behavioral.

leaf ii · ledeWhat behaviourism was for.

Between roughly 1913 and 1965 a school of psychologists insisted that the proper subject of their science was not the mind but the body — specifically, what the body did and how reliably it could be made to do it again.

The wager paid off in techniques. Token economies. Systematic desensitisation. The animal-training methods every modern dog trainer uses. The principles that make slot machines hard to walk away from. Reinforcement learning, the dominant paradigm in machine learning's recent leap. The whole apparatus of applied behaviour analysis. None of it requires the experimenter to know what is happening inside the head.

It also paid off in losses. Behaviourism could not explain how a child learns language. It misread its own animal data. It produced one of the most-cited ethical disasters in the discipline's history. By 1970 the cognitive revolution had taken its central claim to pieces. What remained, refined, is in this deck.

leaf iii · definitionTwo behaviourisms, not one.

The word behaviourism covers two related but distinguishable positions. Methodological behaviourism, associated with John B. Watson, holds that scientific psychology must restrict itself to publicly observable events. Whatever is going on inside the organism — sensations, thoughts, fears — is real but methodologically off-limits, because no one else can see it.

Radical behaviourism, associated with B. F. Skinner, accepts internal events as legitimate data. Skinner's move was to insist that they are themselves behaviours — covert, private, but governed by the same conditioning laws as overt ones. There is no separate mental realm doing the causing; there is only behaviour, some of it loud, some of it quiet.

What both share

A commitment to functional analysis: explain a response by the environmental conditions under which it is reinforced or punished. A commitment to experimental control: a finding that cannot be reproduced with another animal in another lab is not yet a finding. And a near-total mistrust of cognitive constructs (memory, attention, expectation) as causes — these were, on the behaviourist view, things to be explained, not things that explain.

leaf iv · figureIvan Pavlov.

Ivan Petrovich Pavlov did not set out to reform psychology. He was studying how dogs digest. To collect saliva quantitatively he ran a tube from the parotid duct to a graduated cylinder. Standing in the lab one day, he noticed that his dogs began salivating before food touched their tongues — at the sound of footsteps, the sight of the assistant.

This was, in 1901, an embarrassing artefact. Pavlov spent the rest of his career on it. By 1927, when his lectures appeared in English as Conditioned Reflexes, he had laid out a complete experimental vocabulary that psychology still uses. The vocabulary describes classical or respondent conditioning: the formation of a learned association between a previously neutral stimulus and a reflexive response.

Pavlov saw the work as physiology, not psychology. He scorned American behaviourism in his later years. The Americans, more pragmatic, took his findings as the foundation of their science.

leaf v · machineryUCS · UCR · CS · CR.

Classical conditioning is built from four named elements.

The unconditioned stimulus ( UCS ) is the thing that produces the reflex without any learning — food in the mouth produces salivation. The unconditioned response ( UCR ) is that reflex.

The conditioned stimulus ( CS ) starts as a neutral signal — a bell, a light, a metronome — and is paired repeatedly with the UCS. After enough pairings the CS alone produces a learned response: the conditioned response ( CR ), usually a weaker version of the UCR.

What does and does not condition

Classical conditioning works best on reflexive, autonomic responses — salivation, heart rate, the eyeblink, fear, sexual arousal, nausea. It works less well on voluntary movement. Garcia and Koelling (1966) showed it has biological constraints: rats acquire taste aversions in a single trial after eight-hour delays, but cannot learn to associate taste with shock. Evolution shaped what could be conditioned to what.

leaf vi · phenomenaAcquisition, extinction, recovery.

Acquisition is the strengthening of the CR over successive CS–UCS pairings; the curve is negatively accelerated and asymptotic.

Extinction is what happens when the CS is presented repeatedly without the UCS — the CR weakens. Crucially, extinction does not erase the original learning. It overlays it with new inhibitory learning.

Spontaneous recovery demonstrates the point: an extinguished CR returns, weaker, after a rest period. Bouton's reinstatement studies (1990s) further showed that an extinguished response can be brought back by a single re-pairing or even by a context change. This finding underwrites every modern relapse-prevention protocol in addiction medicine.

Generalisation — a CR appears to stimuli similar to the CS (a dog conditioned to a 1000 Hz tone will also respond, more weakly, to 900 or 1100 Hz). Discrimination — a CR is restricted to a specific CS by training the animal that other, similar stimuli go un-paired.

leaf vii · manifestoWatson's psychology.

John Broadus Watson published "Psychology as the Behaviorist Views It" in Psychological Review in 1913. The paper opens: "Psychology as the behaviorist views it is a purely objective experimental branch of natural science. Its theoretical goal is the prediction and control of behavior."

The manifesto rejected introspection — the dominant method in psychology at the time, descended from Wundt and Titchener — as unscientific. It rejected consciousness as a research object. It insisted that psychology be modelled on physics: behaviour is the dependent variable, environment the independent variable, internal states a black box.

Watson's career was cut short in 1920 — not by science but by scandal. An affair with his graduate student Rosalie Rayner ended his Johns Hopkins professorship. He spent the rest of his working life in advertising at J. Walter Thompson, where he was a pioneer of conditioning-based brand marketing. The 1924 textbook Behaviorism, written outside the academy, was his most influential.

leaf viii · scandalThe Albert experiment.

In late 1919 Watson and Rayner began conditioning fear in an infant they called "Albert B." They presented the boy with a white rat (CS); each presentation was paired with a sudden loud bang made by striking a steel bar (UCS). After seven pairings, the rat alone produced crying and avoidance. The conditioned fear generalised to a rabbit, a dog, a fur coat, and a Santa Claus mask.

The 1920 paper, "Conditioned Emotional Reactions," was Watson's most cited. It was also a methodological mess (n = 1, no control, observer-coded responses), and an ethical disaster. The boy was released to his mother before any extinction or counter-conditioning was attempted. Whether the conditioning persisted beyond the hospital is unknown.

The Albert study is now taught as much as a cautionary tale as a foundational experiment. It would not pass any modern Institutional Review Board. It also could not be replicated: ethics committees have refused every attempt since.

leaf ix · precursorThorndike's puzzle box.

Edward Lee Thorndike's 1898 Columbia dissertation, Animal Intelligence, set the template for half a century of American animal learning research. He built wooden cages — "puzzle boxes" — from which a hungry cat could escape, to a fish reward, by performing a small mechanical action: pulling a loop, pressing a lever, stepping on a pedal.

The cats did not solve the problem by insight. They thrashed, scratched, and clawed; eventually they hit the release; with each successive trial the time-to-escape dropped, smoothly, in a long negatively-accelerating curve. Thorndike called this trial-and-error learning.

The principle he extracted, the Law of Effect, would become the central claim of behaviourism: actions followed by satisfying consequences become more probable; actions followed by annoying consequences become less probable. Thirty years later Skinner would call this operant conditioning and turn it into a research programme.

leaf x · figureSkinner.

Burrhus Frederic Skinner is the figure most non-psychologists picture when they hear the word behaviourism — and they picture him correctly. From 1930 to 1990 he was the field's central architect, its most prolific experimentalist, and its most aggressive public philosopher.

Skinner's central technical move was to redefine the unit of analysis. Pavlov's conditioning concerned respondents — reflexes elicited by a prior stimulus. Skinner saw that most interesting behaviour is not elicited; it is emitted. A rat presses a bar because it has done so before and got food, not because anything in the moment caused the press. He called this class of behaviour operant — operating on the environment to produce a consequence.

The other move was the apparatus. The operant chamber, soon nicknamed the Skinner box, gave the experimenter complete control over the contingency between response and reinforcer, and the cumulative recorder gave a continuous quantitative trace of behaviour over hours.

leaf xi · frameworkReinforcement and punishment.

Operant conditioning has four operations, defined by two crossed axes — whether a stimulus is added or removed, and whether the behaviour gets stronger or weaker.

The most common confusion is between negative reinforcement and punishment. Negative reinforcement increases behaviour by removing something aversive when the behaviour occurs. Punishment, in either form, decreases behaviour. Skinner argued punishment is unreliable and produces side-effects (avoidance, aggression); reinforcement-based shaping is preferable wherever possible.

leaf xii · schedulesWhy the slot machine works.

Ferster & Skinner's Schedules of Reinforcement (1957) is one of the strangest and most consequential books in twentieth-century psychology. Seven hundred and forty pages of cumulative records, comparing every possible pattern of reinforcement contingency, on pigeons.

The central finding: intermittent reinforcement (rewarding only some responses) produces behaviour that is more resistant to extinction than continuous reinforcement. Among intermittent schedules, the four canonical forms behave very differently.

Fixed-ratio ( FR-5: reward every 5th response) produces a high steady rate with a brief post-reinforcement pause. Variable-ratio ( VR-30: average of 30 responses per reward, but unpredictable) produces a high steady rate with no pause and is the most extinction-resistant schedule known.

This is exactly the contingency a slot machine implements. So does a video-game loot drop. So does an email inbox refresh. So does a notifications feed. Any behaviour reinforced on a VR schedule is, mechanically, hard to stop. Skinner saw the implications by 1953 and was uneasy about them.

leaf xiii · the language fightSkinner vs Chomsky.

Skinner's Verbal Behavior (1957) attempted to extend operant principles to language: a child learns to say milk because saying it produces milk; grammatical structures are chains of operants under stimulus control; meaning is reducible to the contingencies that selected the utterance.

Noam Chomsky's review, two years later, demolished the project. The argument turned on what he called the poverty of the stimulus. Children produce sentences they have never heard, and rapidly extract grammatical regularities from impoverished input. They do this on a developmental schedule indifferent to reinforcement. They acquire syntax in environments where parents do not selectively reward grammatical productions. None of this is consistent with an operant account.

The review's deeper move was philosophical: it argued that Skinner's terms (response, stimulus, reinforcement) only appeared rigorous; applied to language they had to be stretched until they became metaphors. The cognitive revolution that followed — Miller, Bruner, Neisser, Chomsky himself — took the failure of behaviourist linguistics as a starting point.

leaf xiv · the politicsSkinner the social planner.

Walden Two is a novel about a thousand-person community engineered on operant principles: differential reinforcement of cooperation, planned-economy labour credits, child-rearing in shared facilities, no money, no praise, no blame. Skinner wrote it in seven weeks during the summer of 1945 and never quite reconciled himself to literature; the prose is wooden but the scheme is exact.

Beyond Freedom and Dignity (1971) made the philosophical claim openly: there is no autonomous self that chooses freely, only an organism whose history of reinforcement makes some responses more probable than others. To insist on freedom and dignity, in Skinner's argument, is to insist on the right to be controlled by inadequately-understood contingencies rather than by well-designed ones.

The book sold 200,000 copies and made Skinner the most-attacked psychologist of the decade. Spiro Agnew denounced it. Ayn Rand wrote a chapter against it. Noam Chomsky published a long, angry New York Review of Books demolition. Skinner remained calm and unshaken. The actual experimental communities — Twin Oaks in Virginia (1967), Los Horcones in Mexico (1973) — survive into the present in modified form.

leaf xv · neo-behaviourismHull's drive-reduction theory.

Clark Hull at Yale was the most ambitious neo-behaviourist — the figure who tried to formalise the field as a hypothetico-deductive science with axioms and theorems. Principles of Behavior (1943) and A Behavior System (1952) are the central texts.

Hull's central claim was drive-reduction: an animal learns to do whatever reduces a primary biological drive (hunger, thirst, pain). Reinforcement is drive reduction. Habit strength accumulates with each rewarded trial; the strength of a response depends multiplicatively on habit, drive, incentive, and stimulus intensity.

The system collapsed by 1960. Animals learn things drive-reduction predicts they should not (curiosity, exploration, manipulation in the absence of reward); they fail to learn things that should plainly reduce drive. By the time of Hull's death his students had begun to defect — to Skinner's atheoretical empiricism, to cognitive accounts. The lesson stuck: even within behaviourism, attempts to formalise too soon could harden the field around equations its data could not bear.

leaf xvi · the bridgeTolman's cognitive maps.

Edward Chace Tolman, a Berkeley behaviourist, was the strain of the school that pointed out of it. His 1948 paper "Cognitive Maps in Rats and Men" reported that rats trained to run a particular path through a maze, when given a novel shortcut, took it directly. They had learned not a sequence of stimulus-response associations, but a representation of the maze's spatial layout.

This was a cognitive claim wearing behaviourist clothes. Tolman called the inferred internal structure a cognitive map; he called the variables intervening between stimulus and response intervening variables; he insisted the experiments could be run rigorously despite the inferences. The orthodox behaviourists hated him, but his data were good.

By 1971, when O'Keefe and Dostrovsky discovered place cells in the rat hippocampus, Tolman's cognitive map turned out to have a neural correlate. May-Britt and Edvard Moser's discovery of grid cells (2005) and the 2014 Nobel Prize that followed are downstream of Tolman's stubbornness. Behaviourism's most useful child grew up to be neuroscience.

leaf xvii · therapyWolpe and the phobia.

Joseph Wolpe, a South African psychiatrist working with traumatised soldiers in the 1940s, was the first person to turn classical conditioning into a workable clinical method. His 1958 book Psychotherapy by Reciprocal Inhibition introduced systematic desensitisation: a graded, paired exposure procedure for treating phobias.

The patient and therapist build a hierarchy of feared situations — for an elevator phobia, perhaps: looking at a picture of an elevator, walking past one, standing inside a stationary one, riding one floor, riding ten. Relaxation is trained, often via Jacobson's progressive muscle technique. The patient imagines or confronts each stimulus while relaxed; the relaxation is incompatible with the fear; the conditioned fear extinguishes through repeated non-reinforcement.

This is, in effect, controlled exposure — the active ingredient in every modern phobia and anxiety-disorder treatment. Modern protocols (Foa for PTSD, Barlow for panic) drop the relaxation step and use exposure alone, which works at least as well. The method has the highest evidence base of any intervention in clinical psychology.

leaf xviii · applicationToken economies.

A token economy is the operant principle scaled to an institution. Tokens — plastic chips, points, stamps — are dispensed contingently on target behaviours and exchanged later for back-up reinforcers (food, privileges, free time). The tokens function as conditioned reinforcers — they have no intrinsic value but acquire it through their exchange relation.

The system was developed in the 1960s in psychiatric hospitals, where chronic patients had often become unresponsive to ordinary social reinforcement. It has since been used in special-education classrooms, residential treatment for delinquent adolescents (Achievement Place / Teaching-Family Model, founded 1967), prisons, and corporate settings.

The empirical record is mixed. Token economies produce reliable behaviour change in-program. Generalisation outside the program is poor unless explicit fading and transfer procedures are built in. Critics object to the imposed contingencies as paternalistic; advocates point out that the alternative is usually informal contingencies that nobody designed and nobody can examine.

leaf xix · professionThe applied field.

Applied Behaviour Analysis emerged from Skinner's experimental analysis of behaviour as a clinical-educational discipline in the 1960s. The Journal of Applied Behavior Analysis launched in 1968. The field's foundational claim is that the operant principles established with rats and pigeons apply, with no important modification, to humans in real-world settings.

ABA practitioners are now credentialed via the BCBA (Board Certified Behavior Analyst, Behavior Analyst Certification Board, founded 1998); there are roughly 65,000 BCBAs worldwide as of 2025. They work in autism services, special education, organisational behaviour management, sport psychology, traumatic-brain-injury rehabilitation, and behavioural medicine.

The field's identity is built on two methodological commitments: functional analysis (identifying the contingencies maintaining a problem behaviour before intervening) and single-case design (using each subject as their own control across A-B-A-B reversal or multiple-baseline patterns). These are the methods, not the techniques, that ABA contributes most to other clinical fields.

leaf xx · clinical fightEarly-intensive intervention.

Ole Ivar Lovaas, a Norwegian-American psychologist at UCLA, published in 1987 a study in the Journal of Consulting and Clinical Psychology: 19 autistic children given 40 hours per week of one-on-one operant intervention before age four. Nine reached normal-range IQ and mainstream first-grade placement. The control group: two of nineteen.

The Lovaas study became the founding evidence base for Early Intensive Behavioural Intervention (EIBI). Insurance coverage for autism services in the United States was won, state-by-state, on the back of it. The 1987 numbers have not fully replicated — meta-analyses show real but smaller effects — and the children who participated have spoken in adulthood about the cost.

Lovaas's original protocol included aversives (slaps, shouts, brief electric shock) that the field has since abandoned. The current debate is about whether discrete-trial training, even without aversives, teaches compliance at the expense of authentic communication, and whether naturalistic developmental approaches (PRT, ESDM, JASPER) get most of the same gains with less harm. The honest answer is that the evidence does not yet decide it.

leaf xxi · depressionThe behaviour-only depression treatment.

Most evidence-based depression treatment is now cognitive-behavioural — but a strain of research has shown that the behavioural half does most of the work. Behavioural Activation (BA), as developed by Neil Jacobson, Christopher Martell, and Sona Dimidjian in the late 1990s, treats depression by directly increasing rewarding activity, on the operant premise that depressed mood follows from a thinning of contingent reinforcement in the patient's life.

The technique is unspectacular and effective: activity monitoring (when does the patient feel even slightly better?); graded scheduling of pleasurable and mastery-yielding activities; functional analysis of avoidance patterns; explicit working with the consequences that maintain withdrawal.

Jacobson's 1996 component-analysis study had already shown that BA alone matched full cognitive-behavioural therapy. The 2006 Dimidjian trial replicated it. The 2009 NICE guidelines and the 2017 Veterans Affairs / Department of Defense guidelines list BA as a first-line treatment for major depression. It is one of the cleanest practical wins behaviourism has produced.

leaf xxii · principlePremack's law.

David Premack's 1959 paper in Science ("Toward Empirical Behavior Laws: I. Positive Reinforcement") generalised the concept of reinforcement in a way Skinner's framework had not. A reinforcer, Premack argued, is not a thing — pellet, water, money. It is a relation. Any high-probability behaviour can reinforce any lower-probability behaviour.

The classic demonstration: children who normally chose pinball over candy could be made to eat more candy if pinball-time was contingent on candy-eating. Children who normally chose candy over pinball showed the reverse. The reinforcer changed because the relative probability of the two activities changed.

The Premack principle is the behaviourist origin of "eat your vegetables before dessert." It is also the principle behind the better forms of self-management — schedule the task you want to do more of (exercise, writing) immediately before a high-probability daily behaviour (coffee, lunch). Timothy Pychyl's procrastination research and James Clear's Atomic Habits (2018) are unacknowledged Premackians.

leaf xxiii · historyAversion therapy.

If reinforcement is the field's success story, aversion therapy is its scandal. The technique pairs an unwanted behaviour with an unpleasant stimulus — emetic drugs, electric shock, foul odours — to create a conditioned aversion. It was used through the 1960s and 1970s to treat alcoholism, smoking, paedophilia, fetishism, and — most damagingly — homosexuality.

The gay-conversion programs of that era, conducted at universities and hospitals across the United States and United Kingdom, were behaviourally rigorous and morally indefensible. They produced no convincing evidence of changed orientation, considerable evidence of psychological harm, and an apology from the American Psychological Association in 1997. Most professional bodies now consider conversion therapy unethical regardless of the patient's stated wishes.

Aversion therapy survives in two narrow medical applications: disulfiram (Antabuse) for alcohol-use disorder, which produces nausea on alcohol consumption; and the contested use of contingent electric stimulation at the Judge Rotenberg Center in Massachusetts, where the FDA's 2020 attempt to ban the practice was overturned in court. The field has not, by any honest accounting, fully reckoned with this history.

leaf xxiv · everydayHabit formation in real life.

Behaviourism's clearest contemporary research programme is the experimental study of habit. Wendy Wood, Bas Verplanken, and Phillippa Lally have built an empirical literature showing that habits — context-cued automatic behaviour — are governed by the same operant variables as laboratory bar-presses, scaled up to phones, gyms, kitchens, and commutes.

The findings are sharp. Context matters more than motivation: behaviour that recurs in a stable physical setting with stable cues is performed with little conscious effort. Friction matters: a fitness goal pursued at a gym 5 km from home fails three times as often as the same goal pursued at home. Repetition matters: Lally et al. (2010) found a median of 66 days for a new self-chosen behaviour to reach asymptotic automaticity, with wide variation.

The most useful operational claim from this literature: do not try to change a habit by trying harder. Change the cues — move the chips off the kitchen counter, remove the apps from the home screen, lay out running clothes the night before. The Skinnerian point is that the contingency does the work, not the will.

leaf xxv · craftPryor and the clicker.

Karen Pryor's Don't Shoot the Dog! (1984) is the most accessible book ever written about operant conditioning. It is also a manual. Pryor was a working dolphin trainer in the 1960s — at Sea Life Park, where she wrote the program — when she realised that everything she was doing was Skinner with marine mammals. She came back to the field and rewrote it for laypeople.

The technique that defines her name in the discipline is marker-based positive reinforcement. A neutral signal (a clicker, a whistle, a verbal yes) is paired hundreds of times with food until it becomes a reliable conditioned reinforcer. The trainer can then mark the precise moment of the desired behaviour, even in mid-jump, even at distance — a precision impossible with a treat alone.

Clicker training revolutionised dog training in the 1990s and most of the rest of animal handling — zoo-keeping, livestock management, service-animal training — over the following decade. It works on horses, octopuses, chickens, fish, and the elephant. The technique is pure Skinner; the cultural impact came from a writer who could explain it.

leaf xxvi · pedagogySniffy.

The decline of live-rat operant labs in undergraduate teaching — driven by ethics committees, costs, and time pressure — produced a small, distinctly behaviourist solution. Sniffy the Virtual Rat is a simulation, written originally in HyperCard for the Macintosh, that runs an entire operant chamber on a computer.

Sniffy magnification-bars his way into a Skinner box, sniffs around, and (with appropriate shaping) learns to press a lever. The student delivers reinforcement with a key-press; the cumulative record updates in real time on the screen; schedules can be selected (CRF, FR-5, VR-30, FI-30s, VI-30s); extinction can be run; spontaneous recovery can be observed; the student writes up the results.

The Sniffy curriculum is a quiet testament to behaviourism's pedagogical robustness: the laws of operant behaviour are general enough that a simulated rat behaves like a real rat well enough to teach the principles. A cognitive psychologist could not write an equivalent simulation of, say, working memory; the underlying theory is not specified at the level required.

leaf xxvii · synthesisWhat survived.

By 1980 behaviourism had lost the theoretical war but not the clinical one. The dominant synthesis that emerged — cognitive-behavioural therapy (CBT) — keeps the operant techniques (exposure, behavioural experiments, activity scheduling, contingency management) while adding cognitive techniques (Socratic dialogue, schema work, cognitive restructuring) borrowed from Aaron T. Beck and Albert Ellis.

The third wave of CBT — Steven Hayes's Acceptance and Commitment Therapy (1986), Marsha Linehan's Dialectical Behaviour Therapy for borderline personality disorder (1993), Mindfulness-Based Cognitive Therapy (Segal, Williams & Teasdale 2002) — is more genuinely behaviourist than its name suggests. ACT is built on a behaviour-analytic theory of language called Relational Frame Theory (Hayes 2001) that is in direct lineage from Skinner's Verbal Behavior.

The honest assessment, in 2026: nobody is a pure behaviourist anymore. Almost every effective therapy in current clinical use contains a substantial behaviourist core. The school is gone; its techniques run the building.

leaf xxviii · descendantFrom Skinner to AlphaGo.

The most surprising afterlife of behaviourism is in machine learning. Reinforcement learning (RL) — the subfield of AI in which an agent learns by interacting with an environment and receiving scalar rewards — is, mathematically, an extended Skinnerian framework.

Richard Sutton and Andrew Barto's textbook is explicit about this. The Bellman equation generalises the Law of Effect; temporal-difference learning (Sutton, 1988) generalises Pavlovian acquisition; the actor-critic architecture mirrors operant-respondent integration. Wolfram Schultz's discovery (1990s) that midbrain dopamine neurons fire as TD-prediction-error signals connected the mathematics back to the brain.

RL is the technology behind DeepMind's AlphaGo (2016), AlphaZero (2017), and the RLHF training stage of every modern large language model. The behaviourist programme — that complex behaviour can be shaped from a scalar reinforcement signal, given the right scheduling — turned out to be true in silico, even where it failed in vivo. Skinner, who lived to see the first symbolic AI, did not live long enough to see this vindication.

leaf xxix · booksTwenty-two books.

• 1898Animal Intelligence — the puzzle box.Thorndike
• 1913"Psychology as the Behaviorist Views It" — the manifesto.Watson
• 1924Behaviorism — Watson outside the academy.Watson
• 1927Conditioned Reflexes — the English Pavlov.Pavlov
• 1938The Behavior of Organisms — Skinner's first major book.Skinner
• 1943Principles of Behavior — Hull's hypothetico-deductive system.Hull
• 1948"Cognitive Maps in Rats and Men" — the cognitive opening.Tolman
• 1948Walden Two — the utopian novel.Skinner
• 1953Science and Human Behavior — the introductory textbook.Skinner
• 1957Schedules of Reinforcement — 740 pages of pigeon data.Ferster & Skinner
• 1957Verbal Behavior — the language extension.Skinner
• 1958Psychotherapy by Reciprocal Inhibition — desensitisation.Wolpe
• 1959"Review of Verbal Behavior" — the demolition.Chomsky
• 1968The Token Economy — applied operant on a ward.Ayllon & Azrin
• 1971Beyond Freedom and Dignity — the philosophical brief.Skinner
• 1976Particulars of My Life — autobiography vol. 1.Skinner
• 1984Don't Shoot the Dog! — Pryor for laypeople.Pryor
• 1987"Behavioral Treatment and Normal Educational Functioning"Lovaas
• 1998Reinforcement Learning — the AI textbook.Sutton & Barto
• 2001Relational Frame Theory — the third-wave foundation.Hayes et al.
• 2004B. F. Skinner: A Life — the Bjork biography.Bjork
• 2019Good Habits, Bad Habits — habit science.Wood

leaf xxx · viewingThe lab on film.

↑ Pavlov's classical conditioning · introductory animation

More on YouTube

Watch · B. F. Skinner and operant conditioning — original lab footage
Watch · The Little Albert experiment, 1920 — surviving film

Read

Start with Skinner's Science and Human Behavior (1953) — the cleanest single-volume statement of the radical-behaviourist position, available free from the B. F. Skinner Foundation. Then Karen Pryor's Don't Shoot the Dog! (1984) for the techniques in plain English. Then Chomsky's Verbal Behavior review (1959) for the major counter-argument. For the modern picture, Wendy Wood's Good Habits, Bad Habits (2019) and Sutton & Barto's Reinforcement Learning (2018) — the latter is technical but the prose chapters are surprisingly readable.

Where to keep reading

The Journal of the Experimental Analysis of Behavior (since 1958) and the Journal of Applied Behavior Analysis (since 1968) are the field's two main organs. Behavior and Philosophy, formerly Behaviorism, carries the conceptual debate. The Association for Behavior Analysis International annual conference proceedings are open access from 2010.