Memory

Multi-Store Model of Memory (Atkinson & Shiffrin 1968): 

●      The Multi-Store Model (MSM) is a structural, linear model of memory which focuses on how the different parts of the memory store information, and how information flows from one section to another.

●      The Sensory Register is the place where information is held in the senses. Coding is modality-specific since information is stored from the sensory register as it is. For example, there is the: Iconic register (eyes: vision), Echoic register (ears: sound), Haptic register (skin: touch)

●      Duration 250 milliseconds. If a piece of information doesn’t receive attention it is lost through decay and if it does receive attention, it moves to the short-term memory. But it does have a large capacity. 

●      The function of short-term memory (STM) is to hold information that has just entered the sensory register so that it can be used to perform immediate tasks. It codes information acoustically. It has a limited capacity (7±2) and has to be rehearsed through maintenance rehearsal otherwise it will disappear through the process of decay. Can also be displaced.

●      Long-term memory (LTM) stores limitless information (duration and capacity unlimited) and uses semantic coding in which it is coded based on the meaning of the information. Information can be lost through several methods including: interference, retrieval failure and maybe through decay but some say we just lose access to it.

Example Question: Outline the main features of the multi-store model of memory. (6 marks)

“The MSM is a structural, linear model of memory. It states there are several unitary stores within the memory: the sensory register, the STM and the LTM. The sensory register has a huge capacity: it receives modality specific information from the five senses. This information decays after 250 milliseconds, unless attention is paid to it, in which case it passes to the STM. The STM codes information mainly acoustically, that is, by sound, has a very limited capacity of 7±2 and a duration of only up to 30 seconds with maintenance rehearsal. After this, the memory will be lost through decay; it can also be lost through displacement owing to the small capacity. If information needs to be retained for a longer duration, it will be passed on to the LTM via repetition or elaborative rehearsal. The LTM codes mainly semantically, and has a potentially limitless capacity and duration. Information is theoretically stored here for the remainder of our lives, though it may be lost through decay, interference or retrieval failure.”

Evaluation: 

Positives:

–       Baddeley provides evidence that STM and LTM are distinct stores (as the model suggests) because they code differently. The highest number of errors were made by the STM when words were acoustically similar and semantically similar for LTM. Similarity increases interference. They must be distinct stores or they would code the same (same number of errors for each).

–       Baddeley (1966) had a strong methodology. The lab experiment had high control by operationalising the word list type and the number of errors. They also controlled for extraneous variables such as distractions. High internal validity shows cause and effect – can be sure the number of errors is a result of the list type. Good research so we can conclude they are distinct stores. 

Negatives:

–       It is far too simple, STM and LTM are not unitary stores. Milner’s study of HM showed that his muscle memory (procedural memory) allowed him to play tennis but he could not tell someone he knew how to play tennis (semantic memory). If a unitary store, HM would exhibit damage to all types of LTM. Baddeley & Hitch used the dual-task paradigm. When the CE and AL were both used, task 1 was slower as attention was shared between the slave systems. They would all be the same speed if STM was a unitary store. 

Types of Long-Term Memory (Tulving 1972):

●      MSM says Long-Term Memory is unitary and semantic.

●      Episodic memory: Knowing that something has happened – stored as an event. They’re personal memories with emotional significance because they are personal experiences. E.g. picturing when I was six and carrying my goldfish on my lap in a plastic bag when I moved house 

●      Procedural memory: Knowing how (non-declarative). It is implicit knowledge and unconscious so if asked how to do it you cannot answer. Skills-based and muscle memory due to repetition. E.g. knowing how to tie your laces.

●      Semantic memory: Knowing the meaning of things/facts. An awareness of shared information – it’s not personally significant. It is declarative/explicit and conscious. E.g. knowing that WW2 ended in 1945.

Evidence:

(Point 3 could be that HM is a unique case study, who has undergone rare surgery for already neurodivergent epilepsy, and is not generalisable to neurotypical individuals)

●      Point 1: HM can be used to illustrate individuals having different types of LTM. HM couldn’t add to semantic or episodic memory but could add to procedural (he could play tennis). If LTM were a unitary store he would be equally incapable of making any form of new memory. 

●      Point 2: Whilst HM is far from conclusive due to being a case study, brain scans further support the findings of HM’s case study. Tulving et al (1994) offer nomothetic evidence that different areas of the brain are active when the 3 types of LTM are activated. Episodic memory causes the frontal lobe, parts of the temporal lobe and hippocampus to be activated and combine and form a memory of a particular episode. Parts of semantic memory are also linked to the temporal lobe whereas procedural memory is linked to the motor cortex and cerebellum, which are both involved in the control of motor skills. 

The Working Memory Model (Baddeley & Hitch 1974):

●      The WMM is a processing model of STM. It focuses on the cognitive operations enabling us to process memories within our short-term memory, and the parts of the STM that perform these operations.

●      Central executive: It has overall control of working memory. Processes sensory information and directs attention to particular tasks and allocates ‘resources’, the 3 slave system, to tasks. Has limited capacity (can’t attend to too much at once and cannot store data) but slave systems are used as temporary storage systems and are all independent so different tasks can be performed at once.

●      Phonological loop: Has limited capacity – hard to listen to 2 things at once. Deals with auditory information and maintains the order of information. Baddeley (1986) subdivided into: the phonological store which holds the words you hear and an articulatory process used for words that are seen or heard (they’re repeated in a form of maintenance rehearsal).

●      Visuo-spatial sketchpad: Has a limited capacity. Used to plan a spatial task and it is where the visual (what things look like) and/or spatial (physical relationship between things) information is temporarily stored. Logie (1995) suggested it could be divided into a visual cache storing info about visual items (e.g. colour) and an inner scribe which stored the arrangement of objects in the visual field.

●      Episodic buffer: This part of the WMM was added at a later point because Baddeley realised the model needed a general store. It has limited capacity. It integrates info from the CE, PL and VSSP and also LTM. Also maintains time sequencing, recording events/episodes that are happening. It sends info to the LTM.

Evidence:

Baddeley & Hitch Dual Task Paradigm – independent groups, random allocation. Condition 1 drew on CE and then AL only, and condition 2 drew on CE and then AL and CE (slower). 

–       Positive: Good control because task 1 kept the same & operationalised DV by recording time for task 1 (cause and effect). 

–       Negative: Low ecological validity (artificial) as you wouldn’t have to list random numbers or repeat ‘the’ (can’t generalise).

Shallice & Warrington – case study, using observation, of KF whose short-term forgetting of auditory info was greater than visual stimuli. Auditory problems were limited to verbal materials (letters) but not meaningful sounds (phone ringing). Verbal information is impaired but visual memory is mostly unaffected. Showed there are separate components of STM for visual (VSSP) and verbal information (PL).

–       Positive: Case studies are in-depth and even though an idiographic approach has been taken, can allow a generalisable nomothetic approach to be taken.

–       Negative: only investigates KF so the findings can’t be generalised to other individuals. As KF suffered an accident which impaired his memory, it may not be possible to generalise the findings to individuals who have not experienced an incident.

Brain scans Todd & Maris – more activity left side of the brain doing visual memory tasks and the right for spatial tasks. The prefrontal cortex involved with CE is doing a reasoning task (Braver et al) which supports that there are multiple stores. 

Baddeley et al (1975) – Phonological loop

Average correct recall showed a superiority for the shorter, monosyllabic, words. “The word length effect” – the capacity of the PL is determined by the time taken to say a word (rather than the number of items). It estimated this was roughly 1.5 seconds.

Klauer and Zhao (2004) – Visuo-spatial sketchpad

A spatial memory task was more disrupted by spatial than visual inference and a visual memory task was more disrupted by visual than spatial inference. This provides evidence for the VSS having distinct visual and spatial components.

Baddeley and Wilson (2002) – Episodic Buffer

Individuals with intact CE functions showed unimpaired immediate recall of prose, which would require the VSSP and the LTM for semantic meaning. They didn’t have any recall of the prose after a short time delay. The info from this episode (the prose) was temporarily buffered (stored), so it was only available for a short period of time.

Explanations for Forgetting:

●      Interference theory: competing memories fistful the recall of one memory. The more similar, the more competition. Ceraso (1967) suggests memories are temporarily inaccessible, not unavailable/lost.

●      Proactive interference: Past learning disrupts current attempts to learn something. Old memories prevent accurate recall of new ones such as making the same mistake in an exam because past, incorrect learning disrupts recall of the newly learned, correct information.

●      Retroactive interference: New learning disrupts previously learnt material such as forgetting your best friend’s maiden name after they have been married 20 years.

●      Retrieval failure due to the absence of cues: necessary triggers to access memory are not present. The encoding specificity principle (Tulving and Thompson 1973) suggests information present at the time of coding acts as a cue, making recall easier. 

●      Cues are triggers which have been coded into our memory at the time of learning so have a significant link to the material.

●      Context-dependent forgetting: implicitly coded cues from the environment being absent at the time of recall. E.g. forgetting a maths equation in an exam hall because you do not have the cues in the maths classroom.

●      State-dependent forgetting: emotional, mental or physiological cues at the time of coding. E.g. stress in an exam hall disrupting recall, as you are calm when revising.

Evaluation:

Proactive Interference: Keppel & Underwood (1962)

●      Lab experiment, repeated measures. PPTs were asked to learn consonant trigrams, and then perform the Brown-Peterson technique interference task of counting backwards in 3s for varying time periods. Forgetting increased with the time interval, but there was little or no forgetting of trigrams from the earlier trials. Past learning of the earlier trigrams disrupts the learning of current trigrams. As the time intervals increase, there have been more trigrams learnt so old memories disrupt new learning and prevent the accurate recall of new information.

Positives:

–       Repeated measures remove individual differences which may confound the results.

–       Highly controlled lab experiment – cause and effect

Negatives:

–       Lab experiment artificial (low ecological validity & generalisability) as it’s unlikely you’d be asked to learn trigrams and count backwards. 

–       Participants may have guessed the aim of the study (due to repeated measures) and do intentionally worse. This invalidates the study.

●      Underwood (1957) meta-analysis found that PPTs who were asked to memorise 10 or more lists recalled only c.20% after 24 hours, whereas PPTs who were asked to memorise 1 list recalled over 70%. When there were more lists to learn, the older lists (learnt first) disrupted attempts to learn further lists. The older lists would have accounted for the 20% most likely as the earlier lists prevented accurate recall of later lists

–      Positive:  Large sample size increases validity.

–       Negative: Relies on data being good quality.

Retroactive interference: Muller & Pilzecker (1900)

●      Lab experiment, independent groups. All PPTs were given 6 minutes to learn nonsense syllables. In condition 1 PPTs were then presented with an interference task of viewing and describing 3 paintings. In condition 2 PPTs were given no interference task during this time. Both groups were then asked to recall the syllables.PPTs who had performed the interference task performed notably less well than those who had not. The new learning of viewing and describing the paintings during the interference task disrupted previously learnt nonsense syllables, preventing accurate recall.

Positives:

–       They manipulated the interference task so the number of nonsense syllables recalled accurately is as a consequence of the interference task, rather than extraneous variables. High internal validity allows a causal relationship to be seen. 

–       Independent groups so participants only experienced one condition of the IV. This prevented participants from performing better in the second condition due to practising remembering the syllables.

Negatives:

–       Artificial as it’s unlikely that you’d be asked to learn nonsense and then asked to view and describe 3 paintings. Low ecological validity so not generalisable.

–       Independent group design is possibly confounded by individual differences, questioning the validity of the findings. 

Interference Theory:

●      McGeoch & McDonald (1931) Lab exp, independent groups. All PPTs asked to learn list A: set of 10 adjectives. Both groups were given a rest period of 10 minutes and then asked to learn list B, which was made up of synonyms of the adjectives, nonsense syllables or numbers. Those in the list of synonyms condition recalled only 12%, those in the nonsense syllables condition recalled 26% & those in the numbers condition 37%. The more similar the items are, the more competition there is between them, so there is a greater likelihood that interference will disrupt recall. This is seen in the synonyms condition which has the lowest recall.

Positives:

–       Lab so manipulated type of word list B. Cause & effect can be seen.

–       Independent groups prevent doing better due to practice. 

Negatives:

–       Artificial lab exp. Unlikely to learn random word lists, has low ecological validity and lacks generalizability.

–       Individual differences may confound the results if someone performs well due to their job for example.

●      Baddeley & Hitch (1977) Interviewed members of a rugby team. Asked players to recall the names of the teams they had played. Players who had been injured (fewer games) had better recall. Those who had played every match had lower recall so greater amounts of information (i.e. interference) disrupted the accuracy of recall. Memories compete against each other so the ability to recall some of the memories is disrupted.

–       Positives: Conducted outside of the lab (it is a natural experiment) so it is not artificial, instead it has mundane realism. High ecological validity can be generalised. 

Retrieval Failure due to Absence of Cues: 

●      Tulving & Pearlstone (1966) Lab experiment, independent groups. PPTs were asked to learn 48 words froThen asked to recall as many words as possible in one of two recall conditions. In condition 1 PPTs were asked to undertake free recall (i.e. they were given no cues), whereas in condition 2 PPTs were given the categories as cues. Free recall had an average recall of 40% of the words whereas cued recall had an average recall of 60%. When participants were provided with the category as a cue they performed far better as cues make recall of memories far easier. Also shows the memories are present (as people who are given cues could recall them) however they are not always accessible without cues. 

Positives:

–       Lab experiment manipulated whether a cue was present – cause and effect.

–       Independent groups reduced demand characteristics (if they guessed the aim).

Negatives:

–       Artificial lab experiment as it is unlikely you’d need to learn a word list. Low ecological validity, cannot be generalised.

–       Individual differences may cause one person to be better at recall which would question the validity of the research.

●      Godden & Baddeley (1975) Context-dependent forgetting: Field experiment, independent groups. Four experimental conditions: Scuba divers learned a set of words in one of two conditions: on land or underwater. The scuba divers were then either tested for recall on land or on water. Those who recalled the words in the same environment in which they had learned them had the most accurate recall regardless of whether this was on land or underwater! External cues were present making memories far easier to recall, due to the encoding-specific principle.  

Positives:

–       Field experiments have higher ecological validity so findings can be generalised.

–       No order effects due to independent group design (e.g. if repeated measures – interference second time a word list was recalled so less accurate).

Negatives:

–       Field so hard to control extraneous variables, reducing internal validity.

–       individual differences.

●      Goodwin et al (1969) State-dependent forgetting: Lab experiment. Male volunteers were asked to learn a group of words whilst either drunk (3 times over the UK limit) or sober. 24 hours later, PPTs were then asked to recall the list whilst either drunk or sober. Those who were in the same mental state as when they learned the list recalled far more accurately than those in a different state…even those who both learned and recalled the list drunk! Their mental state acted as a cue which made memories far easier to recall, due to the encoding-specific principle.

Positives:

–       Lab allowed them to manipulate whether participants were drunk or sober. High internal validity as the words recalled correctly was due to mental state – cause and effect.

Negatives:

–       It is artificial as you would not usually be asked to learn/recall a word list whilst drunk or sober. Lacks mundane realism and ecological validity so lacks generalisability.  

Factors Affecting the Accuracy of Eyewitness Testimony:

Misleading Information, Leading Questions Loftus & Palmer (1974):

Leading questions are questions that prompt a particular, desired answer.

1)    Lab experiment, independent groups, questionnaire. Shown 7 clips of a car accident and asked ‘how fast were the cars going when they __ each other?’ Each group was given a different verb and mean speed calculated. Contacted: 31.8mph & smashed 40.8mph. Leading Q biases recall – response-bias explanation.

2)     Lab experiment, repeated measures. 3 groups were shown a clip of a car accident for 1 min and asked about speed. A week later, they were asked if they saw any glass. No broken glass but when they thought the car was faster, they reported that there was (smashed). Q changed their actual memory – substitution explanation. 

Evaluation: 

Negatives:

–       Artificial, highly controlled laboratory experiment.  A video clip of a car crash would not have the same emotional impact as witnessing a real-life crash would. Participants may not have paid such close attention to the video as they would in real life because they are aware that they are in an experiment. Results are confounded, reducing the validity of the experiment. Low ecological validity of the research as it lacks mundane realism, therefore not generalisable.

–       The sample used consisted of students, rather than a range of age groups. Students are used to remembering lots of information so exercise their memory more frequently to those who are not in education.The sample is not representative of the whole population. The findings cannot be confidently generalised. 

Post-event Discussion Gabbert et al (2003):

A conversation between co-witnesses after a crime which may alter a witness’ memory of the event.

●      Lab experiment, independent groups. Each partner watched a different video from the same event. In one condition pairs were encouraged to discuss the event before recall. 71% recalled aspects they had acquired during the conversation, 0% in the control group. Agree for social approval or because they believe others are correct (memory conformity).

Evaluation:

Negatives:

This study was a lab experiment so participants were aware they were in a study. Likely less diligent they were when reporting. Foster et al. found that if participants thought they were watching a real robbery/crime, and thought their responses would influence a trial, their identification of the robber was more accurate. This study is artificial, lacking mundane realism and ecological validity, so conclusions can’t be generalised. Foster et al’s study indicates that if the circumstances in Gabbert et al’s study were to occur naturally, the events would be recalled far more accurately ( fewer than 71%). This questions whether memory conformity really would occur in a real setting, or just in a laboratory experiment. 

Anxiety Affecting Eyewitness Accuracy:

Anxiety reduces accuracy: 

●      Johnson & Scott (1976) – weapon focus effect (Loftus), the weapon creates anxiety and detracts attention from other aspects. Lab experiment, independent groups. A man runs out of the office after an argument with either a pen covered in grease (low anxiety) or a letter opener covered in blood (high anxiety). Accuracy in identifying the man was 49% (pen) and 33% (letter).  

–       Negative: Results may be confounded as Pickel (1988) found that accuracy was least accurate with high surprise items (e.g. a raw chicken). Lab experiment, independent groups. J&S results are not due to anxiety, but surprise.  

Anxiety improves accuracy:

●      Christianson & Hubinette (1993) suggest there’s an evolutionary argument e.g. to remember how you reacted to being chased by a lion on the EEA. Used victims of bank robberies (bank tellers) and bystanders (employees/customers) and interviewed them 4-15 months after. All had 75% accuracy but victims had the best recall. 

–       Positive: In the context of real life so far more sound than J&S because not artificial. Supported by Yuille & Cutshall (1986) who interviewed victims of violent crime and found that they had more accurate recall, despite leading Q’s. Anxiety has a different impact in real life compared to the lab. 

–       Negative: Participants had time to rehearse the information which may be the reason accuracy is more accurate.

Middle ground: 

●      Deffenbacher’s (1983) meta analysis of 21 studies, found that 10 showed improved accuracy and 11 showed the opposite. Yerkes-Dodson curve states that some anxiety improves accuracy, however, too much anxiety reduces it.

–       Positive: Large sample which reduces participant variables and strengthens Deffenbacher’s conclusions. 

Improving the Accuracy of Eyewitness Testimony:

The cognitive interview (Geiselman et al 1984) developed four techniques to enable witnesses to access memory traces that remain out of reach during standard interviews. 

●      Mental reinstatement of context: witnesses are asked to mentally recreate the physical conditions, and their thoughts and feelings, at the time of the incident. With a mugging, witnesses may be asked to ‘picture the scene of the attack; try to get an image of your surroundings in your mind.’ Open questions are designed to provide contextual and emotional cues to recall. 

●      Report everything: even details that may seem inconsequential as these may act as a cue to other, more important, memories. Sometimes small details can be important in themselves or may help create a more accurate account of the incident. 

●      Change the order of the events:  For example, ‘Tell me about the night you were mugged, beginning with when your attacker was chased away by the group of bystanders’. This avoids schemas influencing the witness’ recall of events; rather than saying what they would expect to happen in a mugging, they have to focus on the details of this specific mugging to tell the events out of chronological order. 

●      Change perspective: intended to prevent schemas by asking the witness to try and recall the events from someone else’s point of view. E.g. ‘Try to remember the incident from the perspective of one of the group of bystanders who chased your assailant away. What might he have seen?’

Evaluation:

–       Positive: Fisher et al (1987) found that standard police interviews often bombarded witnesses with questions, interrupted and discouraged them from talking freely. Loftus & Palmer shows this often causes leading questions which reduce the accuracy of recall. Using cognitive interviews improve the accuracy in comparison as the interview is more structured and standardised.

–       Negative: Whilst Fisher demonstrated the effectiveness in real life settings, Kohnken et al (1999) found that inaccurate information rose considerably (61%) in the cognitive interview as well as a 81% increase in correct information. This is concerning when it is applied to real life, convicting criminals, potentially wrongly.

–       Positive: Fisher & Geiselman (1987) suggested an amended version; the enhanced cognitive interview. The interviewer used open-ended questions, didn’t interrupt the witness, aiming to reduce the anxiety of the witness. Must be a strong model otherwise it could not be improved. 

–       Negative: Kebbel et al (1999) found that officers were concerned about the time it takes (use in conclusion).