their beliefs regardless of their inherent value. In this series, I will explore research that reveals the feebleness of the human mind, especially with regards to the cognitive blunders associated with belief. 

Part I: Confirmation Bias

Consider the following sequence of numbers: 2, 4, 6.

Now, take a moment before reading on and try to think of a pattern that governs the sequence.

. . .

One possibility that immediately presents itself is 'consecutive even numbers.'

Now, suppose that you were able to test this hypothesis, by proposing another sequence of three numbers, and learning whether or not your sequence fits the pattern.

Go ahead. What three numbers would you choose to test the hypothesis?

. . .

If you are like most people, you probably came up with something like 8, 10, 12.
Good job, this is absolutely correct! This fits the pattern perfectly. Now, test the hypothesis once more and propose another three numbers.

. . .

If you thought of 14, 16, 18, or any other consecutive sequence of even numbers, you'd be correct once again.

By this point, you are probably quite confident that you have solved the puzzle. After all, the data is perfectly consistent with your hypothesis!

But in fact, 'consecutive even numbers' is not the pattern I had in mind. The pattern I was thinking of was, 'any sequence of ascending numbers' like 1, 2, 3; 12, 52, 54; or 7, 231, 4378.

But don't despair; nearly everyone makes the same mistake. In a study conducted 50 years ago (Wason, 1960), the vast majority of participants approached the puzzle exactly the same way - by selectively searching for affirming evidence. That is, if their hypothesis was 'consecutive even numbers,' the participants only proposed sequences of consecutive even numbers. And after a sufficient number of affirmative trials, they tried to solve the puzzle.

A small number of participants chose a more effective strategy. Instead of trying to confirm their hypotheses, they immediately proposed a disconfirming sequence like 10, 13, 21. The result was that these participants reached the correct answer in a few short trials. The majority however, continued their futile quest, only changing their strategy once they learned that their hypothesis was wrong. Some participants gave up altogether.

The tendency to test one's beliefs by considering confirming evidence alone (while ignoring disconfirming evidence) is called the confirmation bias and it is a folly we all commit. For any given position in a dispute (God vs. the big bang, Roe vs. Wade, The Beatles vs. The Rolling Stones), advocates tend to be sensitive to information the supports their position and blind to anything else. The tragedy is that there is an abundance of information consistent with any particular view, leaving us vulnerable to believing anything we so desire.

The daily horoscope takes advantage of its readership's bias for confirming information, offering vague insights that can be corroborated by practically any occurrence. "Expect difficult challenges at work today'' is just as valid if you sign a new client, forget your lunch at home, or get fired.

The lesson is this: Whether you are a God-fearing believer or a staunch atheist, you've probably amassed oodles of confirming evidence for your position. But that won't get you any closer to the truth. Search instead for that one devastating fact that brings down your entire worldview. Now, you're getting somewhere.

No comments:

Post a Comment