The multistore model of memory

Multi-store model of memory - Atkinson & Shiffrin (1968)

Forgetting = decay,displacement, interference

Sensory Register:

• duration: 1/4 to 1/2 second (< 1 second)
• capacity, potentially unlimited - all sensory experience
• Encoding: sense specific
• if attention isn't paid to the information then it is lost through decay.

Short term memory store:

• duration: 18-30 s max (Peterson and Peterson)
• capacity: 7+/- 2 items (Miller)
• encoded phonologically
• stored auditory
• lost information through displacement
• info maintained through light maintenance rehearsal
• if info is understood semantically through deep elaborative rehearsal then info moves into the long term memory store.

Long term memory store:

• duration: unlimited
• capacity: unlimited
• encoding: mainly semantic (but can be visual and auditory)
• lost information through interference

MSM evaluations - supporting studies

• STM - Jacobs (1887) found the average digit span for numbers was 9.3 items whereas it was 7.3 for letters. hence, the type of information can affect the STM. Maybe this was due to 26 letters being more confusing than 9 numbers.
• STM - Miller devised lab-based studies to investigate the number of words that can be recalled and found the magic number (7 +/- 2 items) was almost universal in retention for STM.
• STM- Peterson and Peterson conducted a lab experiment in which 24 P's had to recall trigrams. Found that after 18 s, just 10% of trigrams could be recalled and by 30s, almost no info was retained by STM.
• LTM- Bahrick et al (1975)- 392 former high school students of various ages were given a photo-recognition test w/ 50 photos from the P's yearbook. Using free recall, accuracy: 35% after 15 yrs of graduating, 20% after 48 yrs of graduating. Using photo-recognition, accuracy 90% after 15 yrs, 70% after 48 years. High levels of accuracy of photo-recognition condition suggest that a large amount of info remains in LTM after many years. Also suggests that the lower level of accuracy seen for…