Memory (Edexcel A-Level Psychology): Revision Notes
The Multi-Store Model of Memory
Overview of the model
The Multi-Store Model (MSM) was proposed by Richard Atkinson and Richard Shiffrin in 1968 as a theoretical framework for understanding how human memory operates. The model distinguishes between the permanent structural features of memory (similar to computer hardware) and the control processes (similar to software programs). Control processes refer to the conscious decisions we make about what to attend to, how we encode information, and how we rehearse and retrieve memories.
The computer analogy helps us understand that memory has both fixed structural components (like hardware) and flexible processes (like software) that we can control and modify based on our needs.
The MSM describes memory as consisting of three separate stores: the sensory register, short-term memory, and long-term memory. Information flows through these stores in a linear sequence, with each store having distinct characteristics in terms of duration, capacity, encoding format, and retrieval mechanisms.
The structure of human memory
Sensory register
The sensory register is the first component of the memory system. When we experience the world through our five senses (seeing, hearing, touching, tasting, smelling), this sensory information initially enters the sensory register, where it is held very briefly before either decaying or being transferred to short-term memory.
Duration: The sensory register can only hold information for an extremely limited time—approximately 50 milliseconds to a few hundred milliseconds. If information is not attended to within this brief window, it is lost.
Capacity: The sensory register can hold approximately 3-4 items of information at any given moment.
Encoding: Information in the sensory register is encoded in its raw sensory form. For example, visual information is registered in its visual format, auditory information in its auditory format, and so forth. There is a separate sensory register for each sensory modality (visual, auditory, haptic, olfactory, gustatory).
Retrieval: Information is retrieved from the sensory register through a rapid scanning process.
The sensory register holds information for less than a second. If you don't pay attention to sensory information immediately, it will be permanently lost. This is why attention is crucial for memory formation.
Research on sensory memory
The visual sensory register has been extensively researched using the whole or partial report technique. In this experimental procedure, participants are briefly shown a visual array (an arrangement of letters or digits) using a tachistoscope (a device that presents visual information for precisely controlled durations).
Sperling's Classic Research (1960, 1963)
Sperling conducted pioneering research using this technique. Participants were shown a visual array of letters for a very brief moment, and then either asked to recall the entire array (whole report) or given a direction to recall just a specific row (partial report).
On average, participants recalled 4.32 letters from the whole array. When recall was tested immediately after presentation, performance was reasonably accurate, but this decayed rapidly if there was any delay before the recall instruction. This demonstrates that the sensory register can hold only a limited amount of information for a very brief period.
Short-term store
The short-term store (also called short-term memory or working memory) is the second structural feature of Atkinson and Shiffrin's model. Information that is given attention from the sensory register is transferred into the short-term store, where it is held temporarily.
Duration of short-term memory
The short-term store can maintain information for approximately 15-30 seconds before it decays completely, unless rehearsal (the process of consciously repeating information) is used to maintain it for a longer period.
Peterson and Peterson's Study (1959)
Peterson and Peterson investigated the duration of short-term memory using an experimental design that prevented rehearsal. Participants were asked to remember a single trigram (a set of three letters or digits, such as 679, 676, or 673) for intervals of 3, 6, 9, 12, 15, and 18 seconds. To prevent rehearsal, participants were required to perform an interference task—counting backwards in threes from a given number.
The findings showed that correct recall of the trigram was likely after a short interval, but performance dropped rapidly after 15-18 seconds. With the interference task preventing rehearsal of the trigram, it can be concluded that decay occurs in short-term memory over a period of approximately 15 seconds.
Capacity of short-term memory
Atkinson and Shiffrin proposed that the capacity of the short-term store was around five to eight items of information. This estimate was refined by Miller (1956), who described the capacity as "the magical number seven, plus or minus two" (7±2). This means the short-term store can be viewed as having between five and nine slots in which information can be stored.
The capacity of short-term memory is often measured using digit span experiments. Digit span refers to how many digits can be retained and recalled in sequential order without mistakes. Through rehearsal, we can maintain between five and nine items in short-term memory. As more information is input into the store, older information or information with a weaker memory trace is displaced and quickly decays.
The capacity limit of short-term memory (5-9 items) is fixed and cannot be increased. However, we can use strategies like "chunking" to group information together, effectively increasing the amount of information stored within each slot.
Encoding of short-term memory
The format in which information is held in the short-term store does not depend on its original input form. For example, we may see the image of a pineapple in its visual form, but it is held in the short-term memory in verbal form as the word "pineapple".
Atkinson and Shiffrin believed that a memory trace in the short-term store was held in an auditory or verbal form because of the phonological similarity effect. This phenomenon occurs when letters and words with similar sounds are more difficult to recall than dissimilar sounding letters and words. The similarity of sounds leads to confusion in short-term memory, suggesting that encoding in this store is primarily acoustic (auditory or verbal).
The phonological similarity effect explains why phone numbers with similar-sounding digits (like 2, 3, 6, 8) are harder to remember than those with distinct-sounding digits. This acoustic confusion in short-term memory is evidence that information is stored in an auditory format, regardless of how it was originally presented.
Retrieval from the short-term store
Retrieval of information from short-term memory is largely based on a rapid sequential scan of the stored information. Rehearsal plays an important role in maintaining information in the short-term store, increasing the strength of the memory trace and ultimately facilitating its transfer to long-term memory. Digit span experiments indicate that through rehearsal, we can maintain between five and nine items. As more information enters the store, older information or information with a weaker memory trace gets knocked out (displaced) and rapidly decays.
Long-term store
The long-term store is the third and final component of the Multi-Store Model. Information can be transferred from short-term memory to long-term storage, particularly through the process of rehearsal, although other encoding strategies can also facilitate this transfer.
Retrieval from the long-term store
Atkinson and Shiffrin believed that long-term memories existed for all sensory modalities—we have memories for taste, sound, smells, and so forth. In the 1968 model, they proposed that multiple copies of a memory were retained in long-term storage. This proposition was largely based on the "tip-of-the-tongue" phenomenon (Brown and McNeill, 1966), which demonstrated that people were able to accurately predict whether they could recognise the correct answer even if they could not immediately recall it at that moment in time.
The individual may feel that a correct answer is on the "tip of their tongue" and may even be able to recall some features of the correct answer, such as the initial letter or its number of syllables. Atkinson and Shiffrin suggested that these results indicate that a long-term memory is not stored as one complete memory trace, but rather that multiple copies, in various forms and fragments, are stored. When we experience the "tip of the tongue" phenomenon, we are retrieving a partial copy of the memory trace, and this partial retrieval can help us gain access to a more complete copy of the long-term memory through some associative process.
Encoding in the long-term store
Encoding information into long-term storage can depend on the rehearsal process or some form of association between the new information and pre-existing knowledge stored in long-term memory. If information is linked to pre-existing knowledge, it makes the search for that information far easier, as a random search of such a large store would be exhaustive.
Example: Association in Long-Term Memory
Atkinson and Shiffrin (1965) explained this through an example of quizzing a graduate student about the capital cities of US states. The student could not immediately recall that the capital of Washington state was Olympia, but when he later recalled that the capital of Oregon was Salem, he immediately remembered that Olympia was the capital of Washington.
When asked how he remembered this, it was found that he had learned the capitals together—they were recalled as an associated pair that was semantically (by meaning) or temporally (by time) related.
Duration of long-term memory
The duration of long-term memory is potentially a lifetime. Bahrick et al. (1975) investigated what they referred to as Very Long-Term Memory (VLTM) using memory tests on the names and faces of students in their high school yearbooks. Four hundred participants between the ages of 17 and 74 years old were tested.
The findings revealed that identification of names and faces was 90 per cent accurate within 15 years of leaving school and remained between 70 to 80 per cent accurate 48 years after leaving school. This demonstrates that although memory deteriorates over time, long-term memory for faces and names is reasonably resilient over the passage of time.
Capacity of the long-term store
The capacity of long-term memory is potentially infinite. Brady et al. (2008) demonstrated the vast capacity of long-term memory by showing participants 2,500 objects over the course of 5.5 hours. Participants were then shown pairs of objects and asked to identify which of the two they had seen.
When participants saw the original object paired with a very different object, identification was 92 per cent accurate. When the object paired with the original was similar, the identification rate was 88 per cent, and when the original object was depicted from a different angle, identification was 87 per cent. This demonstrates that thousands of images can be successfully maintained in long-term storage.
Transfer of information between stores
Atkinson and Shiffrin described the relationship between the sensory register and long-term memory as being important for the identification of sensory information. In order to transfer information received by our sensory register to short-term memory, we must use our long-term memory to make sense of the information and assign it a verbal label.
Example: The Role of Long-Term Memory in Identification
We may register the image of a horse using our visual sensory system, but this cannot be stored as an auditory-verbal short-term memory until we have identified it as such using our long-term memory of what the object represents.
Transfer of information from short-term to long-term storage can occur as a result of rehearsal, although this would leave a relatively weak memory trace.
According to Atkinson and Shiffrin, a more durable memory trace can be achieved by using a mental operation, such as a mnemonic (a system for remembering something, such as an association or a pattern of letters), to increase the strength of the transfer.
While rehearsal can help transfer information to long-term memory, it creates a relatively weak memory trace. Using deeper processing strategies like mnemonics, meaningful associations, or elaborative encoding creates much stronger and more durable long-term memories.
Evaluation
Summary table of the memory stores
| Feature | Sensory register | Short-term store | Long-term store |
|---|---|---|---|
| Encoding | One register for each sensory modality: visual, auditory, haptic, olfactory, gustatory | Largely acoustic and verbal | Semantic and temporal |
| Storage duration | Limited – approximately 50 milliseconds to a few seconds | 15-30 seconds | Potentially a lifetime |
| Storage capacity | 3-4 items | 5-9 items | Potentially limitless |
| Forgetting | Decay | Decay through displacement | Decay and interference |
| Retrieval | Scanning | Sequential search | Semantic or temporal search |
This table summarises the key differences between the three memory stores. Notice how each store has distinct characteristics in terms of encoding format, duration, capacity, and the way information is retrieved. These differences provide strong evidence for the existence of separate memory systems.
Studies of patients with brain damage
Evidence supporting the distinction between short-term and long-term memory stores comes from case studies of brain-damaged patients and experimental evidence from memory studies.
Henry Molaison (HM) suffered amnesia following brain surgery for epilepsy, resulting in severe impairment to his long-term memory, but his short-term memory remained largely intact. This case study demonstrates that the short-term and long-term memory stores were differentially affected by the brain damage, perhaps because they are located in different regions of the brain.
Similarly, Clive Wearing, a musician and chorus master, could not recall past events in his life, but he could remember how to play the piano and conduct an orchestra. Following a motorcycle accident that caused memory loss, Kent Cochrane (patient KC) could recall facts but showed severe memory impairment in remembering personal events in his life before the accident.
These cases suggest that long-term memory is not one single unitary store, but that perhaps we have different long-term stores for:
- Procedural memory (of practised skills and abilities)
- Other long-term stores for factual information and autobiographical events
In both the cases of HM and KC, the patient was unable to transfer information from short-term memory to long-term memory. Case studies such as these demonstrate the separation between short-term and long-term memory and support the distinction proposed by the Multi-Store Model.
Evaluation of case study evidence: However, the subjects of case studies are unique, and the nature of the brain injury sustained by each individual is equally unique, so we may not be able to generalise such a distinction based on individual cases alone. Despite this problem, amnesia research offers reasonably convincing evidence for the distinction between short-term and long-term memory.
Serial position effect
Compelling evidence for the existence of separate short-term and long-term memory stores comes from research on the serial position effect or primacy-recency effect. Glanzer and Cunitz (1966) conducted an experiment to investigate whether the position of a word in a list affected recall.
Procedure: Participants were presented with a list of words and asked to recall them. The researchers examined which words were recalled most accurately.
Findings: Participants recalled more words from the beginning (primacy effect) and end (recency effect) of the word list, but recalled fewer words from the middle of the list.
Explanation of the Serial Position Effect
It was thought that words recalled at the beginning of the list had had the chance to be rehearsed, and memory for these words would have been strengthened and transferred to long-term storage.
Words at the beginning of the list are being processed whilst words in the middle of the list are filling up the slots in short-term memory. Words at the end of the list acted to displace the older memory trace for the middle words, leaving only words at the end of the list in the short-term memory slots.
This explains why recall from the middle of the list was poorest—those words had been displaced from short-term memory but had not been sufficiently rehearsed to transfer to long-term memory.
The serial position curve demonstrates clear evidence for the difference between short-term and long-term memory stores, with the primacy effect representing long-term memory and the recency effect representing short-term memory.
Coding differences
The difference in the type of coding used by short-term and long-term memory also indicates that we have two separate memory stores. Baddeley (1966) conducted a laboratory experiment on the sequential recall of ten words in a list that were either acoustically or semantically related.
Findings: Baddeley found that semantically related words were easier to recall from long-term memory compared to acoustically related words. In contrast, 9.6 per cent of similar sequences were recalled from short-term memory as opposed to 82.1 per cent of dissimilar words.
Conclusion: This suggests that encoding in short-term and long-term stores is different, which leads to the assumption that different memory stores exist independently of one another. Additionally, similar sounding letters and words are less well recalled than dissimilar sounding letters and words. This suggests that there is acoustic coding in short-term memory, explaining why similar sounding phonemes are confused.
Alternative explanations and criticisms
Reductionism
The Multi-Store Model has been criticised for being overly simplistic in that it underplays the interconnections between the different memory systems by proposing that memory has distinctly different stores. Artificially breaking up the memory stores makes it easier to study memory experimentally, but the model can be criticised for being reductionist.
Dual task experiments
The multi-store model has been criticised for failing to address the dynamic nature of short-term memory and our performance on dual task experiments. Dual task experiments involve two tasks that either compete with each other for the same cognitive resource (because they are similar tasks, such as two verbal or two visual tasks) or involve different cognitive resources (because they are different tasks, such as one verbal and one visual).
Research shows that we perform poorly when trying to deal with similar tasks but perform well when trying to do two tasks of a different nature. For example, we tend to perform poorly if asked to do two verbal tasks or two visual tasks together, but perform well when given one verbal and one visual task together.
Dual task performance cannot be explained by the short-term store, which assumes that capacity is unaffected by type of task. A different explanation of short-term memory, proposed by Baddeley and Hitch in 1974, can explain dual task performance and is seen as a more dynamic model of short-term memory.
Emphasis on rehearsal
The multi-store model has also been criticised for the emphasis given to rehearsal in the transfer of information from short-term to long-term storage. Although we do use rehearsal as a memory strategy, it is not essential for permanent learning to take place. In fact, we are often able to learn new information without consciously trying to learn it.
Using imagery is one example of a memory strategy that leaves a strong long-term memory trace without the need for rehearsal. Craik and Lockhart (1972) offer an alternative explanation for the transfer of information from short-term to long-term storage. They describe different levels of processing—structural, phonemic, and semantic—suggesting that the greater depth of processing we give information, the more durable the memory trace formed. This Levels of Processing Approach provides an alternative framework for understanding memory transfer.
Strengths of the model
Despite these criticisms, the Multi-Store Model of memory proposed by Atkinson and Shiffrin has been a valuable framework in understanding human memory and heuristics. It has stimulated a substantial body of memory research. Because of the development of this model of memory, better and more precise theories of memory have since been proposed.
Wider issues and debates
The use of psychological knowledge in society
Rehearsal, or the refreshing of information, can reinforce learned information. Butler and Roediger (2007) simulated a classroom environment where participants viewed three different lectures on consecutive days. Following each lecture, some participants were given a short recall test to refresh their memory, and some did not receive these tests.
One month later, the participants were tested for their memory of the lectures. The researchers found evidence to suggest that the short recall tests reinforced knowledge for long-term memory of the lecture content. This evidence can be used in educational practice to help students learn more effectively.
Ethical issues in case study research
Case studies of brain-damaged patients are often anonymised by using the initials of the patient rather than their full name. For example, until his death, Henry Molaison was only known as HM. This helps to protect patients' identities and maintains their right to privacy.
In fact, the identity of Henry Molaison was fiercely defended by the researchers involved in his care. Despite being the most widely cited case study in the history of psychology, his identity was protected for over 55 years. However, in high-profile cases, such as Clive Wearing who was in the public eye, these individuals cannot be anonymised. This can lead to issues of privacy being violated, as research concerning the case is available within the scientific community and public arena.
Ethical considerations in case study research:
- Anonymity helps protect patient privacy and dignity
- High-profile cases may not allow for complete anonymity
- Researchers must balance the advancement of scientific knowledge with the rights and wellbeing of participants
- Informed consent and confidentiality remain crucial ethical principles
Key Points to Remember:
-
The Multi-Store Model proposes three separate memory stores: sensory register, short-term memory, and long-term memory, each with distinct characteristics.
-
Short-term memory has a duration of 15-30 seconds, a capacity of 5-9 items (Miller's 7±2), and uses primarily acoustic encoding.
-
Long-term memory has potentially unlimited capacity and duration, with semantic and temporal encoding.
-
Evidence supporting the model includes:
- Case studies of brain-damaged patients (HM, KC, Clive Wearing)
- The serial position effect (Glanzer & Cunitz)
- Coding differences (Baddeley)
-
The model has been criticised for:
- Being overly simplistic and reductionist
- Failing to explain dual task performance
- Overemphasising the role of rehearsal in memory transfer
- Not accounting for different types of long-term memory stores