Why Smart People Make Bad Decisions

The smartest people in the room

In the late 1990s, the hedge fund Long-Term Capital Management was as close to a room of certified geniuses as finance has produced. Its partners included Myron Scholes and Robert Merton, who shared the 1997 Nobel Memorial Prize in Economics for their work on pricing risk. Within a year, in 1998, the fund lost most of its capital and came so close to collapse that the Federal Reserve organized a rescue to prevent wider damage. The people who best understood risk on paper had built a position that could not survive a scenario their models treated as nearly impossible. As Roger Lowenstein documented, the failure was not a failure of intelligence. It was a failure of judgment under uncertainty, made by people whose intelligence gave them extraordinary confidence in the very models that undid them.

This is the pattern that should unsettle every executive. If brilliance were protection, the smartest organizations would make the fewest mistakes. They do not. The question the psychology of decision-making answers is not why foolish people fail. It is why capable people, acting in good faith, make errors that are predictable in advance.

Intelligence is not rationality

Herbert Simon established that human decision-making is boundedly rational: real people do not optimize across all options, they satisfice within cognitive and time limits, working with simplified models of a complex world. Amos Tversky and Daniel Kahneman then showed that the shortcuts people use to manage those limits, the heuristics, produce systematic and repeatable biases rather than random noise. Kahneman later framed this as two systems: a fast, intuitive System 1 that generates immediate impressions, and a slow, effortful System 2 that is supposed to check them but frequently does not, because checking is expensive and System 1 feels certain.

The uncomfortable finding is that raw intelligence offers little defense. Keith Stanovich and Richard West have shown that many thinking biases are largely independent of cognitive ability, and that intelligent people are not meaningfully better at avoiding them. Worse, in research on the bias blind spot, West, Meserve, and Stanovich found that cognitive sophistication does not reduce the tendency to see bias in others while missing it in oneself, and can slightly increase it. The capable executive is not immune to bias. The capable executive is often the person most certain they are being objective.

Intelligence is the ability to solve the problem in front of you. Rationality is knowing which problem you are actually solving, and whether your certainty is earned. They are not the same faculty, and organizations keep confusing them.

The errors are predictable

Behavioral economics gives the errors names, and the names matter because a bias you can name is a bias you can design against. Kahneman and Tversky's prospect theory showed that people feel losses roughly twice as intensely as equivalent gains, so leaders defend failing bets to avoid booking a loss. Barry Staw's work on escalation of commitment demonstrated that decision-makers pour more resources into a losing course precisely because they chose it, converting a bad decision into a worse one. Hal Arkes and Catherine Blumer showed that the sunk cost effect, honoring money and effort already spent that cannot be recovered, drives people to persist against their own interest. None of these is a sign of low ability. Each is a feature of normal, capable cognition operating exactly as designed.

Why organizations amplify the error

Individual bias would be manageable if organizations reliably corrected it. They often do the opposite. Hierarchy concentrates the most consequential decisions in the few people with the most confidence and the least direct exposure to disconfirming detail. Deference suppresses the dissent that would surface the flaw. Shared incentives and shared narratives turn one person's bias into a group's consensus, so the error is not caught, it is ratified. An organization can take a predictable individual mistake and, through its own structure, scale it into a strategic one. The bias is human. The amplification is organizational, which means it is also designable.

What this means through the Four A's

Research explains the phenomenon. The Four A's turn it into something an executive can act on, and this one touches several conditions at once. It is primarily a question of Authority: decision quality depends on where a decision sits, who is empowered to challenge it, and whether the structure forces a second, slower look before commitment. It is also a matter of Attention, because bias worsens under overload, when System 2 has no capacity left to check System 1. And it reaches Adaptability, because escalation and sunk cost are failures to update, the refusal to reconfigure a decision once the evidence has moved. Smart people making bad decisions is not a talent problem to solve by hiring. It is a condition to design, by building the decision architecture that assumes capable people will still be predictably wrong.

From bias to constraint

The pattern beneath this is general: an individual bias, left unexamined, is adopted, reinforced, and encoded until it becomes a structural limit on the whole organization. The Four A's are where a leader interrupts the chain.

Evidence matrix

Executive reflection questions

• Which of your current commitments are you defending because of what they have already cost, rather than what they will produce?
• On your most important recent decision, who was empowered to tell you it was wrong, and did they?
• Where in your organization does the most confident voice also carry the least exposure to the detail?
• When your judgment feels most certain, what would it take for you to treat that certainty as a signal to slow down rather than proceed?

Builder actions

Design the decision, not just the decision-maker. Before a major commitment, run a premortem: assume the decision failed and have the team explain why, which Gary Klein has shown surfaces risks that confidence hides. Separate the recommendation from the decision so the person who proposes is not the only one who approves. Set stop criteria in advance, the conditions under which you will reverse course, so escalation has a tripwire before loss aversion takes over. Force a reference view: ask how similar efforts have actually turned out, not how this one feels. And protect the attention of your decision-makers, because a depleted System 2 cannot check anything. None of these assume smarter people. They assume capable people who will be predictably wrong, and they build the conditions that catch it.