A student can leave school knowing many things and still be poorly prepared for the next thing they have never seen before.
That problem sits behind the phrase learning to learn.
Learning to learn refers broadly to the capacities, beliefs, strategies and forms of regulation that help people approach new learning effectively. It appears in educational psychology, assessment, lifelong-learning policy and competence frameworks. It has particular importance in Europe and Finland, where researchers and public institutions have spent decades trying to define and assess it.
The breadth is both its value and its scientific difficulty.
Learning to learn is not one clean latent variable hiding behind every successful learner.
Finland gives the construct an unusually long empirical history
The University of Helsinki's Centre for Educational Assessment reports that it has researched and developed learning-to-learn assessment since 1996, initially at the request of Finland's National Board of Education.
Its framework combines cognitive tasks with attitudes and beliefs related to learning. The Finnish work treats learning to learn as more than strategy recall. It includes thought and problem-solving skills as well as conceptions that guide engagement with learning.
The 2002 Assessing Learning-to-Learn: A Framework by Jarkko Hautamäki and colleagues situates the construct inside educational outcomes and societal capacity building.
This history gives the Institute a legitimate Nordic evidence base to engage with. It should not be turned into a claim that Finland has solved the measurement problem.
Europe treats learning to learn as a competence umbrella
European policy has also given the term a formal role.
The Joint Research Centre's LifeComp framework develops the European Union's personal, social and learning-to-learn competence area. In LifeComp, the learning-to-learn area includes growth mindset, critical thinking and managing learning.
The JRC explicitly describes LifeComp as a conceptual, non-prescriptive framework rather than a diagnostic psychological instrument.
That distinction is important.
A policy competence framework organises educational aims. It is not automatically evidence that the listed elements form one psychometrically coherent trait.
The Institute should preserve the difference between normative competence architecture and empirical construct validation.
The phrase bridges cognitive and sociocultural traditions
A 2008 European Commission report by Ulf Fredriksson and Bryony Hoskins reviewed attempts to define and measure learning to learn and highlighted different understandings emerging from cognitive-psychological and sociocultural paradigms.
That pluralism remains useful.
From a cognitive perspective, learning to learn can involve metacognitive knowledge, monitoring, strategy use, problem solving and regulation. From sociocultural perspectives, learning is also participation in practices, access to tools, relationships and cultural expectations.
A learner's capacity is therefore not separable from the environment in which it must operate.
Someone can possess effective strategies and still fail in a context that withholds feedback, time, safety or access.
Learning to learn is broader than metacognition
Metacognition is central but not sufficient.
A person may know which strategy would improve retention and accurately monitor weak understanding, yet fail to regulate learning because of low motivation, anxiety, exhaustion or competing demands.
Self-regulated learning frameworks make this broader architecture explicit by including motivational, behavioral and affective processes alongside metacognition.
Learning to learn sits at an even broader level when used as a lifelong-learning competence. It can include willingness to engage, transfer across contexts, help seeking, collaboration and adaptation to new learning environments.
The exact scope depends on the framework.
That is why a citation must specify which version of "learning to learn" is being discussed.
It is not the same as intelligence
One attraction of learning to learn is that it names something education can cultivate rather than treating performance as a fixed property.
But the construct overlaps with cognitive ability, prior knowledge, motivation and executive function. Measuring it without simply reproducing those neighboring constructs is difficult.
The 2008 EU report explicitly examined what learning to learn is not, including intelligence, problem solving and learning strategies considered separately.
The point is not that these factors are irrelevant. It is that learning to learn is intended to describe how they are coordinated in learning, not to rename any one of them.
Any strong assessment therefore needs evidence of incremental validity and a clear construct model.
Transfer is the hard test
A learner may master an effective strategy in one subject and fail to use it elsewhere.
This is the classic transfer problem.
Learning to learn becomes most meaningful when a person can recognize structural similarities and differences across tasks, select strategies conditionally and adapt when old routines fail. Teaching one strategy is easier than demonstrating that the learner will transfer the underlying principle to a new domain.
This matters for programmes that promise generalized "learning agility."
Near transfer to a similar task and far transfer to a different domain are not equivalent outcomes. Claims should specify what moved, how far, and under what support.
Assessment can easily become circular
If learning to learn is defined as whatever successful learners do, then a test of successful learning risks becoming a test of the definition itself.
Strong assessment needs independent components and explicit theory.
The Finnish tradition uses specially designed tasks alongside attitudinal items. European work has explored multiple national assessment approaches. These efforts show that the construct can be operationalised, but also that no single instrument exhausts it.
Questionnaires capture beliefs and reported behavior. Cognitive tasks capture performance under specific conditions. Logs can capture study choices. Interviews can reveal strategy knowledge.
The methods answer different questions.
A composite score is only as meaningful as the model that justifies combining them.
Digital and AI environments change what learners need to regulate
Learning to learn was important before generative AI. AI changes its object.
Learners now need to decide when to retrieve from memory, when to ask a tool, how to verify generated material, how much productive difficulty to preserve, and how to build knowledge that remains usable without immediate external assistance.
These are plausible extensions of metacognitive and self-regulated learning problems. Evidence directly establishing a new, universal "AI learning-to-learn competence" is not yet mature.
The Institute can contribute by treating this as a research agenda rather than branding an old construct with new technology language.
A key question is whether tools increase access to feedback and explanation while inadvertently removing forms of practice needed for durable learning.
The educational system is part of the learner
Learning-to-learn discourse can become individualising: if a person struggles, the conclusion is that they lack resilience, strategy or motivation.
The broader literature gives reasons to resist that move.
Learning is shaped by instruction quality, curriculum, assessment, language, social support and material conditions. A learner's ability to regulate depends on the opportunities available for regulation.
At institutional scale, learning to learn therefore becomes partly a design question: does the system teach strategies explicitly, provide feedback, allow revision, make uncertainty safe, and help learners understand why a strategy works?
The learner matters. So does the architecture around the learner.
Strategy instruction works best when learners understand the mechanism
A learner can follow a technique because a teacher requires it without understanding when the technique will transfer.
Learning-to-learn education becomes deeper when strategy use is paired with causal explanation: why retrieval differs from rereading, why spacing changes later accessibility, why worked examples help novices, why feedback timing matters.
Mechanistic understanding can support conditional selection rather than ritual compliance.
This is also where metacognitive knowledge and learning to learn overlap most clearly. The learner builds a model of how learning works and uses that model to choose actions.
Lifelong learning changes the institutional horizon
School assessment often asks whether a student learned a defined curriculum under known conditions.
Learning to learn asks what happens after the curriculum ends.
Adults encounter new tools, occupations, health information, civic problems and technologies. They may need to learn without a teacher, identify credible sources, create practice, request feedback and revise strategies.
That makes learning to learn relevant to workforce and civic policy, but it also makes measurement harder. Long-term adaptive learning cannot be reduced to one school-age test.
European competence frameworks are valuable partly because they keep this broader horizon visible.
Cultural and linguistic context can change the meaning of an assessment
A task that appears to measure strategy may also measure familiarity with test conventions, language proficiency or schooling norms. An attitude item can carry different social meanings across countries.
Nordic evidence is therefore most useful when it is treated as located evidence, not exported as universal by default.
The Finnish tradition offers unusually rich longitudinal and institutional experience. Replication and adaptation elsewhere remain empirical tasks.
For the Institute's Northern European authority ambition, this is a strength: credibility comes from knowing what regional evidence can support and where its boundaries lie.
Learning to learn is partly learning when not to use a familiar strategy
Adaptive learners do more than accumulate techniques.
They abandon strategies that no longer fit.
A method that worked in school may fail in professional practice. A note-taking routine may become inefficient when material changes. An AI workflow that speeds drafting may weaken independent recall when the goal shifts from production to mastery.
This ability to revise the strategy itself is one of the construct's deepest features.
It turns learning to learn from "more effective study" into a capacity for updating the process of learning as conditions change.
Learning-to-learn claims need time horizons
A strategy can improve performance today and weaken independent performance tomorrow. Another can slow initial work while building a representation that transfers later.
Learning to learn therefore cannot be evaluated only through immediate task completion.
Relevant outcomes may include delayed retention, transfer, ability to choose strategies without prompts, persistence under new conditions and quality of help seeking. Different programmes may succeed on one horizon and fail on another.
This becomes especially important with intelligent tools. Assistance can raise current performance so dramatically that it obscures whether the learner has become more capable when assistance is absent, reduced or changed.
Assessment should include adaptation, not only possession
If the construct concerns learning under novelty, fixed familiar tasks capture only part of it.
One promising assessment logic is to observe how learners respond when the first strategy stops working: do they notice, seek information, revise and transfer? Such tasks are harder to standardize, but they better represent adaptation than static self-report alone.
The evidence base for any particular adaptive assessment still requires validation. The principle is conceptual: a competence about learning should reveal something about change in response to learning conditions.
Policy frameworks and research constructs can inform one another without merging
LifeComp can identify socially valued competence areas. Psychological research can test mechanisms and measurement. Educational systems can study implementation.
Confusion begins when a policy category is cited as if it were proof of a mechanism, or a laboratory effect is treated as if it settles what education should value.
The Knowledge Library should make those layers legible rather than flatten them.
What the evidence does not show
Learning to learn is not one universally agreed psychological construct.
European competence frameworks are not, by themselves, psychometric validation studies.
Finnish assessment traditions do not establish that one measure captures learning-to-learn capacity across all cultures, ages or domains.
Learning to learn is not interchangeable with intelligence, metacognition, executive function, motivation or self-regulated learning.
Teaching a strategy in one context does not prove far transfer.
A questionnaire about being a good learner does not demonstrate effective regulation in real tasks.
And the existence of AI tools does not yet establish a validated new learning-to-learn capacity specific to generative AI.
Open questions
Which components of learning to learn transfer most reliably across subjects? How should cognitive performance, strategy knowledge, motivation and learning beliefs be combined without creating an opaque composite? Which aspects are teachable at scale? How do age, language and culture change the meaning of assessment items?
For Nordic and European research, the historical data create an opportunity to study change over time without assuming that the construct is timeless. For AI-era education, the challenge is to distinguish helpful cognitive offloading from premature outsourcing of learning.
The most difficult question may be institutional: how can education cultivate adaptive learners without turning adaptability into a demand that individuals compensate for poorly designed systems?
The next unfamiliar problem
A durable education cannot predict every problem a person will meet.
It can, however, shape what happens when the unfamiliar arrives.
Does the learner know how to diagnose the task? Can they notice what they do not understand? Do they possess more than one strategy? Can they seek feedback? Can they revise a plan? Can they tell the difference between ease and learning?
At the individual level, these are capacities for adaptation.
At the institutional level, they are outcomes worth designing for.
At the European level, they have become part of a long-running competence conversation rather than a passing educational slogan.
Learning to learn should remain broad enough to reflect real learning and precise enough not to mean everything good about education.
Its seriousness begins where the slogan ends.
Research references
- Hautamäki J, Arinen P, Niemivirta MJ, Eronen S, Hautamäki A, Kupiainen S, Lindblom B, Pakaslahti L, Rantanen P, Scheinin P. Assessing Learning-to-Learn: A Framework. Helsinki: Opetushallitus; 2002. ISBN 952-13-1423-0. https://researchportal.helsinki.fi/en/publications/assessing-learning-to-learn-a-framework/
- University of Helsinki Centre for Educational Assessment. Learning to learn. Current institutional overview of Finnish assessment work begun in 1996. https://www.helsinki.fi/en/centre-educational-assessment/assessment-and-research/learning-learn
- Fredriksson U, Hoskins B. Learning to learn: what is it and can it be measured? Luxembourg: Office for Official Publications of the European Communities; 2008. https://doi.org/10.2788/83908
- Sala A, Cabrera Giraldez M. LifeComp: A European competence framework for better lives in our uncertain world. European Commission Joint Research Centre; 2022. https://doi.org/10.2760/491876
- Panadero E. A Review of Self-regulated Learning: Six Models and Four Directions for Research. Frontiers in Psychology. 2017;8:422. https://doi.org/10.3389/fpsyg.2017.00422
- Veenman MVJ, van Hout-Wolters BHAM, Afflerbach P. Metacognition and learning: conceptual and methodological considerations. Metacognition and Learning. 2006;1:3-14. https://doi.org/10.1007/s11409-006-6893-0
- Järvelä S, Nguyen A, Hadwin A. Human and artificial intelligence collaboration for socially shared regulation in learning. British Journal of Educational Technology. 2023;54:1057-1076. https://doi.org/10.1111/bjet.13325