Due to both the ambiguity of the subject and the difficulty of researching it, memory has provided a great deal of controversy in the scientific world. This has been due to issues such as the type of encoding the mind employs in memory tasks, the time limits of storage and other such issues. It was only as recently as the late 1950’s however that the question was raised over whether there could exist a separate long term and short term memory system and if so what the various functions of the two are. Peterson and Peterson (1959)1 were the first to come up with this Duplex Hypothesis as the result of an experiment into human memory.
In this experiment they gave their participants a three consonant trigram (such as DNP) which was spoken to them to the beat of a metronome. Immediately after this the participant was presented with a three-digit number (in the same way) which they had to count back in threes from, again in time with the metronome. When a set signal was given the participant ceased counting and attempted to recall the original consonant trigram. In this experiment the counting was used as a distracter task which would prevent rehearsal of the original trigram by the participant (numbers were used instead of letters to avoid retroactive interference).
Various different retention intervals were tested, ranging from three to eighteen seconds in length. The results showed that if the recall was only three seconds after presentation, the participant accurately recalled the trigram 80% of the time. At nine seconds this accuracy dropped dramatically to 25 %, and at 18 seconds it was only 10%. Peterson and Peterson said that this showed that memory data is lost very quickly due to the passage of time alone (rather than associative interference) without rehearsal or the opportunity for semantic encoding.
Moreover they said that mere rehearsal only postpones the onset of data decay. When the results of these tests were compared to the results of more conventional memory tests the conclusion many psychologists came up with was that there had to be two different types of memory governed by different principles. These two types were labelled ‘short term memory’ and ‘long term memory’. Soon after this difference was discovered structural theories followed, theorising the links and differences between long term and short term memory. The most popular of these is that of Atkinson and Shiffrin (1968)2.
Their theory was based around the concept of information transferral from one form to another in strict order; information starts in the ‘sensory memory’ (a form of immediate storage) and instantaneously moves to short term memory and from there is encoded into the (relatively permanent) long term memory. Moreover they made some important distinctions between long term memory and short term memory. Short term memory in their opinion is distinguishable by its very limited capacity, its acoustically based encoding system and its very short retention span.
Long term memory has a more or less unlimited capacity, a semantic encoding system and a very considerable retention span. These functional distinctions will now be examined more closely, but first it is important to understand the basic stages of memory; Encoding, storage and retrieval3. Encoding is the initial stage in which information is taken form its initial form (speech, writing, vision etc) and transformed (or encoded) into one of a number of forms which memory can accept. Storage is simply the stage in which the information is retained in either the short or long term memory store until needed.
Retrieval is the stage in which the information is recalled to be used (such as in speech). At each of these stages there is a recognised difference in the performance of long term memory and short term memory. Craik and Lockhart (1972)4 claimed that there was no difference between long term memory and short term memory except in their forms of encoding. Short term memory is shown to use a form of acoustic encoding (encoded by the sounds of the information), where the information can be rehearsed to be kept active. This is particularly seen with information containing verbal items such as digits, letters and words5.
Conrad (1964)6 performed an experiment by giving participants a list of six consonants (eg; RLBKSJ), taking them away, and within a few seconds asking them to write the list in order. Participants only made occasional mistakes and it was observed that these mistakes were still acoustically similar i. e. a T instead of a B. This (and further tests comparing the accuracy of participants when the lists were made up of acoustically similar consonants, i. e. TBCGVE) showed that participants were encoding the information phonologically into short term memory.
In comparison, long term memory is seen to generally encode information semantically. This means the information is encoded based on its meaning, so that a sentence can be remembered for content but the exact wording is seldom retained (Sachs, 1967)7. Kintsch and Buschke (1969)8 showed participants in a test 16 typed words over 32 seconds followed by a ‘probe word’, which was a repetition of one of the earlier words. The participants then had to say which word in the list had followed the probe word. Some of the lists consisted of eight pairs of synonyms, while others consisted of acoustically similar words (eg day, pay, say).
Similar meaning was seen to cause confusion when subjects tried to recall words early in the lists, whereas acoustic similarity only produced confusion with the most recent words. Craik and Lockhart (1972) tried to explain this effect by saying that duration of memory was directly linked with depth of encoding, and that acoustic encoding is far less deep than semantic encoding. Craik and Tulving (1975)9 went on to experiment into this by asking participants semantic or acoustic questions about various words, and then giving them an unexpected test on these words.
Performance was proportional to the supposed depth of encoding that the questions caused. The difference in storage between long term memory and short term memory can be seen in two main ways; duration of memory and its capacity. The differences here are marked and as such seem to disprove the earlier theory that the only differences between the two forms of memory lie in encoding. The difference in duration of information is inferred by the two names, short term memory rarely retains information for more than ten seconds.
The short retention period of short term memory is what has led to it sometimes being referred to as working memory. This is because its function is to ‘keep active’ information that is presently being used, such as stages in mental arithmetic. The capacity of short term memory is also incredibly small. Thompson, Muir and Lawrence (1984)10 showed that children of three only have a short term memory capacity of 3 items or ‘chunks’ (groups of items familiar enough to a person to be grouped as one, such as the letters making up a word).
However, by the time we reach our adult level (usually around the age of twelve) humans have an estimated short term memory capacity of 7 (+/- 2) items or blocks (Miller, 1956)11. Thompson, Muir and Lawrence (1984) also pointed out that the increases in span with age precisely parallel the increase in the rate which humans learn to speak rapidly. As such the capacity of short term memory may be a fixed time length of information rather than amount of information and is therefore dependent on speed sub-vocal rehearsal.
Anderson (1983)12 pointed out that this limit on capacity has a great effect on the duration of short term memory, as any new information entered will displace an old item. In comparison long term memory retains accurate information over tens of years, and has a relatively limitless capacity. For example, humans do not (under normal conditions) forget how to ride a bike, drive a car or construct a sentence. Bahrick (1984)13 tested 800 people on a Spanish reading-comprehension test who had studied Spanish for three years in college and gained an ‘A’ grade.
The participants ranged from one week since their last Spanish course to fifty years. The test was marked out of 40, and there was a drop from an average score of 33 to 25 through the first 4 years since the last course. After that there was only a drop from 25 to 20 from 4 years to 50 years, thus proving that long term memory retention can be incredibly long. Retrieval can be seen as two completely different concepts between long term memory and short term memory. It can be argued that short term memory does not involve any real act of retrieval at all.
Rather the information remains active for as long as it is stored and so access is almost immediate. Sternberg (1966)14 did, however show that there is a very small retrieval time involved that depends on the amount of information in the short term memory being considered. He showed participants ‘memory lists’ ranging from 1 to 6 digits, he then removed them and presented a probe digit that the participant had to decide was or was not on the previous list. There were very rarely any errors, but the length of time over the decision was seen to increase in direct proportion to the amount of digits on the original list.
Each item added to the memory list adds approximately 40 milliseconds to the retrieval process. Moreover these results remained constant when the lists consisted of letters, words, auditory tones or faces (Sternberg, 1975)15. Long term memory retrieval, as previously mentioned, differs firstly in the respect that the information has been encoded semantically and so can be remembered in meaning for many years after an event but rarely for exact content. Moreover, due to its content and duration it is far more prone to being ‘forgotten’ (at least in part).
Jenkins and Dallenbach (1924)16 stated that this is rarely a matter of unavailability of the information (it is still in long-term store), but merely inaccessibility to the information at the time of recollection. In cases where the information is not present at all, it is not due to temporal decay as with short term memory but rather due to some form of transferral interference at the time of encoding that stopped the information being stored in the first place. Short term memory and long term memory are therefore fundamentally distinct in function.
They are seen to be based on different principles and employ different methods in all of the basic, conventional stages of memory. Both the information itself and the use it is put to remain functionally distinct. Short term memory stores acoustic information to be used almost immediately for active mental processes, long term memory stores information semantically to use later for a multitude of functions from skill employment (writing, cycling) to simply reminiscing on past experience. The only area in which the two interact is in the transferral of information from one to the other.