The multistore model of memory was proposed by Atkinson and Shiffrin (1968) They proposed that memory consisted of three stores: a sensory register, short-term memory and long-term memory. Information enters into the sensory register, and if attended to will become a short-term memory, then, through rehearsal, it will become a long-term memory. This long-term memory can be retrieved back into a short-term memory. If rehearsal does not occur, the information in the short-term memory will be forgotten. The three stores are different in terms of Coding, Duration and Capacity. Short-term memory has a capacity of 7 +/- 2 items, a Duration of around 18 seconds, and its Coding is mainly auditory. Long-term memory is essentially unlimited in terms of Duration and Capacity, with its coding being mainly semantic. The sensory register has a Duration of around half a second and a wide capacity as it takes in any experience.
Duration of long-term memory was researched by Bahrick et al (1975). The participants were about 400 graduates from an American high school who were shown pictures from their yearbook. Participants were divided into 2 groups: in one they selected the name from a list (recognition), in the other, they had to recall the name by themselves. The recognition group were 90% accurate 14 years after graduation, dropping to 60% accuracy 47 years after graduation. The recall group were 60% accurate 7 years after graduation and less than 20% after 47 years. Therefore he concluded that the Duration of Long-term memory was essentially unlimited due to it lasting a whole lifetime.The duration of short-term memory was researched by Peterson & Peterson (1959). They presented participants with trigrams and asked them to recall them after intervals of 0, 3, 6, 9, 12, 15, and 18 seconds. they asked them to count out loud backwards in threes from a certain number to stop rehearsal. They found that the accuracy of recall declined until less than 10% of trigrams were recalled after 18 seconds. So they concluded that the
Duration of short-term memory was up to 18 secondsMiller had participants remember a set of numbers and recall them correctly, increasing the number of digits with each answer. He found that most adults are able to remember 7 numbers, give or take two. So, he concluded that the capacity of short-term memory was 5-9. Miller suggested that we remember in ‘chunks’ with smaller chunks being easier to remember, and larger chunks being harder to remember. There have also been some evidence to suggest that Atkinson and Shiffrin are correct, at least in having memory separated into different stores. Patient KF is an example of this. After the patient had suffered an injury, and could no longer store Short-term memories, yet his Long-term memories were fine, this indicated that there were separate stores for memory, like Atkinson and Shiffrin had suggested.
Clive Wearing is also another person who supports this model, as he is unable to form long-term memories, yet he has short-term memory and can remember up to thirty seconds of things. This again suggests that short-term memory and long-term memory have different stores since Clive Wearing Long-term memory has been damaged but his Short term memory has notHowever, the model does not explain why Patient KF is able to create short-term memories, as the model would indicate that his long-term memory is also impaired. This is due to the model suggesting that for Long-term memories to be created there needs to be a rehearsal of short-term memories. Due to Patient KF not being able to do said rehearsal, he should not be able to create long-term memories, yet, since he can, it suggests that the model is flawed. It also fails to include the fact that there is more than one type of Long-term memory, as the Working memory model by Baddeley and Hitch showed that instead of having one solid unit for long-term memory, there are multiple different components such as the central executive and the visuospatial part. The model is flawed as it does not show the complexity of the long-term model, and assumes it is relatively simple, which it is not.