6 ideas
Human reason is not a general-purpose tool for improving individual beliefs; it evolved to produce arguments that persuade others and to evaluate arguments others offer. Its apparent flaws when used alone are features of a mechanism designed for social justification, not defects of a truth-seeking faculty.
When producing arguments, people search almost exclusively for reasons supporting their own position and apply lax standards to them. This is efficient division of labor in dialogue: the speaker builds the case cheaply, and the listener's scrutiny catches the weak parts, so one-sided production is only harmful when no one pushes back.
People are lenient with their own reasons but sharp critics of others' arguments, readily spotting flaws they would never notice in their own. Reason's quality therefore depends on which role it is in, making evaluation the reliable half of the faculty.
On the Wason four-card task, only about 10% of individuals typically give the correct logical answer. When small groups discuss it, most groups reach the correct answer, often because one person who has it convinces the others through argument, showing that truth wins in exchange even when solitary reasoning fails.
Because communication makes people vulnerable to deception, humans evolved mechanisms that assess the source's trustworthiness and the coherence of a message with what they already believe. Arguments evolved as a way for speakers to get past this vigilance by showing how a claim follows from things the listener already accepts.
Most of the reasons people give for their beliefs and actions are constructed after the fact to justify themselves to others, not the actual causes of their decisions, which come from intuitive inference. Seeing reasons this way explains why solitary reasoning often rationalizes and polarizes: without an interlocutor, it simply piles up support for whatever intuition came first.