Defining Learning

Learning is described as a relatively enduring change in the potential to engage in a particular behaviour as a result of experience with environmental events (Domjan). It reflects modifications in behavioural potential rather than momentary performance.

Different forms of learning involve different underlying processes:

• Non-associative learning involves alterations in attention or responsiveness to stimuli (habituation, sensitisation).
• Classical conditioning involves learning contingencies between two stimuli (S–S associations).
• Operant conditioning involves learning contingencies between behaviour and outcomes (R–O associations), often dependent on contextual stimuli (S).

Behaviourism and the Early Learning Theories

Behaviourism argued that learning consisted of forming associations between stimuli and responses, strengthened through reinforcement. Internal mental states were dismissed because they were not directly observable.

Major Behaviourist Figures

Theorist	Contribution
John Watson (1878–1958)	Emphasised observable behaviour; argued behaviour is shaped entirely by environmental experience; rejected mentalistic explanations.
B.F. Skinner (1904–1990)	Developed operant conditioning, reinforcement schedules, and the Skinner Box as a controlled environment for studying behaviour.
Edward Tolman (1886–1959)	Introduced purposive behaviourism; argued behaviour is goal-directed and influenced by internal variables such as expectations and thirst.

Challenges to Behaviourism

Behaviourism struggled to explain phenomena involving internal structure, planning, or sudden learning. A central question emerged: Can learning occur without reinforcement?

Tolman’s Contributions

Tolman proposed that organisms form cognitive maps—internal representations of spatial layout and contingencies—and that learning can occur without reinforcement (latent learning).

Latent learning occurs when knowledge is acquired without reinforcement and only expressed when a reward becomes relevant. In Tolman and Honzik’s (1930) maze studies, rats learned the layout without food but performed optimally once food was introduced.

Insight Learning

Insight learning involves sudden restructuring of understanding rather than gradual trial-and-error learning. Wolfgang Köhler’s studies with chimpanzees demonstrated that animals could reorganise their perception of a problem (e.g., stacking boxes to reach bananas), producing an abrupt solution (“aha!” moment).

Classic insight problems include:

• The candle problem, requiring functional reorganisation.
• The nine-dot puzzle, requiring breaking implicit assumptions (“think outside the box”).

These findings challenged behaviourist accounts of incremental S–R strengthening.

Limits of Behaviourism and the Cognitive Revolution

Behaviourism could not account for language, reasoning, or complex problem-solving.

Chomsky’s Critique

Noam Chomsky (1959) argued that:

• Children acquire grammar too rapidly to be explained by reinforcement alone.
• Language learning requires innate cognitive structures.

This critique catalysed the Cognitive Revolution, shifting focus from S–R associations to information processing.

Cognitive Approaches to Learning

Cognitive psychology conceptualises the mind as an information processor, similar to a computer. It encodes, stores, and manipulates information using internal symbols and representations.

Key Cognitive Theorists

Theorist	Contribution
George Miller	Identified the 7±2 limit of working memory; advanced information-processing models.
Herbert Simon	Developed symbolic models of cognition and problem-solving.
Rumelhart & McClelland	Created connectionist (PDP) models; emphasised distributed learning across networks of simple units.

Connectionism and Artificial Neural Networks

Connectionism proposes that knowledge is encoded in networks of interconnected nodes. Learning occurs through alterations in the strength of connections (weights).

Artificial Neural Networks (ANNs) model biological neurons by:

• integrating multiple inputs,
• adjusting connection weights based on experience,
• emphasising distributed representation rather than rule-based logic.

Prediction Error

Learning depends on the difference between expected and actual outcomes:

Prediction Error Type	Effect on Learning
Positive error	Outcome better than expected → strengthens learning
Negative error	Outcome worse than expected → weakens learning
Zero error	Outcome matches expectation → no further learning

These principles parallel the Rescorla–Wagner model of classical conditioning.

Biological Basis of Learning

Biological learning depends on long-lasting synaptic changes such as:

• Long-Term Potentiation (LTP): Strengthening of synapses following high-frequency stimulation.
• Long-Term Depression (LTD): Weakening of synapses after low-frequency stimulation.

Both processes depend on NMDA and AMPA receptors and support Hebb’s principle: “Cells that fire together, wire together.”

LTP and LTD reflect structural strengthening or weakening of neural circuits and underpin memory formation and behavioural change.

Social Learning

Not all learning requires direct reinforcement. Social learning theory emphasises learning through observation.

Forms of Social Learning

Type	Description
True imitation	Copying actions with the same goal and technique (e.g., macaques washing potatoes; cockatoos opening bins).
Contagion	Automatic, reflexive mimicry (e.g., yawning).
Observational conditioning	Learning emotional responses by observing others (e.g., fear acquisition).
Stimulus enhancement	Attention drawn to an object due to another’s interaction.
Social transmission	Knowledge passed socially without action imitation.

Miller & Dollard (1941)

Proposed that imitation follows operant conditioning principles: behaviours are imitated if reinforced.

Generalized Imitation

Repeated reinforcement of imitation leads to imitation becoming a learned behavioural class. Children often imitate new behaviours spontaneously after earlier imitation has been rewarded.

Bandura and the Cognitive Expansion of Social Learning

Albert Bandura challenged behaviourist accounts by showing that imitation is cognitively mediated.

Bobo Doll Experiment (1965)

• Children imitated aggressive behaviour when the model was rewarded.
• Punishment reduced performance but not learning.
• When later offered a reward, children demonstrated the learned aggressive actions.

This showed that learning ≠ performance and that cognitive evaluation of consequences influences behaviour.

Bandura’s Four Processes of Observational Learning

Process	Role
Attention	Focusing on the model’s behaviour
Retention	Storing observed behaviour in memory
Motor reproduction	Ability to reproduce the behaviour
Motivation	Anticipated reinforcement guiding whether behaviour is performed

Summary

Learning is not solely based on conditioning or reinforcement. Internal representations, cognitive processes, and biological mechanisms all contribute to adaptive behaviour. Learning can occur:

• without reinforcement (latent learning),
• abruptly through insight,
• socially through observation,
• or biologically through synaptic modification (LTP/LTD).