Multi-Store Model (MSM)

Model consists of three stores :short-term store, long-term store, and sensory register.

Sensory register contains information collected by senses like the eyes, ears, or skin. It has separate stores for different sensory inputs: echoic store (auditory information), iconic store (visual information), haptic store (tactile information), and olfactory store (smell informations). This information is kept in the sensory store for a brief perios before decaying, or if given attention the data transfers to the short term memory. 

Data in the STM is fragile unless rehearsed and can be lost through displacement (pushed out by new incoming information) or decay (memory fades comepletely). The STM has a limited capacity of 5-9 items and a limited duration of about 2 seconds. Information is typically encoded acoustically.

The LTM has an unlimited capacity and a duration of a lifetime, however data can be lost through decay (memory fades), retrieval failure (material is available but not accessible and so a cue may be needed for retrieval), or interferemce (confusion with other memories). Information is typically encoded semantically.

1 of 30

Key Study

Baddely (1966)

Investigated whether encoding in STM and LTM is mainly acoustic or semantic. 

75 participants slip into 4 groups and given list of 10 words. Group A given acoustically similar words, Group B given acoustically dissimilar words, Group C given semantically similar words, and Group D given semantically dissimilar words. 

When asked to recall immediately, Group A had worst recall at 10%, with other groups recall being at 60-80%. When asked to recall after 20 minute interval, Group C had worst recall at 55%, with other groups recall being at 75-80%. Suggests that STM is encoded acoustically due to acoustic confusion, and LTM is encoded semantically due to semantic confusion.

(-) However, Brandimonte et al (1992) found that individuals used visual coding in STM if given visual tasks and prevented from doing verbal rehearsal.  

(+) Lab study which means cause and effect established, but also means study lacks ecological validity.

2 of 30

Key Study

Jacobs (1887)

Investigated capacity of STM.

Used serial digit span technique, read out 4 digits and asked participants to recall. Increased number of digits until participants could no longer recall list properly. 

Found capacity of 9 items for numbers and 7 items for letters, and also found that number of items increased with age. Concludes that the STM has a capacity of 5-9 items and that we develop better recall strategies as we get older.

(+) Miller (1956) supports this and said magic numbr was 7 (+/- 2 items), and also found that chuncking can increase capacity.

(+) Lab study so cause and effect can be established but lacks ecological validity. 

(-) Also no consistency as different capacity for numbers, letter, and words.

3 of 30

Key Study

Peterson and Peterson (1959)

Investigated duration of STM when rehearsal prevented.

24 university students took part in 2 practice trials and 8 real trials. Given trigram to remember, and then given 3 digit number to count backwards from for 3, 6, 9, 12, 15, or 18 seconds to prevent rehearsal. Then asked to recall what the trigram was.

More trigrams recalled with 3 second interval than 18 second interval. Average recall 90% with 3 second interval, but average recall dropped to about 2% with 18 second interval. Suggests that information remains in STM for less than 18 seconds if verbal rehearsal pervented.

(-) Lacks ecological validity and mundane reality as not a situation you would come across in everyday life.

(-) Lacks population validity as only university student sample.

4 of 30

Key Study

Bahrick et al (1975)

Investigated existence of very long term memory and see difference between recognition and recall.

392 American graduates shown pictured from high school yearbook. Recognition group asked to match photos with names given, while recall group asked to name people in the photos without the list of names.

In recognition group, 90% correct 14 years after graduation, 80% correct 25 years after graduation, 75% correct 34 years after graduation, and 60% correct 47 years after graduation. In recall group, 60% accurate after 7 years, and less than 20% accurate after 47 years. Concludes that certain people can remember information for a lifetime, and LTM better measured in recognition tests than recall tests.

(+) High ecological validity as meaningful stimulus material tested for individual's own memories.

(-) Unclear whether drop in accuracy after 47 years is due to limit in duration or more general decline in memory with age. 

5 of 30


(+) Glanzer and Cunitz (1966) gave participants 20 words, one by one, and then asked to recall. Serial position effect observed: Primacy effect (words at beginning rehearsed and transferred to LTM) and recency effect (words remain in STM prior to recall). Suggests STM and LTM are different stores.

(+) Found prefrontal cortex active with STM, and hippocampus active with LTM. Supports idea of separate stores.

(+/-) Clive Wearing unable to create new long-term memories due to destroyed hippocampus, suggesting separate STM and LTM stores. However, could still play piano which suggests procedural memory intact, which can't be explained by MSM.

(+/-) H.M. unable to form new long-term memories after hippocampus removed, suggesting separate STM and LTM stores. However, able to aquire new skills which suggests procedural memory intact, which can't be explained by MSM.

(-) Doesn't explain processes in encoding. Too much emphasis on structures of memory sytem. Flow of information may not be as rigid as model suggests

6 of 30


(-) Rehearsal may not be only way that memories commit to LTM. May be that information is tranferred to LTM because of distinctiveness or significance to the person.

(-) Too reductionist. Reduces complex ideas of memory down to simple ideas, and is oversimplification of structures and processes. Working memory model is a more suitable and detailed model of memory.

7 of 30

Working Memory Model

Baddely and Hitch (1974) proposed alternative model to explain short term memory and that it is not just one store, but several.

STM is an 'active store' which holds several pieces of information while being worked on:

1. Central executive allocates slave systems.

2. Phonological loop consists of phonological store and articulatory process.

3. Visuo-spacial sketchpad

4. Episodic buffer (added in 2000)

8 of 30

Working Memory Model Diagram

9 of 30

Working Memory Model Components

Central Executive - Most important component of WMM, and acts as a filter to decide which information is received by sensory organs and which is ignored. It directs information to and from 'slave systems' and LTM. Also involved in problem solving and decision-making, but has limited capacity so can only cope with one strand of information at a time.

Phonological Loop - Processes speech and has a limited capacity. Consists of Phonological Store and Articulatory Control Process. Phonological store stores acoustically coded items, and Articulatory Control Process allows sub-vocal repetition (rehearsal) of items stored.

Visuo-Spatial Sketchpad - Stores visual and spatial information, has limited capacity and duration, and consists of visual cache and inner scribe. Visual Cache stores information about form and colour, and Inner Scribe deals with spacial/movement information and rehearses and tranfers information to the Visual Cache that is then sen to the Central Executive.

Episodic Buffer - Additional storage system that binds information from different components into chunks/episodes, including information from the LTM. It also has a limited capacity. 

10 of 30


(+) Explains how it is easier to do two tasks that are different than two tasks that are similar. Each slave system has a limited capacity so can only do so much at one time. Two similar tasks would put too much strain on one slave system resulting in a slower processing speed.

(+) Research evidence from brain damage patients. LH performed better on spatial tasks than visual imagery tasks, suggesting there are separate visual and spatial systems. Provides evidence for existence of two main slave systems as well as sub-systems within visuo-spatial sketchpad. 

(-) Central Executive is vague concept. Isn't clear what it is or how it works, as well as lacking research despite being most important component. So not all areas of WMM are supported.

(+) PET scans show different areas of brain are activated when doing verbal and visual tasks. Suggests phonological loops and visuo-spacial sketchpad are separate stores.

(+) PET scans have shown activation in left half of brain with visual tasks and right half of brain with spatial tasls. Supports existence of visual cache and inner scribe.

(-) WMM only concerns STM and ignores sensory register and LTM, so not comprehensive model. 

11 of 30

Episodic LTM

Episodic memory - Explicit (declarative) and is the memory of the events you have experienced. Have three elements: specific details of event, context, and emotions. Helps individuals distinguish between real and imagined events.

Influenced by emotions present at time of coding, so traumatic experiences often recalled well due to high emotional content. Also influenced by degree of processing of information during coding. 

Episodic memories associated with prefrontal cortex during encoding, neocortex during consolidation, and then in the hippocampus once stored.

(+) Supported by advanced scientific techniques so objective and results more likely to be valid.
(+) Research support by Tulving (1989). 6 participants took part in 4 episodic and 4 semantic tasks, then injected with radioactive gold and scanned to detect location in brain. Found during episodic tasks, activation in frontal lobes. Shows biological basis for episodic memory.
(-) Extent to which episodic and semantic memory differs in unclear and overlap between two systems.
(-) Semantic memories often originate in episodic memory (memory of being taught is semantic but what you learnt is semantic.

12 of 30

Key Study

Tulving (1989)

Investigated differences in processing episodic and sematic memory tasks

6 volunteers performed 4 semantic tasks and 4 episodic tasks in random order. Each task separated by 2 minutes of rest where participant would try retreive episodic or semantic memories. They were then injected with small amounts of radioactive gold and scanned to detect its location in the brain.

3 participants provided inconclusive data. Other 3 participants showed difference in blood flow between episodic and sematic thinking. Greater activity in frontal lobes when thinking about episodic memory and greater activity in posterior regions when thinking about semantic memory. Episodic and semantic LTM invloves different brain areas which suggests biological basis to differences in LTM.

(-) Only 3 participants showed differences in episodic and semantic memory so findings may not generalise to population.

(-) Tulving and wife were both participants so hard to be unbiased and objective.

13 of 30

Semantic LTM

Semantic Memory - Explicit (declarative) and contains knowledge and facts. Associated with degree of processing during coding. Linked to episodic memories, as episodic memory of learning it. Episodic memories seem to gradually shift to semantic memories.

Disagreement over parts of brain associated with semantic memory as hippocampus seems to be involved but evidence suggests coding is mainly associated with frontal and temporal lobes.

(+) Supported by Tulving research (1989). Found that greater activation in posterior region during semantic retrieval. Show biological basis for semantic LTM.

(+) Case Study of CL who suffered brain damage after tumor removal. Showed issues with episodic memories, but still able to create and recall semantic memories. Suggests episodic and semantic memories are separate systems using separate brain areas.

(-) Seems to be overlap between episodic and semantic memories. Research suggests episodic memories may be gateway to semantic memories.

(-) May be case that semantic memory is not single type of memory.

14 of 30

Procedural LTM

Procedural LTM - Implicit (non-declarative) form of memory so doesn't require hippocampus and consists of memory for skills. Many procedural memories learnt in early life including important motor skills like walking, dressing, language, etc, and usually learnt through repetition and practice. Mainly associated with motor cortex, cerebellum, and prefrontal cortex.

(+) Clive Wearing unable to form new LTM after hippocampus destroyed by viral infection. However, still able to play piano and sight read which suggests procedural memory was intact. Provides support for idea that there are different types of LTM stores.

(+) H.M. was bale to learn howto draw a figure by looking at its reflection (mirror drawing) which is procedural memory, however he had no memory that he had learnt this. Provides support for idea that there are different types of LTM stores.

(-) Lack of research that demonstrates which areas of the brain are associated with procedural memory. This is because most supporting research are case studies with brain damage patients which are rare.

15 of 30

Interference Theory

Forgetting is failure to retrieve information due to it being unavailable or inaccesible. 

Interference theory- Forgetting being due to information being confused or disrupted by other information. Proactive interference works forwards in time, where old memories make it difficult to retain new memories. Retroactice interference works backwards in time, where new memories make it diffuclt to remember old memories correctly. Interference is worse when memories are similar.

(-) Only explains forgetting if two pieces of information are similar, so cannot explain all instances of forgetting. Also not common to be presented with two pieces of similar information so can't explain most instances of forgetting in real life situations. 
(+) Support from Schmidt et al. (2000). Found positive association between retroactive interference and recalling street names. Suggests that retroactive interference can explain forgetting in real life situations.
(+) Research highly controlled lab studies. Means high internal validity and easy to establish cause and effect. However, low ecological validity so difficult to apply results to everyday life and situations.
(-) Studies don't and can't identify cognitive processes so inferences must be made. This reduces validity as no direct empirical evidence and inferences may be incorrect as processes not directly observable.

16 of 30

Key Study

Schmidt et al. (2000)

Investigated retroactive interference on memory of street names.

211 participants who were previously students at Dutch elementary school, and given map of area where school is located with street names removed and replaced with numbers. Asked to name as many street names as possible, and gave questionnaires about number of times they had moved town or city. Retroactive interference measured by number of times individual had moved (and therefore learnt more street names).

Found positive association between number of time participant had moved town or city, and number of street names correctly recalled. Suggests retroactive interference can explain forgetting in real life situations.

(-) Extraneous variables. Those who walked to school or spent more time in area would have had more time to learn and remember street names. 

(+) Methodology shows possiblity to conduct research in retroactive interference in real life settings so high ecological validity.

17 of 30

Retrieval Failure

Cue-dependent forgetting is when information is available in LTM, but can't be accessed, so recall is dependent on retrieval cues (reminder/prompt to aid recall such as state or context). 

Tulving (1973) proposed encoding specificity principle which states recall is likely poorer if context of recall is different to coding. Also suggested that effectiveness of retrieval cue is based on how overloaded it is, how deep the processing was, and how well the cue fits with the information.

Effect of retrieval cues are tested by looking at the effects of category headings on recall as the headings act as cues to help the individual remember. Tulving and Pearlstone (1966) gave participants 48 words which were split into groups of 4 and given a heading. Those presented with headings remembered more words as headings acted as cues to trigger recall.

18 of 30

Context-dependent failure

Context-dependent failure is when external environment is different at recall from where the information was learnt. 

Abernathy (1940) found participants had worse recall when tested by unfamiliar teacher in unfamiliar room. Supports context-dependent failure as explanation of forgetting as forgetting more likely to occur if lack of context-dependent cues.

Godden and Baddely (1975) asked scuba divers to learn list of words in and out of water and found participants had worse recall when testing in different context than coding. Supports context-dependent failure as explanation of forgetting as shows that environment acts as retrieval cue.

19 of 30

State-dependent failure

State-dependent failure is when internal retrieval cues (physical or emotional states) are different at the point of coding and recall.

Overton (1972) asked participants to learnt material sober or drunk, and asked to recall half in same state and half in different state. Found participants peformed better in same state as recall and coding. Supports state-dependent failure as explanation for forgetting because recall was poorer if states were different at recall and coding. 

Darley et al. (1973) gave participants weed and then asked them to hide money. Found they were less able to recall location when not under the influence of the drug, but could locate it when they were high again. Supports state-dependent failure as explanation of forgetting as participant's internal state acted as cue.

20 of 30


(+) Highly controlled lab study so more objective and empirical methods used, which increases validity of study and allows cause and effect to be established. However, also means that it lacks ecological validity which means results are necessarily applicable to everyday life and are therefore less valid/lack mundane realism.

(-) Godden and Baddely's findings only occured when during free-recall and not during recognition test. This suggests that cues only prevent forgetting in recall scenario, but have no effect when being tested on recognition. Suggests cue-dependency explanations for forgetting can't explain all incidences of forgetting.

 (+) Cue-dependent forgetting fits with Levels of Processing theory. Suggests information that is more deeply processed during coding is less likely to be forgotten. This is because more links and associations created between items in LTM which provides more retrieval cues. 

(+) Smith (1979) showed that imagining the room you were in when info is learnt is just as effective as being in same room during recall. 

(+) Real-world applications. Police reconstruc crimes to enhance eyewitness testimony by mentally recreating context of incident using retrieval cues (cogntive interview).

21 of 30

Misleading Information on EWT

Eyewitness testimony - Evidence provided by a person who witnessed an event. Three stages: encoding information into LTM, retention (memories may be lost or distorted), retrieval (accuracy may be affected by questions asked). Leading questions and post-event discussion are both forms of misleading information that can makes eyewitness testimonies inaccurate.

Leading questions - Question that suggests a certain answer and often activates particular schema which may lead to eyewitness giving desired response. 

Post-event discussion - Misleading information is added to the memory after the event. Can occur when eyewitnesses discuss what they saw and compare views of event.

22 of 30

Key Study

Loftus and Palmer (1974)

Investigated whether leading questions distort eyewitness recall accuracy.

45 students shown film of car crash and asked critical question. "How fast were the cars going when they...... each other?" Five groups given different verbs (contacted, hit, bumped, collided, or smashed). Then asked to give estimate of the speeds that the cars were going in mph. 

Average speeds: contacted (32%), hit (34%), bumped (38%), collided (39%), and smashed (41%). Shows that misleading information in form of misleading questions can affect memory recall.

(-) Lab study so lacks ecological validity. It's focused on watching clips which isn't the same as witnessing an event so lacks evidence.

Conducted second experiment to see if memory was distorted by post-event information. Participants divided into three groups and shows film of car crash.  Participants asked how fast they thought car was going when they 'hit' or 'smashed' (third group was control). Participants asked a week later whether they saw broken glass. Group with verb 'smashed' more likely to see broken glass when there wasn't any. Shows post-event information can affect memory recall of eyewitnesses.

23 of 30

Post-Event Discussion

Gabbert et al. (2003) put participants in pairs and showed them videos of the same event but from different perspectives so they saw different items. Half of pairs were encouraged to discuss event before each partner individually recalling the event to researcher. 71& of those who discussed event incorrectly recalled items. This suggests that there is confusion where information is from and incorrect information may incorporate into memory.

24 of 30


(+) Lab study so highly controlled. Risk of extraneous variables reduced and increased internal validity. However, this means it lacks ecological validity and not accurately represent real life EWT as some may not take experiment seriously or experience same emotions as witnessing real event. Therefore difficult to generalise to everyday life.

(+) Real world application. Highlights unreliability of EWT so beneficial to justice system. Therefore, more awareness of problems with relying on EWT, especially when there's no other evidence.

(-) Demand characteristics. Participants in Loftus' study were aware they were being studies so may display demand characteristics. Therefore, behaviour may not be natural which reduces internal validity.

(-)Participants don't expect to be mislead by researchers. Therefore, inaccurate recall should be expected as participants believe researcher to be telling the truth.

(-) Ethical issues. Loftus intentionally misled participants and used potentially distressing stimuli. This raises deception and psychological harm issues.

25 of 30

Anxiety on EWT

Loftus (1987) proposed 'weapon focus' which is the idea that if a weapon is present, individuals pay more attention to the weapon than other details and also increase levels of anxiety. Therefore, negatively affects eyewitness testimony as diverts attention away from important details and onto weapon.

Yuille and Cutshall (1986) investigated EWT on the proximity to the event, use of misleading questions, and levels of anxiety. Interviewed 13 witnesses of real life shooting durign armed robbery. Accurate accounts given several months later and those closer gave more accurate accounts. Misleading questions had no effect on accuracy. Challenged Loftus as suggests people with more anxietry had better recall.

Christianson and Hubinette (1993) questioned 110 witnesses who had seen bank robberies. Found that those threatened were more accurate and detailed in recall which suggests emotional arousal impoves memory recall.

Foster et al. (1994) tested consequentiality. Showed two groups a video of a bank robbery and told first group that their responses would influence the trial. First group identified robber more accurately, suggesting that consequences of identifying someone is important factor in eyewitness testimony. 

Yerked-Dodson inverted U hypothesis suggests recall is best at medium anxiety levels.

26 of 30

Key Study

Loftus et al. (1979)

Investigate effects of anxiety on identifying individuals in an event.

First group overheard low-key discussion in lab about equipment failure. Person then emerged holding pen with grease on hands (less anxiety). Second group overheard heated and hostile discussion in lab. After sound of breaking glass and crashing chairs, man emerged from lab holding knife covered in blood (more anxiety). Then given 50 photos and asked to identify person who came out of lab. 

49% identified man in first group, and 33% identified man in second group. Shows that witnesses concentrate on weapon which draws attention away from perpetrator. Concludes that fear or anxiety caused by sight of weapon narrows focus which gives more accurate recall of central details but less accurate recall of peripheral details.

(+) Research from Loftus and Burns (1982). Participants watched violent or non-violent short film of crime. Those who saw violent version were less accurate recalling information.

(-) Ethical issues as participants deceived and may be upset witnessing bloodstained knife. 

27 of 30


(+) Research mainly lab studies so high control, objective, and allows replicablity.

(+) Studies on real-life witnesses have high ecological validity as finding have been in real-life environment so findings can also be applied to real-life situations.

(-) Inconsistent findings. Lab studies tend to show that less stress and anxiety results in better recall, but real-life witness studies tend to show that more stress and anxiety results in better recall. Therefore, unclear whether anxiety improves or decreases accuracy.

(-) Weapon focus may not be caused by anxiety, but actually surprise. Pickel (1998) arranged participants to watch thief enter salon carrying scissors (high threat, low surprise), handgun (high threat, high surprise), wallet (low threat, low surprise), or whole raw chicken (low threat, high surprise). Identification least accurate with high surprise rather than high threat. Suggests weapon focus is due to surpise rather than anxiety.

(-) Ethical issues. Participants not always fully informed and can;t give fully informed consent. Also put under stress so could cause psychological harm.

(-) Individual differences such as mental health, relationship to victim of suspect, etc.

28 of 30

Improving EWT Accuracy

1. Mental reinstatement of conext - Mentally return to environmental and emotional context of crime scene. Improves accuracy because provides conext-based retrieval cues. "What were you feeling?"

2. Report everything - Recall all information even if appears to have little relevance. Improves accuracy because may provide information that was previously overlooked. "Can you tell me everything you remember, even if it seems insignificant?"

3. Change of order - Recount scene in different order e.g. backwards or from certain part. Improves accuracy as checks for inconsistencies and helps verify accuracy. "Can you tell me what happened, working backwards from when the robber left to when he arrived in the building?"

4. Change of perspective - Recount scene from different perspective. Improves accuracy because should provide witness with retieval cues to acces more information, and promotes holistic view. "What would someone from the other side of the road have seen?"


29 of 30


(+) Geiselman et al (1985) found CI procedure produced more accurate recall them SPT technique or interviews conducted under hypnosis.

(+) Kohnken et al (1999) conducted meta-amalysis of 55 studies and found CIs produced more accurate detail but also more inaccurate detail too. However, most participants college students (lack population validity) who were also tested in lab conditions (lack ecological validity).

(+) Geiselman (1988) showed violent crime to students and then interviewed 48 hours later using CI or SPI. Found cognitive review produces more correct items but also more incorrect items. Shows that CI increase amount of correct and incorrect information recalled.

(+) Milne and Bull (2002) suggested 'report everything' and 'mental reinstatment' are key for increasing recall.

(-) Interview no longer just one procedure, time-consuming, trainsing issue, and individual differences.

30 of 30


No comments have yet been made

Similar Psychology resources:

See all Psychology resources »See all Memory resources »