Learning

In short

Learning is a relatively permanent change in behaviour or behavioural potential that results from experience or practice, excluding changes due to maturation, fatigue, drugs or injury. It is the core explanatory principle of behavioural psychology and the base on which behaviour therapy, training design, advertising and classroom instruction all rest. In Indian BA/B.Sc psychology courses under UGC CBCS, Learning is a compulsory unit of the first-year Foundations/General Psychology paper, usually carrying a full essay question on conditioning paradigms. It is also examined in UGC-NET Psychology (Paper II, Unit on learning and motivation), in entrance tests for M.A. Psychology (DU, TISS, Ambedkar University, Christ) and in CTET/B.Ed. papers. Employers in HR, L&D, instructional design, ed-tech content, behavioural health and market research probe reinforcement principles and transfer of training in technical rounds.

Explained

Definition, criteria and types

• Learning: a relatively permanent change in behaviour or behavioural potential arising from practice or experience (Kimble's classical definition). Three defining criteria: change in behaviour, the change comes from experience, and it is relatively durable.
• Behavioural potential: learning may occur without immediate performance. Tolman's latent learning (rats learned a maze without reward and showed it once reward appeared) proves learning ≠ performance.
• Performance = learning + motivation + ability + opportunity. Exam-favourite distinction.
• Not learning: reflexes, instincts, maturation (a child walking), temporary states (fatigue, illness, intoxication), habituation to a repeated stimulus is usually treated as the simplest form of non-associative learning.
• Non-associative learning: habituation (declining response to a repeated harmless stimulus) and sensitisation (heightened response after an intense stimulus).
• Associative learning: classical conditioning (S-S), operant conditioning (R-S).
• Cognitive learning: insight, latent learning, cognitive maps, observational learning.
• Verbal learning, motor/skill learning, concept learning, discrimination learning are the standard content-based classification in Indian syllabi.

Classical (Pavlovian) conditioning

• Ivan Pavlov (1849-1936), Russian physiologist, studying salivation in dogs; Nobel Prize 1904 for digestion work. A neutral stimulus paired with a biologically potent one comes to elicit a similar response.
• Terms: UCS (unconditioned stimulus, e.g. food) → UCR (salivation, unlearned). CS (neutral stimulus, bell) paired repeatedly with UCS → CR (salivation to bell alone).
• Acquisition: the CS-UCS pairings; CR strength grows and reaches asymptote. Steepest early, then flattens — the classic negatively accelerated learning curve.
• Temporal arrangements (order of effectiveness):
  1. Forward delayed — CS begins, continues, UCS arrives before CS ends. Most effective.
  2. Trace — CS ends, gap, then UCS; works if the trace interval is short.
  3. Simultaneous — CS and UCS begin at the same moment; weak, because the CS adds no predictive information.
  4. Backward — UCS precedes CS; least effective, often produces inhibition.
• Optimal CS-UCS interval for most responses is roughly 0.2-0.5 s; eyeblink conditioning peaks near 0.5 s.
• Extinction: CS presented repeatedly without UCS → CR declines. Spontaneous recovery: CR reappears after a rest interval, showing extinction suppresses rather than erases.
• Stimulus generalisation: responding to stimuli resembling the CS (gradient falls with dissimilarity). Discrimination: differential responding after one stimulus is reinforced and a similar one is not.
• Higher-order conditioning: an established CS₁ serves as a UCS for a new CS₂.
• Blocking (Kamin): if CS₁ already predicts the UCS, a new CS₂ added in compound is not learned — conditioning needs informational value, not mere contiguity. Basis of the Rescorla-Wagner model (learning is proportional to prediction error, i.e. surprise).
• Applications: Watson and Rayner's "Little Albert" (conditioned fear of a white rat, 1920); systematic desensitisation and flooding in phobia treatment; aversion therapy for alcohol dependence (disulfiram); conditioned taste aversion (Garcia effect) with long CS-UCS intervals and single-trial learning; chemotherapy-related anticipatory nausea; advertising pairing products with attractive models.

Operant (instrumental) conditioning

• Thorndike's connectionism: cats in a puzzle box; trial-and-error with gradually falling escape times. Law of Effect — responses followed by satisfying consequences are strengthened; also Law of Exercise and Law of Readiness.
• B.F. Skinner: behaviour is shaped by its consequences. Skinner box (operant chamber), cumulative recorder, rate of response as the dependent variable. Three-term contingency: antecedent (discriminative stimulus Sᴰ) → behaviour → consequence (the ABC of behaviour analysis).
• Four contingencies:

• Reinforcer types: primary (food, water — biologically potent), secondary/conditioned (money, marks, tokens, praise), generalised (money, exchangeable for many reinforcers). Premack principle: a higher-probability activity can reinforce a lower-probability one.
• Schedules of reinforcement:
  • Continuous (CRF): every response reinforced; fastest acquisition, fastest extinction.
  • Fixed ratio (FR): every nth response; high steady rate with a post-reinforcement pause (piece-rate wages).
  • Variable ratio (VR): average of n responses; highest, steadiest rate and most resistant to extinction (slot machines, gambling, cold-calling sales).
  • Fixed interval (FI): first response after a fixed time; scalloped pattern (monthly salary, weekly quiz).
  • Variable interval (VI): first response after an average time; moderate, very steady (checking email, surprise tests).
  • Partial reinforcement effect (PRE): intermittently reinforced responses extinguish more slowly than continuously reinforced ones.
• Shaping: reinforcing successive approximations to a target behaviour; used in autism intervention and animal training. Chaining: linking units of a behaviour sequence (forward or backward).
• Escape vs avoidance learning: escape terminates an aversive stimulus already present; avoidance prevents its onset after a warning signal. Two-factor theory (Mowrer): classical conditioning of fear plus operant reinforcement of avoidance — explains why phobias and compulsions persist.
• Behavioural momentum, response cost, satiation, deprivation modulate reinforcer effectiveness.
• Limitations of punishment: suppresses rather than teaches, is situation- and punisher-specific, models aggression, produces fear and avoidance of the punisher, and risks habituation at low intensity. Effective punishment is immediate, consistent, moderate, explained, and paired with reinforcement of an alternative. Corporal punishment in Indian schools is prohibited under the Right of Children to Free and Compulsory Education Act, 2009 (Section 17) and by Juvenile Justice (Care and Protection of Children) Act, 2015 provisions on cruelty by a person in charge of a child.
• Instinctive drift (Breland and Breland) and preparedness/CS-UCS belongingness (Seligman) show biological constraints on conditioning — organisms are not equipotential.

Cognitive and social learning

• Köhler's insight learning: chimpanzees (Sultan) solved stick-and-box problems suddenly, after apparent restructuring of the problem — not gradual trial and error. Features: suddenness, complete solution, immediate transfer to similar problems, retention.
• Tolman's purposive behaviourism: rats form cognitive maps and expectancies; place learning over response learning; latent learning and reward-shift experiments.
• Bandura's observational (social) learning: the Bobo doll studies; learning by watching a model, no direct reinforcement needed. Four sub-processes: Attention → Retention → Reproduction (motor) → Motivation/reinforcement.
• Vicarious reinforcement/punishment: consequences to the model change the observer's behaviour. Reciprocal determinism: person, behaviour and environment interact. Self-efficacy beliefs predict effort and persistence.
• Learned helplessness (Seligman and Maier): dogs given inescapable shock later failed to escape when escape was possible; a model for depression and for passivity in low-performing students; reformulated with attributional style (internal-stable-global attributions for failure).
• Gagné's hierarchy (signal learning → S-R → chaining → verbal association → discrimination → concept → rule → problem solving) is a widely taught instructional sequence in Indian B.Ed. and training design.

Verbal learning, transfer and memory-linked phenomena

• Methods: serial anticipation, paired associates, free recall, serial recall — Ebbinghaus's nonsense syllables to control prior meaning.
• Serial position effect: best recall at the start (primacy) and end (recency) of a list.
• Massed vs distributed practice: spaced practice generally produces better retention; massed may give quicker short-term gains. Whole vs part learning: whole suits short, meaningful material; progressive-part suits long material.
• Overlearning: practice beyond the point of mastery; improves retention and resistance to stress, with diminishing returns.
• Knowledge of results (feedback): immediate, specific feedback accelerates skill learning.
• Transfer of training: effect of prior learning on new learning. Positive (typing helps keyboard use), negative (left-hand driving habits in a right-hand-drive country), zero. Theories: identical elements (Thorndike), generalisation, transposition, learning-to-learn (Harlow's learning sets).
• Learning curves: negatively accelerated (rapid early gain — most common), positively accelerated, linear, and S-shaped; plateau = a flat period caused by fatigue, loss of motivation or reorganisation of skill, not a ceiling.
• Learning styles / VARK: popular in training but the matching hypothesis lacks robust evidence — state this as a critique, not a finding, if asked.

Measurement, applications and the Indian practice context

• Dependent measures: number of trials to criterion, errors, latency, rate of response, percentage correct, time taken, resistance to extinction.
• Behaviour modification: token economies in special schools and psychiatric wards, contingency contracting, differential reinforcement of other behaviour (DRO), extinction plus reinforcement of alternatives, time-out.
• Applied Behaviour Analysis (ABA) for autism spectrum disorder: discrete trial training, prompting and fading, task analysis. Delivered in India by clinical psychologists and special educators; RCI (Rehabilitation Council of India) registration is required to practise as a clinical psychologist/special educator, and the Rights of Persons with Disabilities Act, 2016 mandates inclusive education and reasonable accommodation.
• Programmed instruction: small steps, active response, immediate feedback, self-pacing — Skinner's teaching machines are the direct ancestor of adaptive ed-tech.
• Classroom practice: NCF/NEP 2020 emphasis on experiential, activity-based and competency-based learning aligns with reinforcement of desired behaviour and continuous formative feedback rather than punishment.
• Ethics: informed consent, right to withdraw, minimal harm, and avoiding aversive procedures — studies like Little Albert would not clear a modern Institutional Ethics Committee; Indian psychologists follow ICMR National Ethical Guidelines for Biomedical and Health Research Involving Human Participants (2017) and institutional IEC review.

Key points

• Learning - relatively permanent behaviour change from experience, excluding maturation, fatigue and injury.
• Learning vs performance - learning may stay latent; performance also needs motivation and opportunity.
• Classical conditioning - a neutral CS acquires the power to elicit a response by pairing with a UCS; response is elicited, involuntary, visceral.
• Operant conditioning - behaviour is emitted voluntarily and controlled by its consequences (reinforcement and punishment).
• Reinforcement - any consequence that increases behaviour; positive adds a stimulus, negative removes an aversive one. Negative reinforcement is not punishment.
• Punishment - decreases behaviour; suppresses without teaching an alternative and is legally restricted in Indian schools.
• Extinction and spontaneous recovery - withholding the UCS or reinforcer weakens the response, but it can return, so the original learning is not erased.
• Generalisation and discrimination - responding to similar stimuli vs responding differentially after selective reinforcement.
• Variable ratio - produces the highest, most persistent response rate; the schedule behind gambling.
• Partial reinforcement effect - intermittent reinforcement makes behaviour harder to extinguish than continuous.
• Shaping - reinforcing successive approximations to build a behaviour that never occurs spontaneously.
• Blocking and Rescorla-Wagner - conditioning depends on the CS's predictive information, measured as prediction error, not mere pairing.
• Insight learning - sudden perceptual restructuring with complete, transferable solutions (Köhler).
• Latent learning and cognitive maps - Tolman's evidence that learning can occur without reinforcement.
• Observational learning - attention, retention, reproduction, motivation; vicarious reinforcement (Bandura).
• Learned helplessness - uncontrollable aversive events produce passivity; a model of depression (Seligman).
• Transfer of training - positive, negative or zero carry-over of prior learning to a new task.
• Biological constraints - preparedness, taste aversion and instinctive drift show species are not equally conditionable to all pairings.

Formulas and examples

• Worked example 1 — Rescorla-Wagner acquisition. A tone (α = 0.5) is paired with a shock (β = 0.4, λ = 100). Start V = 0.
  • Trial 1: ΔV = 0.5 × 0.4 × (100 − 0) = 0.2 × 100 = 20; V = 20.
  • Trial 2: ΔV = 0.2 × (100 − 20) = 16; V = 36.
  • Trial 3: ΔV = 0.2 × (100 − 36) = 0.2 × 64 = 12.8; V = 48.8.
  • Trial 4: ΔV = 0.2 × (100 − 48.8) = 0.2 × 51.2 = 10.24; V = 59.04.
  • Interpretation: increments shrink each trial, producing the negatively accelerated learning curve, and learning stops when V reaches λ (prediction error = 0) — which is exactly why blocking occurs.
• Worked example 2 — Savings score. A student needs 24 trials to master a 15-word Hindi-English paired-associate list. After 48 hours she relearns it in 9 trials.
  • Savings = (24 − 9)/24 × 100 = 15/24 × 100 = 62.5% retention by the savings method.
  • Forgetting = 100 − 62.5 = 37.5%. Note the savings method detects retention that free recall would miss entirely.
• Worked example 3 — Percentage transfer. Control group learns task B in 40 trials. The experimental group, which first learned task A, needs 28 trials.
  • Transfer = (40 − 28)/40 × 100 = +30% positive transfer.
  • If instead the experimental group had needed 50 trials: (40 − 50)/40 × 100 = −25%, negative transfer (prior habits interfere).
• Worked example 4 — Schedule arithmetic and reinforcer cost. A token economy in a Bengaluru special school pays 1 token per 5 correct maths items (FR 5); each token is exchanged for ₹2 worth of canteen credit. A student completes 180 correct items in a 45-minute session.
  • Tokens earned = 180/5 = 36 tokens; cost = 36 × ₹2 = ₹72 per session.
  • Response rate = 180/45 = 4 correct items per minute.
  • Thinning the schedule to FR 12 gives 180/12 = 15 tokens = ₹30, a 58.3% cut in reinforcer cost; it also moves the behaviour toward intermittent reinforcement, so it should resist extinction better — but thinning too fast risks ratio strain and the behaviour collapsing.
• Worked example 5 — Hull's multiplicative rule. A trainee has habit strength ᶳHʀ = 0.8, drive D = 5, stimulus dynamism V = 1.0, incentive K = 2.
  • ᶳEʀ = 0.8 × 5 × 1.0 × 2 = 8.0 units of reaction potential.
  • If the incentive is withdrawn (K = 0), ᶳEʀ = 0 — the habit is fully intact but is not performed. This is the formal statement of the learning-performance distinction.

Where it is used

• Clinical and rehabilitation psychology: systematic desensitisation and exposure for phobias and OCD, aversion therapy in de-addiction centres, ABA for autism at institutions such as NIMHANS Bengaluru, AIIMS and RCI-registered centres; contingency management in substance-use programmes.
• Education and ed-tech: programmed instruction, spaced-repetition and adaptive practice engines in Indian ed-tech products, formative feedback and competency-based assessment under NEP 2020; teacher-training modules on reinforcement versus punishment for B.Ed. and DIET programmes.
• HR, L&D and organisational behaviour: behaviour modelling training, on-the-job training design using shaping and knowledge of results, incentive and variable-pay schemes that are effectively FR/VR schedules, behavioural safety programmes in manufacturing (BBS observation and feedback loops at Tata Steel-type plants).
• Marketing, consumer research and UX: classical conditioning in brand-celebrity pairing, loyalty points as token economies, variable-reward loops in gaming and short-video apps — a live policy debate on online real-money gaming in India.
• Sports and skill training: massed versus distributed practice schedules, part-whole practice, feedback timing in SAI and NIS coaching curricula.
• Animal training and wildlife management: shaping and clicker training for dogs and elephants; conditioned taste aversion used experimentally to deter crop-raiding and livestock predation.

Interview questions

• Q: How is negative reinforcement different from punishment? A: Negative reinforcement removes an aversive stimulus and therefore increases the behaviour; punishment adds an aversive stimulus or removes a pleasant one and decreases it.
• Q: Which reinforcement schedule produces the most persistent behaviour and why? A: Variable ratio, because the learner cannot predict which response will pay off, so non-reinforced responses do not signal that reinforcement has ended — this is the partial reinforcement effect.
• Q: Does spontaneous recovery mean extinction failed? A: No — it shows extinction inhibits the conditioned response rather than erasing the original association, which is why relapse prevention is built into exposure therapy.
• Q: What does the blocking effect tell us about classical conditioning? A: That contiguity alone is insufficient; a CS is learned only if it adds predictive information about the UCS, which the Rescorla-Wagner model formalises as learning driven by prediction error.
• Q: Give a workplace example of shaping. A: Training a new call-centre agent by first reinforcing the correct greeting, then greeting plus verification, then the full call flow — reinforcing successive approximations to the target behaviour.
• Q: How does Tolman's latent learning challenge Skinner? A: Rats learned a maze without any reinforcement and demonstrated it only once reward was introduced, showing learning can occur without reinforcement and that reinforcement affects performance rather than acquisition.
• Q: Why is punishment generally a poor management tool? A: It suppresses behaviour only in the punisher's presence, teaches no alternative response, generates fear, resentment and avoidance, and models aggression; reinforcing an incompatible desirable behaviour works better.
• Q: What are the four processes in observational learning? A: Attention to the model, retention of what was observed, motor reproduction, and motivation or reinforcement to perform it.
• Q: What is conditioned taste aversion and why is it theoretically important? A: A single pairing of a novel food with illness produces lasting avoidance even with hours between CS and UCS, contradicting the equipotentiality and short-interval assumptions of classical conditioning and supporting biological preparedness.
• Q: Distinguish escape from avoidance learning. A: Escape terminates an aversive stimulus that is already present; avoidance prevents its onset following a warning signal, and is highly resistant to extinction because the absence of the aversive event is itself reinforcing.
• Q: What is negative transfer and how would you design training to avoid it? A: Prior learning that interferes with new learning; reduce it by making the new task's stimuli and responses clearly distinct from the old, teaching the discrimination explicitly, and over-practising the new response.
• Q: A trainee performs well in practice but poorly on the floor. How would learning theory explain it? A: Learning may be intact while performance fails, because performance also needs drive, incentive and supporting stimulus conditions — Hull's multiplicative equation predicts zero performance if incentive or drive is zero.

Common mistakes

• Calling negative reinforcement "punishment" — the commonest single error in exams and interviews; "negative" refers to removing a stimulus, not to an unpleasant outcome.
• Confusing fixed/variable (predictability) with ratio/interval (responses versus time), and then naming the wrong real-world example.
• Writing that classical conditioning involves voluntary behaviour, or that operant conditioning works on reflexes.
• Treating extinction as unlearning and forgetting to mention spontaneous recovery, renewal and reinstatement.
• Describing insight learning as fast trial-and-error instead of perceptual restructuring, and omitting its transfer and retention features.
• Reporting Little Albert or the Bobo doll study without the ethical and methodological critiques (single subject, no debriefing, artificial aggression measure).
• Mixing up stimulus generalisation (similar stimuli, same response) with response generalisation and with transfer of training.
• Giving Pavlov's and Skinner's procedures correctly but failing to answer the actual question asked — which is usually a comparison or an application, not a narration of the experiment.

Related topics

Memory · Motivation and emotion · Schools of psychology · Biological bases of behaviour · Personality theories · Intelligence