Notes · PsychologyUK · A-Levels

Memory

Memory is the mental capacity to encode, store and retrieve information. This chapter examines the multi-store and working memory models of how memory is structured, the types of long-term memory, two explanations of why we forget, and how misleading information and anxiety affect the accuracy of eyewitness testimony - together with the cognitive interview designed to improve it.

6 sections~21 min reading time3 competenciesLevel Foundation 1 · Standard 3 · Advanced 2

T·0222 / 17

Exam profile

AO1 · Describe the models of memory, the types of LTM, and the explanations for forgetting and eyewitness testimonyAO2 · Apply the models and explanations to novel memory scenarios and to research dataAO3 · Evaluate the models, the forgetting explanations and the eyewitness-testimony research

Operators:describeoutlineexplainapplyevaluatediscuss

basic level

AS-Level covers the multi-store model, coding/capacity/duration, the working memory model, forgetting and the factors affecting eyewitness testimony.

higher level

The full A-Level adds the types of long-term memory, a deeper evaluation of the models using case studies (e.g. KF, HM) and a fuller treatment of the cognitive interview.

Depth

Text

Contents · 6 sections

Memory

The multi-store model and the features of the stores#

FoundationAQA 7182 3.1.2DfE GCE Psychology - the multi-store model

The multi-store model of memory

Fig. 1Information flows one way through three stores; attention and rehearsal move it forward, decay and displacement lose it.

The multi-store model (Atkinson and Shiffrin, 1968) describes memory as a flow of information through three separate, permanent stores that differ in how they code information, how much they hold (capacity) and how long they hold it (duration). Information from the environment first enters the sensory register, is passed to short-term memory (STM) if we pay attention to it, and is transferred to long-term memory (LTM) through rehearsal. The model is a linear, structural account: information moves in one direction and each store is a distinct 'box'.

The three stores have contrasting features. The sensory register holds a fleeting impression of everything the senses detect: its coding is modality-specific (a visual iconic store, an acoustic echoic store), its capacity is very large but its duration is under half a second, so information is lost unless we attend to it. STM codes mainly acoustically (by sound), has a limited capacity of about 7 plus or minus 2 items (Miller's 'magic number') and a duration of roughly 18-30 seconds unless the material is rehearsed. LTM codes mainly semantically (by meaning), has a potentially unlimited capacity and can last up to a lifetime.

The transfer between stores depends on two processes. Attention moves information from the sensory register into STM. Maintenance rehearsal (repeating information over and over) keeps it circulating in STM and, if prolonged, transfers it to LTM; without rehearsal STM information is lost by decay or displacement. Retrieval is the reverse flow, bringing information from LTM back into STM to be used. Evidence for separate stores comes from studies of duration and capacity and from patients such as HM, whose ability to form new long-term memories was destroyed while his STM remained intact.

The model is evaluated as a useful but oversimplified account. Its strength is that it was the first testable model and is supported by experimental evidence for the different coding, capacity and duration of the stores, and by brain-damage cases showing STM and LTM can be separately impaired. Its main weakness is that it treats STM and LTM as single, unitary stores: the case of patient KF, who had a poor STM for verbal information but a normal STM for visual information, shows STM is not one store, and the model's reliance on maintenance rehearsal alone is challenged by evidence that elaborative rehearsal (linking to meaning) transfers information to LTM far more effectively.

\text{STM capacity} \approx 7 \pm 2 \ \text{items}

Miller's magic number

The span of immediate memory is about five to nine items; chunking (grouping items into meaningful units) increases the effective capacity.

The decay of short-term memory

Fig. 2Without rehearsal, recall from STM decays rapidly over about 18 seconds (illustrative curve reflecting Peterson and Peterson's pattern).

Worked example

Chunking and STM capacity

A person can recall the string 1 9 4 5 1 9 6 6 2 0 1 2 as three dates (1945, 1966, 2012). Explain, using capacity, why chunking helps.

State the raw load
The 12 individual digits exceed the STM capacity of about 7 plus or minus 2 items, so recalling them one by one is unreliable.
Apply chunking
Grouping the digits into three meaningful dates creates just 3 chunks, well within capacity.
Explain
Because capacity is measured in chunks rather than raw items, meaningful grouping increases how much can be held without increasing the number of stored units.

Result: Chunking reduces 12 items to 3 chunks, bringing the load within STM capacity.

Typical mistakes

Confusing coding, capacity and duration - coding is the format (acoustic/semantic), capacity is how much, duration is how long.
Saying STM lasts a few minutes - it is under about 30 seconds without rehearsal.

Active revision

Outline the multi-store model of memory and explain one limitation supported by the case of patient KF.

Active recall

Recall the key points — then reveal.

Sources: GCE AS and A level subject content for psychology (Department for Education) · AQA A-level Psychology 7182 specification (AQA)

Types of long-term memory#

StandardAQA 7182 3.1.2DfE GCE Psychology - types of long-term memory

The three types of long-term memory

Fig. 3Episodic and semantic memory are declarative (conscious); procedural memory is non-declarative (automatic).

Tulving (1985) argued that the multi-store model's view of LTM as a single store was too simple, and proposed that LTM is divided into three types that differ in what they store and whether recall is conscious. Episodic memory stores personal events ('episodes') from our lives - what happened, where and when - and is time-stamped and recalled with conscious, deliberate effort. Semantic memory stores our knowledge of the world: facts, concepts and the meanings of words, not tied to a particular time or place, and also recalled consciously. Procedural memory stores our motor skills and how to do things, such as riding a bike or driving; it is recalled automatically, without conscious awareness, and is hard to put into words.

The distinction matters because it explains patterns of memory loss that a single-store view cannot. Episodic and semantic memories are both 'declarative' (they can be consciously declared or stated), whereas procedural memory is 'non-declarative'. In amnesia, the different types are often affected separately: patients such as HM and Clive Wearing lost the ability to form new episodic memories yet retained procedural skills - HM could learn new motor tasks even though he could not remember doing them, and Clive Wearing could still play the piano. This dissociation is strong evidence that the types are supported by different systems.

Brain-scanning evidence supports the distinction too: episodic and semantic memories are associated with activity in different regions (episodic memory drawing on the hippocampus and the right prefrontal cortex, semantic memory on the temporal lobe and the left prefrontal cortex), while procedural memory involves the cerebellum and motor areas. This physical separation strengthens the claim that these are genuinely different stores rather than convenient labels.

The types can also be evaluated and applied. A real-world benefit is that identifying which type is impaired allows targeted help - for example, older people show a decline in episodic memory but relatively preserved semantic memory, so interventions can focus on the weaker system. A criticism is that the boundaries are not always clear: some psychologists (Cohen and Squire) argue that episodic and semantic memory are better grouped together as one declarative store, distinct only from procedural (non-declarative) memory, so the three-way split may be finer than the evidence requires.

Worked example

Classifying memories by type

A stroke patient can no longer recall recent events or learn new facts but can still ride a bicycle. Identify which types of LTM are impaired and which is spared, and state what this dissociation shows.

Impaired types
The loss of recent events points to impaired episodic memory, and the inability to learn new facts points to impaired semantic memory - both declarative types.
Spared type
Retained cycling shows intact procedural (non-declarative) memory.
Interpretation
Because declarative memory is lost while procedural memory is preserved, the two must be supported by different systems, supporting Tulving's distinction.

Result: Episodic and semantic memory are impaired, procedural memory is spared - a dissociation supporting separate LTM systems.

Typical mistakes

Confusing episodic and semantic - episodic is a personal event you experienced; semantic is impersonal knowledge.
Describing procedural memory as consciously recalled - it is automatic and hard to verbalise.

Active revision

For each of the following, name the type of LTM involved: remembering your last holiday; knowing that water boils at 100 degrees Celsius; tying a shoelace.

Active recall

Recall the key points — then reveal.

Sources: GCE AS and A level subject content for psychology (Department for Education) · AQA A-level Psychology 7182 specification (AQA)

The working memory model#

AdvancedAQA 7182 3.1.2DfE GCE Psychology - the working memory model

The working memory model

Fig. 4The central executive coordinates the phonological loop (inner ear and inner voice), the visuo-spatial sketchpad and the episodic buffer, which links to LTM.

Baddeley and Hitch (1974) argued that STM is not a single store but an active system of several components that work together to process information we are using right now, and called it the working memory model. Its strength over the multi-store model is that it explains how we can perform two tasks at once when they use different components, but struggle when they compete for the same one.

The central executive is the control centre. It directs attention, decides which component to use for a task and coordinates the others, but has a very limited capacity and no storage of its own - it is described as 'attentional' rather than a memory store. It is the least well specified part of the model, which is one of the model's criticisms.

The central executive controls two 'slave systems', each modality-specific. The phonological loop deals with auditory (sound-based) information and preserves word order; it is subdivided into the phonological store (the 'inner ear', which holds spoken words briefly) and the articulatory process (the 'inner voice', which allows maintenance rehearsal by silent repetition), and has a capacity of about two seconds' worth of speech. The visuo-spatial sketchpad ('inner eye') deals with visual and spatial information - what things look like and where they are - and was later subdivided into a visual cache (form and colour) and an inner scribe (spatial relations). The episodic buffer, added by Baddeley in 2000, is a temporary general store that integrates information from the other components and from long-term memory into a single, coherent 'episode', giving the model a link to LTM it previously lacked.

The model is well supported. The dual-task studies of Baddeley and Hitch show that two tasks using the same component (e.g. two visual tasks) interfere with each other, whereas two tasks using different components (a visual and a verbal task) can be done together with little loss - exactly as separate slave systems predict. The case of KF, whose verbal STM was impaired while his visual STM was intact, supports separate phonological and visuo-spatial stores. Criticisms are that the central executive is too vaguely defined to test properly, that the model explains only STM (not the whole of memory), and that much of the evidence comes from brain-damaged patients whose other deficits make firm conclusions difficult.

Worked example

Predicting dual-task performance

A participant must simultaneously repeat 'the-the-the' aloud and describe from memory the layout of their kitchen. Predict the outcome and justify it using the working memory model.

Identify the components used
Repeating 'the-the-the' occupies the phonological loop; describing a spatial layout occupies the visuo-spatial sketchpad.
Apply the dual-task logic
Because the two tasks use different slave systems, they should not compete for the same resources.
Predict
Performance on both should be reasonably good - unlike two verbal tasks, which would both need the phonological loop and interfere.

Result: The tasks use different components, so both can be performed with little interference.

Typical mistakes

Saying the central executive stores information - it directs attention and has no storage of its own.
Forgetting the episodic buffer (added in 2000) or confusing the visual cache and inner scribe.

Active revision

Explain why a driver can listen to the radio (verbal) while driving (visual/spatial) but struggles to read a map at the same time as driving.

Active recall

Recall the key points — then reveal.

Sources: GCE AS and A level subject content for psychology (Department for Education) · AQA A-level Psychology 7182 specification (AQA)

Explanations for forgetting: interference and retrieval failure#

StandardAQA 7182 3.1.2DfE GCE Psychology - explanations for forgetting

Interference and retrieval failure

Fig. 5Interference: old and new memories disrupt each other. Retrieval failure: the cue at learning is absent at recall.

Forgetting from long-term memory is explained in two main ways on the specification, and both assume the memory is still stored but cannot be accessed. Interference theory says forgetting occurs because other memories get in the way. Proactive interference is when an older memory disrupts a newer one (an old phone number interfering with recall of a new one); retroactive interference is when a newer memory disrupts an older one (learning a new language making it harder to recall a previously learned one). Interference is worse when the two sets of information are similar.

The classic evidence is McGeoch and McDonald's (1931) study, in which participants learned a list of words and then learned a second list that varied in similarity to the first. Recall of the original list was worst when the second list was made of synonyms (similar meanings) and best when it was unrelated, showing that similarity increases interference. The strength of this account is its strong laboratory support; its weakness is that the artificial word-list tasks lack ecological validity, and interference may explain forgetting only in a narrow set of conditions where similar material is learned close together.

Retrieval failure due to the absence of cues is the second explanation, captured by Tulving's encoding specificity principle: a cue present at learning must also be present at recall to help retrieval. Context-dependent forgetting occurs when the external environment differs between learning and recall - Godden and Baddeley (1975) found divers who learned and recalled word lists in the same environment (both on land or both underwater) recalled about 40% more than those whose environments were mismatched. State-dependent forgetting occurs when the internal state (mood, or being sober versus intoxicated) differs between learning and recall.

Retrieval failure has good real-world support and application: revising in conditions similar to the exam, or mentally reinstating the context of an event, improves recall - a principle used in the cognitive interview. A limitation is that the encoding specificity principle is difficult to test because we cannot be sure a cue was encoded in the first place, making the theory hard to falsify, and the size of context effects is often smaller in everyday life than in the striking diver study.

Worked example

Applying encoding specificity

Learners recall a list better in the room where they learned it than in a different room. Explain this using retrieval failure, and suggest one way to use the principle before an exam.

Identify the effect
This is context-dependent forgetting: the external environment is a retrieval cue encoded at learning.
Apply encoding specificity
When the recall context matches the learning context, the cue is present and aids retrieval; when it differs, the cue is absent and recall falls.
Suggest a strategy
Because you cannot revise in the exam hall, mentally reinstate the revision context during the exam, or revise in varied contexts so recall does not depend on one setting.

Result: Matching contexts restores the cue and improves recall; context reinstatement is the practical application.

Typical mistakes

Reversing proactive and retroactive - 'pro' points forward (old affects new); 'retro' points back (new affects old).
Saying forgotten memories are erased - both explanations assume the memory is stored but cannot be accessed.

Active revision

A student who revised for French, then Spanish, finds the Spanish words intruding when she tries to recall French. Identify the type of interference and explain why the effect is strong here.

Active recall

Recall the key points — then reveal.

Sources: GCE AS and A level subject content for psychology (Department for Education) · AQA A-level Psychology 7182 specification (AQA)

Eyewitness testimony: misleading information and anxiety#

AdvancedAQA 7182 3.1.2DfE GCE Psychology - eyewitness testimony

Leading questions in the Loftus and Palmer study

Fig. 6The critical verb altered the mean speed estimate - stronger verbs produced faster estimates (Loftus and Palmer's published means, mph).

Eyewitness testimony (EWT) is the evidence given by a witness to a crime or accident, and psychologists study the factors that reduce its accuracy because mistaken testimony has led to wrongful convictions. Two such factors are on the specification: misleading information and anxiety. Misleading information includes leading questions - questions whose wording suggests a particular answer - and post-event discussion, where witnesses' memories become contaminated by talking to one another.

Loftus and Palmer (1974) demonstrated the effect of leading questions. Participants watched a film of a car accident and were asked how fast the cars were going when they 'hit' each other, with the verb varied between groups ('smashed', 'collided', 'bumped', 'hit', 'contacted'). The estimated speed rose with the severity of the verb: those given 'smashed' estimated about 40.8 mph, those given 'contacted' about 31.8 mph. In a follow-up, participants given 'smashed' were more likely to report (falsely) having seen broken glass a week later - showing the wording had altered the stored memory (a substitution/response-bias effect).

Post-event discussion also distorts memory. Gabbert et al. (2003) found that participants who discussed a witnessed event with a co-witness who had seen a different version often incorporated details they had not themselves seen into their own account - memory conformity. This matters in real investigations, where witnesses frequently talk before being interviewed.

Anxiety has a more complex effect. High anxiety at the scene of a crime can reduce accuracy through the 'weapon focus' effect (Johnson and Scott), in which a witness fixates on a weapon and encodes less about the perpetrator's face. Yet other research (Yuille and Cutshall's real robbery study) found that highly stressed witnesses were accurate months later, suggesting stress can enhance memory for a genuine event. The two patterns are reconciled by the Yerkes-Dodson law: memory accuracy is best at a moderate level of arousal and falls when arousal is too low or too high, giving an inverted-U relationship. The research is evaluated for its practical importance (informing police interview practice) but also for the low ecological validity of laboratory studies using film clips, which lack the emotional impact of a real crime.

Anxiety and eyewitness accuracy (Yerkes-Dodson)

Fig. 7The Yerkes-Dodson law: EWT accuracy peaks at a moderate level of arousal and falls when arousal is too low or too high (illustrative shape).

Worked example

Interpreting the effect of a leading question

The mean speed estimate was 31.8 mph for 'contacted' and 40.8 mph for 'smashed'. Calculate the difference and the percentage increase, and explain what it shows.

Find the difference
40.8 - 31.8 = 9.0 mph.
Express as a percentage of the lower value
9.0 / 31.8 x 100 = 28.3%.
$\frac{40.8 - 31.8}{31.8} \times 100 = 28.3\%$
Interpret
A single word raised the estimate by over a quarter, showing that the wording of a question can substantially alter what a witness reports - evidence that leading questions distort EWT.

Result: A 9.0 mph (28.3%) increase caused by one verb, demonstrating the power of leading questions.

Typical mistakes

Saying anxiety always worsens memory - moderate arousal can improve it (Yerkes-Dodson).
Confusing leading questions (misleading wording) with post-event discussion (contamination between witnesses).

Active revision

Two witnesses discuss a robbery before being interviewed, and one is later sure she saw a detail the other described. Explain this using post-event discussion, and evaluate the reliability of her testimony.

Active recall

Recall the key points — then reveal.

Sources: GCE AS and A level subject content for psychology (Department for Education) · AQA A-level Psychology 7182 specification (AQA)

Improving eyewitness testimony: the cognitive interview#

StandardAQA 7182 3.1.2DfE GCE Psychology - the cognitive interview

The four techniques of the cognitive interview

Fig. 8The cognitive interview applies memory principles through four techniques to maximise accurate recall.

The cognitive interview (Fisher and Geiselman, 1987) is a police interviewing technique designed to increase the accuracy of eyewitness testimony by basing questioning on psychological principles about how memory works, especially retrieval failure and the effect of leading questions. It replaces the standard interview, which used rapid, closed and sometimes leading questions that could distort memory.

It has four main techniques. Report everything: the witness is encouraged to include every detail even if it seems trivial or irrelevant, because small details may act as cues that trigger other memories and because the witness cannot judge what will be important. Reinstate the context: the witness mentally returns to the scene, recalling the environment and their feelings at the time - this applies the encoding specificity principle, providing context cues to aid retrieval. Reverse the order: the witness recounts the event in a different chronological order (e.g. from the end backwards), which disrupts the expectations and schemas that can lead to reconstructing rather than recalling. Change perspective: the witness describes the event from another person's point of view, again to reduce the influence of expectations on recall.

An enhanced cognitive interview added social and communication elements - building rapport, minimising distractions, letting the witness control the flow, reducing anxiety and avoiding leading questions - because the effectiveness of memory retrieval also depends on the interaction between interviewer and witness.

The cognitive interview is evaluated as effective but demanding. Meta-analyses show it produces substantially more correct information than the standard interview, particularly the 'report everything' and 'reinstate context' elements, which are the most useful. However, it can also increase the amount of incorrect information recalled, so accuracy (the proportion correct) does not always improve; it is time-consuming and requires special training that many forces cannot fully provide, so in practice a partial version is often used. It is also less suitable for some groups, such as young children.

Worked example

Justifying a technique from memory theory

Explain why 'reinstate the context' should improve recall, referring to a forgetting explanation from earlier in the topic.

Link to retrieval failure
Earlier we saw that forgetting can be due to the absence of cues (encoding specificity principle).
Apply to the technique
Reinstating the context re-creates the external and internal cues that were present at encoding, so the missing cues become available again.
Predict the effect
With the cues restored, more of the stored memory becomes accessible, increasing correct recall - as Godden and Baddeley's diver study demonstrated.

Result: Context reinstatement works by supplying encoded cues, directly applying the retrieval-failure explanation of forgetting.

Typical mistakes

Listing the techniques without explaining the memory principle behind each.
Claiming the cognitive interview always improves accuracy - it increases correct recall but can also increase errors.

Active revision

Explain how two techniques of the cognitive interview draw on principles of memory covered earlier in this topic.

Active recall

Recall the key points — then reveal.

Sources: GCE AS and A level subject content for psychology (Department for Education) · AQA A-level Psychology 7182 specification (AQA)

References & sources

Sources

Department for Education

GCE AS and A level subject content for psychology

AQA

AQA A-level Psychology 7182 specification

Memory

6 sections~21 min reading time3 competenciesLevel Foundation 1 · Standard 3 · Advanced 2

Memory

Key points

Chunking and STM capacity

Key points

Classifying memories by type

Key points

Predicting dual-task performance

Key points

Applying encoding specificity

Key points

Interpreting the effect of a leading question

Key points

Justifying a technique from memory theory

Sources

Memory

Key points

Chunking and STM capacity

Key points

Classifying memories by type

Key points

Predicting dual-task performance

Key points

Applying encoding specificity

Key points

Interpreting the effect of a leading question

Key points

Justifying a technique from memory theory

Sources