Abilities refer to the underlying traits or skills a person possesses, while capabilities describe what someone can actually do with those abilities in a given context. The distinction sounds subtle, and in everyday conversation the two words are often swapped without anyone noticing. But across psychology, military strategy, workplace health, artificial intelligence, and human development, the difference carries real weight. An ability is something you have; a capability is something you can deploy when circumstances allow.
Why the Two Words Feel Interchangeable
In casual English, saying “she has the ability to run a marathon” and “she has the capability to run a marathon” land the same way. Dictionaries reinforce this overlap by defining each word partly in terms of the other. The confusion is understandable because the concepts genuinely share territory: both involve what a person, machine, or organization can do. The meaningful split emerges when you add context, conditions, and constraints. A person might have the raw physical ability to lift heavy objects but lack the capability to do so at work because of an injury, a poorly designed workstation, or company policy. The ability hasn’t vanished; the capability has. This context-dependence is where nearly every specialized field draws the line.
Abilities as Stable Traits
Psychologists treat abilities as relatively fixed internal attributes. Cognitive abilities, including general intelligence, reasoning, working memory, and processing speed, rank among the most stable psychological traits researchers have measured. A large meta-analysis drawing on over 200 longitudinal studies and more than 87,000 participants found that the rank-order stability of cognitive abilities at age 20, measured again five years later, averaged around 0.76 on a zero-to-one scale. That is remarkably consistent: people who score high relative to their peers at one time point tend to score high years later.1PubMed Central. The stability of cognitive abilities: A meta-analytic review of longitudinal studies
Physical abilities follow a similar logic. Aerobic capacity, often measured by VOâ‚‚ max, is treated as a core physical ability that varies between individuals and across sports. Research comparing athletes in different disciplines has found that peak aerobic capacity values differ by sport, with endurance-heavy sports demanding higher baselines than those relying more on bursts of power.2PubMed Central. Aerobic capacity as an indicator in different kinds of sports The ability itself is partly genetic and partly trained, but it lives inside the individual. Whether that ability translates into match-day performance depends on a cascade of additional factors, from injury status to altitude to mental state.
This framing matters because it means abilities, cognitive or physical, are something you carry with you across situations. They shift slowly, driven by aging, training, or neurological change. Capabilities, as we’ll see, shift much faster because they depend on what surrounds you as much as what’s inside you.
Capabilities as Context-Dependent Power
In occupational health, the distinction between ability and capability is baked into the way researchers assess whether someone is fit for a particular job. Work ability has been defined as an individual’s capacity to match job demands according to their physical and mental conditions and work circumstances.3PubMed Central. Relation of the work ability index to fitness for work in healthcare and public employees in a region of Northeastern Italy Notice the phrasing: it’s not just about what the worker can do in a vacuum, but whether their internal condition meets what the specific job demands. Two people with identical physical strength might have different work capabilities if one works on a loading dock and the other sits at a desk.
An interesting finding from occupational research makes this concrete. A study of male nursing staff in Iran measured both their physical work capacity (an objective lab measure of what their body could produce) and their Work Ability Index (a broader score reflecting how capable they felt in their specific role). The two showed no statistically significant relationship.4PubMed Central. A Study on Work Ability Index and Physical Work Capacity on the Base of Fax Equation VO 2 Max in Male Nursing Hospital Staff in Isfahan, Iran In other words, knowing someone’s raw physical ability didn’t reliably predict how capable they considered themselves in their day-to-day job. That’s the gap between ability and capability in action: the lab measure captures what the body can do, while the workplace measure captures what the whole person, in that environment, with those demands, can actually pull off.
How Military Planners Think About Capability
Military strategy offers one of the most precise definitions of capability in any field. Defense planning literature defines capability as the power to achieve a desired operational effect in a specific environment within a specific time, and to sustain that effect for a designated period.5Security and Defence Quarterly. Military capabilities and the strategic planning conundrum That definition packs in environment, time, and sustainability, all context variables. A tank division has the ability to move across terrain and fire projectiles. Whether it has the capability to secure a particular city depends on logistics chains, communication networks, intelligence, personnel readiness, and dozens of other factors.
The military also acknowledges something worth noting: capability can include “undeveloped or unused faculty,” a latent potential that hasn’t been activated yet. This sense of capability as dormant power waiting for the right conditions to surface recurs across multiple disciplines, from AI research to human development. It’s a reminder that capability isn’t always visible. Sometimes the capacity to do something exists but hasn’t been tested, triggered, or measured.
The Same Split in Linguistics
Linguists stumbled onto a parallel distinction decades ago. Noam Chomsky famously separated “competence” from “performance.” Competence is a person’s internalized knowledge of their language’s rules, the mental grammar that lets you construct and understand sentences you’ve never heard before. Performance is what actually comes out of your mouth in real conversation, complete with stumbles, false starts, and interruptions. As one summary of Chomsky’s framework puts it, competence is one’s capacity to use language, while performance is the actual application of that capacity in speaking and listening.6ResearchGate. The Relationship between Competence and Performance: Towards a Comprehensive TG Grammar
Swap “competence” for “ability” and “performance” for “capability in context” and the parallel is almost exact. You have the linguistic ability to produce a grammatically perfect sentence. Whether you actually do, in a noisy bar, while tired, with a word on the tip of your tongue, depends on performance conditions. The ability remains even when the capability to express it falters. Anyone who has ever struggled to say something they knew perfectly well has felt this distinction viscerally.
Where AI Capability Gets Tricky
The ability-versus-capability question has become surprisingly important in artificial intelligence, though AI researchers tend to use the terms differently than psychologists or military planners do. When AI researchers talk about a model’s “capabilities,” they generally mean the full range of tasks the system can successfully perform. But a persistent and unsettling finding in recent AI research is that models often have latent capabilities that aren’t apparent until someone discovers the right way to unlock them. Large language models frequently possess capabilities that lie dormant unless explicitly elicited through fine-tuning or specific prompting techniques.7NeurIPS Proceedings. Large Language Models and Latent Capabilities
This matters for AI safety because evaluators need to know what a model can actually do, not just what it appears to do on standard benchmarks. Previously hidden capabilities have been drawn out of models long after their initial release, which creates a moving target for regulators trying to assess risk.8arXiv. The Elicitation Game: Evaluating Capability Elicitation Techniques In human terms, it’s as if you tested a job candidate on ten tasks, deemed them incapable of task eleven, and then discovered months later that a different set of instructions prompted them to perform it perfectly. The ability was always there; it just hadn’t been surfaced.
A related problem is that high benchmark scores don’t necessarily reflect genuine underlying ability. AI systems have exceeded human performance on many tests designed to measure general cognitive capacities, yet benchmark performance often does a poor job of predicting how well those systems handle real-world tasks.9AI Magazine. Six principles for evaluating cognitive capabilities in AI models The analogy to human test-taking is striking. A student might ace a standardized exam (demonstrating ability on a narrow measure) while struggling to apply that knowledge in an unfamiliar situation (lacking real-world capability). For AI, the stakes are higher because safety decisions rest on these evaluations.
The Capability Approach in Human Development
In economics and political philosophy, the word “capability” carries a very specific meaning that lines up neatly with the broader distinction. The capability approach, most associated with Amartya Sen and Martha Nussbaum, focuses not on what people are able to do in some abstract sense but on what they are actually free and equipped to do. Your capabilities, in this framework, are the real freedoms you have to live the kind of life you value. A person might have the physical ability to work, but without access to education, transportation, or a nondiscriminatory labor market, their capability to pursue meaningful employment is limited.
Research in this tradition has found that skills and capabilities are dynamically linked. Skills embodied in people empower them, and enhanced skills expand opportunities, which in turn promote capabilities.10Taylor & Francis Online. Capabilities and Skills The arrow runs both ways: gaining a new skill (an ability) opens doors (capabilities), and having those doors open motivates further skill development. This feedback loop helps explain why small interventions, like providing someone with a wheelchair, literacy training, or internet access, can produce outsized effects on what that person ends up being able to achieve.
Assistive technology offers a practical illustration. Devices like hearing aids, mobility aids, and adaptive software are designed to bridge the gap between a person’s internal abilities and what their environment demands of them. Research on older adults has shown, however, that assistive technologies don’t always fully close that gap between personal capabilities and environmental demands.11PubMed Central. The Impact of Disability and Assistive Technology Use on Well-Being in Later Life: Findings From the National Health and Aging Trends Study A wheelchair gives someone mobility capability they lacked, but if the local bus system has no ramps, the capability is still constrained. The ability to move exists. The technology enhances it. But capability depends on whether the surrounding systems cooperate.
How Abilities Transform Into Capabilities
Neural plasticity is the mechanism by which the brain physically rewires itself in response to experience, and it provides a biological lens on how raw abilities become functional capabilities. Plasticity underlies both learning in an intact brain and relearning in a damaged one during rehabilitation.12PubMed. Principles of experience-dependent neural plasticity: implications for rehabilitation after brain damage A stroke patient, for example, might retain the neural architecture for speech (the ability, at a structural level) but lose the functional connections needed to produce words reliably. Rehabilitation aims to rebuild those connections, effectively converting a latent ability back into a usable capability. The raw neural material is the ability; the trained, context-responsive performance is the capability.
This maps onto what happens in everyday skill acquisition too. Learning to drive a car involves building new neural pathways until what starts as effortful, clumsy performance becomes automatic. Your initial ability, basic motor control and visual processing, doesn’t change much. What changes is your capability to coordinate those abilities in a specific, demanding context. Once the brain has rewired through practice, the capability persists even when you haven’t driven for a while, though it may be rusty. The underlying ability was always there; the capability required shaping it through experience.
When the Distinction Matters in Practice
For most casual conversations, swapping “ability” and “capability” won’t cause problems. But there are real situations where getting the distinction right changes what you do. In hiring, a résumé lists abilities: programming languages, degrees, certifications. An interview tests capabilities: can the candidate actually solve a problem under time pressure, communicate with a team, adapt to an unfamiliar codebase? Employers who hire purely on abilities and ignore capability gaps end up with credentialed people who underperform in context.
In disability rights and workplace accommodation, the distinction is even more consequential. A worker may have the cognitive and physical ability to perform a job’s core functions, but architectural barriers, inflexible scheduling, or missing technology suppress their capability. The legal frameworks around accommodation in many countries essentially exist to close that gap, to ensure that people’s abilities aren’t wasted because of removable environmental obstacles.
In medicine and rehabilitation, clinicians use the distinction constantly, though they might not always use these exact words. A physical therapist assessing a patient after knee surgery cares about the joint’s range of motion (an ability measure) but cares more about whether the patient can climb stairs, get into a car, and walk to the grocery store (capability measures). The treatment goal is capability recovery, not just ability restoration on a clinical test.
Capabilities of Machines Versus Capabilities of Organisms
An interesting wrinkle emerges when you compare how the terms are applied to biological organisms and to engineered systems. In robotics and manufacturing, a machine’s capabilities are defined by what it can physically execute given its design specifications. Research on automated manufacturing, for instance, describes how a “process domain” bridges the gap between what a product requires and the specific capabilities of the equipment on the factory floor.13Robotics and Computer-Integrated Manufacturing. Rapid and automated configuration of robot manufacturing cells A robot arm either can or cannot reach a certain point in space with a certain payload. Its capability is deterministic: given the inputs, the output is predictable.
Biological organisms are messier. Research in animal cognition has found that skills like tool use, cooperation, and perspective-taking evolve independently of each other in response to specific environmental pressures, appearing in distantly related species with very different overall cognitive profiles.14PubMed Central. Old and New Approaches to Animal Cognition: There Is Not “One Cognition” A crow can use tools but may not cooperate well; a wolf cooperates brilliantly but doesn’t use tools. Their abilities are domain-specific, shaped by ecological pressures rather than a general “intelligence” dial. Their capabilities in any given situation depend on whether that situation calls for the particular abilities their lineage developed. There is no single “capability score” for a species, just as there’s no single score for a person.
This matters because people often assume that being highly capable in one domain signals high capability across the board. In reality, both organisms and AI systems exhibit patchy profiles: strong in some areas, weak in others, and dependent on context for how any given ability translates into real-world performance. The distinction between ability and capability is, at bottom, a reminder that what you can do always depends on where, when, and under what conditions you’re trying to do it.