A child says they will remember all ten objects.
Then remembers four.
The interesting fact is not only that memory failed. It is that the child had an idea about memory before the test began, and that the idea could be compared with what memory actually did.
That gap between cognition and cognition-about-cognition became part of the territory associated with John H. Flavell, one of the most influential developmental psychologists of the twentieth century and a foundational figure in modern metacognition research.
Flavell died on 13 March 2025 at the age of 96. Stanford University, where he became the Anne T. and Robert M. Bass Professor of Psychology and later professor emeritus, describes him as a leading scholar of children's cognitive development and credits him with coining the term metacognition. He joined Stanford in 1976 and retired in 2002.
His legacy is often compressed into that one word.
It is more useful to recover the problem the word was trying to name.
Development came before the label
Flavell's intellectual formation was deeply connected to the work of Jean Piaget. His 1963 book The Developmental Psychology of Jean Piaget helped introduce and interpret Piagetian thinking for English-speaking psychology at a time when cognitive development was becoming a major research field.
But Flavell did not simply preserve Piaget. His own research increasingly examined how children understand mental processes: memory, perspective, appearance and reality, communication, intention and the conditions under which one person's knowledge differs from another's.
This developmental orientation matters for metacognition.
The question was not originally “How can adults optimize their studying?” It was closer to: how does a person develop knowledge about cognition itself? What does a child know about what is easy or difficult to remember? Can they predict when a strategy will help? Do they notice when understanding has failed? Can knowledge about the mind guide action?
Metacognition, in this lineage, is not an ornament placed on top of cognition. It develops as part of becoming able to participate in one's own thinking.
Metamemory was one of the early laboratories
Before metacognition became an umbrella term, metamemory provided a more specific research problem.
Flavell and colleagues studied what children know about memory: which tasks are difficult, which strategies might help, how memory changes with age, and whether children can monitor what they know or are likely to remember.
These questions separated at least two things that everyday language often merges.
A person can possess memory ability without possessing accurate knowledge about that ability. And a person can know that a strategy is useful without reliably deploying it at the right moment.
That distinction remains central to contemporary research. Knowledge, monitoring and control are related, but they are not the same operation.
The later Knowledge Library entries on metacognitive knowledge, monitoring and control inherit this architecture even though contemporary models have refined the constructs considerably.
The 1979 paper gave the field a durable map
Flavell's 1979 American Psychologist article, “Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry,” is one of the field's canonical statements.
The article describes metacognition broadly as knowledge and cognition about cognitive phenomena and organizes the territory around several interacting classes: metacognitive knowledge, metacognitive experiences, goals or tasks, and actions or strategies.
Metacognitive knowledge includes beliefs and knowledge about people, tasks and strategies. A learner may know that they remember meaningful material better than arbitrary lists, that one task demands more attention than another, or that a particular strategy is useful under certain conditions.
Metacognitive experiences are more immediate conscious experiences related to a cognitive undertaking: the sense that something is confusing, the feeling that an answer is close, the recognition that a plan is not working.
Goals and strategies connect those forms of monitoring to what the person is trying to do.
This was not yet today's unified computational theory. Its importance was architectural. It made the territory researchable without reducing it to one score.
“Knowing about knowing” is only the beginning
Popular summaries often define metacognition as “thinking about thinking.” The phrase is memorable but incomplete.
Flavell's framework already points beyond detached reflection. Metacognitive knowledge can influence what a person attempts. Metacognitive experiences can interrupt or redirect activity. Strategies can be selected in response to what monitoring reveals.
Modern models make the monitoring–control loop more explicit, but the developmental insight remains: knowing that something is happening in one's cognition matters because it can alter what happens next.
A learner who notices that a paragraph has become unintelligible can reread, slow down, ask for clarification or change strategy. A learner who fails to notice can continue producing the performance of reading without comprehension.
The difference is not simply intelligence. It is access to information about one's own cognitive process and the ability to use it.
Flavell's “person, task, strategy” distinction still travels well
One of the most durable elements in Flavell's metacognitive knowledge framework is the distinction among knowledge about persons, tasks and strategies.
Person knowledge concerns cognitive characteristics of oneself and others. Some may be accurate; some may be stereotypes or mistaken self-beliefs.
Task knowledge concerns what a cognitive task demands: whether material is familiar, ambiguous, dense, time-sensitive or likely to overload memory.
Strategy knowledge concerns possible ways of approaching the task and the conditions under which a strategy may help.
These categories are useful precisely because they expose why generic learning advice fails. A strategy is not effective in the abstract. It is effective for a person facing a task under particular conditions.
That contextual structure also places a boundary around self-knowledge. Knowing “I am bad at names” is not enough. The useful metacognitive question is what kind of name-learning task is difficult, under what conditions, and what action changes the outcome.
Developmental research complicated the image of the transparent mind
Flavell's wider work on cognitive development repeatedly showed that children do not begin with adult-like understanding of mental representation.
Research on perspective-taking, appearance versus reality and the distinction between one's own knowledge and another person's knowledge helped make mental representation itself a developmental object of study.
That matters because metacognition can sound, in adult philosophical language, as though consciousness simply turns inward and reads its own state.
Developmental psychology makes the situation less romantic. Knowledge of mind has to be built. It can be partial. It can lag behind performance in one domain and lead it in another. It depends on language, experience, task structure and social interaction.
The mind is not perfectly transparent to itself.
That proposition is now ordinary in metacognition research, but its developmental roots are worth remembering.
Influence should not be mistaken for final validity
Researcher profiles easily become hagiography.
Flavell's influence is substantial. That does not mean every category in his early framework has been validated as a clean, separable psychological variable across populations.
Later researchers developed different architectures. Nelson and Narens formalized relations between object-level cognition and meta-level monitoring and control. Educational psychology developed models of self-regulated learning that include motivation, behavior and context. Cognitive neuroscience introduced signal-detection and computational measures of metacognitive sensitivity. Contemporary confidence research distinguishes multiple levels of uncertainty representation and subjective report.
These later developments do not make Flavell obsolete. They show what foundational work is for: it makes questions stable enough for later researchers to disagree more precisely.
The Institute therefore treats Flavell as historically foundational, not as the final authority on what metacognition is.
The phrase “coined metacognition” needs a little care
Stanford's official obituary says Flavell coined the term metacognition, and this attribution is widespread.
Historical concepts, however, rarely appear from nowhere. Philosophers, educators and psychologists had long studied self-knowledge, reflection, memory monitoring and strategic regulation. Flavell's contribution was not that no human being had previously noticed that minds can reflect on themselves.
His importance lies in giving modern developmental psychology a productive research vocabulary and programme around cognition about cognition.
That distinction protects against invented origin stories. Scientific fields are usually assembled from older questions, new terminology, new methods and communities that make the questions tractable.
From the child at the desk to the institution
The developmental framing has implications beyond individual learners.
If metacognitive capacity develops, then environments help determine what opportunities people have to practise it. A classroom that only rewards correct answers may offer little reason to articulate uncertainty. A workplace that punishes revision may train people to conceal monitoring failures. A system that supplies answers before questions are formed may reduce some occasions for learners to notice what they do and do not know.
Those are Institute synthesis claims. Flavell did not provide a ready-made theory of AI-era institutional design.
But his work helps make a deeper question available: what kinds of environments help people become more accurate participants in their own cognition?
That question belongs to education, organizational learning and responsible technology design as much as to developmental psychology.
Metacognition is not introspective infallibility
One legacy risk is that the term itself can suggest a supervisory faculty sitting above thought, capable of inspecting cognition objectively.
Contemporary evidence does not support that picture.
Confidence can be biased. Judgments of learning can be miscalibrated. Self-report inventories can diverge from task-based measures. People can be confident when wrong and uncertain when right. Different metacognitive measures do not always agree.
Flavell's framework is compatible with fallibility. Metacognitive knowledge includes beliefs, and beliefs can be wrong. Metacognitive experiences are experiences, not guarantees.
The field becomes scientifically interesting precisely because monitoring varies in accuracy.
Education inherited both the promise and the problem
Metacognition became central to education because the idea offers something attractive: learners who can plan, monitor and regulate their own learning may become less dependent on external direction.
There is substantial evidence that metacognitive strategies and self-regulated learning processes matter. But educational language can easily convert metacognition into a virtue word.
“Be more metacognitive” is not an intervention.
Effective support may involve teaching strategy knowledge, prompting prediction, creating opportunities for delayed checking, helping learners interpret errors or structuring feedback so that monitoring has useful information to work with.
The mechanisms need to be specified.
This returns to Flavell's original precision. Metacognition is not one generic act of reflection. It involves different forms of knowledge, experience and strategic response.
The AI era changes the availability of cognitive evidence
When a person thinks without assistance, certain internal signals are hard to avoid. They know when recall is slow. They feel when a word is missing. They discover that they cannot yet explain a concept.
External systems can change those signals by supplying fluent output before the person reaches the point of detectable difficulty.
The Institute calls this a metacognitive evidence problem: if tools alter the situations in which people encounter their own uncertainty, they may alter the evidence available for self-monitoring.
This is a contemporary synthesis, not a conclusion from Flavell's developmental studies.
It suggests a research agenda rather than a policy answer. Some tools may improve monitoring by providing feedback, counterexamples and comparison. Others may mask gaps by making task performance smoother. The effect will depend on design and use.
What the evidence does not show
Flavell's work does not establish that metacognition is one unitary ability.
It does not show that people have direct, unbiased access to their own cognitive processes.
It does not show that knowing a strategy guarantees using it effectively.
It does not show that childhood developmental findings generalise unchanged to every adult domain or culture.
It does not validate any contemporary questionnaire merely because that questionnaire uses the word metacognition.
It does not make self-report equivalent to metacognitive accuracy.
And historical influence does not establish the evidential strength of every later claim made under the metacognition label.
Open questions left by a foundational programme
How does metacognitive knowledge become calibrated to experience? Which aspects of monitoring are domain-general and which are task-specific? How do social environments shape what people are willing to admit they do not know? How should metacognition be measured when eliciting a judgment can itself change performance? What forms of monitoring remain essential when external systems can provide answers, memory and decision support?
The field now has models and measures Flavell did not have.
It still lives inside the problem he helped name.
The child who predicts ten and remembers four
The useful scientific moment is not the mistake.
It is the possibility that prediction, outcome and later strategy can become connected.
A person can form a belief about their cognition. Reality can answer back. The belief can be revised. Action can change.
That loop is one of the roots from which contemporary metacognition research grew.
Flavell's legacy is not a slogan about thinking about thinking.
It is the decision to make that loop visible enough to study.
Research references
Flavell JH. Metacognition and cognitive monitoring: A new area of cognitive-developmental inquiry. American Psychologist. 1979;34(10):906–911. https://doi.org/10.1037/0003-066X.34.10.906
Flavell JH. The Developmental Psychology of Jean Piaget. Van Nostrand; 1963.
Flavell JH, Wellman HM. Metamemory. In: Kail RV, Hagen JW, eds. Perspectives on the Development of Memory and Cognition. Erlbaum; 1977.
Stanford University. John H. Flavell, influential scholar of developmental psychology, has died. Stanford Report. 2025. https://news.stanford.edu/stories/2025/06/john-flavell-developmental-psychology-obituary
Nelson TO, Narens L. Metamemory: A theoretical framework and new findings. In: Bower GH, ed. The Psychology of Learning and Motivation. Vol 26. Academic Press; 1990:125–173. https://doi.org/10.1016/S0079-7421(08)60053-5