Models of Memory

Models of Memory

Multi-Store Model

(Atkinson & Shiffrin, 1968) • They suggest 3 component s of memory:• Sensory Memory (SM)• Short Term Memory (STM)• Long Term Memory (LTM)

• Your Task: As we go through each component note down the MOST important points – be selective!

Component 1: Sensory Memory

• All information comes to us through our senses.• We `register` all of this information in SM.• `Attend` to some and neglect others. – largely

unconscious process.• DURATION: very short approx half a second.• CAPACITY: all sensory info.• CODING: Sense/Modality –specific: info held in same

sense that it was registered. e.g. eyes = info held in iconic form (visual form)

INFO

SENSES SM

Component 2: Short Term Memory (STM)

• Info that we attend to in SM is passed onto STM

• CAPACITY – between 5 and 9 on average• DURATION – 18 to 30 seconds• CODING – Mainly Phonological

(auditory/sound based).

SM STM

Capacity of STM

• Carry out your own experiment to test the capacity of STM.

Have a look at these TRIGRAMS

GER SOE GEF LET

REM DIT PIB LAF

NID CUR FES SEM

KEJ LOR TEF QUS

Capacity of STM

• Miller (56) presented participants with a list numbers. He then tested their immediate recall of thjese numbers.

• FOUND average capacity (number of items recalled) = 7 items, but people’s capacity was very commonly between 5 and 9.

• This evidence suggests that… EVALUATIVE POINT?• Capacity can be extended through CHUNKING.

• Through REHEARSAL – info maintained in STM and then transferred to LTM.

Component 3: Long Term Memory (LTM)

• If we rehearse info in STM (e.g. repeating it to ourselves, using it a lot) it gets transferred to LTM

• DURATION: Unlimited.• CAPACITY: Unlimited.• CODING: Mainly semantic (based on meaning) • If we need that information again at a later

date we can RETRIEVE it (call it back) into STM.

SM STM LTM

Coding

• Remembering directions that you’ve just worked out from a map

• In SM (takes iconic form – image)• In STM (stored in auditory/sound

format e.g. `turn left, turn right)• In LTM (transformed to semantic

code – becomes a remembered route – rather than a list of turns etc)

“ Turn left at the traffic lights, turn right, then straight on….”

My route to work, passes the bus

stop I used to wait at… and the road I used to live on…

Make Your Own Model

• You can use any of the materials that are in front of you to create your own representation of the model.

- Your model must in some way demonstrate the capacity and duration of each memory component.

- Explain your model to another group.- YOU HAVE 10 MINUTES!

You Make the Summary!

SM STM LTM

CAPACITY

DURATION

CODING

Finishing Task

• To summarise you need to use your summary to write out a section of prose to answer the Question

• “How does Multi-store Model of Memory work?”

Multi-store Model (AO1s)- Suggests 3 separate stores of information

and that different processes (attention, rehearsal etc) determine where that material goes.

- Firstly info enters SM (Capacity, duration & coding)

- Info that is attended to in SM is transferred to STM, info not attended to is lost. STM (capacity, duration and coding)

- Info that is important to us is rehearsed (what’s this?) and this information will then be transferred into LTM (capacity, duration and encoding)

Application

• The hospital told John over the phone the number that he needed to call to make an appointment. When John went to phone the next day he couldn’t remember the number.

• Why might the Multi-Store Model suggest that John is unable to remember the number?

Evaluation: is the Multi-Store Model any good?

• EVIDENCE

• Research has focussed on FUNCTIONAL SEPERATION – whereby the two stores are separate and do different jobs.

• If research supports functional separation this supports the idea of separate STM and LTM and thus supports M.S. Model.

STM LTM

Is the Multi-Store Model any good?

• Other Evaluative points?- The model states that rehearsal is needed for info

for it to be passed into LTM.- The dilemma presented by Patient KF: motorbike

accident – brain surgery – STM storage of only 2 items, but was able to learn and recall new information (LTM = normal). Why would multi-store model have trouble explaining this?

Model Number 2:…

Levels of Processing(Craik and Lockhart, 1972)

“How can you ensure you have learned information so that you can remember it for the exam?”

The Theory Bit

• Alternative to structural models – It does accept separate STM and LTM, BUT focussing most on memory processes.

• Info can be processed on different levels, this level determines likelihood of it being retrieved in future.

• Deeper processing = longer lasting & more likely to be retrieved

• Shallow = less long-lasting & less likely to be retrieved.

The Different levels

Shallow/ Structural Level

Can be Graphemic (e.g. by processing word by looking at letters that make it up or Orthographic, by processing the shape of the word

e.g. Deciding deciding whether the letters are upper or lower case

Intermediate/ Phonetic Level

Processing information based on the sounds of the words

e.g. deciding on how it would be pronounced/ or whether it rhymes with another word.

Deep/ Semantic level

Processing information based on the meaning of the words.

e.g. deciding if that information is describing or evaluating a theory

Continued…

• DON’T JUST WRITE THIS OUT – put it into most important bullet points:

• “In order to carry out structural processing, it is necessary

to simply process the words structurally, by scanning it visually. To process it phonologically, it is necessary to mentally `sound out` the words, whilst to process it semantically it is necessary to think about the meaning of the word, by relating it to the rest of the sentence or by putting it into a meaningful category, this involves relating the new information to information already in LTM”

Question?

• As such: In comparison to Multi-Store Model, what do C&L think is important regarding how well information is remembered?

• MAKE SURE YOU HAVE AN ANSWER WRITTEN DOWN!

Research Question

• Discuss in small groups – How can we test Levels of Processing Theory?

• Propose a research study that we could use.

How can we research L.o.P?

• Lists of words – diff tasks performed with them:• Think of an opposite for each word.• Does the word have an `th` sound in it?• Is the word presented in capital letters?

• Don’t tell pp’s that there will be recall after – WHY? (make sure you have an answer written down!)

• STAR STUDY – Craik & Tulving provide support for L.o.P using this method.

Evaluation

• We can test the model – and we overcome one of the vital problems in in M.S. Model research – what’s this problem and how is it overcome?

• There is some ambiguity in this theory – where is that ambiguity and why is it a problem?

• Which Mem Model can better explain why we may have strong recall for vivid events even without rehearsing them? Why?

• Multi-store model could explain findings like `Craik & Tulving’s?` How could this be (think about time involved in each task)?

Evaluative points• Morris et al (77) found phonetic processing may

sometimes be more effective than semantic.• Oversimplified – suggesting one LTM store but fails to

describe diff storage processes and systems for diff types of memories (e.g……)

• Neurological evidence to suggest that higher brain activity is associated with deeper processing. Brain imaging studies using PET and fMRI techniques have shown that higher levels of processing correlate with more brain activity, and activity in different parts of the brain than lower levels. This difference in `brain activation` may be a sign of `better` memory formation.

Model 3The Working Memory Model

***

INTRODUCTION: Take down the key bullet points:

• Whilst multi-store model suggests that there is one STM and that this is a very passive store of info, the working memory argues instead that Short term memory should be viewed instead as an active part of memory which has separate sub-systems. This model views that STM is a system that not only maintains, but also manipulates information.

Components: Expand on these points

• Made up of:

1) Phonological Loop:- phonological store and articulatory rehearsal

process

2) Visuo-Spatial Sketchpad:- visual and spatial component

3) Central Executive:- (attention, co-ordination, connects to LTM

Phonological Loop

- Info enters component – transformed to phonological code.

1) Phonological store – holds as it enters component.

2) Articulatory Rehearsal Process – Sub-vocal speech (holds 2/4 items) – rehearses info to maintain it.

Visuo-Spatial Sketchpad

1) Visual component – deals with objects and features

2) Spatial Component – deals with locations and movements

Central Executive

• `The Attentional System`

- Focusses and switches attention

- Co-ordinates sub-systems (deployment… and integration…)

- Link to LTM…

Revision

• The model was revised by Baddeley (2000), and a fourth component was added – WHY?

• What are the problems with 3 component model?• Watch this:• http://www.youtube.com/watch?

v=DeWsZ2b_pK4

Revision• The model was revised by Baddeley (2000), and a

fourth component was added – WHY?• Problems with 3 component model:- Struggled to explain how info moves to and from

LTM and WM.- No mechanism for allowing interaction of the two sub-

systems (in ordered chronological fashion) – we experience visual and sound info integrated not as separate `bits`.

- New component called: EPISODIC BUFFER

The Episodic Buffer

• The EPISODIC BUFFER is an additional sub-system which is able to integrate information from the p.Loop and V.Sketchpad in a chronological/ordered fashion.

• As such this enables model to explain our conscious experience (e.g. you understand a film as images and noise simultaneously – this is our conscious experience of a film.)

Make YOUR OWN Model of WM

• It needs to somehow represent:

• Central Executive

• Phonological Loop

• Visuo-spatial Sketchpad

• Central Executive

• Be CREATIVE!!

Evaluation

• Research Evidence :

1) Evidence from Interference tasks:

- If given two tasks to do using same system then performance should be impaired because the system only has enough resources to process one task at a time.