1.3 Learning theory

Four models of learning

Classical conditioning pairs two stimuli, whereas operant conditioning pairs a behaviour with its consequence.

Classical conditioning

Pavlov originated classical conditioning through his work on salivation in dogs, and Watson later applied it to human emotion.

• Classical conditioning acts on reflexes, so conditioned responses are involuntary and often autonomic.

• Its vocabulary of stimuli, responses and pairing is separate from the operant terms for consequences and schedules.

• In Little Albert (Watson and Rayner, 1920), a white laboratory rat was paired with a noise, so the rat was the conditioned stimulus.

• Acquisition strengthens across paired trials towards a plateau.

• Prediction matters more than simple pairing (Rescorla): a tone predicts little if shocks also occur without it.

Tap to enlarge

Timing of the stimuli

Tap to enlarge

Extinction and related processes

• Spontaneous recovery shows that extinction suppresses rather than erases learning: an inhibitory memory forms beside the original association and competes with it.

• Generalisation turns a fright in a lift into fear of all enclosed spaces, and spread Little Albert's fear to other furry objects.

• Habituation involves one stimulus, whereas extinction weakens a learned association between two.

Tap to enlarge

Classical conditioning in clinical problems

• A phobia can be acquired when a neutral stimulus is paired, even once, with a sufficiently frightening event.

• Preparedness (Seligman) means humans are evolutionarily readied to fear ancestral dangers, such as snakes, spiders and heights, far more easily than modern hazards.

• Incubation (Eysenck) is the strengthening of a conditioned fear when the feared stimulus is met only briefly and escaped while anxiety is rising, so extinction never occurs.

• Anticipatory nausea or vomiting to cues repeatedly paired with chemotherapy is classical conditioning.

• Taste aversion can follow a single pairing of a flavour with nausea hours later.

Operant conditioning

• Operant conditioning is learning in which the consequences of a person's own voluntary behaviour decide how likely it is to be repeated (Skinner), as in reward charts and token programmes.

• Thorndike's law of effect states that satisfying consequences strengthen a response; his puzzle-box cats improved by trial and error.

• Skinner's contingency runs antecedent, behaviour, consequence, where the antecedent signals that a consequence is available.

Contingencies have two independent axes.

• Positive or negative says whether a stimulus is added or removed, and neither term means pleasant or unpleasant.

• Reinforcement or punishment says whether the behaviour increases or decreases.

• Negative reinforcement maintains escape, avoidance, safety behaviours, compulsions and reassurance seeking in anxiety disorders.

• Pain relief is both a primary and a negative reinforcer, because the two terms describe different properties.

Reinforcers and timing

• Learning is strongest when reinforcement follows within about half a second, is disrupted from 0.5 to 5 seconds and largely fails beyond about 5 seconds.

• Taste aversion is the exception and tolerates long delays.

• The Premack principle states that a more probable, preferred activity reinforces a less probable one when access to it depends on doing the less preferred one first, such as pudding after vegetables.

• Selective reinforcement of a response already in the repertoire over an alternative is plain operant conditioning, not shaping.

Escape and avoidance

• Mowrer's two-factor theory holds that fear is classically conditioned to the warning signal and avoidance is negatively reinforced by the fall in fear.

• Active or ritualised avoidance maintains OCD and agoraphobia through relief, and blocks extinction because the feared outcome is never disconfirmed.

• In obsessive-compulsive disorder (OCD), washing relieves contamination anxiety and earns credit for preventing catastrophe; attention, reassurance and reduced responsibility can add reinforcement.

Extinction and punishment

• Operant extinction is the decline of a previously reinforced response when the reinforcer is withheld, after a brief extinction burst in which the behaviour increases.

• Punishment works best when immediate, consistent and paired with reinforcement of an alternative, and reinforcing an alternative is preferred where possible.

• Punishment suppresses behaviour without teaching a replacement or meeting the underlying need, and the suppression may last only while the punisher is present.

• Punishment can cause fear, avoidance and counter-aggression.

Schedules of reinforcement

• Continuous reinforcement gives fast learning and fast extinction, whereas intermittent reinforcement is learned more slowly but extinguishes more slowly.

• Behaviour is best established with continuous reinforcement, which is then thinned.

Tap to enlarge

Ratio schedules reward the count of responses and so encourage fast responding; interval schedules depend on time elapsed.

Shaping and chaining

• Shaping reinforces successive approximations to a target behaviour not yet in the repertoire, with a moving criterion so that cruder attempts stop being rewarded.

• Chaining builds a complex skill from separate responses that are taught and linked, each completed step cueing the next.

• Backward chaining teaches the last step first, so the learner always completes the task and is reinforced.

• Shaping creates a new behaviour and always runs forwards, whereas chaining links responses into a sequence and runs forwards or backwards.

• Prompts cue the desired behaviour and are then faded until natural cues control it.

Observational learning

• Social learning theory (Bandura) holds that people learn by watching a model, without direct reinforcement; modelling is Bandura's contribution, not a conditioning theorist's.

• In the Bobo doll experiments, children imitated a model's aggression more when the model was rewarded than when punished.

• Vicarious punishment reduced imitation, but offering incentives equalised performance, so the punished acts had been learned and only their performance was suppressed.

• Bandura's four processes are attention, retention, motor reproduction and motivation, where motivation includes vicarious reward or punishment.

• Imitation is more likely when the model is high in status, prestige or competence, similar, warm and visibly rewarded.

• Gender identity and gender-typed behaviour are, in social learning theory, learned through observation, imitation, reinforcement of models and cognitive processing.

• Kohlberg's cognitive-developmental theory instead puts gender constancy first.

Cognitive learning

Cognitive accounts stress expectations about how events relate, not strengthened reflexes, and insight is restructuring rather than gradual trial and error.

Behavioural treatments for fear

• Counterconditioning replaces a conditioned fear by pairing the feared stimulus with a pleasant, incompatible response, as in Mary Cover Jones's treatment of a boy's fear of a rabbit.

• Counterconditioning differs from extinction, which presents the feared stimulus without the original harm.

• Reciprocal inhibition (Wolpe) is the suppression of anxiety by an incompatible response, such as deep relaxation or eating, evoked with the feared cue; Wolpe showed it by feeding cats.

• Systematic desensitisation pairs learned relaxation with a graded hierarchy of feared situations, worked from least to most feared, moving on when anxiety at each step is low.

• Systematic desensitisation rests on reciprocal inhibition and is the behavioural treatment of choice for specific phobia.

• A specific phobia from a single conditioning event, without features of post-traumatic stress disorder, is treated with exposure-based behaviour therapy.

• Implosion is prolonged exposure in imagination, ungraded and starting at the top of the hierarchy, with escape blocked and no relaxation, until anxiety extinguishes.

• Flooding is the same procedure in vivo.

• Traditional exposure aimed for anxiety to fall, but exposure can build inhibitory learning without anxiety falling within the session.

• Exposure and response prevention for OCD combines exposure to the obsessional trigger with complete prevention of the compulsive response.

• Habit reversal training, built on awareness training and a competing response, is the best-evidenced behavioural treatment for trichotillomania.

• Biofeedback covers muscle, heart and breathing, and extends to central nervous system function through electroencephalographic feedback (neurofeedback).

Aversive techniques

• Aversive conditioning pairs an unwanted behaviour, or its cues, with a real noxious stimulus so that the behaviour becomes less frequent.

• Aversive conditioning pairs the cue with something unpleasant, whereas counterconditioning pairs it with a pleasant, incompatible state.

• Covert sensitisation pairs an unwanted urge with an imagined unpleasant consequence, so both halves happen in imagination.

• Shame aversion therapy treats paraphilic behaviour by pairing it with the embarrassment of performing the act before observers.

• Supervised disulfiram maintains abstinence by negative reinforcement, removing the anxiety of deciding repeatedly whether to drink, not by aversive conditioning; most patients never have the reaction.