We like to think of ourselves as open-minded, but we're not. The problem is not that, once we've found a solution to a problem, we refuse to think of alternatives. It's that we don't even realize that there are alternatives to consider.
If you remember someone having a name "like Megan," it's going to be hard to shake the actual name out of your head. If you think that the diabolical Count is the murderer in an Agatha Christie mystery, it's going to be hard to think of anyone else committing the crime even if he's shown to be innocent. If you're pretty sure you left your keys in that one old jacket you have, you're going to keep circling back to it because it's hard to think of other places your keys might be.
That's not the congruence bias. The congruence bias so completely dominates our minds that we can't even realize there are alternative theories. We can't find the real solution because we're not looking for it. One researcher tested this by giving people lists of numbers that followed a certain rule. (The numbers given were 2, 4, and 6, and they were simply ascending numbers.) People assumed that they were even numbers, or numbers that increased by two, which is a perfectly understandable guess, and not an example of bias. The bias came when people were told that their guess regarding the rule was wrong. Instead of thinking of alternate solutions, they began re-wording their guess. The problem couldn't be with the concept they'd thought of, just the way they expressed it.
A bigger problem with the congruence bias is that testing it and asking questions often won't help. One study put forward a plausible but incorrect hypothesis about a complicated social issue, and backed it up with a list of questions and answers. Subjects looked over the hypothesis and the answers, and rated the questions and the answers that seemed to confirm the hypothesis as very important, while downplaying the questions and the answers that cast doubts on it. As for testing, people tend to set up tests that will yield a positive answer if the hypothesis is right, not ones that will disprove it if it's wrong. They also don't set up tests that might indicate the validity of alternate theories. We want to be told we're right so much, we don't even think we could be wrong.
The congruence bias might have real world repercussions. I remember posters all over town which featured a picture of a homeless alcoholic man sitting dejectedly on a street corner and said, "It doesn't always end here." The lower third of the poster featured a fresh-faced teen smoking a joint and said, "But it often starts here." Studies had found that a lot of drug users and homeless people once smoked pot. Therefore pot is a "gateway drug." No further hypotheses needed.
[Via Heuristics and Biases in Diagnostic Reasoning, Philip Copitch]