The echoic store however can hold sounds for up to 4 seconds, which therefore enables a sentence to be held in this store.
The multistore model of memory can be drawn and represented in quite a few ways, here are a couple of them: -
Information flows from the sensory register/store to STM and then onto a permanent LTM. The transfer of information from STM to LTM is controlled by top down processes such as attention (visual, auditory, tactile).
In the sensory store, information lasts for a very short period of time (1-2 seconds) and if not selected then the information decays very rapidly.
STM has a limited capacity of about 7 items (Miller 1956 Magic number 7/ + or – 2) and unrehearsed items are forgotten, whereas LTM has unlimited capacity and contains very diverse information.
Atkinson & Shiffrin’s multistore model sees STM as a crucial part of the memory system and that without it, information cannot get into or out of LTM.
In terms of the multistore model, rehearsal has a very important part to play. It increases the time that information can be held in STM and it is also vital for transferring information to LTM. Atkinson & Shiffrin suggested that the longer information was held in STM the more likely it was to be transferred to LTM, and there is some experimental evidence to show this, when Rundus (1971) asked participants to rehearse a list of items out loud and found that the more frequently an item was rehearsed the more likely it was to be recalled.
But then in the early seventies, Craik & Lockhart (1972) proposed an alternative to the multistore model. They suggested that the strength of a memory trace can be determined by the level of processing for that trace. This theory can be best described by comparing the levels of processing in a pyramid: -
The bottom levels represent preliminary, shallow processing and are concerned with physical and sensory features, like lines, angles, brightness, pitch etc. The top levels of the pyramid represent deep processing and are concerned with the extraction of meaning.
The information may undergo elaboration at any level e.g. the rehearsal of a phone number involves elaboration of a string of numbers at a very shallow level of processing, so although the information undergoes elaboration, it is not processed deeply and therefore will not be remembered for more than a few minutes.
This is an important concept because it illustrates the fact that simple rehearsal will not facilitate long-term recall.
The levels of processing framework were presented as an alternative to theories of memory that postulated separate stages for sensory, working and long-term memory. According to the levels of processing framework, stimulus information is processed at multiple levels simultaneously depending on its characteristics.
Craik & Lockhart (1972) were disillusioned with the metaphor of memory as a container – if memory was a container or a set of containers then one might expect to be able to reliably measure the capacity of the containers, and measure the rate at which information stored in these containers was lost if rehearsal was prevented. By contrast Craik & Lockhart proposed that memory performance of a given stimulus was directly predicted by the level of processing to which that stimulus was analysed.
However in 1974 Alan Baddeley (along with a little help from Graham Hitch) proposed the ‘working memory’ model.
This model concerned STM only, and suggested it has 3 components; the visuo-spatial sketchpad, the phonological loop and the central executive. Baddeley said STM was multicomponent.
The visuo-spatial sketchpad encodes visual and spatial information and also it analyses and manipulates information. Visual information refers to what things look like, spatial information refers to the layout of things, like one object being on top of another for example.
The phonological loop rehearses information and consists of two parts; the phonological store and the articulatory control process. The capacity of the phonological loop was demonstrated by the following experiment (Baddeley et al., 1975). Participants were given lists of short words and long words to recall. They recalled more of the short words than the long words. The researchers found that the phonological loop can hold the number of items that can be said in about 2 seconds therefore since short words can be said in a shorter time than long words, more short words were recalled, This is known as the word-length effect.
The central executive controls all of working memory and it decides which particular store is needed and coordinates the retrieval of information from LTM.
There is evidence for working memory, for example the dual task hypothesis (Baddeley 1986). This study challenges the traditional view of STM and is evidence for multistore STM. He presented the participants with two tasks at one time (one digital and one reasoning). This made no difference and showed STM wasn’t full and that there must be separate stores.
Other evidence for the existence of working memory is that K.Fs visual store was unaffected, suggesting separate stores and the fact that the phonological loop only rehearses things is another criticism of Atkinson & Shiffrin (1968).
Short-term memory and long term memory have been seen as separate stores for over 100 years now, and although at the time Atkinson & Shiffrin’s multistore model of memory was seen as a major breakthrough in cognitive psychology, everyone now agrees with Baddeley.
0ther criticisms of the multistore model are that it is now seen as passive and simplistic whereas Baddeley’s working memory model is more complex.
Shallice & Warrington (1974) looked at K.F whose visual/acoustic encoding was having problems, but his semantic encoding was untouched.
Eysenck (1995) criticised ‘rehearsal’. He said that they were saying only things that are rehearsed could go into LTM. He said this isn’t the case. We are taking in stimuli from everywhere into the STM, not necessarily rehearsed, unless see life as one big rehearsal. He said very rarely in everyday life do we rehearse everything.
Baddeley and Hitch also ignored the influence of LTM on STM (Van de Goot 1966). Also the multistore model is seen as limited in explanatory power when compared with other models such as the working memory model.
Over the years the views on the models of memory and the actual models of memory themselves have gradually and sometimes greatly changed. Who knows what the future holds for the study and analysis of memory? Another model could be introduced or one could be ruled out completely, we will just have to wait and see.