# MODELS OF MEMORY

unit 1

?

## Multi-store model of memory

MSM is also known as Modal Moder and Linear Moder

Made by Atkinson and Shiffrin in 1968

1 of 42

## sensory

Sensory Memory

Duration -> fraction of second (after physical stimulus is gone)

3 types of stimulus

1. iconic store -> visual input

2. echoic store -> auditory input

3. haptic store -> semantic

1. Allows to interrogate visual information so that (concious level) experience smooth, continuous experience. decide what images to pay attention to!

2 of 42

## evidence for sensory memory

Spearling 1960 lab experiment

He used a chart with 3 rows of letters which he showed for v.brief exposures (50 milliseconds) to his p's.

P's were immediately asked to recall as many of the letters a possible. but most could only remember 4/5. They frequently reported being aware of more letters even though they could no longer recall them.

From this, he decided to alter his experiment, slightly. He trained participants to distinguish three tones; participants would be instructed to recall the top row when signalled a high tone, the middle row for a medium tone and the bottom row if a low tone was sounded. P's recalled on average 3 items from whichever row had been cued by the tone; so 9-10 out of 12.

3 of 42

## STM

STM (short-term memory)

Duration -> brief duration (few minutes) temporary store

Capacity -> 7/+-2 chunks

Encoding ->  acoustic

Forget -> decay/ displacement

4 of 42

## evidence for STM capacity

IMMEDIATE DIGIT SPAN

Involves reading out a list of random digits and having p's repeat them back in the correct order. the sequence starts with 3 and gradually increases in length until it became impossible to recall them in serial order.

Over a number of trials, the sequence length at which p is correct for 50% of the time is defined as their digit span. Most people have a digit span of '7/+-2' (Miller 1956 'Miller's magic seven', according to him chunking occurs when we combine individual letters/ numbers into a larger meaningful unit)

5 of 42

## more evidence for capacity in STM

The reading speed of the p's was measured, than they were presented with sets of 5 words on a scree, the were 2 sets of words one with 1-syllable words and the other polysyllabic, p's were asked to write the 5 words in serial order immediately after presentation, they recalled several lists of both short and long words.

P's could recall considerably more short words than long. They were able to recall as many words as they were able to articulate in 2 seconds. There was a strong positive correlation between reading speed and memory span.

However it might be simply that short words are easier to recall than long as more familiar. Baddeley et al said in later versions of the study and it showed pronunciation time is more important over familiarity.  Miller had not been able to account for the findings of research which sowed immediate memory span to depend on nature of the stimulus.

6 of 42

## Factors affecting capacity of STM

• Influence of long-term memory- there are difficulties in using immediate digit span as a measure of STM capacity. It is difficult to exclude the influence of LTM; Bower and Winzenz 1969; found that digit strings that are repeated within a series of immediate memory span trials become progressively easier for p's to recall, this indicates information being stored in the LTM which is helping to increase STM capacity temporarily. Cowan 2000 believes that Miller may of over-estimated the no.of chunks that can be held in STM, (due to rehearsal in LTM), he said STM is 4 chunks when these factors are controlled.
• Reading aloud- If p's read the digits aloud before attempting to recall them, performance is better than when they simply read without doing so sub-vocally. Baddeley 1922 suggested as digits are briefly in echoic store in strengthens memory trace.
7 of 42

## carrying on factors affecting capacity in STM

• Pronunciation time- Other researchers have found that pronunciation time may be more important indicator of STM capacity than Miller's digit span. Schweikert and Boruff 1986 tested immediate span for a number of different type of stimulus, e.g. letters, colours, shapes, they found that people consistently remembered as many items as they were able to pronounce in approx 1.5-2 secs. Naveh-Benjamin and Ayres 1986 have tested immediate memory span for speakers of various world languages, they found, e.g that the digit span for native English speaks is considerably greater than Arabic - the only explanation for this finding if Arabic numbers take longer to pronounce than English; these results confirmed earlier finding in the experiment by Baddeley etal 1975 in Capacity of STM. Hich, Halliday and Littler 1984 found that the immediate memory span of young children was related to length to articulate words. Therefore the capacity is determined by the time constraints rather than structural limitations.
• Individual differences- Macleoad and Donnellon 1993 said people who are highly anxious haver shorter span
8 of 42

## evidence for STM duration

Bayrick 1975

He tested the memory of 392 graduates of an American high school for their former class mates, using various memory tasks- recognition of faces in photos, matching name to photo, attempting recall name with no photo cue.

P's did remarkably well up to the age of 34 but their performance was better on recognition than recall tasks. There was a dip after 47 years but was difficult to decide if was either due to time/decay or ageing effect on elder p's.

9 of 42

## more evidence for duration in STM

Peterson and Peterson 1959 lab experiment

This test was used to see how long STM lasts when rehearsal is prevented. P's were briefly shown a constant trigram (e.g. CPQ), than asked to count backwards in 3's from a specific number to stop them rehearsing the letters, after intervals of 3,6,9,12,15 or 18 seconds; p's were asked to recall the trigram

There was a negative correlation (line graph) of percent correctly recalled and interval in seconds.

However trigrams are artificial things to remember and may not reflect day to day memory. It is possible however the interference from the earlier trigrams ( not merely decay) cause the poor recall.

10 of 42

## Factors affecting duration in STM

• Maintenance rehearsal- The key point about findings based on the Brown-Peterson technique is that items disappear from STM only when rehearsal is prevented. New items can only take their place if existing items move on- either to LTM or because they are forgotten. If we want to remember something for a short period of time, we tend to repeat it subvocally or to ourselves. This repetition serves as a method of continually reinserting the information into STM and thereby strengthening the memory. Without rehearsal the duration in STM is v.brief.
• Deliberate intention to recall- Sebrechts et al. 1989 briefly presented p's with lists of 3 common English nouns and then gave them an unexpected test where they had to recall the words in the serial order (serial recall test). Correct recall of the items fell to 1% after only 4 seconds! Studies such as these demonstrate that information can vanish from STM in a matter of few seconds if people do not make a conscious effort to retain it.
11 of 42

## carrying on factors affecting duration in STM

• Amount of information to be retained- Murdock 1961 presented p's with either a single 3 letter word e.g cat or 3 unrelated words like hat;pen;lid. He then followed the same procedure as in Brown-Peterson technique. Forgetting for the 3 unrelated words was the same as for the consonant trigram in Perterson and Peterson study. However the single 3 letter word cat (which could be processed as 1 chunk) was remarkably resistant to forgetting, and accurate recall level was still at 90% after 18 seconds.This shows that, as in Miller's digit span the important factor is the number of chunks to be remembered rather the number of individual items i.e. letters.
12 of 42

## evidence for encoding in STM

The aim was to explore the effects of acoustic and semantic encoding in STM. P's were split into 4 groups; each group heard a list of 5 words drawn from either: -acoustically similar words (e.g man,mad,map), -acoustically dissimilar words (e.g pen,day,few), -semantically similar words (e.g great,big,large), -semantically dissimilar words(e.g hot, old, late). Immediately after hearing the words,they were asked to recall in serial order. this procedure was carried out 4 times

Acoustically similar words were harder to recall than dissimilar sounds. Similarity of meaning had only slight effect. The effect of sound similarity disappeared when he tested p's long-term learning, suggesting that a major factor affecting encoding is whether the items in STM or LTM.

This supports Conrad; STM relies more on sound than meaning in STM

13 of 42

## more evidence for encoding in STM

He looked to see if items that acoustically similar are easier to recall, he did this by showing random 6 constants quickly. 1 condition the letters were similar the other not.

The p's found the recall task quite easy but many made errors; the majority of errors was for the similar as must convert the visual to acoustic in STM and therefore get acoustic confusion.

14 of 42

## factors affecting encoding in STM

• The sound of words- Conrad's 1964 study suggested that STM relies heavily upon acoustic coding; other researchers thought that the type of stimulus to be remembered might affect the way it was encoded.Baddeley 1966, wondered whether Conrad's results were dependent on using sound than in any other characteristic such as visual form or meaning. He set about testing this idea by presenting p's with words rather than consonants and comparing the effect of similar in meaning and acoustic.
• Other ways of encoding- It seems likely that acoustic coding is the preferred method for STM, but there are several studies, which show that other modes of representation are possible.
15 of 42

## ltm

LTM (long-term memory)

Duration -> forever

Capacity -> no limit

Encoding ->  semantic (meaning)

Forget -> decay/ interference

16 of 42

## evidence for duration in LTM

Bahrick et al. 1975

The aim was to establish the existence of very long-term memory & to see whether there was any difference between recognition and recall. Investigators tracked down the graduates from a certain high-school in America over a 50 year period. 392 graduates were shown photo from their yearbook.

Recognition group: for each photo,p's were given a group of names and asked to select the name that matched the person in the photo. Recall group: p's were simply asked to name the people in the photos with no list.

P's did remarkably well but better in recognition. In conclusion people can remember certain types of information for almost a life time, very long-term memory appears to be better when measured by recognition test than recall.

It is unclear however whether the drop-off in accuracy after 47 years reflect the limit of duration or just a decline in age.

17 of 42

## factors affecting duration in LTM

• Experimental techniques- people remember things from a distant past much better with a cue. Bayrick et al.1975
• Depth of learning- People likely to remember if learnt well in 1st place. Bayrick and Hall tested LTM for algebra and geometry, people who had only done math to high-school showed a steady decline in recall over the years, however students who had gone on further in math showed a higher accuracy to as far as 55 years later.
• Pattern of learning-Bayrick 1987 looked at people who had learnt Spanish and found that vocab items learned in sessions over a space of time retained better than learnt in intensity.
• Nature of material to be learnt- If interest high will recall easier, Conway tested open uni students and found that statistics was are particularly well retained as involved skills not facts.
• Childhood amnesia- Very young children are incapable of laying down well-organized and integrated memories and so they are not available for later recall
18 of 42

## evidence for encoding in LTM

The aim was of the study was to explore the effects of acoustic and semantic coding in LTM. P's were divided into 4 groups and shown a list of 10 words drawn from 1 of the following: acoustically similar, acoustically dissimilar, semantically similar and semantically dissimilar. After an interval of 20 minutes during which they were given another tasks to do,p's were asked to recall in correct order,then 10 words they had heard. this procedure was carried out 4 times

Recall was much worse for semantically similar than semantically dissimilar. Recall for acoustic was similar for dissimilar and similar. This shows LTM uses semantic coding.

19 of 42

## factors affecting encoding in LTM

• Acoustic coding- our ability to recognize sounds, such as sirens shows that we can store material in an acoustic form.
• Visual coding- We can also easily bring to mind pictorial images of people/ places that suggest some visual coding in LTM
20 of 42

## Distinction between STM and LTM

To distinguish distinction between the 2.

Free recall ->  a way of testing memory, where p's can recall items from a list in any order.  This is below is the serial position curve.

21 of 42

## Significance in neuropsychological evidence for ST

• Case Study HM - had severe epilepsy than had brain surgery to remove parts of his temporal lobe and hippo-campus. The operation led to severe memory deflects, but his IQ was still average.He could recall events early in life but no within last 10 years of surgery. He could learn and retain new information so STM relatively intact but LTM not.
• Alzheimer's disease- serious disorder in brain and early symptoms include: sever memory impairment. Neurotransmitters (specialised chemicals in brain) is shown to be of low level in those who have alzheimers, especially acetylcholine. Drachman and Sahkin 1979 gave drug to p's that blocked the chemical and did various memory tasks testing STM and LTM and compared with control group, the result was low LTM but normal STM..
• PET- Positron Emision Tomography. FMRI- Function Magnetic Resonance Imaging. both modern brain scanners provides support to STM and LTM
22 of 42

## more significance in neuropsychological evidence f

• Case Study KF- young man had severe head injuries due to motorbike accident, intact LTM (so could learn new information and retain) STM effect so had recency effect of 1 item.
23 of 42

## carrying on distinction between STM and LTM

Glanzer and Cuntiz 1966

The aim was to find evidence for the distinction. P's had to do free recall and were divided into 2 groups: 1. Immediate recall group: p's recalled the words immediately after presented. 2. Delayed recall group: p's counted backwards for 30 seconds before recalling.

The immediate recall group remembered the 1st and last words best. The delayed group remembered the words form beginning of list best. Both groups had difficulty with middle.

Both groups found beginning of list easy as was in LTM, words at end in STM, when recall was delayed with distraction task p's were unable to rehearse words in STM and therefore forgot.

24 of 42

## Strengths and Weaknesses

1. MSM made an important contribution to memory research, information-processing approach has enabled psychologists to construct testable models of memory and provided foundation for future important work. research shows basic distinctions from STM to LTM with plenty of evidence.

2. Model if v.simplistic and reductionist, outlines components of memory but fails to explain process and complexity of the human memory.

3.It doesn't take account of different types of information that of individual processes. For example some easier due to humour, interest and relevance.

4. Role of rehearsal in the model is central for STM to LTM  but simple rehearsal is not most effective. Craik and Lockhart researched and found that if memories have meaning they are more likely to be remembered. Kulik and Brown 'flashbulb memory' highly emotion/ significant even are imprinted in LTM. We can also develop various strategies to help us remember things and msm cannot account for this.

25 of 42

## More Strengths and Weaknesses

5. Atkinson and Shiffrin believed the 1 way system, but it is clear that need to use LTM to make  sense of STM. Ruchkin et al. 1999 measured brain activity in p's who had been presented aurally with suggests that semantic information stored in LTM was used. a set of words and pseudo-words and asked to recall in serial order. If people only process information acoustically in STM there should be no difference in brain activity when processing the word, however it was found that there were considerable differences, which

6. Clinical evidence shows there is a difference between STM and LTM. KF had severely impaired STM but LTM was good, shows flow is interactive not sequential.

7. All experiments are lab based therefore doesn't reflect on how memory works everyday, data is open to interpretation. Some data MSM doesn't explain for.

26 of 42

## The WMM- What is workimg memory?

in the early 1970's, there was a huge interest in the topic of STM & many experimental techniques were devised to investigate it. The research became highly theoretical & lab based with little apparent relevance for everyday life. In addition, it was becoming clear that traditional multi-store models could not account for some of things we know about memory.

Baddeley & Hitch (1974) decided to try & approach STM memory research in a novel way by asking the question: 'What is is for?' One feature central to A&S's model was the idea of STM being a unitary store. B&H contested this & pointed out that some of the research data undermined this idea. Specifically they refereed to the cast if KF who had a digit span for 2 & yet could transfer new information to his LTM. This suggested that there had been selective disruption of his STM, In other words his digit span was impaired but other aspects of STM must have continued to function

27 of 42

## The WMM- What is workimg memory?...

B&H concluded that the STM must have more than 1 component & must be involved in processes other than simple storage. They envisaged STM as a sort of workspace where a variety of operations could be carried out on both old & new memories. Crucially for the model, they concluded that 2 tasks can be carried out simultaneously in STM provided that they are being dealt with by different parts of the memory system. they envisaged LTM as a more passive store that maintains previously learned material for use by STM when needed.

28 of 42

## Components of the WMM

Supervisory component central executive has overall control. Limited capacity but can process information from any sensory modality. Responsibility for  a range of important control processes include:

• monitoring & correcting errors
• starting the rehearsal process
• inhibiting irrelevant information
• retrieving information from LTM
• switching retrieval plans
• coordinating activity needed to carry out more than 1 processing task at a time.

the core component is supported by 2 'slave' systems, which can be used as storage systems, thereby freeing its own capacity to deal with more demanding information processing tasks. Slave systems have separate responsibilities & work independently of one another

29 of 42

## Components of the WMM

the phonological loop inner voice is a limited capacity, temp storage system for holding verbal information in speech based form

the visuo-spatial sketchpad inner eye is a limited capacity, temp storage system for holding visual information and or spatial information.

• Baddeley & colleagure (Logle 1999) have improved aspects of WMM since first made in 70's. the phonological loop now has a passive storage system -> the phonological store, linked to an active rehearsal system called the articulatory loop where words are kept by subvocal repetition. the visuo-spatial sketchpad has a passive visual store called the visual cache linked to an active inner scribe (rehearsal mechanism)

30 of 42

## Evidence for the phonoligical loop

aim to find evidence for the phonological loop

procedure they gave visual presentations of word lists for v.brief exposures & then asked p's to write the down in serial order. In 1 condition, the list consisted of 5 words taken from a pool of familiar, 1-syllable English word such as harm, wit. In the 2nd cond, the 5 words came from a list of polysyllabic words e.g organisation

result average correct recall overall several trials, showed a marked superiority for the short words (word length effect)

conclusion Word length effect, the capacity of the loop is determined by the length of time it takes to say words rather than by no. of items. They estimated this time to be 1.5 secs

31 of 42

## Evidence for the phonoligical loop

Methodological issues

lab experiment using a repeated measures design. it was well-controlled. however a criticism is that longer words are simply less familiar than shorter words are, so harder to recall

Ethical issues

no serious ethical issues in this kind of study but investigators always have to gain consent from p's & to debrief afterwards

32 of 42

## ...

B& colleagues produced more evidence for an articulatory loop when they investigated the word length effect under conditions of articulatory suppression (a p is given a task that would usually make use of the articulatory loop but they are simultaneously asked to repeat aloud a meaningless chant, e.g lalala.

They found that the word length effect disappeared (ie. short words were recalled no better than long). this suggest the advantage in recalling short over long words depends critically on having a verbal rehearsal system. ie. the articulatory loop. If this loop is filled up with irrelevant material ie. lalla. then it seems likely that short & long words are being processed elsewhere, probably in central executive

33 of 42

## Evidence for the visuo-spatial sketchpad

Shepard & Feng 1972

aim to find evidence for the visuo-spatial sketchpad

procedure S&H used shapes such as plan of box. p's were asked to imagine folding these flat shapes, to form a cube with shaded area as base, than to decide whether in finished cube the arrows would meet head on.

result they found that the time taken to make the decision was related to the no. of fold that would of been required if the p's had actually been doing the folding. in other words visual images work in v.similar ways to real life perception

conclusion  it is believed that the VSS is used in tasks like the above, for the temp storage & manipulation of visual patterns & spacial movement. it is difficult for 2 such tasks to be carried out at the same time

34 of 42

## Evidence for the visuo-spatial sketchpad (another

Baddeley, Grant, Wright & Thomson 1973

aim to find evidence for the visuo-spatial sketchpad

procedure giving p's a simple tracking task (following a spot of light with a pointer as it moved round a circular path) while carrying out a simultaneous visual imagery task, had to imagine looking at an angular block capital e.g F/H. p's were asked to hold the image in their head & than starting at bottom left hand corner to respond to each angle as a 'yes' if it included bottom/ top line & no if did not. BOTTOM YES. TOP NO

result p's had enormous difficulty in tracking the spot of light and accurately classifying the corners. this seems to be because the 2 tasks were competing for same limited resources of the VSS

conclusion the p's could successfully carry out the tracking task at same time as performing verbal task so 2 task simultaneously is difficult due to temp storage of VSS

35 of 42

## Evidence for the visuo-spatial sketchpad (another

Methodological Issues

highly controlled lab experiment although the task was rather artificial, the investigators used a repeated measures design to eliminate the effects of individual differences

Ethical issues

as with all studies, the investigators had to gain consent from the p's & debrief afterwards. p's might have felt frustrated at being unable to complete both tasks effectively & it is important that they were given reassurance that this is perfectly normal

36 of 42

## .

It has been suggested (Logie, 1995) that the visual cache stores information about visual form & colour, & that the inner scribe processes spatial & movement information. Klauer & Zhao 2004 supported this idea by asking p's to carry out 1 o 2 primary tasks:

At the same time as doing one of these tasks, they were asked to do either:

• a visual interference task or
• no secondary task (control condition)

they found that performance of the spatial task was much poorer for people who were simultaneously carrying out the spatial distractor task than for people who were doing the visual distractor task & vice versa.

37 of 42

## Evidence for the central executive

The Central executive has the responsibility for a range of important control processes. its responsible for switching attention between tasks, setting task goals, monitoring & correcting errors, starting the rehearsal process etc

this aspect of WMM is highly complex & B attempted to investigate the functions given to the central executive of selective attention & switching retrieval plans. he asked p's to generate a string of digits, 8 in total which have no relevance to each other & no kind of pattern emerging.

e.g 48 16 3 12 72 85 29 81

• recite the alphabet
• counting from 1
• alternating between letters & number e.g a1 b2 c3 d4
38 of 42

## Evidence for the central executive

This resulted in the digit string becoming less random especially in COND 3. B concluded that the random number generation & alternation task were competing for the same central executive resources.

There has been other studies investigating the regions of the brain that are activated during central executive processes

This study shows how the visual-sketch pad & phonological loop compete for the same central executive resources

39 of 42

## Strengths & weaknesses of the WMM

The model can be applied to previous research data E.g acoustic confusion effect, digit span etc & reinterpret it within the framework of WM. it can also account for findings that are difficult fro MSM to explain e.g some of the selective memory deficits that have been found in brain damaged patients such as KF

The model attempts to explain how memory functions Baddeley et al 1989 have presented evidence that the phonological loop e.g plays a key role in the development of reading & that phonological loop is not operative in some children with dyslexia. While it seems to be less crucial for fluent, adult readers it still has an important role in helping to comprehend complex text. It also helps in the learning of new spoken vocab

The model can be applied to various real life setting high correlation between WM span & performance on various tasks, it has been suggested that WM capacity might be used as a  measure of suitability for certain jobs, specifically there has been investigations into its use as a recruitment tool for the US airforce Kyllohen 7 Christal 1990

40 of 42

## Strengths & weaknesses of the WMM more

Individual differences in memory processing are considered. Turner & Eagle 1989 devised a test to measure the capacity of WM. they asked p's to hold a list of words in memory while simultaneously working out mental arithmetic problems. the number of words correctly recalled in a subsequent test was called the working memory span. this measure of WM capacity has been shown in a no. of studies (Engle, Kane & Tuholski 1999) to be linked to the ability to carry out various cognitive tasks such as reading, comprehension, reasoning, spatial navigation, spelling, note-taking etc.It has also been found that there can be individual differences within the components of WM. e.g Shah & Miyake 1996 have shown that an individual can score high on spatial WM yet low on verbal WM & vice versa.

41 of 42

## ...

NEGATIVE. The WMM does not offer a complete understanding in how memory works. the exact role played by the CE remains unclear & other researchers (e.g Shah & Myake 1996 have questioned whether it can be a single component or whether there are separate verbal & spatial WM systems. Cowan 1998 has suggested that in order to explain abilities such as text comprehension, WM should also encompass some kind of LTM activation, Berz 1995 has also criticised the model for failing to account musical memory cause we able to listen to an instrument without impairing performance on other acoustic tasks

42 of 42