Multistore model of memory

• Who created the multistore model of memory

  Atkinson and Shiffrin
• What is the MSM

  An information processing model:

  • Linear - info is passed in one direction
  • Passive stores - stores hold onto info before being passed on or lost.
• The stores 3 of MSM

  • Sensory register
  • Short term memory
  • Long term memory
• Features of each store

  • Coding - the format Info is stored in
  • Capacity - how much the stores can hold
  • Duration - how long the info can be held
• Sensory Register

  • The store that directly receives sensory information
  • Not under cognitive control
  • All info from STM and LTM was initially gathered by the Sensory Register
  • Info is passed onto the STM by paying attention
  • Any info that isn't paid attention to is lost
• Coding of SR

  • Coding is different for each sense, meaning it is modality specific.
• Capacity of SR

  • Very large, potentially unlimited, as the brain needs to detect all sense info we receive.
• Duration of SR

  Very short, as low as 250 milliseconds, because so much info is held, it cannot be retained for very long.
• Short term memory

  • Receives info from the SR by paying attention.
  • Passes info to the LTM through rehearsal.
  • Info is passed back from the LTM with retrieval.
  • Info can be lost via displacement (new info) or decay (lost over time).
• Types of rehearsal

  • Maintenance rehearsal - repeating the info.
  • Elaborative rehearsal - linking to info already in the LTM.
• Coding of STM

  • Stored acoustically, meaning in the form of sound / spoken words
• Capacity of STM

  • Approx. 7 +/- 2 items (5-9) - suggested by Miller.
  • This can be improved by chunking.
• Chunking


  Making small sets/groups of items, reducing number of items overall.
• Duration of STM

  • Short, 18-30 seconds
  • This can be extended by Verbal rehearsal
• Long Term memory

  • Info comes from the STM via rehearsal.
  • To use info stored in LTM, it needs to be passed back to STM via retrieval.
• Coding of LTM


  Info is stored semantically meaning in the form of "meaning"
• Capacity of LTM

  • No limit to how much the LTM can hold has been found.
  • Info can be lost.
  • However, it seems that it has not completely been lost but cannot be accessed.
• Duration of LTM

  • Potentially unlimited.
  • As recall of childhood is normal for older people
• Research evidence for the MSM - Glanzer and Cunitz

  • Asked ppts to free recall word lists.
  • Found that recall was much stronger at the start and end of the list.
  • Suggesting that there are separate short and long term memory stores.
  • due to the primary effect, words first heard entered the LTM and were recalled.
  • and the recency effect, middle words were in STM but were displaced by later words.
• Research evidence for the capacity of SR - Sperling.

  • PPTS were quickly (1/20th of a second) presented with a 3x4 grid containing 12 letters.
  • Were then asked to recall one row
  • Found that recall for a row was over 75%
  • Suggesting that all rows were contained within the large capacity of the SR.
• Research for duration of SR

  In Spearling's study, he found that participants could only recall the first 4-5 letters, suggesting the letters faded from the sensory register, suggesting duration for SR is <1s
• Research Evidence for Coding of STM - Baddely

  • Gave 4, 10 word lists to 4 PPTS.
  • A: Acoustically similar, B Acoustically dissimilar, C Sematically similar, D Semantically dissimilar.
  • Found that immediate recall was worse for the Acoustically similar list, after 20 mins was the worst with Semantically dissimilar words.
  • Suggesting that coding in the STM is acoustic, as recalling Acoustically similar words was most difficult as recalling similar sounds caused confusion in recall.
• Acoustically similar means words that sound the same
• Acoustically dissimilar means words that sound different
• Semantically similar means words that have a related meaning
• Semantically dissimilar means words that are unrelated
• Research Evidence for the Capacity of STM - Jacobs

  • PPTs were presented with a list of letters or numbers.
  • PPTs then had to recall the list.
  • Found that the capacity was on average around 7 times for letters and 9 for numbers.
  • Suggests that capacity for STM is very limited
• Research evidence for Duration of STM - Peterson and Peterson

  • Showed PPTs 3 letter trigrams
  • PPTs then had an interference task, to count backwards for a few seconds to stop maintenance rehearsal.
  • Found that after 18 seconds recall was less than 10%
  • Suggests that unless maintained, info is held in STM for only a few seconds.
• Research evidence for the coding of LTM - Baddeley

  • Baddeley found that immediate recall was the worst for acoustically similar words and recall after 20 mins was the worst with Semantically dissimilar words.
  • Suggesting that coding in the LTM is Semantic, as recalling Semantically similar words was most difficult as recalling similar meanings caused confusion.
• Research evidence for the capacity of LTM - Wagenaar

  • Created a diary of 2400 events during the course of 6 years, including who, what, where and when
  • found that when testing these cues, he had 75% recall of one particular detail after 1 year, 45% after 5 years.
  • His retention judgement was 80% after 5 years.
• Research Evidence of Duration of LTM - Bahdrick

  • 392 PPTs aged 17-74
  • Tested for memory of old photos and names of their school friends.
  • Found that recall in matching names to faces was 90% after 15 years, and 80% for names after 48 years. Suggesting the duration of LTM is very large.
• Eval of MSM

  • The experiments of the MSM are often highly artificial, lacking in external validity
  • Also due to the low ecological validity, the experiments may not be generalisable to other naturalistic situations.
  • Also lacks mundane realism, not testing memory accurately to how people would use it in real-life situations
  • However, these are the only/most effective ways of studying memory.
• Eval of memory 2

  • Models of memory cannot be directly observed, meaning that researchers have to make inferences about the memory structure.
  • The inferences could be educated guesses, biased or overall incorrect.