The multi-store model (Atkinson and Shiffrin, 1968)

The MSM treats memory as an information-processing system with three separate stores, linked by attention and rehearsal.

• Sensory register — huge capacity, duration under about two seconds, modality-specific coding. Information is lost unless we pay attention to it.
• Short-term memory — limited capacity (Miller, 1956: about 7 ± 2 items), limited duration (Peterson and Peterson, 1959: roughly 18–30 seconds without rehearsal), coded mainly acoustically.
• Long-term memory — potentially unlimited capacity and duration, coded mainly semantically (by meaning). Maintenance rehearsal transfers material from STM to LTM; retrieval brings it back.

Evaluate it: it is supported by HM (a damaged STM–LTM link) and by the serial position curve, but it is too simple: it treats STM and LTM as single, unitary stores, and it cannot explain why rehearsal is not always needed (a flashbulb memory is stored instantly).

The working memory model (Baddeley and Hitch, 1974)

Baddeley and Hitch replaced the single STM with an active, multi-component system.

• Central executive — attentional control system; limited capacity; allocates work to the slave systems. Modality-free.
• Phonological loop — deals with sound; made of the phonological store (the inner ear) and the articulatory process (the inner voice, which rehearses).
• Visuo-spatial sketchpad — the 'inner eye'; visual and spatial information.
• Episodic buffer (added by Baddeley in 2000) — a limited-capacity temporary store that integrates information from the other components and links to LTM.

Key evidence — dual-task studies: people can do a visual task and a verbal task at the same time with little cost, but two tasks that use the same component interfere badly. That is very hard for the MSM's single STM to explain.

Which component?

Types of long-term memory (Tulving, 1972)

Tulving argued LTM is not one store. Edexcel requires episodic and semantic memory.

• Episodic memory — personally experienced events, time-stamped and context-rich ("my first day at school"). Consciously recalled (declarative) and vulnerable to forgetting.
• Semantic memory — facts and general knowledge ("Paris is the capital of France"), not tied to a time or place, and more resistant to forgetting.
• Procedural memory — skills such as riding a bike; recalled without conscious effort. HM could learn new procedural skills while denying he had ever done the task before.

Why it matters: if brain damage or dementia can strip out one type while sparing another, LTM cannot be the single unitary store the MSM claims.

Which store or component?

Tap a description, then tap where it belongs.

Classic study: Baddeley (1966b)

Baddeley tested how short-term and long-term memory encode information, using word lists that were either acoustically similar, semantically similar, or unrelated (control).

• Acoustically similar words: man, cab, can, cad, mad. Semantically similar words: great, large, big, huge, broad.
• Immediate recall (testing STM) was clearly worse for the acoustically similar list — the sounds got confused.
• Recall after a delay with an interference task (testing LTM) was worse for the semantically similar list, while the acoustic effect had largely disappeared.

Conclusion: STM codes largely by sound, LTM largely by meaning — supporting separate stores and the phonological loop. Evaluate it: a well-controlled lab experiment, but word lists are artificial (low ecological validity) and the effect may not hold for all material.

Encoding

Reconstructive memory and schema theory (Bartlett, 1932)

Bartlett argued that remembering is an active reconstruction, not a replay. We fit new material into existing schemas — packets of prior knowledge and expectation.

• In The War of the Ghosts, English participants repeatedly recalled a Native American folk tale. Their versions became shorter, unfamiliar details were omitted or changed into familiar ones (canoes became boats), and the story was rationalised to make sense within their own culture.
• Levelling (losing detail), sharpening (exaggerating some elements) and assimilation (making it fit our culture) describe what happens.

Applications: reconstructive memory explains why eyewitness testimony is unreliable, and it directly contradicts the idea that memory is a faithful recording. Criticism: Bartlett's procedure was not well controlled and his scoring of 'distortion' was subjective.

Schema in action

Case study: HM · and developmental memory

Henry Molaison (HM) had large parts of the medial temporal lobes, including most of the hippocampus, removed in 1953 to control severe epilepsy.

• He was left with severe anterograde amnesia — he could not form new long-term memories — while his STM and much of his older memory and his procedural learning were intact.
• Strengths of the case study: rich, detailed, longitudinal data on a rare condition; it provided strong evidence for separate memory stores.
• Weaknesses: a single unique case cannot be generalised; the damage was not experimentally controlled; and much of the data is qualitative and open to researcher interpretation.

Developmental: memory span is not fixed. Sebastián and Hernández-Gil (2012) mapped digit span across childhood in a Spanish population and found span increases with age through childhood into adolescence. Dyslexia and Alzheimer's disease both affect memory span and working memory, showing individual differences.

Match the term to its meaning

Tap a term on the left, then its correct definition on the right.

Designing an experiment

The Topic 2 method is the experiment.

• Lab vs field: lab experiments give strong control over extraneous variables and high internal validity; field experiments have higher ecological validity but weaker control.
• Designs: repeated measures (same participants — no participant variables, but order effects, so counterbalance), independent groups (no order effects, but participant variables, so randomly allocate), and matched pairs.
• Variables: operationalise the IV and DV; control extraneous variables before they become confounding ones.
• Threats: demand characteristics, experimenter effects; consider internal, predictive and ecological validity, and reliability.

Choosing a design

Choosing and interpreting an inferential test

Edexcel names two non-parametric tests of difference for this topic.

• Mann-Whitney U — test of difference, independent groups, ordinal data or above.
• Wilcoxon signed ranks — test of difference, repeated measures or matched pairs, ordinal data or above.
• Significance: the standard level in psychology is p ≤ 0.05 — a 5% or smaller probability that the result is due to chance. Compare the observed value with the critical value from the tables (checking one- or two-tailed, N, and the level of significance).
• Type I error — accepting a difference that is not really there (more likely if you set p at 0.10). Type II error — rejecting a real difference (more likely if you set p at 0.01).

Careful: for Mann-Whitney U and Wilcoxon, the observed value must be equal to or less than the critical value for the result to be significant.

Which test?

Contemporary study

How Topic 2 is assessed

Cognitive psychology is examined in Paper 1: Foundations in psychology and is compulsory.

• Expect to compare the multi-store model and the working memory model, and to use HM as evidence for both.
• Expect data-response on an experiment: identify the IV and DV, name the design, spot a confounding variable, choose the right inferential test and justify it, and compare an observed value with a critical value.
• You must be able to describe and evaluate Baddeley (1966b) and your chosen contemporary study.

The reliable mark-winner: whenever you evaluate a memory model, name a specific piece of evidence it cannot explain (the MSM cannot explain dual-task performance; the WMM says little about the central executive itself).

The big ideas to know

That is the whole of Topic 2 — a compulsory topic on Paper 1. Press Finish to see your score.