David Crow
david.crow @ a c m . o r g

Section B - Question 1
Working memory versus Short term Memory

1. Compare the Atkinson & Shiffrin model of short term memory with the Baddeley working memory model. Describe the research considerations that motivate the Baddeley model over the Atkinson & Shiffrin model.

The main consideration for the move from the Short Term Memory (STM) Model to Baddeley’s Model of Working Memory was that in order for information to become part of long term memory (LTM) in the short term memory model this information must be rehearsed in the STM before it could become part of LTM.

Craik and Lockhart (1972, in Anderson, 1995) argued that it was not how long a piece of information was rehearsed in memory as proposed by Atkinson & Shiffrin (1968, in Anderson, 1995) but rather the depth to which it was processed. This theory, held that rehearsal improved memory only if the material was rehearsed in some meaningful way. Experiments have shown that the passive rehearsal of information results in little improvement in memory performance. Glenberg, Smith, & Green (1977, in Anderson, 1995) has subjects study a four-digit number and then rehearse a word for 2, 6, or 18 seconds. The participants thought that the task was to recall the number and the words were used to fill time, when in fact it was the words that the experimenters were interested in. The participants were asked to recall the words they had rehearsed, not the numbers and the participants recalled only 11, 7 and 13 percent of the words for the 2, 6, and 18 second conditions respectively. This experiment provides support for the Baddeley model of working memory in which there is no time requirement for information to become integrated into LTM. Therefore it appears that there is no short-term halfway station on the route to LTM but rather it is the way the information is processed that is critical for setting up the LTM trace.

Baddeley’s proposed model of working memory also addressed the issue of memory span. In the Atkinson & Shiffrin short-term model of memory, it was proposed that there was a limit on the number of items that short-term memory could hold called the memory span. In terms of short term memory this number was thought to be approximately 7 +/- 2 (Miller’s Magic Number) (Broadbent, 1975, in Anderson, 1995; Miller, 1956). Baddeley proposed what limited the length memory is the speed at which we can rehearse information. Baddeley, Thomsom, & Buchanan (1975, in Anderson, 1995) had participants read 5 words and try to repeat them back without looking at the page. They found that with a shorter set of words, such as, Chad, Burma, Greece, Cuba, Malta, that participants could recall 4.17 out of 5 words, with a set of longer words, Czechoslovakia, Somaliland, Nicaragua, Afghanistan, Yugoslavia, participants could only recall 2.80 out of the 5 words correctly. The crucial factor was not how many words they present but how long it took to say each word. Vallar and Baddeley (1982, in Anderson, 1995) measured how many words of various lengths participants could say in a second. The percentage correct of recalled words exactly mirrors the reading rate. This provides further support of it is the time required to rehearse information not the number of items that limits the capacity of the articulatory loop.

Baddeley’s model of working memory contained not just the articulatory loop to work with verbal information but also what he called the Visuo-Spatial Sketch Pad (VSSP) and the Central Executive. The Central Executive is responsible for the control of the slave systems (the articulatory loop and the VSSP). The central executive can put information into any of these slave systems or retrieve information from other systems. The VSSP is used for rehearsing images and information that is visual or spatial in nature. Baddeley speculates that there could be more slave systems. This separation of working memory into separate systems matches much of the neurological and behavioral evidence. The work by Kosslyn et al. (1993, in Anderson, 1995) on the activation of visual areas of the cortex during visual imagery and the work by Roland & Frieberg (1985, in Anderson, 1995) which showed increases of blood flow to different areas of the cortex during imaging for a spatial route task and for a verbal jingle task. Specifically the temporal and occipital regions were more active during the spatial task, these regions are associated with spatial and visual processing of the cortex and the parietal area associated with language was more active during the verbal jingle task. This work and the work of Goldman-Rakic’s (1988, 1992, in Anderson, 1995) work on the role of the frontal cortex in working memory suggest that the frontal cortex is involved in working memory. Then depending on the type of information and the task other parts of the cortex become involved specifically the temporal, occipital regions for visual spatial information and the language areas of the left parietal cortex for verbal information. These results suggest that there is not just one short term memory system that is limited in capacity but that the memory system is more like the working memory model proposed by Baddeley with the frontal cortex being the central executive and other systems being used as the articulatory loop and the VSSP, in particular the language centers of the brain for the articulatory loop and the visual cortex being used as part of the VSSP.

Anderson, J. R. (1995). Cognitive Psychology and its Implications (4th ed.). W.H. Freeman and Company, New York.
Miller, G. A. (1956). The Magical Number Seven Plus or Minus Two: Some limitations on our capacity for processing information. Psychological Review, 63(2), pp.81-96.