It is widely accepted that memory can best be understood in terms of three essential stages (encoding, storage, and retrieval of information) that involve the flow of information through memory system. In general, short-term memory (STM) refers to limited capacity that can store information for short periods of time. On the contrary, long-term memory (LTM) has unlimited capacity that can hold a vast quantity of information which can be stored for long periods of time. However, it has been argued that there are dichotomous systems for processing the three essential stages for STM against LTM (Nee, Berman, Moore & Jonides, 2008). According to Peterson and Peterson (1959), evidence showed that the stores of STM and LTM are separate, and yet Melton (1963) argued that the stores are a unitary system. This essay will first examine the evidence for STM and LTM in separate stores, and then discuss whether they are a unitary system in neuropsychology. Lastly, the similarities and differences of STM and LTM will be addressed.
Peterson and Peterson (1959) used the Brown-Peterson paradigm (Brown, 1958; Peterson & Peterson, 1959) to prove the idea of two-store memory. Participants were first shown a trigram, followed by a given number, then asked to repeat the number, and then to count backwards from it until a recall cue was given, whereupon participants attempted to repeat back the consonants (Baddley, 1997). They found that participants were able to recall 80 percent of trigrams after a 3 second interval. Progressively fewer trigrams were recalled as the time intervals lengthened, and fewer than 10 percent of the trigrams were recalled correctly after 18 seconds. Therefore, it was proved that if rehearsal is prevented, information vanishes rapidly from STM. Broadbent (1958) suggested that decay is the mechanism for forgetting in STM. However, it was illustrated that counting backwards in numbers is completely different from the to-be-remembered letters (McGeoch & McDonald, 1931). Conrad (1958) suggested that forgetting is due to the decay of STM instead of interference. Therefore, Hebb (1949) suggested that the division between STM and LTM is that forgetting is due to decay in STM, but in LTM, forgetting is due to interference.
According to Baddeley (1966), there were some tasks that can show two separate components. The immediate free recall task is one of the examples that show two components. In this task, participants were asked to listen to list of letters, and then recall immediately in any order (Glanzer & Cunitz, 1966). Participants were more likely to remember the last few items easily when they were asked to recall immediately. If participants were asked to perform filler tasks before recalling the items, they were more likely to recall the first few items instead of the later items (Postman & Philips, 1965). Therefore, it was suggested that the later items were held in STM while earlier items were stored in LTM (Murdock (1961). Glanzer (1972) found that later items (STM) had nothing to do with word familiarity, speed of presentation, the age of participants and types of filler tasks, and yet these factors could affect earlier items (LTM).
Moreover, STM and LTM have differential coding for words aa‚¬” phonological code for STM and semantic code for LTM (Baddeley, 1966). Set of words that are phonologically similar (e.g. mat, map) are less likely to be recalled immediately than non-phonologically similar ones (e.g. lay, gray), and also similarity of meaning (e.g. big, humungous) had a slight effect (Baddeley, 1976). Nevertheless, semantic similarity is more likely to take over acoustic similarity when a delay occurs between presentations and recall (Conrad & Hull, 1964; Bransford, Barclay & Franks, 1972; Kintsch & Buschke, 1969). Hence, these findings support that STM and LTM are functionally different due to differential coding.
Furthermore, there are some neuropsychological evidences shown that STM and LTM damaged independently. According to Milner (1966)aa‚¬a„?s case study, amnesic patient HM who had damage to the medial temporal lobe (MTL) had normal digit span, STM intact but had problem in remembering events since operation, which indicated that his LTM was impaired. On the contrary, in Shallice and Warrington (1970)aa‚¬a„?s case study, patient KF had impairment in STM, could only remember 1 or 2 items in digit span immediately, but he had normal LTM, and good paired-associate learning. Based on the double dissociation as neuropsychological evidence, it shows support to two models in division of STM and LTM.
On the other hand, memory was thought to be a unitary system for STM and LTM (Melton, 1963). The interference theory can explain about forgetting that associated with information from STM and LTM, in which indicated they were a single system. The Brown-Peterson paradigm was used to prove that counting backwards could affect the recall of STM instead of the decay of STM (Waugh & Norman, 1965). Moreover, Keppel and Underwood (1962) demonstrated that new information can be interfered by old information. Also, no forgetting was showed in the first consonant trigram of Brown-Peterson paradigm due to no interference. Therefore, interference is the main key for forgetting in STM and LTM.
Moreover, the recalls increases when a random number sequence is repeated every three presentations, in which means long-term learning is emerged in STM tasks, and it also indicates an existence of a unitary system (Melton, 1970). Participants showed an increasing retention interval of recalls in the Brown-Peterson paradigm after repeating an item several times (Melton, 1967). Therefore, Papagno and Vallar (1992) proposed that different memory systems are unnecessary when both of them work together. Further evidence revealed that patients with amnesia showed good memory on recalling of later items (STM) but poor performance on recalling of earlier items (LTM) in immediate and delayed free recall tasks (Baddeley & Warrington, 1970). They found out that patients were actually learning on the repeated sequence, so that they could perform a repeated digit sequence task with a good result as people with normal memory. With the evidence, of long-term learning, the finding is against the separate stores of memory, and the uncertainty of long-term learning existing in amnesic patients. Also, it is hard to evaluate the differences of memory ability among the patients, and yet still unsure to find out what part of brain is damaged.
With respect to the previous mentioned neuropsychological evidence, it was reported that the activation of MTL is responsible for early-presented items (LTM), and the activation of right inferior parietal is responsible for late-presented items (STM) (Talmi, Grady, Goshen-Gottstein & Moscovitch, 2005). However, Ranganath and Blumenfield (2005) argued that STM can be influenced by damage of the MTL. Patients with MTL lesions showed deficits in STM when the information was novel and the activity of MTL did not change. Reversely, patients with large frontal lobe lesions showed normal performance on span task, which suggests that the frontal cortex is not only responsible to STM (Daa‚¬a„?Esposito & Postle, 1999).
In addition, patients with perisylvian damage as STM deficits might have phonological problem rather than memory problems (Martin, 1993). A patient IR, who was described as an impaired STM patient with LTM intact, showed good semantic LTM but poor phonological LTM in a revisited task (Belleville, Caza & Peretz, 2003). Hence, perisylvian damage does not only form deficits to STM, it could also affect LTM by generating phonological deficits. Moreover, different test materials that used to examine patients could make differences in STM and LTM (Martin & Saffran, 1997). For instance, LTM tasks reply on semantic variables; STM tasks are more likely to depend on phonological variables.
Furthermore, there are no certain criteria used in localising the areas of the brain to memory in different studies (Wager & Smith, 2003). For instance, patient IR is still categorised as classic STM patient, even he had lesions in the temporal lobe, more than just perisylvian damage. Without the exact localisation of which brain area that is responsible to some particular memory types, it may lead to different theories and point of views to support different memory areas. Therefore, having clear and precise criteria would be helpful for a fuller picture of distinction of STM and LTM.
In conclusion, STM and LTM are functionally different but due to differential coding. However, there is still no clear evidence to show whether STM is particularly distinct from LTM, or whether STM and LTM can be a single system. Although some evidences showed that there were difference processes involved in remembering words over short and long delays, STM and LTM are not complete separation of representations. Moreover, standardised criteria are needed to localise memory area in the brain in order to gain a better understanding of the memory.