The Multi Store model

Memory has 3 basic processes:

- Encoding is the creation of knowledge to be stored.
- Storage is integrating and organizing the existing knowledge.
- Retrieval is re acessing the previously learned material and being able to apply it to present activity.


Atkinson and Shiffrin (1968) multi store model of memory

They saw memory as a series of processing systems which followed in a sequence. External stimuli comes in from the environment and is registered in the sensory memory where it stays for a split second before decaying. This component has a small capacity and is fragile. Info can be lost if not rehearsed.
Items in STM are usually sounds although other kinds of encoding are positive if the material is sufficiently rehearsed it gets passed to LTM.
+/-:
- Doesn't state how memory works or why some things are easier to remember than others.
- It's simplistic for psychology as a science.
- Deterministic as no account for individual differences e.g. people with brain abnormalities.
- Reductionistic as doesn't explain complex processes such as forgetting.
- This would be hard to test.
+ It does show there's distinctions between STM and LTM. Milner (1966)'s case study of 'HM' who had brain damage reported he was able to recall early life but unable to remember ten years beforehand or retain any new information.

Sensory memory

- Stimuli going into memroy system from external environment registered firstly in sensory store.
Atkinson & Shiffrin proposed 3 different memory stores.
- iconic (see) duration - 0.2-0.4 secs.
- Echoic (hear) duration - 3-4 secs.
- Haptic (touch)
They all overlap to perceive continuation.

Sperling (1960)
aim- to find out evidence of sensory memory.
procedure- 3 rows of letter stimulus in whole report technique. In this condition they could only recall 4 or 5. p's said they knew the letters but couldn't recall so he tested this.
He changed the procedure (partial report technique) where p's distinguished between 3 tones with the row of letters.
P's could only recall 3 letters from each line. They didnt know what would be asked so all rows were available in iconic memory but image fades. They performed better on second procedure as they had the echoic information available too.


Short and Long term memory


evidence for a distinction between stm and ltm is to conduct a 'free recall task'. The researcher would then plot a graph to show the relationship between where a word appeared in comparison to how often it would be recalled.
This produces 'characteristic serial curve':
- recency = last few words= easily retrieved
- primary= words at beginning passed onto LTM.
- asymptote= words at middle = displaced.


Glanzer and Cunitz (1966) 
Aim: Investigate how distractions effect free recall. 
procedure:
- used free recall task
- condition 1: recall immediately
  Condition 2: distracter task, 30 secs, recall.
Results:
- Condition 1: found expected serial curve
- Condition 2: distrater disrupted recency effect.
Conclusion:
- last few words couldn't be retrieved. (primary) first words were displaced to LTM.
Evaluation:
- lab. Controlled. Low ecological validity.
- Repeated trials to avoid unrepresentative results, and get the average score.
-  informed consent.


Neuropsychological evidence of STM and LTM.

Milner (1966) 
Studies people with seveer memory loss.
Case study on HM. He had epilepsy and an operation to alleviate it left him with severe memory loss. He could only recall early memories, but not the most recent 10 years and he couldn't recall or retain new info. His STM was intact, but LTM was not effective.
Shallice and Warrington (1970) 
Case study on KF. After motorcycle accident, he suffererd brain injuries. Appeared to have an intact LTM as he was able to learn new info. However STM was affected, so that he had a recency effect of only one item

Capacity:


LTM: Unlimited, lost (decay, interference).
STM: Limited storage space.
Use 'digit span technique' (repeat back a string of digits in order of presentation) to work out the capacity of STM.
Jacobs:
Found on average 7 digits could be recalled immediately.
Miller (1956):
Proposed we could hold 7 items in STM
Immediate memory span determined by number of chunks of info rather than individual letters/ numbers. The chunks have to be meaningful to be remembered efficiently.
- Chunking increases memory span, we recall 7+/-1 of chunks at  any one time.


Factors effecting capacity of STM
- Pronunciation time
- Individual differences
- reading aloud: digit span increases if p's read the digits aloud instead of subconsciously.  Baddeley suggests this is because digits are stored in echoic store which strengthens memory.
Cowan (2000)
Believes performance on span tasks is affected by rehearsal and LTM doesn't reflect the capacity of STM. He estimated that the capacity of STM is actually 4 chunks when such factors are controlled.
Barner
found digit strings that were repeated within a series of immediate memory span trials become easier for p's to recall.
This suggests that the strings have been gradually rehearsed and stored in LTM, which temporarily increases the capacity of LTM.


Duration:
LTM can hold unlimited info. 
Bahrick (1975) attempted to explore the length of time memories can be retained.
He used 392 American graduates and tested them on their memory of cues such as matching former classmates pictures to names.
p's performed well until 34 years.
The performance was better on recognition task than recall.
Factors effecting LTM duration:
-Experiments like Bahrick's show that we recall better when cued.
-Depth of learning= people likely to remember things for longer if they have learnt it very well in the first place.
-Pattern of learning= Bahrick 91987) said people who had learned spanish over a period of spaced sessions retained vocab longer than when learned in intensive sessions.
-The nature of material to be learned- It's meaning.


STM: temporary store
According to A+S we need to rehearse the info to recall it.
Peterson and Peterson (1959)

Aim: To find out how long items would remain in STM without rehearsal
Procedure: Trigram. Count back in 3's (stop rehearsal) Stop counting. Recall trigram. 
Results: P's able to recall 80% of trigrams after 3 sec interval without rehearsal bu it became progressively worse after. At 18 secs they could recall fewer than 10%.
Conclusion: information decays rapidly from the STM when rehearsal or repetition is prevented.
Evaluation: Lab. however used repeated measures design to avoid individual differences. However trigrams are artificial. Low ecological validity. 
- Loww of info was more to do with capacity limitations than duration. Subsequent counting task might have displaced the trigram.
- Also possible trigrams presented on earlier trials caused confusion for p's resulting in trigrams being incorrectly cued. 

Factors effecting duration of STM:
-rehearsal
-intention to recall: making conscious effort to recall increases retrieval.
- Amount of info to recall.

Encoding: Length of time held
A+S said stimulus is encoded, visually, acoustically and semantically. 
Baddeley (1966) -show that encoding in LTM is semantic.
Aim: investigate the encoding for the sub systems, STM and LTM.
Procedure:  1. pool of short, familiar words in 4 categories; acoustically similar, acoustically dissimlar, semantically similar, semantically dissimilar.
P's had to write down the random sequence of  5words from each categort after presentation in serial order. 
2. he then modified experiment  to test LTM. He extended the length of the word lists to 10 and prevented rehearsal. Recall was tested after 20 min interval. 
Results: acoustically similar words harder to remember than in other categories. STM codes acoustically.
Under second condition, the psychologist found acoustic similarity had no effect but semantically similar words were poorly recalled.
Conclusion:LTM codes semantically.
Evaluation: Lab. Low ecological and population validity. 

Conrad (1964):
Aim: to find it people are likely to confuse items that are acoustically similar.
Procedure: Sequence of 6 consonants.
- Condition 1: letters acoustically similar
- Condition 2: letters acoustically dissimilar
P's immediately asked to write down in serial order.
Result: Hard to recall similar letters as would substitute. 
Conclusion: We must convet visually presented material to an acoustic code in STM that we then find it hard to distinguish between similar sounding words. Acoustic confusion.


+/- OF MULTI STORE MODEL


+ Evidence the encoding is different in both memory stores. In STM, memory is encoded acoustically, in LTM, memory is encoded semantically.
+ Model supports the differences between STM and LTM.
+ Important contribution to memory research.
+ There's a difference in the duration of info in both memory stores. Duration in STM can last up to 30 secs whilst the duration in LTM can last a life time.
+Evidence the capacity of both memory stores is different. In STM only 7+/-2 chunks/items can be stored. Whereas in LTM, the capacity is different.
-A+S believed info flows through a one way system however sometimes LTM has to be activated first for STM to work.
-They focused on rehearsal a lot however sometimes rehearsal isn't needed to retain info. Craik +Lockhart found things are remembered better if they are processed semantically.
-It is over simplified and fails to reflect the complexity of the human brain.
-Takes no account of the different types of things we remember and emphasises on the amount of info we can handle and disregards the nature of the info.
-Most of the evidence comes from lab experiments.