I had an interesting conversation yesterday with a dean of education at an American university. She asked: what is the evolutionary explanation for why human beings are so willing to believe things that are at odds with their day-to-day experience? What evolutionary advantage to we gain from ignoring reality? You'd think it would be a disadvantage. What makes this particularly puzzling is that the more educated and intelligent a person is, the more likely they are to exhibit this cognitive dissonance. This is what Claude says: The short answer most evolutionary psychologists would give is that her premise contains a hidden assumption: that the function of belief is to track reality. For much of what we believe, it isn't. Several explanations work together here. **The cost of being wrong is usually tiny; the cost of being out of step is large.** Dan Kahan's work on "identity-protective cognition" makes this point well. If I'm wrong about climate sensitivity or the origins of a pandemic, the consequences for me personally are close to nil, since my belief changes nothing in the world. If I hold a view my social group regards as heretical, I may lose friends, status or my job. For ancestral humans, whose survival depended on their band, being expelled was close to a death sentence. Natural selection would favour minds that prioritise coalitional loyalty over accuracy wherever accuracy carries no immediate penalty. Philosophers such as Dan Williams call these "socially adaptive beliefs." **Absurd beliefs can be useful precisely because they are absurd.** If a belief is obviously true, holding it says nothing about your loyalty, because everyone holds it. A belief that runs against common sense works as a costly signal: it shows you're willing to pay a price to belong. Richard Sosis has documented this pattern in religious communities, where more demanding creeds produce more cohesive and longer-lasting groups. Shibboleths work because outsiders find them hard to fake. **Trusting others over your own experience is mostly a winning strategy.** Joseph Henrich argues that humans succeeded through cumulative culture: copying the practices of prestigious elders even when the reasons are opaque. His example is manioc (cassava) processing in South America. The traditional preparation is long and laborious, and skipping it causes no immediate harm, but it removes cyanide that would otherwise poison you slowly over years. Someone who trusted their own experience over tradition would have died. So we evolved to defer to testimony, and the occasional absorption of nonsense is the price of a system that is usually right. This cuts both ways. The same willingness to believe things that contradict experience is what lets us accept that the Earth orbits the Sun, or that invisible germs cause disease. **Asymmetric errors favour false positives.** Under error management theory (Martie Haselton and colleagues), when one kind of mistake is much costlier than another, selection favours biases toward the cheaper error. Mistaking wind in the grass for a predator costs little. Making the opposite mistake might cost your life. We are therefore primed to see agency, patterns and intention where none exist, which is fertile ground for superstition and conspiracy theories. **Self-deception helps persuasion.** Robert Trivers and Bill von Hippel argue that we deceive ourselves partly to deceive others more convincingly. A man who sincerely believes he is more able, more virtuous or more in the right than he is makes a better advocate for himself. **People are less gullible than they appear.** Hugo Mercier, in *Not Born Yesterday*, distinguishes "intuitive" beliefs, which guide action, from "reflective" beliefs, which we profess but keep in a separate compartment. People who claimed to believe Pizzagate mostly left bad online reviews of the restaurant rather than calling the police. When a belief would actually cost us something in daily life, we tend to quietly stop acting on it. Many beliefs that seem to ignore reality are closer to declarations of allegiance than to working models of the world. **There is also a mismatch problem.** Our deference to consensus evolved in groups of perhaps 150 people, where the shared view had usually been tested against local reality. Mass media and online communities can manufacture a consensus far removed from anyone's direct experience, and our instincts can't easily tell the difference. So the answer to the dean's puzzle is roughly this: we don't ignore reality where it bites. We ignore it where it doesn't, and in those domains belief serves social ends rather than epistemic ones. The interesting question then becomes what happens when those cheap, tribal beliefs are aggregated into votes and public policy, where they start to have real consequences that no individual believer bears.