The short answer
Ability is an underlying capacity that influences learning or performance. Knowledge is what a person understands and can draw on. Skill is learned control of an activity. Competence is the integrated performance of work to an adequate standard in context, with appropriate judgement and responsibility. Expertise is deep, domain-specific capability that supports consistently strong and adaptive performance.
These terms overlap because performance draws on all of them. They remain distinct because each supports a different claim. Knowing how a procedure works does not establish skill in carrying it out. Performing one rehearsed task does not establish competence across the situations in which the work matters. Holding a position for many years does not establish expertise.
Guidebeam uses capability as the umbrella term:
Guidebeam working definition: Capability is the potential and demonstrated power to perform and develop in a specified activity, under specified conditions.
This is Guidebeam’s working definition, not a universal scale. It joins occupational, educational, psychological and professional traditions that organise the vocabulary differently. The page on The Vocabulary of Work defines the tasks, practices, roles and occupations in which capability is developed and used. This page defines the resources and performance claims attached to the person-in-context.
The individual-differences construct map keeps ability distinct from preference, value, aptitude and motivation.
Five terms, five questions
| Term | Guidebeam working distinction | The question it answers | A common overclaim |
|---|---|---|---|
| Ability | A comparatively enduring capacity that influences how someone learns or performs | What underlying capacity is relevant here? | Treating a test score as fixed potential |
| Knowledge | Facts, concepts, principles, procedures and understanding available for use | What does the person know and understand? | Assuming recall guarantees application |
| Skill | A learned capacity to carry out an activity with control | What can the person do? | Calling every attribute a skill |
| Competence | Integrated, adequate performance to a relevant standard in context, including judgement and responsibility | Can the person do this work properly here? | Reducing competence to task completion |
| Expertise | Deep, domain-specific capability expressed in consistently strong and adaptive judgement and performance | Can the person handle the domain’s demanding and unfamiliar cases? | Using years, speed, title or reputation as proof |
The consequential misconception is simple: these terms do not form one universal ladder. A person can possess detailed knowledge without a practised skill, execute a narrow skill without wider competence, meet a competent standard without expert judgement, or hold expertise that an institution has never formally recognised.
Why the frameworks disagree
The vocabulary changes with its purpose. Psychology asks which attributes help explain learning and performance. Occupational systems describe what work requires. Education frameworks specify learning outcomes. Professional bodies and regulators decide what evidence or permission is needed for particular practice.
The US Department of Labor’s O*NET Content Model separates abilities under worker characteristics from skills and knowledge developed through education and experience. It then keeps experience, credentials, tasks and work context in further domains. The model is useful because it prevents one job requirement from silently standing in for all the others.
Qualification frameworks draw the boundaries differently. The Australian Qualifications Framework describes knowledge as what a graduate knows and understands, skills as what a graduate can do, and application as the use of both in contexts involving autonomy, responsibility and accountability. The European Qualifications Framework similarly places responsibility and autonomy beside knowledge and cognitive or practical skills.
Even competence has no single international meaning. Françoise Delamare Le Deist and Jonathan Winterton’s comparison of US and European traditions found functional, cognitive, behavioural and more holistic approaches. Some systems use competency for a behavioural attribute; others use it for a task standard or as a near-synonym for competence. The word needs a local definition before it can carry an assessment or hiring decision.
Guidebeam’s distinctions therefore work as a translation layer. They do not declare one framework correct for every purpose.
Ability, IQ and “natural talent”
Ability is easy to hear as “anything a person can do”. This page uses it more narrowly: a comparatively enduring underlying capacity that influences learning or performance. Reasoning is one kind, not the whole category. The US Department of Labor’s O*NET Content Model distinguishes cognitive abilities from psychomotor abilities such as manual dexterity and reaction time, physical abilities such as strength, endurance and balance, and sensory abilities involving vision, hearing and speech perception.
None of those capacities is capability in the abstract. Its relevance depends on the activity. Spatial orientation matters differently to a pilot, an electrician and a footballer. Manual steadiness may matter greatly for one procedure and little for another. Tools, training, technique and support can change how strongly an underlying capacity affects the performance that matters.
Where does IQ fit? An IQ is a score produced from a standardised intelligence test. Such tests combine performance across specified mental tasks, commonly involving reasoning, comprehension, problem-solving, memory or processing. The APA Dictionary of Psychology describes intelligence tests as standardised sets of mental, verbal and performance tasks. The testing standards used elsewhere in this article supply the boundary: the score supports only those interpretations for which there is appropriate evidence.
An IQ score can therefore be evidence about aspects of cognitive functioning under the test’s conditions. It is not direct evidence of everything a person knows, a practised skill, competent work, creativity, motivation, responsibility or future achievement. A high score does not establish that someone can perform a profession well. A lower score does not describe the full range of what that person can learn or do with relevant knowledge, time, tools and support. As with every assessment, the observed score and the wider capability claim must remain separate.
Sport makes the same point visible in the body. What looks like a single sporting gift may combine perception, reaction time, coordination, balance, strength or endurance with learned technique, tactical knowledge, practice, coaching, motivation, health and access to competition. A 2024 systematic review of talent research in racket sports grouped the evidence across anthropometric, physiological, technical, tactical and psychological characteristics and concluded that talent should be studied through multidimensional and longitudinal approaches.
“Natural talent” can name an observation: someone begins ahead, learns unusually quickly or performs strikingly well. It is not yet an explanation. The label does not reveal which capacities matter, how they developed, how much prior exposure went unnoticed, or whether the advantage will persist as the task and environment change. Turning “this person is ahead now” into “this person was born for it” replaces a question with a destiny.
The better capability questions are more exact. Which underlying capacity appears relevant? What has already become knowledge or skill? Under which conditions was the difference observed? What might practice, teaching, maturation, tools or support change? Ability can influence learning and performance without fixing either. On its own, it does not establish skill, competence or expertise.
The companion page on aptitude takes up the distinct future-looking question: what present evidence can responsibly support a prediction about learning or developing performance in a specified domain?
The companion page on intelligence and IQ separates the theoretical construct, the test instrument and the normed score, then examines prediction, heritability, culture and fairness in depth.
Capability is a relationship, not a ladder
The same person can present a different capability profile across tasks and conditions. An experienced bookkeeper may understand accounting principles and use one system fluently, then need time to learn a new platform. A trainee can show strong reasoning ability and incomplete procedural knowledge. A craft worker can make expert judgements that a written test captures poorly. A qualified graduate can still need supervised participation before taking responsibility for a difficult case.
Several dimensions can develop at different rates:
- underlying capacities relevant to the work;
- breadth, depth and organisation of knowledge;
- fluency and control in particular activities;
- performance across routine, complex and unfamiliar conditions;
- autonomy, judgement and responsibility;
- access to tools, support and people;
- evidence and social recognition.
A promotion may increase authority without deepening technical skill. Repetition may increase speed without improving diagnosis. Formal study may expand knowledge before it changes performance. An expert may remain dependent on a team, instrument or information system because expertise often includes knowing how to use those resources well.
Capability · Resources, performance and context
Capability vocabulary works as a relationship—not a ladder.
Ability, knowledge and learned skill contribute to performance. Competence and expertise make stronger claims in specified contexts. Evidence and recognition remain separate.
Resources
Specified work situation
Performance claims
Separate layer · Evidence and recognition
Qualification · test result · licence · title · reputation
These may support or authorise a claim. They are not the capability itself.
A specified work situation sits at the centre. Ability and knowledge are resources, practice develops skill, competence integrates resources against a standard and expertise adds consistently strong adaptive judgement. Tools, colleagues, supervision, physical conditions and rules surround performance. Qualifications, test results and titles remain on a separate evidence and recognition layer.
How capability develops
Education, practice, apprenticeship, feedback and experience make different contributions.
Education can organise knowledge. It gives names to patterns, explains why a procedure works, supplies cases beyond one person’s experience and creates opportunities to compare alternatives. The AQF treats knowledge, skills and their application as connected learning outcomes while keeping them analytically separate. Completing a course can establish selected outcomes without proving performance in every future workplace.
Practice changes performance. Repeated activity can improve control, accuracy, fluency and the ability to notice errors. Practice quality matters. K. Anders Ericsson, Ralf Krampe and Clemens Tesch-Römer used deliberate practice for structured, effortful activity directed at improving a specific aspect of performance, typically with goals and informative feedback. Their 1993 account was never a claim that any 10,000 hours automatically creates an expert.
Feedback makes error available for learning. A result, teacher, supervisor, colleague, client or instrument can reveal a gap between intended and actual performance. Useful feedback is specific enough to change the next attempt. Repetition without a trustworthy signal can stabilise a poor method as readily as a good one.
Apprenticeship gives access to the practice itself. Jean Lave and Etienne Wenger described learning as participation in a community of practitioners, with newcomers moving towards fuller participation. Allan Collins, John Seely Brown and Susan Newman’s model of cognitive apprenticeship makes expert processes visible through modelling and coaching, then adjusts support through scaffolding, articulation, reflection and exploration.
Supervised participation contributes something a classroom or isolated exercise may not. It reveals local standards, timing, hand-offs, consequences and the moment when a rule no longer fits cleanly. Responsibility can increase as the learner demonstrates performance, while the supervisor remains accountable for work that the learner is not yet ready to carry alone.
Experience supplies encounters. It may broaden the range of cases a person has seen and deepen contextual knowledge. Elapsed time remains a weak description of what was learned. Two people can spend the same years in a field while receiving different tasks, feedback, teaching, permission to act and opportunities to reflect.
The contribution of deliberate practice is a live research dispute. Brooke Macnamara, David Hambrick and Frederick Oswald’s 2014 meta-analysis found that the association between deliberate practice and performance differed sharply by domain and left substantial variation unexplained. Ericsson replied that many included activities did not meet his narrower definition. The responsible conclusion is modest: structured practice and feedback can improve performance; accumulated hours do not guarantee expertise; and development also depends on prior capability, teaching, opportunity, motivation, task structure and the environment.
Competence includes judgement and responsibility
Task completion is observable. Competence asks whether the work was performed to the relevant standard, under the conditions that matter, with consequences handled properly.
Consider a correct calculation. In one setting, correctness may be enough. In another, the worker must choose whether the calculation applies, recognise missing information, explain uncertainty, protect confidential data and know when to seek review. The arithmetic can remain the same while the competent performance changes.
This is why both the AQF and EQF include autonomy and responsibility in their level descriptions. The ILO’s cross-system skills glossary describes competence as demonstrated capacity to use knowledge, know-how and skills in usual and changing occupational situations.
A competence claim depends on five parts:
- the work or outcome;
- the expected standard;
- the conditions, tools and support;
- the range of routine and non-routine situations;
- the autonomy, judgement and responsibility involved.
“Competent” is always competent for something, in some conditions, to some standard. A narrow unit of competence and broad professional competence are different claims.
Expertise is more than speed, completion or years
Experts often work efficiently because familiar patterns no longer demand the same conscious effort. Speed can also come from cutting checks, repeating one case or working only where conditions are predictable. It is useful evidence only when speed belongs to the performance the domain values.
The expert-performance tradition defines its subject through strong, reproducible performance on tasks that represent the domain. K. Anders Ericsson and Neil Charness argued for representative tasks and consistently superior performance rather than selecting experts by reputation and then assuming their performance is superior.
Work domains complicate that clean laboratory test. Outcomes may be delayed, cases differ in difficulty, quality may be disputed and success may depend on a team. Expert performance can therefore require several forms of evidence: difficult work samples, decisions across cases, error detection, peer scrutiny, outcomes adjusted for context and the ability to explain or revise a judgement.
Expertise is also bounded. A surgeon’s expertise in one procedure does not confer expertise in hospital administration. A senior software engineer may be expert in one architecture and a novice in a new safety domain. Deep patterns can transfer where structures are shared, but occupational labels alone do not establish transfer.
Social recognition is another layer. A title, reputation, credential or invitation to advise may identify people worth examining. Recognition can lag behind capability, especially when access to institutions is unequal. It can also exceed current capability when status rests on an old achievement or untested authority. The strongest claim of expertise joins recognition with current performance evidence and clear domain boundaries.
Tacit knowledge and situated performance
Michael Polanyi’s account of tacit knowing begins from the idea that people can know more than they can fully state. A practitioner may notice that a sound is wrong, a conversation is turning or a material is near failure before they can give a complete rule for the judgement.
Tacit knowledge does not make performance mysterious or immune to scrutiny. Demonstration can expose sequence and attention. A learner can ask what the expert noticed. Comparison across cases can reveal cues and exceptions. Cognitive apprenticeship uses articulation and reflection to make some previously hidden reasoning available. Some residue may remain embodied, perceptual or difficult to verbalise.
Context is part of the performance. Tools, language, team routines, physical layout, time pressure, local risks and the person receiving the service can all change what successful action requires. Situated learning explains why becoming capable often includes learning how a community coordinates and judges its work, not only absorbing portable facts.
The practical lesson is to distrust both extremes. Capability is not a context-free possession that appears identically everywhere. Nor is it trapped so completely in one setting that nothing can transfer. The question is what remains similar enough for knowledge and skill to be used, and what must be learned again.
Transfer, decay and obsolescence
“Transferable skill” often sounds like a permanent passport. Transfer is better treated as an empirical question about distance between situations.
Susan Barnett and Stephen Ceci’s taxonomy of far transfer distinguishes changes in knowledge domain, physical and social context, time, function and other dimensions. Evidence for transfer is substantial under some conditions and unresolved under others. A planning method may transfer between projects that share a decision structure even when their subjects differ. A technique that depends on one machine, regulation or team convention may travel poorly.
Capability can also weaken. Winfred Arthur and colleagues’ quantitative review of skill retention found substantial loss after non-use, with the amount varying by task type, retention interval, retrieval conditions and how performance was evaluated. Recognition can outlive current readiness. A licence may remain current while a rarely used procedure requires refresher training; conceptual knowledge may remain while speed and precise execution decline.
Obsolescence is different from forgetting. The person may retain the old capability while tools, standards, markets or problems change around it. Some components survive: causal understanding, judgement, client knowledge or the ability to learn an adjacent method. Others lose value because the work no longer calls for them.
For someone deciding what to develop next, three questions matter:
- Which capabilities are being deepened through current work?
- Which are going unused and may need maintenance?
- Which parts remain useful when the task, tool or occupation changes?
That makes development more precise than collecting a list of “future-proof skills”.
Capability and evidence of capability
A person’s capability and the evidence available about it are different things.
The AQF defines an AQF qualification as formal certification that a graduate has achieved the learning outcomes of an accredited complete programme. That is a meaningful claim about assessed learning. It does not say that every graduate will perform every associated job competently without induction, practice or local support.
Credentials also do different institutional jobs. The US Bureau of Labor Statistics distinguishes educational certificates, professional certifications and licences. A certification attests to a specified occupational capability under the issuing body’s method. A licence or registration conveys legal authority to practise where the jurisdiction requires it. Permission matters even when capability is strong; permission does not demonstrate the whole of capability.
Other evidence can include work products, demonstrations, references, portfolios, employment history, peer recognition and outcomes. Each has a scope and a failure mode. A polished portfolio may omit routine reliability. A reference may reflect opportunity or affinity. A knowledge test may under-sample practical judgement. A live demonstration may be distorted by unfamiliar equipment or anxiety.
Certifications, Licences and the Fragmenting Credential examines issuers, evidence, renewal and legal consequence in depth. The durable distinction here is that a credential is a claim about learning, performance, identity or permission. It is not the person’s capability itself.
What assessment observes and what it infers
Assessment directly records selected behaviour or products under defined conditions. It may record answers, decisions, movements, code, a repaired object, a role-play, a supervisor rating or an outcome. From that sample, someone interprets broader knowledge, ability, skill, competence, expertise or likely future performance.
| Evidence source | What is observed more directly | What is inferred beyond the sample |
|---|---|---|
| Knowledge test | Responses to selected questions | Breadth, organisation and usable understanding elsewhere |
| Ability measure | Performance on specified items under standard conditions | An underlying capacity and its relevance to learning or work |
| Work sample or simulation | Performance on selected tasks with stated tools and constraints | Performance across the wider job, new cases and time |
| Workplace observation | Behaviour in one or more real situations | Consistency, transfer and independent performance elsewhere |
| Portfolio or work history | Chosen outputs and recorded participation | The person’s contribution, current proficiency and unseen routine work |
| Qualification, certification or licence | Completion, assessment or permission recorded by an issuer | Current capability beyond the credential’s actual standard and scope |
The Standards for Educational and Psychological Testing treats validity as evidence supporting an interpretation of scores for an intended use. A measure that supports one inference does not automatically support another. Evidence that helps place a learner in a course may be too narrow for a high-stakes employment decision. A result established in one population or language may need further evidence elsewhere.
What Is Psychometrics? explains how observations become scores and how scores support limited inferences. A capability result becomes easier to understand when it names the target, samples representative work, records the conditions, separates observation from inference and states the uncertainty.
Tools, accommodations and environmental support
Performance belongs to a person in an environment. The World Health Organization’s ICF includes products and technology, physical settings, support, attitudes, services and systems as environmental factors that can facilitate or obstruct functioning.
The question “Can the person do it?” is incomplete without the conditions. A screen reader, adapted control, checklist, lifting device, quiet room, flexible sequence, colleague or AI assistant can change observed performance. So can poor lighting, inaccessible software, hostile supervision or an unnecessary time limit.
Accommodation does not mean lowering every standard. The UN disability-rights committee describes reasonable accommodation at work as individualised modification, adjustment or support enabling a person to perform the inherent requirements of work on an equal basis. The essential question is which requirement belongs to the work and which barrier belongs only to the way access or assessment was designed.
These conditions change the meaning of the result. If the normal job permits calculators, reference material, assistive technology or team consultation, an unaided assessment may measure a different capability from the one used at work. When rapid unaided action is genuinely inherent to safety, that condition needs a defensible reason and appropriate evidence.
What the distinctions change for a career decision
What Is a Career? describes a trajectory through work, learning, relationships and institutions over time. For a person considering their next move, the capability vocabulary replaces the blunt question “Am I qualified?” with a more useful profile:
- What underlying capacities appear relevant, and how certain is that inference?
- What knowledge is already organised and usable?
- Which activities have become skills through practice?
- In which conditions has competent performance been demonstrated?
- Where is judgement deepening towards expertise?
- What can be learned, transferred, refreshed or supported?
- What evidence or formal permission is still required?
This profile connects directly to What Is Career Fit?. A current gap can be a learnable skill, a missing credential, an inaccessible route, an environmental barrier or a genuine mismatch with an inherent demand. Those conclusions lead to different decisions.
The same distinctions expose what can go wrong in hiring. Decisions are more informative when the work is described before a proxy is chosen: its tasks, decisions, conditions, consequences and responsibility. Some requirements are needed on entry; others can be learned through induction or supervised participation. A degree, years-of-experience threshold or familiar job title may be convenient while excluding people who could demonstrate the work more directly.
The distinction also protects identity. Work and Identity explains why institutional recognition can make a practice socially legible without exhausting the person. Someone who loses a role, licence or physical capacity may retain knowledge, judgement and an occupational identity; someone who gains a title may still be developing competence.
What AI changes, and what it does not settle
AI can perform, accelerate or reshape some activities that previously supplied evidence of human skill. It can also become a tool through which a person performs. The result is neither automatic replacement nor proof that capability no longer matters.
The ILO’s 2025 global index of generative-AI exposure scores tasks within occupations and keeps exposure separate from employment outcome. Because most occupations contain tasks that still require human input, the study identifies job transformation as the more likely broad effect. Particular tasks can still be automated, and effects will differ across occupations and institutions.
Four layers keep the new arrangement visible:
- Human performance without the tool matters where unaided action is inherent or foundational learning is being assessed.
- Human performance with the tool includes selecting, directing, checking and integrating it as part of competent work.
- System performance is the result produced by the person, AI, data, workflow and organisation together.
- Human responsibility remains with whoever is answerable for judgement, harm, escalation and correction.
AI can raise output while concealing a learning gap. It can also remove routine work that once gave newcomers practice, feedback and access to harder cases. When organisations automate those entry tasks, another route is needed for supervised participation and the development of contextual judgement. Senior expertise will not reproduce itself automatically after its learning pathway has been removed.
Whether AI assistance was allowed, required or prohibited changes what an assessment result means. A polished AI-supported output does not prove every underlying human capability. An obsolete unaided method may be equally misleading when the real work is performed responsibly with tools.
The durable unit is responsible performance in context. AI changes the distribution of tasks and evidence. It does not by itself decide who is competent, who is an expert or what a career should become.
The vocabulary of capability, tested
The working distinctions can now be read together:
- Ability names a relevant underlying capacity. It influences performance and learning without fixing a person’s future.
- Knowledge names what the person understands and can draw on. Its value appears in use, connection and revision as well as recall.
- Skill names learned control of an activity. Its scope depends on the task and conditions in which it has been practised.
- Competence names integrated, adequate and responsible performance to a stated standard in context.
- Expertise names deep, domain-bounded capability expressed through consistently strong and adaptive judgement and performance.
Capability holds these together with development, tools, support and conditions. Evidence of capability makes selected claims visible. Recognition makes some of those claims socially or institutionally consequential. Keeping the three layers separate shows what is known, what is inferred and what could still be developed.
Sources and further reading
- US Department of Labor, National Center for O*NET Development. O*NET Content Model.
- Australian Qualifications Framework Council. Australian Qualifications Framework, second edition. 2013.
- Council of the European Union. Recommendation on the European Qualifications Framework. 2017.
- International Labour Organization. Glossary of Skills and Labour Migration.
- Delamare Le Deist, Françoise, and Jonathan Winterton. “What Is Competence?” Human Resource Development International 8 (2005).
- Lave, Jean, and Etienne Wenger. Situated Learning: Legitimate Peripheral Participation. Cambridge University Press, 1991.
- Collins, Allan, John Seely Brown, and Susan E. Newman. “Cognitive Apprenticeship: Teaching the Crafts of Reading, Writing, and Mathematics.” Technical report, 1987; revised chapter published 1989.
- Ericsson, K. Anders, Ralf Krampe, and Clemens Tesch-Römer. “The Role of Deliberate Practice in the Acquisition of Expert Performance.” Psychological Review 100 (1993).
- Ericsson, K. Anders, and Neil Charness. “Expert Performance: Its Structure and Acquisition.” American Psychologist 49 (1994).
- Macnamara, Brooke, David Hambrick, and Frederick Oswald. “Deliberate Practice and Performance in Music, Games, Sports, Education, and Professions.” Psychological Science 25 (2014).
- Ericsson, K. Anders. “Summing Up Hours of Any Type of Practice Versus Identifying Optimal Practice Activities.” Perspectives on Psychological Science 11 (2016).
- Polanyi, Michael. The Tacit Dimension. 1966.
- Barnett, Susan, and Stephen Ceci. “When and Where Do We Apply What We Learn? A Taxonomy for Far Transfer.” Psychological Bulletin 128 (2002).
- Arthur, Winfred, Winston Bennett Jr, Pamela Stanush, and Theresa McNelly. “Factors That Influence Skill Decay and Retention.” Human Performance 11 (1998).
- AERA, APA and NCME. Standards for Educational and Psychological Testing. 2014.
- American Psychological Association. “Intelligence Test.” APA Dictionary of Psychology.
- Nijenhuis, Sebastiaan B., Till Koopmann, Jesper Mulder, Marije T. Elferink-Gemser, and Irene R. Faber. “Multidimensional and Longitudinal Approaches in Talent Identification and Development in Racket Sports: A Systematic Review.” Sports Medicine - Open 10 (2024).
- World Health Organization. International Classification of Functioning, Disability and Health.
- UN Committee on the Rights of Persons with Disabilities. General Comment No. 8 on the right to work and employment. 2022.
- US Bureau of Labor Statistics. Credentials and Occupational Outlook Handbook methodology.
- Gmyrek, Pawel, and colleagues. Generative AI and Jobs: A Refined Global Index of Occupational Exposure. ILO, 2025.