Psychologists call it learning: a relatively permanent change in behavior or knowledge that results from experience. The boundary matters. A change counts as learning only when experience produces it and the change persists — fatigue, injury, surgery, drugs, and normal maturation all alter behavior without qualifying. That definition comes from the standard textbook tradition built on Kimble's 1961 criteria, repeated across editions of Myers and DeWall's Psychology and Schacter, Gilbert, and Wegner's Psychology. This guide unpacks the definition term by term, then gives you four concrete tests that separate real learning from its impostors.

The Term-by-Term Definition That Decides What Counts

The classic academic definition is "a relatively permanent change in an organism's behavior or knowledge due to experience," and every word in it is doing work. Skip one clause and you misclassify things. The phrase "relatively permanent" exists because learning can decay or become dormant; it doesn't require a lifetime guarantee, only a change that outlasts the moment of training. "Behavior or knowledge" matters because some learning shows up only in what you can do, and some only in what you can say.

Psychologists distinguish the definition from performance. You can hold knowledge and fail to demonstrate it — a skill learned in one room may not appear in another. Researchers therefore treat behavior as evidence of learning rather than learning itself. The split explains why the term is often glossed as "a change in the potential for behavior," which is the wording Myers and DeWall use across editions of Psychology. Potential covers knowledge you don't happen to express today.

Experience is the causal clause, and the only one that is strictly necessary. A change produced by a drug, a lesion, or a growth spurt is excluded even when it alters behavior dramatically. That produces a genuine edge case. Children who begin walking after months of crawling have changed their behavior permanently, but maturation carries most of the explanation; practice contributes, so the honest answer is that maturation and learning are entangled rather than pure opposites.

The definition has an origin you can use as a credibility marker. Kimble proposed in the 1961 Psychological Review article "Pavlovian Conditioning and the Concept of Learning" that learning be defined by three criteria — a change in behavior or behavioral potential, produced by experience, that is relatively permanent. Textbook editions after 1961 largely adapted Kimble's triad, and that lineage is why the phrasing appears nearly unchanged in modern glossaries.

Three Ordinary Examples, and Why Each One Qualifies

Start with taste aversion. Someone eats a bad scallop, feels sick hours later, and stops eating scallops for years. John Garcia and Robert Koelling's work in the 1960s established that rats learn these aversions in as little as one trial and will associate taste with nausea long after the meal. The change sits in a behavior — refusal — and it outlives the single experience by a wide margin.

Consider a child who learns that words are tools for getting attention. The child, unable to say "milk," points at the refrigerator once, is handed the carton, and points again within days. B. F. Skinner's account treats that as operant learning reinforced by a consequence. The evidence is a rise in pointing, and it persists even when the original listener is replaced by someone new.

Take a driving skill. A learner practices a three-point turn in an empty lot, then drives that maneuver confidently on a busy street months later. Motor learning here is procedural, largely implicit, and slow to emerge. The predecessor years of a much older adjacent puzzle — Ebbinghaus measuring his own retention in 1885 — show why retention curves matter: savings after a delay, not just performance on the day, is the truer sign of permanence.

Each example shows something different about what "experience" does: a single pairing, a reinforcing consequence, and repeated practice. That variety is why psychologists group learning into types rather than treat it as one mechanism. The type is set by the conditions, not by the definition.

Habituation and Sensitization: The Two Simplest Forms

The simplest form of learning is habituation — a decreasing response to a repeated stimulus that carries no consequence. A person who lives near a train line stops waking at the 2 a.m. freight; the sound is unchanged and the person has simply ceased to respond. Eric Kandel's lab studies on the sea slug Aplysia between the 1960s and 1980s localized a cellular basis for this, which is why habituation appears in educational psychology textbooks as a guaranteed-students-kind of example.

The mirror form is sensitization — an increasing response after an intense or dangerous stimulus. After a car backfires near you, a similar bang the following week may startle you more, not less. The reasons are dose and context: repeated mild stimuli wear the response down, while repeated strong ones prime the nervous system. The same person can habituate to lawn mowers and sensitize to gunshot-adjacent sounds.

Non-associative learning is the umbrella label for both, because neither depends on a pairing between events. The distinction earns teaching space because it separates change-by-experience from change-by-fatigue. A fatigued response fades within minutes and returns after rest; a habituated one takes days or weeks to recover and often returns only after a change of context, such as a new house.

Watch the placebo problem here. Students often hear "ignore a repeated sound and the reaction goes away" and treat it as advice. The mechanism is not willpower; the nervous system reduces its response when a stimulus proves harmless. The identical pattern appears in animals that cannot reason about stimuli at all, which is why habituation is the most universal learning type.

Classical Conditioning: How a Neutral Cue Becomes a Signal

In classical conditioning, an organism learns that one event predicts another. The canonical terms — neutral stimulus, unconditioned stimulus, unconditioned response, conditioned stimulus, conditioned response — are worth learning as labels because every textbook exam checks them. The field usually dates the formal beginning to Ivan Pavlov's laboratory work on dogs' salivation, published in English translation in 1927 in Conditioned Reflexes.

The procedure matters more than the vocabulary. A neutral cue (a tone) precedes an unconditioned stimulus (food); the unconditioned response (salivation) begins without training. After repeated pairings the tone alone yields salivation, and the tone has become a conditioned stimulus. Timing is not arbitrary: the cue must precede the food, not follow it. Backward conditioning, where the cue comes after the consequence, is notoriously weak.

What tends to be omitted is the answer at the biological level. Conditioned salivation is not a wish to cooperate; it is a preparatory response that readies the body for the food. Pavlov's own title used "conditioned reflex" for exactly that reason. The same logic explains an office worker's hunger pangs at 11:50 a.m.: the clock, the hallway smell, or a colleague's footsteps become cues for lunch.

Classical conditioning explains many everyday fears, including ones that never tied logically to a danger. Watson and Rayner's 1920 "Little Albert" demonstration is the textbook example and still the error-filled one: the study had a single participant and never fully resolved whether the fear generalized. Cite it as a cautionary tale about ethics and replication, not as proof of how all phobias form.

Operant Conditioning and the Difference That Confuses Everyone

Operant conditioning changes behavior through consequences — reinforcement increases it, punishment decreases it. The distinction from classical conditioning is the learner's role. In classical conditioning, the tone does nothing; in operant conditioning, what the learner does governs what happens next. B. F. Skinner's work from the 1930s onward, and his The Behavior of Organisms (1938), put the consequence at the center of the account.

The classic confusion is the "reinforcement versus reward" trap. A reinforcer is defined by its effect, not its appeal. Anything that makes a behavior more likely, including an unpleasant consequence, counts as reinforcement — because the busy employee who finally gets handed a coffee after weeks of asking was, technically, receiving reinforcement. Positive and negative mean adding and removing a stimulus; reward and punishment mean increasing and decreasing behavior. That's four categories with only two words doing two jobs.

Schedules matter because they make the same schedule produce different persistence. Ratio schedules tie payoff to count; interval schedules tie it to time. The partial reinforcement effect says intermittent reinforcement produces more persistent behavior than continuous reinforcement, which explains why a slot machine keeps pulling attention and a reset password ritual keeps being over-repeated. A table of the common schedules is more useful than a list:

Schedule What governs the reward Typical real-world example Quirk
Fixed ratio A set number of responses Piecework pay Short pause after each payout
Variable ratio An unpredictable average Social media likes Highest resistance to extinction
Fixed interval A fixed time after last reward Weekly salary review Scalloped response pattern
Variable interval An unpredictable average time Email checking Steady, slow responding

If you want to know what makes a consequence sting, remember that "punishment" here is a technical label: it can mean adding a penalty or removing a desired privilege, and neither is magically effective. It teaches what not to do only when the alternative is clear.

Observational Learning and Insight: When the Learner Isn't the One Being Conditioned

Albert Bandura's Bobo doll experiments in the 1960s showed children imitating aggression they watched an adult display, which is the founding evidence for observational learning. The level of aggression mattered less than the demonstrated behavior, which is why the lesson is sometimes misused to blame screens rather than model behavior. The underlying claim is simple: what learners watch shapes what they try.

Bandura's 1977 Social Learning Theory formalized the mechanism into attention, retention, reproduction, and motivation. School revision rarely names the fourth step, but it's the one that explains the common failure. Students who watched the same demonstration differ in whether they act on it. Watching and trying split apart when the reward for trying is missing.

Insight is the other form that breaks the simple model of gradual conditioning. Wolfgang Köhler's chimpanzee observations from roughly 1913 to 1917, published in 1925 as The Mentality of Apes, described the animals fitting boxes and sticks together to reach fruit, apparently after a period of apparent pause. The "sudden understanding" reading is contested; modern accounts note that prior play and repetition often laid groundwork. Treat insight as a distinct description of a pattern, not a guaranteed separate process, because attention and restructure matter.

For practical use, observational learning means demonstration is cheaper than correction. Show the step, watch the learner try, and keep a repeatable example, rather than telling the learner what you'd like to see. It's also the section where "learning" gets most slippery: modeling teaches knowledge and the behavior that goes with it, but whether it holds is tested later.

What Is Not Learning: Drugs, Injury, Maturation, and the Placebo Trap

Here's a failure case the top articles usually skip. A person who has coffee every morning and reports a daily boost is reporting a change in how they feel, not evidence of learning. The same sensory and behavioural changes appear with decaffeinated coffee given in a blind setup, which separates the expected boost from the caffeine itself. A pattern built from expectation collapses if the experience is removed and the behavior returns to baseline, which is the opposite of what a conditioned change does.

The same logic cleans up a common workplace error. A training session that produces a one-day spike in the "knowledge check" and then fades by Thursday hasn't failed at explaining the content; it has failed at long-term retention, and the two are different outcomes. A relatively permanent change requires follow-up. Testing at 24 hours, two weeks, and again at eight weeks separates a real outcome from a mood.

Maturation does the same clean-up job for child development. A child who speaks a few words at 15 months and produces full sentences at 24 months has learned, but maturation of the language systems still explains a lot. Textbooks don't need that caveat, but a working summary does. Assign maturation a floor, treat experience as the delta, and think twice before crediting a single class with a permanent change in how a child speaks.

Injury is a final clean boundary. A person who loses an ability after a neurological event has changed in a way that is not causal learning. Rehab, on the other hand, produces learning in a brain that has been altered, and the relearning is a genuine change made by experience even though the cause of the initial change wasn't.

Four Tests You Can Actually Run on Any Claimed Change

The following sequence is the practical procedure for reporting a behavior change honestly; it takes minutes and needs no laboratory.

1: Name the behavior and timestamp the first measurement. Record what happened, when, and under what conditions before any intervention. Count or code it: number of times the employee left the complaint unaddressed, number of seconds a child stayed on task, number of attempts before a learner succeeded. 2: Change one experience at a time. Introduce the practice, the cue, the demonstration, or the feedback in isolation. If three changes land together, you cannot tell which produced the change, and any claim to permanence is unsupported. 3: Re-test after a delay. Use a window that reflects the goal: next day for short tasks and two weeks to eight weeks for study or work skills. Test in a different room or with a different person if the goal is transfer, because same-context flukes are how training evaluations inflate. 4: Run a contrast or return-to-baseline. Remove the reinforcement, the cue, or the practice and observe. If the behavior falls back to the pretest level, the change was contingent on the experience and not permanent. If it stays changed, you have evidence for a relatively permanent effect. 5: Check the three exclusions. Rule out fatigue, drugs, injury, and maturation by asking whether the same behavior appeared without experience. Also ask whether the behavior happened only because someone was watching.

The bad outcome here isn't "the training didn't work." It's the false positive: an enthusiastic participant tells you the lesson landed, the behavior peaks on Friday, and it quietly disappears. If your goal is a real permanent change, the delayed retest does the work that the cheerful exit survey cannot. A short answer to "how do I prove learning?" is to try to make it come back to zero and see whether it does.

A Short History: Pavlov, Thorndike, Skinner, and Bandura

The definition has a lineage that is worth knowing because it tells you why the word "learning" is shared by so many research traditions. Before Pavlov's Conditioned Reflexes (1927), Edward Thorndike's puzzle-box work, reported in Animal Intelligence in 1898 and later formalized in The Laws of Learning in the 1930s, was the argument that learning is gradual and satisfaction-driven. Thorndike's "law of effect" is the seed of the reinforcement idea.

Skinner supplied the modern operant frame in The Behavior of Organisms (1938) and extended it widely, including to teaching machines and programmed instruction. Bandura's Social Learning Theory (1977) added observational channels that behaviorism had left out, and the children's television debate that followed Bobo made modeling famous. Each framing altered the definition slightly, but none removed the experience clause.

The history is where a course or certification often stops, but it matters for mistakes. A student who treats learning as "when a teacher teaches" loses the boundary between instruction and acquisition. A student who treats learning as "when a consequence changes a response" misses that we learn from watching, reading, and thinking. The definition that holds all of these together is the relatively permanent change one.

FAQ

Is every change in behavior learning?

No. The definition requires the change to come from experience and to persist. Any behavior change caused by a drug, a head injury, fatigue, or maturation falls outside it, even if it lasts months. Textbook definitions also require the change to be relatively permanent, which excludes the brief spikes of a testing session.

What's the difference between learning and memory?

Learning is the acquisition — the relatively permanent change produced by experience — while memory is the storage and later retrieval of what was acquired. You can learn something and then fail to retrieve it under stress, which is a memory failure, not a learning failure. The distinction explains why a student who studied at a desk may blank at an exam table.

Why does the partial reinforcement effect matter for real life?

Intermittent and unpredictable rewards generate behavior that survives longer after the rewards stop. That's the 1960s finding behind variable ratio schedules, and it explains why social media notifications keep attention long after the novelty fades. For a person trying to reduce a habit, an unpredictable reward is a stronger maintenance mechanism than a steady one.

Does classical conditioning explain phobias?

It explains some phobias more convincingly than others. The Watson and Rayner "Little Albert" demonstration in 1920 showed conditioned fear in a single infant, but the study had one participant and ambiguous follow-up. Many adult phobias form without a direct trauma, so learning curves, observational transmission, and biological preparedness all contribute alongside conditioning.

What is the difference between positive and negative reinforcement?

"Positive" and "negative" describe whether a stimulus is added or removed, not whether the consequence feels good or bad. Positive reinforcement adds something that increases the behavior; negative reinforcement removes something, such as a headache, and also increases the behavior. Both increase behavior, which is why the pair confuses nearly every first-year student.

Can you learn something you can't demonstrate?

Yes — that's why psychologists separate learning from performance. You may hold a skill or fact that doesn't show on a specific test because the context is unfamiliar or the reward for demonstrating it is missing. Evidence from retention tests, savings measures, and transfer tasks is often needed before calling knowledge permanent.

Who first proposed the three-criteria definition?

Kimble's 1961 Psychological Review article on Pavlovian conditioning set out the three criteria — change, behavioral potential, and permanence — that most modern glossaries adapt. The phrasing appears in Schacter, Gilbert, and Wegner's Psychology, and in Myers and DeWall's Psychology, which keeps the exact definition in its study guides.