Memory

🎯 Learning Objectives

• Describe and evaluate the multi-store model (Atkinson-Shiffrin 1968), citing coding, capacity and duration evidence (AQA 4.1.2)
• Describe and evaluate the working memory model (Baddeley-Hitch 1974): central executive, phonological loop, visuo-spatial sketchpad, episodic buffer (AQA 4.1.2)
• Distinguish episodic, semantic and procedural LTM with supporting case studies (HM, Clive Wearing) (AQA 4.1.2)
• Evaluate interference (proactive/retroactive) and retrieval failure as explanations of forgetting (AQA 4.1.2)
• Evaluate eyewitness testimony research: misleading information and anxiety (Loftus and Palmer, weapon focus) (AQA 4.1.2; OCR Loftus and Palmer 1974)
• Describe and evaluate the cognitive interview and its effectiveness (AQA 4.1.2)

📋 Key Definitions and Core Concepts

📌 Key Points

The multi-store model (Atkinson and Shiffrin, 1968) (AQA 4.1.2; OCR Baddeley 1966b)

The multi-store model (MSM) proposes three separate memory stores linked in a linear sequence by attention and rehearsal: sensory memory (iconic for visual, echoic for auditory; very large capacity but duration under one second, unless attention is paid, in which case information passes to...), STM (coding mainly acoustic, capacity of 7 ± 2 items - Miller 1956's 'magic number'; duration of 18-30 seconds - Peterson and Peterson 1959, where 90% of three-consonant trigrams were forgotten after 18 seconds of distraction) and LTM (coding mainly semantic, unlimited capacity, potentially lifelong duration). Information moves from sensory to STM via attention and from STM to LTM via rehearsal: maintenance rehearsal (repetition, keeps in STM) and elaborative rehearsal (linking to existing knowledge, transfers to LTM). Retrieval from LTM returns information to STM for conscious use.

Evidence: (1) Glanzer and Cunitz (1966) - the serial position effect: participants recalled words from the start of a list (primacy, LTM via rehearsal) and the end (recency, still in STM) but not the middle; a 30-second counting task before recall abolished the recency effect, supporting a separate STM store. (2) Baddeley (1966) - participants confused acoustically similar words (cat, cap, can) in STM but semantically similar words (great, large, big) in LTM, supporting different coding. (3) Case study KF (motorcycle accident) - impaired verbal STM but intact visual memory, challenging the idea of one unitary STM store. (4) Bahrick et al. (1975) - US graduates recognised yearbook photos up to 48 years later at 90% accuracy, showing LTM duration is potentially lifelong.

Evaluation (AO3):

• Strength - supporting evidence: brain-imaging (different active regions for STM/LTM tasks), case studies (HM, KF) and the serial position effect all converge on multiple stores.
• Limitation - oversimplified: KF's dissociation (impaired verbal, intact visual STM) contradicts a unitary STM; the working memory model supersedes it.
• Limitation - rehearsal not the only route: Craik and Lockhart's (1972) levels-of-processing approach and flashbulb memories show information enters LTM without prolonged rehearsal (deep processing matters more than repetition).
• Limitation - artificial evidence: most support (trigrams, word lists) uses lab tasks with low ecological validity; everyday memory is richer (faces, events).
• Strength - practical: the model underpins revision technique itself (elaborative rehearsal, chunking) and multi-store logic appears in exam questions on encoding/capacity/duration studies.

The working memory model (Baddeley and Hitch, 1974) (AQA 4.1.2)

The working memory model (WMM) replaces unitary STM with an active, multi-component system that manipulates information while performing tasks. Components: (1) the central executive - allocates attention and slave-system resources, has very limited capacity and no modality-specific storage; (2) the phonological loop (PL) - processes speech-based and auditory information, split into the phonological store (holds heard words, 'inner ear', ~2 seconds) and the articulatory process ('inner voice', allows subvocal rehearsal; its limited capacity explains the word-length effect - longer words are harder to recall because they take longer to rehearse, Baddeley et al. 1975); (3) the visuo-spatial sketchpad (VSSP) - processes visual and spatial data, subdivided by Logie (1995) into the visual cache and inner scribe; (4) the episodic buffer (added 2000) - integrates the CE, PL and VSSP with LTM, providing a limited-capacity 'store' for combined information and explaining how we form integrated episodes.

Evidence: (1) Dual-task performance - Baddeley and Hitch found participants could track a moving light (VSSP) while performing a verbal reasoning task (PL) with little interference, but performance collapsed when both tasks used the same component - supporting separate slave systems. (2) Case study KF - impaired verbal STM but intact visual STM supports the PL/VSSP distinction. (3) Brain imaging - different activation areas for verbal (left hemisphere) versus visual (right) tasks. (4) The word-length effect disappears when articulation is suppressed (saying 'the the the'), confirming the articulatory process.

Evaluation (AO3):

• Strength - experimental support: dual-task studies are robust, replicable and directly test the model's predictions; the word-length effect is a reliable laboratory finding.
• Strength - clinical evidence: KF and other brain-damaged patients support dissociable components, giving the model converging validity across methods.
• Limitation - central executive underspecified: the CE is 'the most important but least understood component' - vague, unitary and hard to test empirically; some researchers argue it should be subdivided.
• Limitation - case study generalisability: KF's impairment may involve more than one damaged component (Hitch and others noted other damage); single cases may not generalise.
• Limitation - LTM interaction: the original model said little about how working memory links to LTM; the 2000 episodic buffer was bolted on partly to fix this, showing the model is incomplete.

Types of long-term memory (AQA 4.1.2)

Tulving (1972, 1985) distinguished three LTM systems. Episodic memory: personal, time-stamped memories of events (your last birthday), consciously recalled and prone to distortion. Semantic memory: general knowledge and facts (capital of France), not time-stamped, constantly updated. Procedural memory: implicit motor skills and actions (riding a bike), recalled without conscious effort and highly resistant to forgetting.

Evidence: HM (hippocampus removed to treat epilepsy in 1953) could form no new episodic memories but retained procedural skills and some semantic knowledge; Clive Wearing (herpes encephalitis) similarly lost episodic memory (seconds-long awareness) while remaining able to play the piano - dissociation supporting distinct stores. Tulving's PET scans (1989): episodic retrieval activated the right prefrontal cortex, semantic the left - physical evidence for separate systems. The case studies also link to the MSM debate: LTM is not a single store.

Evaluation (AO3):

• Strength - converging evidence: clinical cases (HM, Clive Wearing), neuroimaging (Tulving) and everyday dissociations (knowing how vs knowing that) all support separate LTM systems.
• Strength - practical application: distinctions guide dementia care (procedural memory often spared) and eyewitness research (episodic memory is what EWT measures).
• Limitation - case study weakness: HM's and Wearing's brain damage was severe and widespread, so clean dissociation is hard to prove; cases may not generalise.
• Limitation - overlapping categories: semantic and episodic systems interact in everyday recall (a personal memory of a fact lesson), so the divisions may be less neat than the model suggests; some psychologists propose more than three systems.

Explanations of forgetting: interference and retrieval failure (AQA 4.1.2)

Interference explains forgetting in LTM: memories compete, with proactive interference (old disrupts new - an old phone number intruding on a new one) and retroactive interference (new disrupts old - new parking spot habits making the old one harder to recall). Evidence: McGeoch and McDonald (1931) - participants learned a word list then a second list; when the second list was semantically similar, recall of the first was much worse, showing similarity drives interference. Baddeley and Hitch (1977) - rugby players who had played every match in the season recalled fewer recent games than those who missed matches, showing new games (retroactive interference) displaced older ones. Müller and Pilzecker (1900) showed retroactive interference occurs when new learning follows closely on old (consolidation time).

Retrieval failure (cue-dependent forgetting) explains forgetting when a memory is available but lacks access cues: Tulving's encoding specificity principle states a cue only aids recall if it was present at encoding. Godden and Baddeley (1975) - deep-sea divers learned word lists on land or 15ft underwater; matching learning and recall contexts produced much better recall (context-dependent forgetting). Goodwin et al. (1969) - heavy drinkers who learned tasks drunk recalled them better drunk (state-dependent cues). Abernethy (1940) - students tested in the same room by the same teacher performed better. Forgetting can also be explained by decay (trace fades - only really applies to sensory memory) and motivated forgetting/repression (Freud - distressing memories pushed into the unconscious, weak empirical support but relevant to anxiety-based forgetting).

Evaluation (AO3):

• Interference - strength: lab studies (McGeoch and McDonald) and field studies (rugby players) converge; similarity effect is reliable.
• Interference - limitations: lab studies use artificial word lists (low ecological validity); interference explains only some LTM forgetting and predicts forgetting should be faster the more you learn, which real-life expertise contradicts; research by Anderson (1974) shows interference is strongest when forgetting tests are cued by the competing material.
• Retrieval failure - strength: robust, replicable evidence (Godden and Baddeley, Goodwin) and clear practical value - the cognitive interview's context reinstatement and exam revision advice (revisit the study environment) derive from it.
• Retrieval failure - limitations: effects are mainly demonstrable in lab word-list recall; in real life cues are rarely absent entirely; and Baddeley himself argued context effects are weak in everyday memory; the explanation is hard to falsify - we cannot prove whether a memory is gone or merely lacking cues.
• A* synthesis: the two explanations are complementary rather than competing - interference is about competing material at encoding, retrieval failure about absent cues at retrieval; exam questions often reward noting this.

Eyewitness testimony: misleading information, anxiety and the cognitive interview (AQA 4.1.2; OCR Loftus and Palmer 1974)

Misleading information (leading questions and post-event discussion) distorts EWT. Loftus and Palmer (1974): 45 US students watched clips of car accidents and were asked how fast the cars were going when they smashed, collided, bumped, hit or contacted - estimates ranged from 41 mph ('smashed') to 32 mph ('contacted'); a week later, 'smashed' participants falsely recalled broken glass, showing response-bias and memory distortion. Loftus and Zanni (1975) - 'the' versus 'a' broken headlight changed false recollection. Gabbert et al. (2003) - pairs watched different versions of the same event, discussed it, and 71% (versus 0% of controls) incorrectly recalled details they could not have seen - memory conformity (going along to fit in, not true distortion).

Anxiety has mixed effects. Weapon focus (Loftus et al. 1987) - participants' eyes fixated on a gun, reducing identification of the holder; anxiety narrows attention onto the threat. Deffenbacher et al. (2004) meta-analysis - an inverted-U (Yerkes-Dodson) relationship: moderate arousal aids accuracy, extreme arousal impairs it. But Christianson and Hubinette (1993) - real bank-employees who were victims of threats recalled more accurately than bystanders, showing real-life emotional involvement can aid memory. Yuille and Cutshall (1986) - witnesses to a real shooting gave accurate accounts 4-5 months later, and misleading questions had little effect - challenging lab findings.

The cognitive interview (CI) (Geiselman and Fisher, 1985) uses four techniques: context reinstatement (recreate the scene, weather, feelings - uses retrieval-failure logic), report everything (all details, however trivial - triggers may cue others), reverse order (recall backwards - disrupts schemas and prevents dishonesty) and change perspective (view events from another witness's position - disrupts schema expectations). The enhanced CI (Fisher) adds social techniques (building rapport, minimising distractions). Evidence: Fisher et al. (1990) - detectives trained in the CI elicited 47% more correct information from real witnesses in Miami; Kohnken et al. (1999) meta-analysis of 50 studies - CI produced significantly more correct details than standard interviews (34% average increase).

Evaluation (AO3):

• Misleading information - limitations: lab clips lack real-world stress (ecological validity), participants know they are in a study (demand characteristics); Loftus's findings reflect response bias as much as true memory change; but Yuille and Cutshall's real-shooting study shows lab effects may not transfer.
• Misleading information - strength: robust replicable evidence with huge practical impact - police now avoid leading questions, and the legal system recognises post-event discussion contamination.
• Anxiety - contradiction: weapon-focus studies suggest anxiety harms accuracy while Christianson and Hubinette suggest it helps - the inverted-U resolves this but individual differences (trait anxiety) moderate effects.
• CI - strengths: strong empirical support (Fisher, Kohnken) and real police adoption; individual techniques are effective in isolation.
• CI - limitations: change perspective demands speculation and can increase recall of incorrect details; the full CI is time-consuming and requires expensive training, so UK forces typically use the enhanced CI with selected elements; Kohnken et al. also found increased errors alongside more correct information.

💡 Worked Example

❓ Practice Questions & Mark Schemes

🎬 Video Resources

• YouTube: AQA A-level psychology memory multi-store working memory eyewitness testimony cognitive interview revision

📄 Past Papers & Exam Resources

• AQA A-Level Psychology (7182) - past papers and mark schemes (AQA website)
• Physics & Maths Tutor - Memory topic notes, past paper questions and model answers

🔗 Further Reading & Resources

• Simply Psychology - Multi-Store Memory Model (Atkinson-Shiffrin)
• Simply Psychology - Working Memory Model (Baddeley and Hitch)
• Physics & Maths Tutor - Psychology revision hub (memory notes, questions by topic)
• AQA - find past papers and mark schemes for Psychology 7182

🧠 Flashcards (Spaced Repetition)

📝 Exam Questions by Topic

🎯 Target Tests (Auto-Graded)

🎓 Smart Lesson (Guided)