Short Term Memory and the Atkinson & Shiffrin Model

The basic structure and processes of the Atkinson & Shiffrin(1968) Multistore Memory Model

• Three Memory Buffers
• Control Processes which move information from buffer to buffer.

• Basically unlimited capacity
• Duration @300 msec
• Forgetting -- Decay
• Code: Low level sensory features

• Limited Capacity: 7 plus or minus 2
• Duration @ 10 seconds if not rehearsed
• Forgetting--Interference
• Code: Primarily verbal/acoustic

• Unlimited capacity
• Duration: Forever(?)
• Forgetting: Retrieval failure
• Code: Primarily semantic

• Attention
• Maintenance rehearsal
• Elaborative rehearsal

The Sensory Information Store

• Sperling (1960) showed subjects a matrix of letters very briefly (50msec)

• Subjects were asked to report all the letters that they could after this brief presentation. Ss averaged 4.5 letters
• Why? Two possibilities:
• SIS has a capacity limit of @ 4.5 items
• Information in SIS decays quickly and so all the letters can't be reported.
• Sperling's Partial Report Technique
• Assign each row of letters a different tone
• When tone sounds Ss only have to report that row
• Vary how quickly after the display the tone goes off
• Multiply by the number of rows to get capacity of SIS
• The results indicate that almost all the letters are available immediately but that the number of letters available decreases rapidly.

Capacity of Short Term Memory

• Digit Span Task: Number of digits that can be recalled in the correct order. (Normal adult can recall 5-9)

Forgetting in Short Term Memory

• Subjects hear three letters (i.e. "C H J")
• Subjects hear a random number (i.e. "506")
• Subject counts backwards from number until they see a light.
• Upon seeing light subjects try to repeat letters
• LKW (435) (5 secs)
• TKR (232) (10 secs)
• HQM (987) (20 secs)
• FJG (238) (30 sec)

• Interference Theory of Forgetting
• Decay Theory of Forgetting
• Probe task

The Code in short term memory

• Ss errors are not random.
• When subjects misreport items in short term memory tasks they tend to make errors which are acoustically related

• This finding led early researchers to conclude that STM only operated with a verbal/acoustic code.
• Although acoustic codes may predominate in STM, other types of information (visual, semantic) can also be stored and manipulated in STM.

Working Memory Model (Baddeley)

• Phonological Loop
• Visuospatial Sketchpad
• Central Executive

Long Term Memory

• Primacy Effect
• Recency Effect
• Rundus & Atkinson (1971)
• Postman (1965)

These lecture notes were written by Jim Lampinen at the University of Arkansas.  Feel free to link to them but please do not copy these notes in whole or in part for use in your own course without my permission.