The word "memory" can be used in different ways. Reber (1985) identifies:
It can be helpful to think of three processes which are involved in memory, namely
Lots of research has been performed to investigate the idea that memory is split into Short Term Memory (STM) and Long Term Memory (LTM). This theory suggests that information is first stored in STM and then transferred via rehearsal (a process of repeating the information over and over) to LTM, and that information which does not get transferred will not be remembered.
Research has been done to understand some of the characteristics of STM:
Peterson & Peterson (1959) (see study sheet) used tests where P's were read a series of random letter sequences and then asked to remember them after a short delay (a few seconds). During the delay, the P's were asked to count backwards, to prevent them from rehearsing (repeating) the sequences to themselves and thus transferring them to LTM. It was found that with a very short delay (a couple of seconds), P's could successfully remember sequences, but that as the delay grew longer, the success rate dropped dramatically, with a success rate of around 6% for delays of 18 seconds. So the evidence from this study suggests that the duration of STM is quite short.
Various research suggests that STM has a capacity of seven items. Miller (1956) referred to seven as "the magic number". He showed that people on average could store seven unrelated chunks of information, such as digits in a telephone number or random words, or phrases in a sentence.
The theory of displacement suggests that as the spare "slots" in STM become full, new incoming information must displace an existing "chunk". Waugh and Norman (1965) studied this and showed that information in the STM is continually being replaced
Murdock (1962) analysed which items were recalled, and found that P's tended to remember more items from the beginning (primacy effect) and end (recency effect) of a list, and fewer from the middle.
Glanzer & Cunitz (1966) (see study sheet) suggest that the primacy effect is as a result of the P's having more time to rehearse the first items on the list, and that the recency effect is due to items still being in STM when P's are asked to recall them. They demonstrated that the recency effect could be eliminated by preventing rehearsal, while this did not alter the primacy effect. From this, they argued that in fact the capacity of STM might be as low as three or four chunks.
Conrad (1964) and others have shown that coding in STM is typically done acoustically (i.e. you're remembering the "sound" of the items). For example, when asked to remember letters that look similar but sound different (e.g. "D","O"), P's do better than when asked to remember letters that sound the similar but look different (e.g. "B","T").
Anecdotally, we know that some memories last longer than 18 seconds, which suggests that there must be a form of LTM. But it is a lot harder to investigate LTM, since studies might need to be performed over periods of years rather than minutes. Most, if not all, studies of LTM suffer from the problem of being unrepresentative of real life situations.
The research that has been done to understand some of the characteristics of LTM shows:
Bahrick (1975) (see study sheet) tested P's and showed that they were able to accurately recall information that was more than 50 years old. It was shown that the ability to recall was helped if retrieval cues (memory joggers) were provided.
It is generally agreed that there is no evidence of any limit to the capacity of LTM. Tulving and Pearlstone (1966) suggest that "forgetting" is due to a retrieval failure rather than a retention failure; information which is "forgotten" may still be in LTM, but just not retrievable; we have subjective evidence of this when long-forgotten memories spontaneously appear in response to a cue (such as a scent or piece of music).
Baddeley (1966) suggests that LTM is coded mainly in a semantic way (based on meaning). Although the results of STM tests were unaffected by whether or not a set of words had meaningful relationships to one another, LTM tests showed significant differences in the results. On the other hand, while STM tests were affected by the acoustic relationship of words, this had no effect on LTM.
To summarise the differences between STM and LTM:
| Type | Duration | Capacity | Coding | Reason for Forgetting |
| STM | 12-18 sec | 3-4 chunks | mainly acoustic | retention failure |
| LTM | unlimited | unlimited | mainly semantic | retrieval failure |
Evidence from case-studies supports the theory of STM/LTM, e.g. Milner (1966) studied HM, a patient who had had his hippocampus removed in an attempt to cure epilepsy, and reported that while HM could remember events prior to the operation, and appeared to have a functioning STM. But he appeared have lost the ability to transfer information from STM to LTM, which meant that he was unable to learn or remember information following the operation.
HM's case is one of anterograde amnesia (forward acting amnesia) in which a patient is unable to learn any new information following some trauma. Such symptoms also occur in patients suffering from Korsakoff's syndrome, which is caused by heavy drinking. This contrasts with retrograde amnesia (backward acting), in which a patient will lose the memory of events before a trauma.
Read handout up to "Models of Memory"
Back to class notes index page