A cognitive test is designed to make interpretation easier. Real life does almost the opposite. Instructions may be unclear, the relevant information may be buried among distractions, and the person solving the problem may be tired, interrupted or under pressure. They may be allowed to use a phone, a calculator, a checklist, a colleague or a search engine. The problem itself may change while they are solving it. Sometimes the most intelligent move is not to solve the problem unaided at all, but to recognize that the situation requires information or expertise that should come from somewhere else.
This is where the familiar distinction between intelligence and performance becomes more than a technical point. Cognitive ability matters outside the testing room: it is associated with educational achievement and later occupational outcomes, and decades of research have shown general cognitive ability to be useful in predicting learning and job performance, particularly where tasks involve complexity (Roth et al., 2015; Schmidt & Hunter, 1998; Strenze, 2007). But an ability is not an outcome waiting to happen. What a person can bring to a problem still has to meet the problem that is actually there.
The task changes what intelligence has to do
Consider something mundane but consequential: managing a new medication. The apparent task is simple—take the right medicine at the right time—but successful performance may require understanding written instructions, distinguishing one medication from another, remembering timing, coordinating doses with meals, noticing changes in symptoms, updating a routine, judging when information is uncertain and knowing when to ask a professional for help.
A failure at any point could look identical from the outside: the medication was taken incorrectly. Yet the mechanisms behind that failure could be completely different. The instructions may have been linguistically difficult. Packaging may have made two medications hard to distinguish. The schedule may impose unnecessary memory demands. The person may misunderstand a concept, understand it but forget at the critical moment, or know exactly what to do but be too unwell to carry it out consistently. The problem may even lie in the healthcare system’s communication rather than in the person.
Calling the final outcome “intelligent” or “unintelligent” tells us almost nothing about which of those things happened.
This is not an argument that intelligence disappears in context. A person with stronger reasoning ability may learn an unfamiliar system more quickly, detect inconsistencies more readily or handle novel complications more effectively. The same principle helps explain why cognitive ability predicts performance across many educational and occupational settings. But the prediction is never produced by ability in isolation. Real tasks determine which abilities are relevant and how heavily they are demanded.
A badly designed form can turn a straightforward decision into a working-memory test. A clear diagram can remove a verbal burden that was never essential to the underlying problem. Time pressure can make processing speed more consequential. A familiar task can reduce the need for novel reasoning because knowledge now carries much of the load. Change the task and you can change the cognitive demands without changing the person.
Ability changes what a person can bring to a task. Performance tells us what happened when that person met this task under these conditions.
Knowledge changes the problem before reasoning begins
Expertise makes this especially visible. A novice and an expert do not merely solve the same problem at different speeds. Often they are not, cognitively speaking, facing quite the same problem.
The expert recognizes patterns that the novice must laboriously construct. Relevant information comes to mind more readily. Some decisions have become routine. Years of learning have organized knowledge so that important features stand out and irrelevant possibilities can be discarded earlier. What once required effortful reasoning may later be handled through well-developed representations and procedures.
This is why intelligence and knowledge cannot sensibly be placed on opposite sides of an explanation. Cognitive ability contributes to learning; learning builds knowledge; knowledge then changes future cognition. Fluid reasoning may be especially important when a problem is genuinely unfamiliar, while accumulated knowledge can transform familiar domains. Cattell’s investment idea was built around precisely this kind of developmental relationship, even though modern accounts do not reduce crystallized knowledge to fluid intelligence alone.
Education illustrates the interaction. Schooling plainly teaches knowledge and skills, but evidence also suggests that additional education can produce gains in measured intelligence (Ritchie & Tucker-Drob, 2018). The causal story therefore does not run only from intelligence to learning. Development unfolds through repeated exchanges among ability, instruction, knowledge and experience.
The same interaction appears in ordinary competence. A skilled nurse, mechanic, teacher or programmer is not simply a person applying a fixed quantity of general intelligence to different content. Their performance reflects cognitive ability operating through a structure of acquired knowledge that changes what they notice, remember and infer.
We rarely think alone
Real-world cognition is also strikingly dependent on things outside the head.
Adults use calendars, notebooks, diagrams, calculators, maps, databases, alarms, checklists and search engines. Professionals build procedures precisely because important work should not depend on unaided memory. Teams distribute expertise among people. A pilot uses instruments; a physician consults records; a scientist writes equations down rather than trying to maintain an entire derivation in working memory.
These supports complicate simplistic judgments about intelligence because a tool can play different roles. Sometimes it removes an irrelevant burden: writing an appointment in a calendar prevents forgetting without doing the planning for us. Sometimes it augments performance by making a difficult operation faster or more reliable. Sometimes it substitutes for an operation the person could not perform independently. Those cases are not psychologically identical.
Nor is using a tool cognitively free. Someone still has to recognize when help is needed, choose an appropriate source, formulate the problem, interpret the output and notice when the answer makes no sense. A calculator does arithmetic; it does not decide whether the calculation answers the question. A search engine retrieves information; it does not guarantee that the information is relevant or true.
This matters because the modern world increasingly evaluates people in environments saturated with cognitive supports. If our question is what someone can accomplish in that world, insisting that “real” intelligence must always be unaided can become artificial. If our question is what a person can do independently, however, the distinction becomes essential. The right interpretation depends on what we are trying to measure.
The same person is not cognitively identical every hour
There is another complication that standardized testing tries to control and ordinary life cannot: state.
Relatively stable differences in cognitive ability do not imply perfectly stable performance from moment to moment. Sleep loss, illness, pain, acute stress, intoxication and other temporary conditions can affect cognition, although their effects vary by task and circumstance. A person can possess an ability and fail to deploy it effectively on a particular occasion.
This fact is easy to abuse. “They were tired” cannot become a universal explanation for every disappointing result, any more than a favorable result should automatically be treated as pure ability. Contextual explanations require evidence too. But ignoring state produces the opposite error: treating every observed performance as if it were a transparent expression of stable capacity.
The distinction is especially important when the stakes are practical rather than psychometric. If someone repeatedly makes errors at the end of a twelve-hour shift but performs well under rested conditions, the useful question may not be “Which performance reveals their true intelligence?” Both performances are real. The more informative question is what combination of person, task and state produces reliable success.
Intelligence in the world is an interaction problem
Some contemporary theories push this idea further. Robert Sternberg’s adaptive-intelligence framework, for example, argues explicitly that intelligent behavior should be understood through interactions among person, task and situation rather than as the expression of a personal trait alone (Sternberg, 2021). That is a theoretical proposal, not a settled replacement for psychometric models of individual differences, and it should be treated as such.
But we do not need to accept any one theory of intelligence to recognize a simpler empirical point: real-world performance is multiply determined.
A useful analysis therefore asks what the person brings, what the task demands, what they already know, what state they are in, which tools or people they can use, and which features of the environment make success easier or harder. These are not competing explanations in which only one is allowed to be true. Their importance changes from problem to problem.
This is also why interventions can improve performance without changing intelligence. Rewrite an instruction and you may remove an irrelevant language burden. Add a checklist and you may reduce prospective-memory demands. Teach the underlying concept and the person now possesses knowledge they previously lacked. Allow more time and a speed constraint may cease to dominate. None of those changes proves that cognitive ability was irrelevant. They tell us something about what was limiting performance.
The distinction becomes powerful precisely when something goes wrong. “They couldn’t do it” is an observation. It is not yet an explanation.
A student who repeatedly fails a mathematics task may lack the underlying concept, misunderstand the language of the problem, lose intermediate steps from working memory, work too slowly under a time limit, or face reasoning demands beyond their current ability. Those possibilities can produce the same mark on a page and require very different responses. The same logic applies to work, healthcare, finance, technology and everyday decision-making.
Asking how intelligence is used in real life therefore changes the question. Instead of treating performance as a simple readout of the person, we ask what had to happen for this person to succeed at this task, here, now.
Intelligence remains part of the answer. It is simply no longer mistaken for the whole causal chain.
References
- Ritchie, S. J., & Tucker-Drob, E. M. (2018). How much does education improve intelligence? A meta-analysis. Psychological Science, 29(8), 1358–1369. https://doi.org/10.1177/0956797618774253
- Roth, B., Becker, N., Romeyke, S., Schäfer, S., Domnick, F., & Spinath, F. M. (2015). Intelligence and school grades: A meta-analysis. Intelligence, 53, 118–137. https://doi.org/10.1016/j.intell.2015.09.002
- Schmidt, F. L., & Hunter, J. E. (1998). The validity and utility of selection methods in personnel psychology: Practical and theoretical implications of 85 years of research findings. Psychological Bulletin, 124(2), 262–274. https://doi.org/10.1037/0033-2909.124.2.262
- Sternberg, R. J. (2021). Adaptive intelligence: Intelligence is not a personal trait but rather a person × task × situation interaction. Journal of Intelligence, 9(4), 58. https://doi.org/10.3390/jintelligence9040058
- Strenze, T. (2007). Intelligence and socioeconomic success: A meta-analytic review of longitudinal research. Intelligence, 35(5), 401–426. https://doi.org/10.1016/j.intell.2006.09.004
