The Multi-Store Model of Memory

The Multi-Store Model of Memory

The Multi-Store Model of Memory

Atkinson and Shiffrin (1968) suggested that our memory is made up of three different parts. These are the sensory register; short term memory and long-term memory and they called this the multi store model of memory.

Each section of this memory model stores the information. We need to consider how it is coded in each store, the duration and the capacity of the store.

The Sensory Register (SR)


Information in the sensory register is coded according to our senses. The two main stores are iconic, visual information, and echoic, auditory information. We also have the hepatic, gustatory and olfactory store, which is a tactile, taste and smell store respectively.


The capacity of the sensory register is rather large. It is storing all the information of the world we sense around us. With hundreds of millions of cells in each eye storing visual information.


We are constantly processing the information around us but unless we pay it attention it disappears quickly through spontaneous decay. The duration of the sensory register is said to be very short but varies between sensory stores

Short Term Memory (STM)


The STM is said to be coded acoustically, auditory information.


The capacity of the STM is limited to around 5-9 pieces of information. This can be extended by chunking information such as grouping letters as words or numbers into sequences.


The STM can hold information for around 18 – 30 seconds before it is lost. With maintenance rehearsal this information can be stored in the LTM

Long Term Memory (LTM)


The LTM is said to be coded semantically, information with meaning.


The capacity of the LTM is said to be unlimited as no human has currently reached the limit of new information as we continue to learn throughout our lives.


Although information can be forgotten by decay and displacements our LTM can last a lifetime and therefore we say that its duration is a lifetime.

Supporting Evidence


Sperling (1960) designed an experiment to investigate the capacity of the SR. Participants were shown a 3 by 4 grid of letters which flashed on for 1/20th second. They either had to try and recall the entire grid or a specific row that was indicated by either a high, medium or low-pitched tone.  When participants and to recall the entire grid, they correctly remembered around 4/5 letters and with a specific row they could recall on average three. This supports that the sensory register likely held all the information it was presented.


Baddeley (1966) showed participants 4 groups of words and asked them to recall them immediately testing the STM. He then asked them to recall 20 minutes later to test their LTM.

The groups of words were as follows:

  1. Acoustically similar (Cat, Cab, Can)
  2. Acoustically dissimilar (Pit, Few, Cow)
  3. Semantically similar (Big, tall, large)
  4. Semantically dissimilar (Good, Huge, Root)

Participants struggled to recall group 1 when asked to recall immediately and group 3 when asked to recall 20 minutes later. Baddeley suggested that this was due to how the specific memory store coded the information and therefore struggled to process the information. So STM is coded acoustically and LTM semantically.


Jacobs (1887) read out a series of numbers or letters to participants who then had to immediately recall. If the recalled successfully they would repeat the experiment increasing the numbers of letters by 1 until they could no longer recall successfully. The average length of numbers recalled was 9.3 and letter 7.3. From his results he concluded that the capacity of the STM was 5 plus or minus 2 pieces of information.

Miller (1956) took Jacobs results further and suggested that information could be grouped together to increase the capacity. So, he suggests that the capacity of the STM is 5 plus or minus 2 “chunks” of information.

Peterson and Peterson

In 1959 this married couple designed an experiment to investigate the duration of our STM. Participants were asked to recall a nonsense trigram. This is a consonant only 3 letter sequence. This was to avoid any chance a word could be used that would help participants remember them.

They were also given a 3-digit number and asked to count down in threes from that number aloud. After 3 seconds they were asked to recall the trigram. This would be repeated up to multiple times in increasing 3 second intervals.

Participants correctly recalled the trigram after 3 seconds 80% of the time and this fell dramatically after 18 seconds. They concluded from their experiment that the duration of the STM was up to 30 seconds.

Bahrick et al

Bahrick et al (1975) asked a group of participants to free-recall the list of names of their ex-classmates and a name or photo recognition test. Within 15 years of leaving school 90% correctly identified names and/or faces and 60% success rate for free recall. This fell to 30% for free recall after and after 48 years an 80% success rate for recalling names and 40% for photos.

This shows that the levels of recall are still high, regardless of free recall or recalling names/faces, after a significant time has passed. Therefore, this supports the suggestion that the duration of the LTM lasts a lifetime when ignoring any diseases that affect the mind.


There is plenty of research that supports the MSM but there are several limitations. Although most of the experiments in memory are highly control laboratory experiments this questions there ecological and external validity. We must question the artificial situations and whether we can successfully generalise the results to real life situations.

Remembering random letters and digits is not true to life. We must consider if results would be different if participants were asked to recall examples such as mobile numbers or shopping lists. Jacobs’s experiment was also conducted in the 1800s, so we must question how robust their scientific method was.

The MSM is also a limited explanation. When studying diseases relating to memory there is not just a single short term and long-term memory store and they may be more complicated than this theory suggested, it is suggested there are three different types of long-term memory and the working memory model elaborates on our short-term memory,