multi store model of memory

Atkinson and Shiffrin 1968

describes the process of memory in three stages: sensory memory, short-term memory, and long-term memory.

linear model (info flows through system 1 direction)

Sensory register

  • Capacity: Limited to what the senses can perceive at any moment.

  • Duration: 0.5 seconds (Sperling)

  • Coding: sense specific → iconic memory (code visual info) + echoic memory (code auditory info)



  • Sperling

    • participants shown 12 items (digits/letters)

    • asked to write down all items or row indicated after a tone

    • recall worse when asked to write down all digits (around 42% recalled) than when asked to write down one row (around 75% recalled)

    • suggests the info had decayed in iconic sensory register before all written down, indicating very short duration


Short term memory

  • Capacity: 7±2 items (Miller's Law) → supported w Jacobs’ digit span test → Miller suggested this could be improved by chunking

  • Duration: 18 seconds without rehearsal (Peterson and Peterson)

  • Coding: acoustically



  • Jacobs digit span test (1887)

    • participants shown sequences of digits to repeat back in same order

    • length of the digit sequence increased progressively until participant can no longer accurately recall digits.

    • found that the average digit span was around 5-9 digits → supports Miller's Law

    • valid study bc replicated in more controlled studies (Bop and Verhaeghen 2005)


  • Peterson and Peterson

    • 24 students

    • participants shown trigram (3 letters)

    • asked to count backward by threes from given number for varying intervals (3, 6, 9, 12, 15, 18 seconds) to prevent rehearsal of trigram

    • After each interval, asked to recall trigram


    • found after 3 seconds, recall around 80%

    • after 18 seconds, it dropped to about 10%


    • concluded duration of STM decreases without rehearsal (eg disrupted by interference like counting)

    • without rehearsal, duration around 18 seconds

    • however, artificial stimuli→ memorising trigrams not everyday use of memory → lack external validity



Long term memory

  • Capacity: unlimited

  • Duration: lifetime (Bahrick)

  • Encoding: semantically (meaning)


    if we want to recall information from LTM, need to be transferred back to STM by retrieval



  • Bahrick 1975

    • investigate how well ppl can remember faces + names of classmates in high school over long periods of time

    • 392 American participants aged between 17-74

    • used high school yearbooks


      recall tested by:

    • photo recognition test → matching faces to names using 50 photos including some from participants high school yearbook

    • free recall test → writing down as many names remembered from graduating class


      photo recognition test findings

    • Participants tested within 15 years of graduation → about 90% accurate in photo recognition.

    • After 48 years, recall declined to about 70% for photo recognition


      free recall test findings

    • Free recall was less accurate than recognition

    • about 60% after 15 years, dropping to 30% after 48 years


    • shows LTM may last up to a lifetime

    • high external validity → tested meaningful memories





    Evaluation of the Multi-Store Model Of Memory

Strengths:

  • Research support : Baddley (1996) → STM and LTM coded differently


    • gave 4 groups different lists of words to remember:

      → Group 1: acoustically similar (cat, cab, can)

      → Group 2: acoustically dissimilar (pit, few, cow)

      → Group 3: semantically similar (great, large, big)

      → Group 4: semantically dissimilar (good, huge, hot)


    • When asked to recall immediately (recall from STM) did worse w acoustically similar words

    • When asked to recall after time interval (retreive from LTM + recall) did worse w semantically similar words


    • Findings suggest info is coded acoustically in STM and semantically in LTM


    • however later research evidence suggest STM + LTM being unitary stores too simplistic as we have diff types of LTM (episodic, semantic, procedural) and STM

    • however, artificial stimuli→ words may have no personal meaning to participants → MSMM not valid model of how we use memory in everyday lives



  • Research support : Case of HM → STM and LTM are separate and independent


    • after surgical procedure, removing hippocampus from both sides of brain to help with severe epilepsy, patient HM experienced amnesia


      → His memory assessed 1955, he thought it was 1953

      → When STM tested → scored high

      → When LTM tested → scored badly and never improved


    • suggests STM and LTM are separate and independent as HM damaged one of these stores with the other remaining unaffected, validating framework of memory proposed by Atkinson and Shiffrin


    • however, case study → individual experience → cant generalise or repeat



Weaknesses:

  • Case of KF → suggests there could be another STM store for non verbal sounds

    • had amnesia

    • STM bad when digits read out to him but fine when he read digits to himself

    • suggests there could be another STM store for non verbal sounds

    • goes against MSMM → MSMM says there are only unitary stores that process diff types of info


  • Emphasis of rehearsal

    • MSMM says the more is info rehearsed, more likely to transfer to LTM (prolonged rehearsal)

    • Craik + Watkins → suggest type of rehearsal more important than amount : maintenance rehearsal and elaborative rehearsal


    • maintenance rehearsal → repeating info to keep it in STM → if rehearsed long enough, passes into LTM

    • elaborative rehearsal → link new knowledge to existing knowledge → means info can be transferred to LTM without prolonged rehearsal


    • MSMM does not fully explain how long term storage achieved



  • The cognitive experiments testing aspects of MSMM r artificial

    • eg use meaningless digits (Jacobs) and words (Baddley)

    • lacks external validity

    • results collected in lab environment may not be generalisable to naturalistic situations

    • lack mundane realism as experimental tasks arent good representations of how ppl use memory in real life scenarios