Why Smart People Make Consistently Bad Choices

Intelligence is not a general-purpose protection against bad decisions. This is one of the findings in behavioral science that people find most difficult to absorb — not intellectually, but operationally. We act as if it weren't true. We assume that the smart people in the room will catch the errors, identify the biases, and arrive at sound conclusions. We give their opinions disproportionate weight. We appoint them to lead. And then they make the same mistakes, in the same categories, for the same underlying reasons, that everyone else makes. Sometimes with more sophistication and confidence. ## Motivated reasoning at high intelligence The most significant finding in this area is about motivated reasoning — the tendency to arrive at the conclusion you wanted to arrive at, using apparently logical reasoning as the mechanism. High cognitive ability does not reduce susceptibility to motivated reasoning. In many studies, it increases it. The mechanism is this: more intelligent people are better at constructing arguments. When the goal is to justify a predetermined conclusion — which motivated reasoning is — being better at constructing arguments means being better at motivated reasoning. The result is a more elaborate, more internally consistent, more confident-sounding justification for a conclusion that was reached through the same emotional and identity-driven processes as everyone else's. Intelligence, in this context, is less a check on bias and more a turbocharger for it. ## The expertise problem Domain expertise creates a different category of decision failure — one that is genuinely paradoxical. Deep expertise in a domain improves decision-making within that domain. It also creates something researchers call "expert overconfidence" — the extension of domain-specific calibration (which is genuinely good) to adjacent domains where it doesn't apply. Experts systematically overestimate the transferability of their expertise. The cardiologist who believes their medical training gives them superior insight into organizational strategy. The engineer who applies deterministic thinking to human behavior problems. The economist who reduces complex social phenomena to rational-choice models. Each of these is a failure mode of expertise — a confident application of the wrong mental model to a domain it doesn't fit. ## What actually improves judgment The research on what improves judgment — as distinct from what improves raw reasoning ability — points to a different set of variables. **Calibration training.** Learning to have accurate confidence — to know the difference between what you know well and what you merely know — is teachable. Forecasting practice, prediction markets, and consistent feedback on confidence scores improve calibration across domains. **Perspective diversity.** Groups with more diverse problem-framings make better decisions than groups with higher average intelligence. This is not a feel-good finding — it's a structural one. Different framing approaches catch different categories of error. **Pre-mortem practice.** Explicitly imagining that a decision has already failed, and working backward to identify why, counteracts the optimism bias that afflicts most confident decision-makers. **Slowing the judgment cycle.** Decisions made quickly on complex, novel problems are systematically worse than decisions made after deliberate delay, in most domains. High-intelligence people are particularly resistant to this finding because speed of thinking feels like accuracy. ## The uncomfortable conclusion The uncomfortable conclusion from this body of work is that the qualities we use to select decision-makers — intelligence, confidence, expertise, fluency — are only loosely correlated with decision quality, and in some domains are negatively correlated. The qualities that actually predict good decisions — calibration, intellectual humility, tolerance for ambiguity, perspective-taking ability, comfort with uncertainty — are rarely what we select for, and rarely what we reward. We teach people to be smart. We do not teach them to have good judgment. And we continue to act as if these were the same thing.