Learning outcomes
By the end of this lesson you should be able to: describe the elements of classical conditioning and the phenomena of acquisition, extinction, spontaneous recovery, generalisation and discrimination; distinguish positive and negative reinforcement from punishment and explain why the distinction matters clinically; compare the four schedules of reinforcement and their effect on resistance to extinction; outline the four processes of observational learning and the concept of reciprocal determinism; and describe attention, language and problem-solving well enough to apply them in behavioural and cognitive therapy.
Prerequisites
Lesson 1 (Motivation and emotion). You need the drive and arousal ideas from that lesson, and above all the distinction between removing an aversive stimulus and applying a punishment.
The big idea
Learning is the mechanism by which psychiatric disorder is built and the mechanism by which it is dismantled. A phobia is a learned association between a neutral cue and intense fear; a panic attack becomes a disorder when the attacks themselves become a conditioned cue for further attacks; a compulsion is maintained because it buys relief. Every psychological treatment with real evidence rests on learning theory: exposure dismantles a conditioned fear response, behavioural activation re-establishes contact with reinforcement, and contingency management changes what behaviour pays. Learn the principles here once and the therapy lessons later will make sense rather than feel arbitrary.
Classical conditioning: learning associations between events
Pavlov's dogs established the canonical paradigm. An unconditioned stimulus (the sight and smell of food) reliably produces an unconditioned response (salivation). A neutral stimulus (a bell) is paired with the unconditioned stimulus; after enough pairings the neutral stimulus becomes a conditioned stimulus, and it now produces a conditioned response on its own.
| Element | Definition | In Pavlov's experiment | In a spider phobia |
|---|---|---|---|
| Unconditioned stimulus (UCS) | Stimulus that reliably elicits a response without prior learning | Food in the mouth | The sight of the spider at close quarters |
| Unconditioned response (UCR) | The unlearned response to the UCS | Salivation | Fear and escape |
| Conditioned stimulus (CS) | Previously neutral stimulus that now elicits the response after pairing | The bell | Any spider, a picture of one, even the word |
| Conditioned response (CR) | The learned response elicited by the CS | Salivation to the bell alone | Fear, avoidance, anticipatory anxiety |
Five phenomena govern how robustly the association forms, and each one maps onto a clinical fact. Acquisition is strongest when the CS reliably precedes the UCS and is close to it in time — a single intense pairing can be enough, which is why one dog bite can produce a lifelong fear. Extinction occurs when the CS is repeatedly presented without the UCS, and is the basis of exposure therapy. Spontaneous recovery is the return of the extinguished response after a delay, which is why patients are taught to expect setbacks after treatment rather than to read them as failure. Stimulus generalisation is the transfer of the response to similar stimuli — after a road traffic accident, cars, engines, motorway signposts and the smell of fuel all acquire the power to frighten. Stimulus discrimination is the learned narrowing of the response to the reliable cue, and it is what a graded exposure hierarchy builds deliberately.
A woman panics in a supermarket queue and is taken to hospital; nothing is found. Two weeks later she panics in a queue at the post office. The queue, the heat, the crowding and the bodily sensation of standing still have all become conditioned cues. This is how panic disorder with agoraphobia develops: the panic is originally physiological, but the situations in which it occurred become danger signals in their own right, and avoidance of those situations is negatively reinforced by the relief it produces.
NoteCounterconditioning is the deliberate replacement of a conditioned fear response with an incompatible one, and systematic desensitisation applies it: the patient is taught relaxation and then exposed to the feared stimulus in a hierarchy ordered from mild to severe, remaining relaxed throughout. It remains the historical template from which modern graded exposure developed.
Operant conditioning: learning from consequences
Edward Thorndike's law of effect stated that behaviour followed by a satisfying consequence is more likely to recur, and B. F. Skinner's operant analysis made consequences the central explanatory unit. The four-way classification below is the vocabulary that every behavioural formulation depends on.
| Contingency | Effect on behaviour | Definition | Clinical example |
|---|---|---|---|
| Positive reinforcement | Increases it | A desirable stimulus is delivered after the behaviour | Attention, comfort and reassurance follow a self-harm attempt, making help-seeking that is entangled with harm more likely to recur |
| Negative reinforcement | Increases it | An aversive stimulus is removed or avoided after the behaviour | Checking the lock removes anxiety, so checking is repeated — and avoidance of leaving the house removes the panic, so avoidance is repeated |
| Positive punishment | Decreases it | An aversive stimulus follows the behaviour | A painful injection after an overdose attempt; rarely justified and rarely effective in sustained behaviour change |
| Negative punishment | Decreases it | A desirable stimulus is removed after the behaviour | Loss of visiting privileges after an incident — the operative mechanism in behavioural contracts on inpatient units |
Watch outNegative reinforcement is not punishment. It strengthens behaviour. The two most common exam and clinical errors in this whole lesson are (a) calling the removal of anxiety a punishment, and (b) thinking that because checking reduces anxiety it must be a reasonable coping strategy. It reduces anxiety today and entrenches the disorder over years — the defining criterion of negative reinforcement.
Escape learning produces a behaviour that terminates an ongoing aversive state; avoidance learning produces a behaviour that prevents it from starting. Avoidance is the more clinically important of the two because it never extinguishes the fear: the patient stops encountering the situation, so the expectation that the situation is dangerous is never tested and never disconfirmed. That single sentence is the rationale for exposure and response prevention.
Schedules of reinforcement account for why some behaviours are extremely resistant to extinction. Continuous reinforcement (every response rewarded) produces rapid learning and rapid extinction. Intermittent reinforcement — and variable-ratio schedules in particular — produces behaviour that persists through long stretches of non-reward.
| Schedule | Rule | Effect on response rate | Effect on extinction |
|---|---|---|---|
| Fixed interval (FI) | Reward for the first response after a fixed time has elapsed | Scalloped: little responding just after reward, rising as the interval ends | Moderate |
| Variable interval (VI) | Reward for the first response after an unpredictable interval | Steady, moderate rate | High |
| Fixed ratio (FR) | Reward after a fixed number of responses | High rate with a post-reinforcement pause | Moderate |
| Variable ratio (VR) | Reward after an unpredictable number of responses | Very high, steady rate | Very high — the most extinction-resistant schedule |
NoteSlot machines are variable-ratio devices, and so is addiction's relationship with the drug-seeking environment. Variable-ratio reinforcement explains why a behaviour that is rewarded only occasionally — gambling, checking, reassurance-seeking out of an intermittent payoff — is harder to give up than one rewarded every time.
Observational learning and social cognition
Albert Bandura showed that much learning occurs by watching, with no direct reinforcement of the learner at all, and that the model's consequences matter to the observer. His Social Learning Theory specifies four processes: attention (the observer must notice the behaviour), retention (it must be encoded in memory, often verbally), motor reproduction (the observer must be physically able to perform it), and motivation (the observer must expect a valued outcome, including vicarious reinforcement — seeing the model rewarded — or vicarious punishment).
Bandura's later construct of reciprocal determinism places person, behaviour and environment in a two-way causal loop: people choose environments, environments shape people, and behaviour changes both. This is why a purely environmental behaviourism is inadequate as a model of psychiatric disorder — and also why a purely trait-based model is inadequate. Modelled mastery, in which the observer watches someone similar succeed, is the strongest single ingredient of self-efficacy, and self-efficacy predicts treatment adherence and outcome in depression, chronic pain and substance dependence.
Attention, language and problem-solving
| Domain | What it covers | Relevance in psychiatry |
|---|---|---|
| Selective attention | Focusing on one stimulus stream while filtering others; the cocktail party effect is the demonstration | Impaired in depression and anxiety, where attention locks onto threat-relevant material; in ADHD; after traumatic brain injury |
| Sustained attention / vigilance | Maintaining focus over time; degrades with fatigue and sleep loss | Deteriorates in delirium and dementia before memory does, and is what acute withdrawal impairs |
| Divided attention | Attending to two streams simultaneously | Impaired in delirium and in dementia; the basis of dual-task gait testing |
| Language | Comprehension and production; Broca's area in the left inferior frontal gyrus for production, Wernicke's in the superior temporal gyrus for comprehension | Expressive vs receptive dysphasia after stroke; pressured speech in mania; poverty of speech in schizophrenia; concrete interpretation in intellectual disability and autism |
| Problem-solving | Converting a goal state from the present state: algorithms (exhaustive, guaranteed) versus heuristics (efficient, imperfect) | Heuristics are the cognitive therapist's target; impaired problem-solving predicts relapse in depression and psychosis |
| Insight learning | Sudden restructuring of the problem rather than trial and error | Kohler's chimpanzees; insight-based psychotherapy is named by analogy but is not the same phenomenon |
Two functional obstacles recur. Functional fixedness is the tendency to see objects only in their usual function, and it is what makes a patient unable to imagine any use for an empty bottle other than the one that harms them. Mental set is the tendency to apply a strategy that has worked before regardless of whether it fits the current problem, and it is the mechanism by which an entrenched avoidance response is applied to new situations that never warranted it.
Common mistakes
Watch outAttributing every behaviour change to conditioning. Conditioning is powerful and general, but it does not explain language acquisition, insight, or the fact that patients deliberately do things they dislike because of a long-term goal. Operant accounts of inpatient behaviour fail when staff assume a contingency is present when it is not.
Watch outDelivering intermittent reinforcement by accident. A ward that responds quickly to reprimands but slowly to calm cooperation has accidentally placed the disturbed behaviour on a variable-ratio schedule and made it more resistant. Consistency, not severity, is what changes behaviour.
Watch outUsing Watson's Little Albert experiment as evidence that phobias are all acquired. The study is ethically indefensible by modern standards, was never replicated with adequate controls, and its arithmetic has been questioned. It is historically important; it is not proof, and the finding that phobias cluster on evolutionarily relevant stimuli (snakes, heights, water) argues for prepared learning rather than blank-slate conditioning.
Check your understanding
A man with obsessive-compulsive disorder checks the gas hob twenty times each night. His partner reassures him each time. Explain in operant terms what is being reinforced, and by whom.
Why does a variable-ratio schedule produce behaviour more resistant to extinction than a fixed-ratio schedule?
Name the four processes Bandura identifies as necessary for observational learning.
Summary and next
Classical conditioning explains how cues acquire emotional meaning, including the pairing of a panic attack with the place it happened; operant conditioning explains how consequences stamp behaviour in or out, and negative reinforcement of avoidance is the mechanism that makes anxiety disorders persistent. Schedules of reinforcement determine how extinction-resistant a behaviour will be. Observational learning adds a social route, and attention, language and problem-solving are the cognitive capacities that determine how a patient can engage with any of it.
Next: memory and thinking. You will see how the associations formed by conditioning are encoded, stored and retrieved, how those systems break down in amnesia and dementia, and how the heuristics of everyday thinking become the cognitive distortions targeted in therapy.
Multiple choice
A patient's panic attack while travelling on a bus now makes her unable to queue anywhere. Which phenomenon best explains the spread from buses to queues?
Which of the following is negative reinforcement?
Which schedule of reinforcement produces the highest resistance to extinction?
According to the two-factor theory of avoidance learning, a phobia is maintained primarily by which mechanism?
Which pairing of cognitive deficit and disorder is most accurate?