What is cognitive dissonance theory?
Definition and explanation
Holding two conflicting beliefs causes us to feel slightly uncomfortable, and to reject (at least) one of them. For instance, you held the beliefs that science is a useful way to discover the truth and that doctors have your best interest at heart and aren’t involved in a widespread conspiracy theory to cause ill-health. You then encounter some evidence, maybe via your friend Andrew, in favour of the idea that vaccines cause autism. (Author’s note: for this example, you might be unsure, but in real life, VACCINES DON’T CAUSE AUTISM. Sorry for the all caps. That is all.) We strive for internal consistency (except when we don’t) and these beliefs appear to conflict with one another, so you have to reject one or more.
The feeling of discomfort increases with:
How important the subject is to us.
How strongly the dissonant thoughts conflict.
Our inability to rationalize and explain away the conflict.
Examples of cognitive dissonance
Cognitive dissonance makes it tough to change our minds, especially when the two beliefs are tied up in our identity. Paul Graham (founder of Y Combinator) recommends we fight this uphill battle by identifying with as few things as we need to. Using the example of vaccines above, if someone was really invested in the belief that vaccines cause autism, went to anti-vaxxer meetings, and told all their friends it was a conspiracy, they’d be unlikely to change their mind when presented with strong evidence to the contrary.
Aumann's agreement theorem is kind of like a multiplayer version of this. Briefly, Aumann showed that we should take disagreement seriously because if two rational agents disagree, at least one of them is wrong, and it might be us.
Results like Condorcet's jury theorem suggest that if many people converge on the same answer to a question (independently of one another, to avoid groupthink), we should treat that as good evidence that it’s the correct one.
Also check out
Cognitive dissonance, Wikipedia
Your brain is not a Bayes net (and why that matters), Julia Galef