Memory: key studies
12 studies. Tap a card to flip it, then say whether you knew it. Cards you miss come back every round; cards you know come back less often. Progress is saved in this browser. Terms deck · Match-up game · Back to the topic
Atkinson and Shiffrin (1968), multi-store model
Aim To describe memory as a series of stores through which information flows.
Method A theoretical model built from laboratory evidence on capacity, duration and coding.
Findings Sensory register (huge capacity, under half a second, modality-specific coding) → attention → short-term memory (5–9 items, about 18–30 seconds, mainly acoustic) → rehearsal → long-term memory (unlimited, up to a lifetime, mainly semantic).
Conclusion Memory is made of separate unitary stores and information passes between them through attention and rehearsal.
Evaluation Supported by Baddeley's coding study, Jacobs, Peterson and Peterson, Bahrick and the serial position effect (Glanzer and Cunitz) and by patient HM; but STM is not unitary (KF), LTM has several types, and maintenance rehearsal alone does not explain transfer (Craik and Watkins).
Examiners say The least well answered 1-mark question in the bank was the duration of LTM: 'up to a lifetime'. Reports also warn that an 'outline of STM according to the MSM' is about its features, not about the whole model.
Baddeley (1966), coding
Aim To find out how information is coded in short-term and long-term memory.
Method Participants learned lists of words that were acoustically similar, acoustically dissimilar, semantically similar or semantically dissimilar, and recalled them either immediately or after 20 minutes.
Findings Immediate recall was worst for acoustically similar words; recall after 20 minutes was worst for semantically similar words.
Conclusion STM codes mainly acoustically and LTM mainly semantically.
Evaluation Artificial stimuli with no personal meaning; the 20-minute delay may not be true LTM; later work shows STM can use visual coding too.
Examiners say Application questions give two lists and ask you to explain the results: name the store, its coding and why similar-sounding words confuse STM.
Jacobs (1887), capacity
Aim To measure the capacity of short-term memory.
Method Digit span: participants heard a sequence of digits or letters and repeated it; the sequence lengthened until they failed.
Findings Mean span was about 9.3 digits and 7.3 letters.
Conclusion STM holds roughly 7 ± 2 items (Miller, 1956), more if chunked.
Evaluation Old study with little control over distraction; Cowan's review suggests capacity is nearer four chunks.
Examiners say 'Outline one way capacity has been investigated' wants the procedure (span task, sequences lengthening), not just the figure.
Peterson and Peterson (1959), duration of STM
Aim To measure how long STM lasts without rehearsal.
Method 24 students were given a consonant trigram (e.g. YCG) and counted backwards in threes from a number for 3, 6, 9, 12, 15 or 18 seconds before recalling it.
Findings About 80% correct after 3 seconds, around 3% after 18 seconds.
Conclusion STM lasts about 18–30 seconds unless rehearsed.
Evaluation Meaningless stimuli (low external validity); the counting task may cause displacement rather than decay.
Examiners say Reports say the overwhelming choice for 'one way of investigating duration', so the procedure must be accurate: trigram, counting backwards, varied intervals.
Bahrick et al (1975), duration of LTM
Aim To test how long long-term memories last.
Method 392 American ex-high-school students aged 17–74 were tested on photo recognition and free recall of names from their yearbooks.
Findings Recognition was about 90% after 15 years and 70% after 48 years; free recall was lower, around 60% after 15 years and 30% after 48.
Conclusion LTM can last a lifetime, with recognition more durable than recall.
Evaluation High external validity because real-life memories were used; confounding variables (people may have stayed in touch) could not be controlled.
Examiners say Evidence for the MSM's claim about LTM duration.
Baddeley and Hitch (1974), working memory model
Aim To replace the single STM store with a system that can hold and work on information at the same time.
Method A model based on dual-task studies: performance suffers when two tasks use the same component.
Findings Central executive (allocates attention, very limited capacity) controls the phonological loop (phonological store and articulatory process, about two seconds of speech), the visuo-spatial sketchpad (visual cache and inner scribe, about three or four objects) and, added in 2000, the episodic buffer that integrates information and links to LTM.
Conclusion STM is an active, multi-component system.
Evaluation Supported by KF (poor verbal STM, intact visual STM) and dual-task studies; the central executive is vague and probably not unitary (EVR); dual-task evidence is from artificial tasks.
Examiners say 'Describe the phonological loop' needs both sub-parts and their jobs. The one-mark 'why is the WMM a better explanation than the MSM' answer is that STM is not a unitary store.
Tulving (1985), types of long-term memory
Aim To show that LTM is not one store.
Method A distinction drawn from case studies and brain-imaging evidence.
Findings Episodic memory (events, time-stamped, conscious recall), semantic memory (facts and meanings, conscious recall) and procedural memory (skills, recalled without conscious effort).
Conclusion LTM is made of at least three qualitatively different stores.
Evaluation Clinical cases (HM, Clive Wearing) show procedural memory intact when episodic is lost; brain scans show the hippocampus active for episodic and the cerebellum and motor cortex for procedural; Cohen and Squire argue for two stores, declarative and non-declarative.
Examiners say Credit depends on the defining features: conscious vs unconscious recall, and whether the memory is time-stamped.
Godden and Baddeley (1975), context-dependent forgetting
Aim To test whether recall depends on the environment matching the learning context.
Method Divers learned word lists on land or underwater and recalled them in the same or the other environment.
Findings Recall was about 40% lower when the learning and recall environments did not match.
Conclusion Forgetting can be retrieval failure caused by the absence of external cues present at encoding.
Evaluation Supports the encoding specificity principle; when recognition rather than recall was tested the effect disappeared, so it only applies to some memory tasks; the contexts must be very different for the effect to show.
Examiners say Application questions about exam rooms (sitting the exam in an unfamiliar hall) want 'absence of contextual cues' explained, not just 'retrieval failure'.
Loftus and Palmer (1974), leading questions
Aim To see whether the wording of a question changes eyewitness recall.
Method 45 students watched film clips of car accidents and estimated the speed when the cars 'hit', 'contacted', 'bumped', 'collided' or 'smashed'. A second experiment asked a week later whether they had seen broken glass.
Findings 'Smashed' produced a mean estimate of 40.5 mph against 31.8 mph for 'contacted'; 'smashed' participants were more likely to report broken glass that was not there.
Conclusion Leading questions distort memory, possibly by altering the stored memory (substitution) rather than just the response.
Evaluation Laboratory films lack the stress and consequences of real events (Yuille and Cutshall's real robbery witnesses were accurate and unaffected by leading questions); students may be better at estimating than the general public; shows the response bias vs substitution debate.
Examiners say 'Describe how post-event discussion affects EWT' is a different study: Gabbert et al (2003). Reports flag students who use Loftus and Palmer for post-event discussion.
Gabbert et al (2003), post-event discussion
Aim To see whether discussing an event with another witness contaminates recall.
Method Pairs watched the same crime from different angles so each saw details the other did not, then discussed it before individually recalling.
Findings 71% of those who discussed recalled details they could not have seen; the control group showed no such errors.
Conclusion Post-event discussion produces memory conformity, through informational influence or memory contamination.
Evaluation Controlled and replicable; does not show whether the memory itself is changed or whether participants simply go along with the co-witness; artificial crime.
Examiners say Students evaluate this poorly: for a 2-mark evaluation, say what the design controlled or what it cannot tell us.
Johnson and Scott (1976), anxiety and the weapon focus
Aim To test whether anxiety reduces the accuracy of eyewitness identification.
Method Participants in a waiting room heard an argument next door and then saw a man walk out holding either a pen with greasy hands (low anxiety) or a bloodied knife (high anxiety); they later identified him from photos.
Findings 49% identified the pen-holder correctly, 33% the knife-holder.
Conclusion Anxiety narrows attention onto the weapon (weapon focus) and reduces accuracy for other details.
Evaluation Ethical issues of deliberately causing anxiety; may measure surprise rather than anxiety (Pickel's chicken and gun in a hairdresser's); contradicted by Yuille and Cutshall, where the most anxious real witnesses were most accurate; the Yerkes–Dodson inverted U reconciles the two.
Examiners say Stem questions (a frightened witness close to a knife) expect weapon focus named and applied, and good answers add the inverted-U qualification.
Fisher and Geiselman (1992), cognitive interview
Aim To improve police interviewing using what is known about memory.
Method Four techniques: report everything, reinstate the context, reverse the order, change perspective; the enhanced version adds building rapport and controlling eye contact.
Findings Kohnken et al's meta-analysis of 50 studies found 41% more correct information than standard interviews, but also more incorrect information.
Conclusion Retrieval improves when cues are reinstated and recall is varied to reduce schema-driven reconstruction.
Evaluation Time-consuming and needs training, so many forces use only parts of it (Milne and Bull found report everything plus reinstate context most useful); increases incorrect recall too; variations between forces make it hard to evaluate.
Examiners say 'Identify and outline two techniques' needs the technique and what the witness is asked to do. A limitation for 2 marks must be developed: 'it takes longer, so police forces short of time use only some of the techniques, reducing effectiveness'.
Work with an AQA psychology tutor
Know the studies but not scoring on the essays? That's what a tutor fixes.