This article will discuss research and evidence in support of a unitary memory store and in support of a dichotomous approach. A unitary store implies that a single system for short-term and long-term memory exists and that they occur along a continuum. They are able to interact and occur within each other, for example, long-term learning can occur within short-term memory tasks. A dichotomous approach views short-term and long-term memory stores as being separate components. They are independent of each other and, although they are able to share information, for example, short-term memories are transferred to long-term memory, they do not occur across a continuum. Studies by Arthur Melton (1963), Donald Hebb (1961, cited in Melton, A. 1963) and Ovid Tzeng (1973), described in this article, provide support in favour of the unitary memory store. Garden Sheds Warrington Experimental evidence from Glanzer and Cunitz (1966, cited in Baddeley, 1997) and Glanzer (1972, cited in Baddeley, 1997) support the approach to separate stores of memory. Studies by Milner (1966, cited in Baddeley, 1997) and Shallice and Warrington (1970) of patients with amnesia due to brain damage have provided data that agrees with the suggestion that separate stores exist.
Short-term memory (STM) is a store that holds a limited amount of information for a limited amount of time, usually a few seconds. The short-term memory can store information that has recently been provided, information that has been retrieved from long-term memory or information that has been recently processed. Long-term memory (LTM) stores information that has meaning and can hold it for any amount of time, from 30 seconds to decades. Rehearsal can transfer information from the short-term to long-term memory store, as long as rehearsal occurs before the information has been forgotten.
Many interference theorists, such as Arthur Melton (1963) claim that STM and LTM are part of a single continuum, or unitary store. Interference is the process of a memory trace being disrupted by another and therefore forgetting or, maybe just alterations, of the disrupted memory trace occurs (Baddeley, A. 1997). Melton (1963) used the Peterson task devised by Peterson and Peterson (1859, cited in Melton, A. 1963) to show that when an object is presented a number of times, i.e. rehearsal is possible, the level of retention is increased. Donald Hebb (1961, cited in Melton, A. 1963) had devised a demonstration that Melton incorporated into his own study to show evidence of long-term learning in STM. In Hebb's study, participants were given number sequences, just above the short-term memory span, and asked to immediately recall them. Every third sequence was a repeated sequence, unknown to the participant.
It was found that the level of recall of this sequence increased with the number of trials, showing long-term learning. Melton (1963) used 80 tests, during which the repeated 9-digit sequence would be intervened by 3, 4, 5 or 8 sequences. His findings were that as repetition increased so did the mean number of digits correct in recall. These were used to support Hebb's findings and to add support to the theory of a continuum of memory stores. Because repetition decreases as the number of intervening numbers increases, retroactive interference is increased in the intervening gap. Retroactive interference occurs when a memory is disrupted due to learning more information during a retention period. Up until this period of time, interference theory had been used to explain forgetting in LTM. Melton argued that the ability to use interference theory to explain reduced retention in STM was evidence that LTM and STM should be focussed on as a unitary, continuous store. However, if more than one underlying system of memory is identified with particular tasks, the stores are not necessarily unitary. If the first 10 letters of the alphabet are recalled correctly, as would be expected, short-term capacity has not suddenly increased as recall would be due to previous long-term knowledge of the alphabet.
Ovid Tzeng (1973) studied the effect of recency in delayed free recall. Four lists of 10 words were used to test free recall. In the 1st condition, the 'initial recall' group, the list was given, the participant counted backwards from 20 then wrote the words from the list in any order. In condition 2, the 'final recall' group, the participants counted back from 20 after each word was given then recalled the words at the end of the list. The recency effect was found in both groups. When the serial position of the words was compared with the percentage recall score, the curves for both groups gave almost identical trends. Initial recall should have had a higher recency effect as words should still be available in STM and not transferred to LTM. Counting backwards in the final recall group should have transferred the words to LTM and produced little recency. These curves could be used to conclude that 'the recency effects...could be attributed to the same long-term processes' (Tzeng, 1973). Hence, STM and LTM can be viewed as a unitary store.
It has since been argued that underlying systems produce the difference in results in different performance tasks. Waugh and Norman (1965, cited in Baddeley, A. 1997) used the term primary memory and secondary memory to refer to short-term and long-term memory systems, respectively. Primary and secondary memories are different to STM and LTM because they refer to the storage of information, rather than the stores themselves that hold the information.
Melton's study has provided important evidence into a unitary memory store, but many studies since have provided evidence for separate memory stores. Glanzer and Cunitz (1966, cited in Baddeley, 1997) showed, using free recall, that items from the beginning and end of a list are recalled better than those in the middle. This is called the primacy-recency effect and can be simply explained by the first words being transferred to the LTM and easily retrieved from there during recall. The end words are still available from the STM and so are recalled easily. When the list is followed by a brief filled delay, the recency effect cannot be seen. This is because the filled delay has resulted in words in short-term storage being unable to be rehearsed. They cannot be transferred to LTM and so they decay. Glanzer (1972, cited in Baddeley, 1997) showed that recency effect is unaffected by many variables including familiarity and presentation rate of the word, the age of the participant or the ability to perform other tasks at the same time. These variables have instead been shown to affect primacy effect.
Some of the strongest evidence in support of separate memory stores comes from amnesia patients. Milner (1966, cited in Baddeley, 1997) studied a patient called H.M who had suffered brain damage after surgery to treat epilepsy. H.M could remember events from early on in his life, but he had severe difficulties with recent memories and new information. He was able to remember events and experience from early life, such as how to mow a lawn, but could not learn ongoing experience or remember recent events, such as where he left the lawnmower. Although he was severely impaired in learning new information, his short-term memory span was intact. This suggests a combination of a defective secondary store and a normal primary store. If memory was unitary, both stores would be defective and there would not be a difference between the LTM memories of early life and ongoing experience. It has been suggested that a seemingly normal STM and defective LTM may be a result of STM tests being easier than LTM tests. Therefore, the STM would be less disrupted than LTM.
Shallice and Warrington (1970) studied a patient, K.F., suffering with lesions on his brain. The Peterson task, free recall and a proactive interference test were used to assess short-term capacity and found this was greatly reduced. The free recall showed primacy effect but no recency effect. Probe recognition and missing scan found that retrieval failure was not the cause of this. K.F.'s performance on LTM related tasks showed normal LTM. The suggestion that ease of STM and LTM tasks affects the results in amnesic patients cannot account for these results, as the STM tasks were harder for K.F. than H.M. A double-dissociation is presented between these results and H.M.'s results. Contrasting discrepancies in STM and LTM on performance tasks lend strong evidence in support of two separate memory systems.
During the 1960's and 1970's much research was conducted to decide whether memory exists along a continuum or as two separate stores. Although Melton and Tzeng provided evidence in support for a unitary system that was widely accepted by interference theorists at the time, it has since been assumed that there are two separate stores. Amnesic patients have provided outstanding supportive evidence that a duplex exists and, due to a larger quantity and quality of support for this theory, the idea of a duplex is now largely accepted as the correct approach to memory stores.