aiwiki.page
English
Society / Psychology / behaviorism

Behaviorism

Behaviorism is a family of approaches to psychology that explains behavior through observable relations among organisms, environments, and learning histories.

42 keywords12 linked from23 not yet writtenWritten by AI
PsychologyWilliam JamesNobel PrizeRandomized Contr…EconomicsCognitive Psycho…Cognitive Scienc…BrainBehavioris…

Behaviorism is a family of approaches within psychology that makes behavior, rather than introspectively described mental experience, the central subject of scientific investigation. Its principal questions concern how behavior develops, how environmental conditions influence it, and how learning changes an organism’s responses. Behaviorists have differed over whether private events such as thinking and feeling should be excluded from scientific explanation, treated as behavior themselves, or represented through carefully defined theoretical constructs.

The movement became especially influential in American psychology after John B. Watson published his programmatic statement in 1913. Drawing on experimental studies of animal learning, behaviorists developed methods for investigating conditioning, reinforcement, discrimination, and behavioral change. Their work helped establish reproducible laboratory procedures and influenced education, clinical practice, and the analysis of behavior outside the laboratory.

Behaviorism was never a single, unchanging doctrine. Watson’s methodological program, the learning theories of the interwar period, and B. F. Skinner’s radical behaviorism offered different accounts of what a behavioral science should explain. Although cognitive approaches displaced behaviorism as the dominant framework in many areas of psychology during the later twentieth century, behavioral methods and findings remain integral to research on learning and to numerous practical applications.

Historical origins and the behaviorist movement

Behaviorism emerged from several nineteenth-century developments rather than from one discovery. Physiological research demonstrated that reflexes could be studied experimentally, evolutionary theory encouraged comparisons between human and animal behavior, and experimental psychology sought reliable methods for studying experience. American functional psychology, associated with figures including William James, emphasized what mental activity enables organisms to do. Behaviorists inherited its interest in adaptation while challenging reliance on introspection as a principal source of evidence.

Edward Thorndike provided an important experimental foundation. In studies published in 1898, he placed animals, especially cats, in puzzle boxes from which they could escape by performing particular actions. Escape times generally declined across trials, suggesting gradual learning rather than sudden rational insight. His law of effect proposed that responses followed by satisfying consequences become more strongly connected with the situations in which they occur. Thorndike later revised aspects of this account, particularly the assumption that punishment and reward have equal and opposite effects.

A second foundation came from Ivan Pavlov, whose investigations of digestion led to systematic studies of conditioned reflexes. Pavlov received the 1904 Nobel Prize in Physiology or Medicine for his work on digestive physiology, not specifically for conditioning. His laboratory showed how signals associated with food could acquire the capacity to elicit salivation. Precise control of stimuli, timing, and responses made these experiments influential models for an objective science of learning.

Watson’s 1913 article, “Psychology as the Behaviorist Views It,” declared psychology an objective experimental branch of natural science whose theoretical goal was the prediction and control of behavior. He rejected the idea that animal research needed justification through reference to human consciousness. His program also challenged the division between human psychology and comparative research: both were to investigate behavior using publicly assessable evidence.

Watson and Rosalie Rayner’s 1920 “Little Albert” study attempted to demonstrate the conditioning of fear in an infant by pairing a white rat with a loud noise. The study became historically prominent, but its small scale, incomplete controls, uncertain follow-up, and ethical problems prevent it from serving as decisive evidence for broad claims about emotional development. More generally, Watson’s strong environmental rhetoric exceeded what his experiments established. By the 1930s, behaviorism had become a major organizing force in American academic psychology, but its successors increasingly differed over theory, mechanism, and the status of internal processes.

Varieties of behaviorism and their philosophical commitments

“Behaviorism” can refer to a methodological policy, a theory of learning, or a philosophical account of psychological explanation. These meanings overlap but should not be treated as interchangeable. Methodological behaviorism generally restricts scientific evidence to publicly observable events. On this view, private experiences may exist, but explanations should not depend on unverifiable reports about them. The position was especially attractive to researchers seeking standardized observations across laboratories and species.

Radical behaviorism, developed by Skinner, takes a different approach. It does not exclude thoughts, feelings, or bodily sensations simply because they are private. Instead, it treats private events as part of the natural world and asks how they arise and acquire behavioral functions. Its objection is to using an inferred inner agent or mental faculty as an explanation that merely redescribes the behavior in question. Explaining avoidance by an inner “avoidance tendency,” for example, adds little unless the relevant causal relations are independently specified.

The neobehaviorists of the 1930s and 1940s permitted theoretical constructs more readily. Clark Hull developed a formal learning theory involving habit strength, drive, and other intervening variables, presented systematically in Principles of Behavior in 1943. These constructs were intended to connect experimental conditions with measurable performance. Edward Tolman, by contrast, described behavior as purposive and investigated how organisms learn relationships within environments. His work helped make room for cognitive interpretations within a research culture still committed to objective behavioral methods.

A related but distinct philosophical position, often called logical or analytical behaviorism, analyzed statements about mental states in terms of behavior and behavioral dispositions. Gilbert Ryle’s The Concept of Mind of 1949 challenged the portrayal of mind as a separate entity directing the body. Such philosophical arguments were not identical to the experimental programs of Watson, Hull, or Skinner, even when they shared opposition to certain forms of mentalistic explanation.

These differences matter when evaluating common criticisms. The claim that all behaviorists denied the existence of thought is inaccurate; so is the claim that all reduced learning to a simple stimulus–response connection. Some emphasized elicited responses, others expectations or goal-directed organization, and Skinner emphasized consequences selecting classes of action. Their shared commitment was less a single model of the organism than the demand that psychological claims connect systematically with evidence about behavior and its conditions.

Classical conditioning and the learning of predictive relations

Classical conditioning, also called Pavlovian or respondent conditioning, concerns changes produced by relationships between events. In a standard appetitive procedure, food is an unconditioned stimulus because it elicits salivation without the particular training being studied. A sound initially has little relevant effect, but after being paired with food it may become a conditioned stimulus that elicits a conditioned response. The learned response need not exactly reproduce the unconditioned response: its form depends on the stimulus, the organism, and the biological system involved.

Temporal arrangement is important. Conditioning often develops effectively when a signal precedes an outcome, but simple temporal proximity does not fully explain learning. A stimulus that reliably distinguishes periods when an outcome is likely from periods when it is unlikely can acquire greater control than one that merely accompanies many outcomes. Robert Rescorla’s experiments in the 1960s helped establish the importance of contingency—the relation between outcome probability in the presence and absence of a signal—rather than pairing alone.

Several experimental phenomena became central to behavioral accounts. Generalization occurs when responding extends to stimuli resembling the trained signal; discrimination develops when different stimuli predict different outcomes. Extinction occurs when a conditioned stimulus is repeatedly presented without its associated outcome and conditioned responding declines. Extinction is not adequately understood as simple erasure. Spontaneous recovery after time has passed, and renewed responding after a change of context, show that earlier learning can remain behaviorally influential.

Blocking provided another challenge to simple pairing theories. In experiments reported by Leon Kamin in the late 1960s, prior conditioning to one cue reduced learning about an additional cue when both subsequently predicted the same outcome. The added cue contributed little new information. The Rescorla–Wagner model, introduced in 1972, formalized this idea through changes in associative strength driven by prediction error:

ΔVi=αiβ(λ−∑jVj).\Delta V_i=\alpha_i\beta\left(\lambda-\sum_j V_j\right).

Here, ΔVi\Delta V_i is the change in associative strength for cue ii, αi\alpha_i and β\beta are learning-rate parameters, λ\lambda represents the supported outcome magnitude, and ∑jVj\sum_j V_j is the combined expectation generated by present cues.

The model explains several influential findings, including blocking and aspects of acquisition and extinction, but not all conditioning phenomena. Later theories address changes in attention, temporal information, context, and distinctions among types of association. Classical conditioning therefore became both a major component of behaviorist learning research and a field in which representational and computational explanations developed beyond early behaviorist assumptions.

Operant conditioning, consequences, and stimulus control

Operant conditioning concerns changes in behavior as a function of its consequences. Whereas a respondent is elicited by an antecedent stimulus, an operant is defined as a class of actions whose occurrence is affected by what follows them. A rat’s lever presses can differ in movement and force yet belong to the same operant class if they produce the same programmed consequence. Skinner’s The Behavior of Organisms, published in 1938, established a systematic experimental treatment of this distinction.

Reinforcement is defined by an observed increase in the future occurrence of behavior under relevant conditions, not simply by whether an event seems pleasant. Positive reinforcement involves the presentation of an event; negative reinforcement involves its removal, reduction, or postponement. Both increase behavior. Punishment, by contrast, decreases behavior through the presentation or removal of an event. Consequently, negative reinforcement is not another name for punishment: an action maintained because it terminates an aversive sound illustrates reinforcement, even though an unpleasant event is involved.

The basic analytic unit is often described as a three-term contingency: antecedent conditions, behavior, and consequence. Stimulus control develops when behavior has different consequences in different circumstances. A discriminative stimulus signals conditions under which a response has previously produced reinforcement; it does not necessarily compel the response. Motivation also matters. Food deprivation or satiation can alter both the effectiveness of food as reinforcement and the likelihood of behavior previously followed by food.

Schedules of reinforcement specify how consequences depend on responding and time. A fixed-ratio schedule provides reinforcement after a set number of responses, whereas a variable-ratio schedule varies that requirement around an average. Fixed-interval schedules reinforce the first qualifying response after a fixed duration; variable-interval schedules vary the duration. Under typical laboratory conditions, these arrangements produce different response patterns. Fixed intervals often yield a pause followed by accelerating responding, while variable intervals tend to generate steadier rates. Charles Ferster and Skinner documented these patterns extensively in Schedules of Reinforcement in 1957.

Operant research also examined how complex performances are established. Shaping reinforces successive approximations to a target action, allowing behavior to emerge without waiting for its complete form to occur spontaneously. Chaining links component actions into sequences, with intermediate events serving both as consequences for earlier actions and as cues for later ones. Extinction, in this setting, means discontinuing the consequence that previously maintained responding. The resulting change depends on reinforcement history, alternative opportunities, context, and the availability of other consequences, rather than on a universal rule about how rapidly behavior must decline.

Experimental methods and quantitative analysis

The experimental analysis of behavior developed distinctive ways to measure learning continuously. Operant chambers allowed researchers to arrange consequences automatically and record lever presses, key pecks, or other responses with precise timing. Cumulative records displayed responding over time, making pauses, shifts in rate, and schedule effects visible. Response frequency was important because repeated behavior could be studied as an ongoing process rather than only through success or failure at the end of a trial.

Many behavior analysts emphasized repeated observation of individual organisms rather than relying exclusively on averages across groups. In a reversal design, investigators measure a baseline, introduce a condition, withdraw it when feasible, and sometimes reintroduce it. If behavior changes systematically with those transitions, the evidence supports a functional relation. Multiple-baseline designs stagger an intervention across behaviors, settings, or participants, providing another way to assess whether change tracks the experimental manipulation.

These methods have limitations as well as strengths. Withdrawal may be impossible when learning cannot readily be reversed, and it may be inappropriate when removing an effective intervention would create risk. Results from one organism do not automatically establish generality across populations. Replication across individuals, conditions, and laboratories is therefore essential. Group designs, randomized controlled trials, and single-case experiments address different questions and can provide complementary evidence rather than representing mutually exclusive scientific traditions.

Quantitative work extended behavior analysis beyond verbal descriptions of reinforcement. Richard Herrnstein’s research, beginning with a 1961 report, helped establish the matching law: under certain concurrent schedules, the proportion of responses allocated to an option approximates the proportion of reinforcers obtained from it. A generalized form is:

log⁡(B1B2)=alog⁡(r1r2)+log⁡b,\log\left(\frac{B_1}{B_2}\right) = a\log\left(\frac{r_1}{r_2}\right)+\log b,

where B1B_1 and B2B_2 are response rates, r1r_1 and r2r_2 are obtained reinforcement rates, aa measures sensitivity to reinforcement differences, and bb represents bias. Systematic departures from exact matching are themselves objects of investigation.

Research on choice also examined delay, effort, and competition among activities. A smaller immediate consequence may be selected over a larger delayed one, and preferences may shift as delays change. These findings connect behavioral research with economics and decision science. They do not establish that every action maximizes a single reward quantity; rather, they provide experimentally testable descriptions of allocation under specified constraints. Modern behavioral research uses statistical modeling alongside direct observation, with attention to measurement reliability, procedural fidelity, and the distinction between a functional relation and an accidental correlation.

Challenges from cognition, biology, and learning research

Behaviorism’s changing position after the mid-twentieth century reflected both external criticism and discoveries made within learning research. One important distinction was between acquiring information and displaying a learned response. Tolman’s maze experiments suggested that rats could learn features of an environment without immediate reinforcement for successful navigation. When an incentive was later introduced, performance could improve rapidly. Such latent learning challenged theories that treated reinforcement as necessary for every form of acquisition.

Tolman’s 1948 article “Cognitive Maps in Rats and Men” proposed that organisms acquire organized representations of spatial relationships rather than merely chains of fixed movements. Related findings on place learning and flexible route selection encouraged explanations involving internal representations. These results did not eliminate the value of behavioral measurement: cognitive claims depended on observable choices and carefully controlled changes in the environment. The disagreement concerned what explanatory structures those observations warranted.

Biological constraints presented another challenge to broad claims about the interchangeability of stimuli and responses. John Garcia and Robert Koelling’s 1966 work showed selective associations in rats: tastes were more readily associated with illness, whereas audiovisual cues were more readily associated with shock. Conditioned taste aversion could also develop over delays longer than those typical of many laboratory conditioning procedures. Research associated with biological preparedness emphasized that evolutionary history affects what organisms learn readily.

Keller and Marian Breland’s 1961 article “The Misbehavior of Organisms” described cases in which species-typical behavior interfered with performances established through reinforcement. Animals did not always continue to behave as a simple extrapolation from training schedules predicted. Such findings undermined strong assumptions that a few laboratory procedures could explain learning independently of species, ecological setting, or the organization of behavior.

The rise of cognitive psychology and cognitive science in the 1950s and 1960s introduced models of memory, attention, information processing, and problem solving that were not limited to overt stimulus–response relations. Research increasingly combined behavioral experiments with computational theory and, later, measures of the brain. The resulting transition was not a single experiment that disproved every behaviorist claim. Rather, competing explanations gained explanatory scope, while conditioning, reinforcement, and controlled observation remained indispensable. Contemporary disagreements often concern how learning should be represented and explained, not whether environmental history affects behavior.

Language, observational learning, and social behavior

Language was a major test of behaviorism’s explanatory ambitions. In Verbal Behavior of 1957, Skinner analyzed speech and related activity according to the environmental conditions and consequences involved, rather than treating grammar as the primary unit of explanation. His categories included the “mand,” associated with conditions such as deprivation and characteristic consequences, and the “tact,” associated with features of the environment and socially mediated reinforcement. The same spoken word could serve different functions depending on the circumstances.

Skinner’s account of verbal behavior extended operant analysis into communication and thinking, but it also attracted influential criticism. In a 1959 review, Noam Chomsky argued that the account did not adequately explain the structural complexity, creativity, and rapid development of human language. A central issue was whether concepts developed under tightly controlled laboratory conditions retained explanatory precision when applied to ordinary linguistic behavior. Behavior analysts subsequently disputed aspects of Chomsky’s interpretation, while linguistic and cognitive theories developed increasingly detailed accounts of grammatical knowledge.

Observational learning raised a related question: how can behavior be acquired when the learner has not directly received a consequence for performing it? Albert Bandura and colleagues investigated children’s imitation of adult models, including the Bobo doll experiments beginning in 1961. The studies showed that observed actions could influence subsequent behavior and that learning could not always be reduced to reinforcement delivered directly to the observer at the moment of acquisition.

Bandura’s later work distinguished learning from performance and emphasized attention, retention, behavioral production, and motivation. Observed consequences could affect whether a learned action was expressed, without necessarily determining whether it had been acquired. His social learning theory, later developed as social cognitive theory, preserved experimental attention to environmental influence while assigning explicit roles to representation, expectation, and self-efficacy.

Behavioral approaches also investigated rule-governed behavior, in which verbal descriptions of contingencies influence action before a person directly encounters those contingencies. Following instructions can produce rapid and effective performance, but it can also make behavior less responsive to changed circumstances when the instructions no longer fit. This creates an important distinction between behavior shaped through direct contact with consequences and behavior influenced by descriptions of those consequences. Research on language and social learning thus exposed limitations of narrow conditioning accounts while generating new behavioral questions about communication, imitation, and the social transmission of practices.

Educational, clinical, and organizational applications

Behaviorist principles influenced education through explicit objectives, frequent assessment, sequenced tasks, and immediate feedback. Skinner’s teaching machines and programmed instruction in the 1950s presented material in small steps and required active responding. The aim was not simply to mechanize lectures but to arrange conditions in which learners could progress through a controlled sequence and receive consequences closely connected to their responses. Later instructional technologies adopted some of these features without necessarily sharing Skinner’s wider philosophy.

In classrooms, behavioral methods have included reinforcement of task engagement, systematic practice, and adjustments to antecedent conditions. Their effectiveness depends on the behavior being measured, the instructional context, and whether the acquired skill transfers beyond the training arrangement. A rise in correct responses on rehearsed items does not by itself establish conceptual understanding or flexible application. Behavioral approaches to pedagogy therefore intersect with broader research on curriculum, motivation, and learning, rather than providing a complete educational theory.

Applied behavior analysis developed as a field concerned with socially significant behavior and experimentally demonstrated relations between interventions and outcomes. The journal Journal of Applied Behavior Analysis began publication in 1968. In an influential article that year, Donald Baer, Montrose Wolf, and Todd Risley described defining dimensions of the field, including that it should be applied, behavioral, analytic, effective, and capable of producing generalizable change. Applications have included communication, everyday skills, classroom behavior, workplace practices, and safety.

Within clinical psychology, behavior therapy drew on learning research as an alternative to explanations centered on unconscious conflict. Joseph Wolpe developed systematic desensitization in the 1950s, while other researchers studied exposure, reinforcement-based interventions, and skills training. Contemporary exposure therapy retains a connection to research on fear learning and extinction, although its theoretical accounts now include expectations, context, and inhibitory learning. Behavioral activation similarly focuses on patterns of activity, avoidance, and contact with environmental reinforcement.

Cognitive behavioral therapy subsequently combined behavioral procedures with approaches directed at interpretations, beliefs, and other cognitive processes. Its development illustrates continuity without identity: a treatment may incorporate behavioral experiments and activity scheduling while rejecting a strictly behaviorist account of mind. Organizational applications show a similar pattern, using measurable goals, feedback, and consequences to investigate performance. Across these settings, the scientific question is whether a specified procedure produces a defined, meaningful change; a procedure’s behaviorist ancestry alone does not establish its effectiveness or suitability.

Ethics, explanatory limits, and continuing influence

Behaviorism’s language of prediction and control made ethical questions especially visible. Experimental demonstrations that environments can alter behavior do not determine who should choose the goals, whose interests should prevail, or which methods are permissible. Skinner’s Beyond Freedom and Dignity of 1971 argued for examining the environmental conditions that shape action rather than relying primarily on praise, blame, or appeals to autonomous agency. The book generated debate about responsibility and social organization, but its philosophical proposals must be distinguished from experimentally established learning effects.

Applied settings raise concrete questions about consent, autonomy, acceptable risk, and the meaning of successful change. A behavior can be inconvenient to an institution without being harmful to the person displaying it. Conversely, the suppression of a visible response may conceal the persistence of distress or leave an important need unmet. These distinctions are particularly important when interventions involve children, people with disabilities, or individuals whose freedom to refuse is restricted. Relevant ethical frameworks extend beyond behavioral science to human rights and professional standards.

The history of behavior modification includes the use of aversive and coercive procedures. The fact that a consequence changes behavior is not an ethical justification for imposing it. Evaluation therefore requires attention to the intervention’s goals, participants’ preferences, adverse effects, and the availability of less restrictive alternatives. Outcome measurement also matters: response frequency, generalization, maintenance, participant experience, and quality of life can yield different assessments of the same program.

At the explanatory level, behaviorism remains strongest where it specifies measurable relations and tests them under controlled conditions. Disagreement persists over whether a sufficiently developed analysis of those relations can explain cognition without representational theories, and over how physiological, cognitive, and behavioral explanations should be integrated. Demonstrating an environmental influence does not exclude a neural mechanism; identifying a neural mechanism does not eliminate the need to explain how experience established its behavioral function.

One important continuation appears in reinforcement learning, a field of machine learning concerned with learning policies through interaction and evaluative feedback. Its vocabulary and some of its problems reflect connections with animal learning, but modern algorithms also rely on mathematical frameworks and computational assumptions not supplied by behaviorism alone. In experimental psychology, the continuing investigation of schedules, choice, stimulus control, and extinction preserves a more direct lineage. These research programs increasingly coexist with models of internal computation, evolutionary constraints, and neural implementation rather than requiring adherence to a single historical school.