The people with the most mental horsepower do not always have the safest lead when the stakes rise.
Give students a difficult arithmetic task under ordinary conditions and those with greater working-memory capacity tend to perform better. Tell them the result will prove whether they can outperform everyone else, however, and that advantage can shrink or even reverse. The same mental system that normally helps them hold rules, intermediate answers and possible solutions in mind becomes crowded with a second problem: what the outcome will say about them.
This is often summarized as “smart people choke more,” but that version outruns the evidence. The experiments measured working-memory capacity, not intelligence as a whole, and the performance losses appeared on demanding tasks under particular forms of pressure. They do not show that highly intelligent people are generally the worst performers in stressful situations.
What they do show is more useful. Pressure does not affect everyone in the same way, and changing the purpose of a task from proving status to developing mastery can change how much cognitive capacity remains available to solve it.
Pressure can consume the advantage it is supposed to reveal
Working memory is the limited mental workspace used to keep information active while manipulating it. It helps a person remember which step of an equation they have reached, compare alternatives in a reasoning problem and resist irrelevant distractions. It is related to academic performance and fluid reasoning, but it is not a synonym for intelligence.
In a landmark experiment published in 2005, psychologists Sian Beilock and Thomas Carr tested people on mathematical problems under low- and high-pressure conditions. Only participants with high working-memory capacity showed a significant performance decline under pressure, and the damage was confined to problems that placed the greatest demands on working memory.
The researchers’ explanation was not that capable participants suddenly knew less. Pressure occupied the workspace they normally used to perform well. Worries about failure, evaluation and consequences competed with the calculations for the same finite resource.
A later study extended the pattern to abstract reasoning. Its authors found that performance pressure reduced scores among people with high working-memory capacity on a measure of fluid intelligence. Participants’ anxious perception of the high-stakes setting helped explain the decline.
Under ordinary conditions, more capacity meant better performance. Under pressure, the extra capacity did not disappear. Access to it was disrupted.
The goal can become a second task
Pressure is not just a property of an exam room, deadline or prize. It is also created by the meaning assigned to the task.
Achievement-goal researchers distinguish between mastery goals and performance goals. A mastery-approach goal asks whether a person can learn, improve or understand. A performance-approach goal asks whether that person can demonstrate superior ability, often by doing better than other people. Performance-avoidance goals focus on not looking incompetent or not doing worse than others.
In two experiments published in PLOS ONE, Marie Crouzevialle, Annique Smeding and Fabrizio Butera asked university students to complete complex arithmetic problems. Some received instructions emphasizing proficiency and a high score relative to others. Others received a mastery-oriented framing without the same social comparison.
The result was striking. Under performance-approach goals, greater working-memory capacity predicted worse arithmetic performance. The pattern was strongest when participants were uncertain about their chances of outclassing others. A measure of accessible thoughts suggested that concerns about status were intruding on the task.
The goal had become a second calculation. Participants were not only solving the problem in front of them. They were monitoring what the result might do to their standing.
This was not simply a test of rewards
The finding is sometimes described as proof that intelligent people crumble whenever goals are reward-based. That is too broad.
The central manipulation in the PLOS ONE experiments was social comparison and demonstrated competence, not merely the presence of a reward. Financial incentives can create pressure, and a review of incentive-induced choking describes evidence that very large rewards sometimes impair cognitive or motor performance. But a bonus, a ranking, public evaluation and fear of disappointing others are not psychologically identical.
Pressure can also help. Higher arousal may improve performance on a simple or well-learned task. The risk is greater when the task is complex, when success depends heavily on working memory and when the mind starts devoting scarce attention to consequences rather than execution.
That is why “rewards are bad” is the wrong lesson. The more precise question is whether the reward changes attention from doing the work to protecting an identity, rank or outcome.
Mastery goals usually offer a safer frame
The benefits of mastery framing extend beyond one experiment. A 2015 meta-analysis assembled 79 effect sizes from 19 papers and 3,482 participants. It found that mastery-approach goals produced better task performance than performance-approach goals overall, particularly when people did not expect feedback and were not under time pressure.
A larger later meta-analysis examined 235 effect sizes from 90 studies. It likewise concluded that experimentally induced mastery-approach goals led to higher performance on average than performance-approach, performance-avoidance and no-goal conditions.
“On average” matters. Goals are not magic phrases, and their effects depend on how they are framed, the task and the person. An unusual pair of experiments found that even a self-improvement goal could reduce the normal link between working memory and math performance when high-capacity participants were told there was very little room left for them to improve. The message may have converted learning into another threatening verdict.
A mastery goal works best when improvement is genuinely possible and mistakes provide information. Renaming an outcome target “learning” while preserving the same threat does not remove the threat.
How to design pressure without wasting ability
For teachers and managers, the research suggests changing the information environment, not simply telling anxious performers to relax.
Before a difficult task, emphasize the process that deserves attention: identify the principle, test an approach, notice the error and revise. Give feedback against a clear standard before turning performance into a public ranking. Separate practice, where errors should be diagnostic, from selection, where outcomes carry consequences.
For the person facing the pressure, one useful strategy is to move worries out of working memory before the task begins. In a recent experiment with highly test-anxious participants, a brief expressive-writing exercise reduced anxiety and eliminated the pressure-related accuracy loss among those with high working-memory capacity. That is one study, not a guaranteed treatment, but it fits the idea that worries compete with the task for limited space.
Reappraising physical arousal, practicing under realistic pressure and using routines that narrow attention to the next action may also help. Still, a systematic review of pressure-coping interventions found wide variation in methods and evidence quality. No single technique has earned universal status.
A high-stakes score can underestimate the mind that produced it
The most important implication is about measurement. A pressure-filled performance is not a transparent reading of ability. It is the product of ability, task demands, motivation, attention and the meaning attached to the outcome.
High working-memory performers may have more to lose precisely because demanding tasks normally let them use more of that resource. When status concerns occupy it, the test can erase part of the advantage it was designed to detect.
That does not make pressure inherently unfair or mastery goals universally superior. It does mean that an organization can spend years recruiting capable people, then evaluate them in a way that diverts their strongest cognitive asset from the work to the consequences of the work.
If the aim is to discover what a person can do, the best question may not be “Can you prove you are the best?” It may be “What can you learn from the next problem?”