Multi-Store Memory Model: Atkinson and Shiffrin

The multi-store model of memory (MSM), proposed by Atkinson and Shiffrin, describes memory as comprising three distinct stores: sensory memory, short-term memory (STM), and long-term memory (LTM).

Information moves through these stores sequentially in a linear process, via attention, rehearsal, and retrieval.

Sensory Memory

Sensory memory (or sensory register) is like a very quick snapshot that your brain takes of everything happening around you.

It captures sights, sounds, smells, tastes, and touches. It fades in an instant, lasting less than a second.

For example, glance at a car, then close your eyes. You’ll still briefly see the image in your mind for about half a second before it fades away.

Sensory memory helps you experience your environment smoothly and continuously, even though it holds each sensation only momentarily.

Key Features

• Duration: between 0.25 seconds and 2 seconds.
• Capacity: All sensory experience (v. larger capacity).
• Encoding: Sense specific (e.g. different stores for each sense).

How It Works

Think about how you briefly see the trail of a sparkler in the dark. Or notice how someone’s words seem to echo in your ears just after they’ve spoken.

These fleeting impressions happen in your sensory memory.

The sensory stores constantly receive information. Most of it never gets attention, so it stays only briefly before fading.

Atkinson and Shiffrin (1968) explained that sensory memory stores information exactly how it’s received from your senses.

For example, images you see are stored visually as iconic memory, and sounds you hear are stored as echoic memory. Both fade fast.

The sensory memory store has a large capacity but a very brief duration. It can encode information from any sense.

Attention is the first step in remembering something: if it is focused on one of the sensory stores, the data transfers to STM.

Sperling (1960) – Evidence for Sensory Memory

• Aim: To test how much information sensory memory really holds, since a whole-report test might underestimate it if items decay before they can all be reported.

• Method: Participants saw a grid of 12 letters for about 50 milliseconds. In the whole-report condition they tried to recall all 12 letters. In the partial-report condition, a tone played immediately after the display told them which single row to report.

• Results: Whole report produced only about 4-5 letters (roughly 35%). Partial report produced about 75-100% of the cued row, whichever row was cued, implying almost all 12 letters were briefly available.

• Conclusion: Sensory memory holds a large amount of information very briefly. The whole-report limit reflects rapid decay during reporting, not a small capacity, exactly as the sensory register the multi-store model proposes.

Short Term Memory

Short-term memory is like your brain’s temporary notepad, holding information for a brief period – usually about 15 to 30 seconds.

It’s what you use when remembering someone’s name you’ve just heard, a phone number you’re about to dial, or directions someone gives you.

Think of it as your mental workspace. Information stays there briefly unless you actively repeat it or give it meaning.

However, short-term memory has limited space, typically holding around 5 to 9 items at once. If you don’t actively keep thinking about or rehearsing this information, it fades quickly.

• Duration: Information is held in short-term memory for a brief period, typically around 15 to 30 seconds, if it is not actively rehearsed.
• Capacity: 7 +/- 2 items
• Encoding: mainly auditory or phonemic coding (meaning it’s coded based on sound).

Maintenance rehearsal

Maintenance rehearsal is verbally or mentally repeating information. It extends how long information stays in short-term memory, beyond the usual 30 seconds.

For example, repeating a phone number until you dial it is maintenance rehearsal.

This method does not give the information meaning or link it to other memories.

Continual rehearsal “regenerates” or “renews” the trace, making it stronger. But if rehearsal stops, information in short-term memory is lost: new items push it out through displacement, or it simply fades through decay.

Long Term Memory

Long-term memory is like your brain’s personal library, storing everything you’ve learned and experienced throughout your life.

It holds memories from days, months, or even years ago. Think of the name of your childhood friend, how to ride a bike, or details from your favorite movie.

Unlike short-term memory, long-term memory has almost unlimited space. It can store information indefinitely.

However, memories usually need to be meaningful or frequently used to last.

Think of long-term memory as a filing system. The better organized the files, the easier they are to retrieve later.

• Duration: Unlimited (information in the LTM can potentially last a whole lifetime).
• Capacity: Unlimited
• Encoding: Mainly Semantic (based on what it means), but can be visual and auditory.

Elaborative rehearsal

Elaborative rehearsal means giving new information meaning by linking it to what you already know in long-term memory.

Instead of just repeating something, you connect it to existing knowledge or create vivid associations.

For example, to remember someone’s name, you might link it to someone with the same name, or picture it vividly.

This is deeper processing. It helps information stick and transfer more reliably into long-term memory than maintenance rehearsal does.

Key Studies

These experiments test the key aspects (duration, encoding, capacity) that define the multi-store model, providing evidence for its distinct memory stores.

1. Murdock (1962) and Glanzer and Cunitz (1966) – The Serial-Position Effect

• Aim: Murdock (1962) first tested whether a word’s position in a list affects recall. Glanzer and Cunitz (1966) then asked whether primacy and recency come from separate memory stores.

• Method: Participants heard word lists and immediately recalled them, showing primacy and recency effects. In some conditions, participants first counted backwards for 30 seconds, which prevents rehearsal.

• Results: Both primacy and recency effects appeared on immediate recall. After the 30-second delay, the recency effect disappeared, but the primacy effect remained.

• Conclusion: Recency reflects a temporary short-term store emptied by the delay, while primacy reflects words already transferred to a durable long-term store. This supports two separate memory systems.

2. Peterson and Peterson (1959) – Duration of Short-Term Memory

• Aim: To measure how long short-term memory holds information without rehearsal.

• Method: Participants briefly saw trigrams (three letters, e.g. “BRT”), then counted backward by threes to prevent rehearsal. Recall was tested after delays of 3 to 18 seconds.

• Results: After just 18 seconds, recall accuracy dropped dramatically to about 10%.

• Conclusion: Short-term memory has a limited duration of roughly 18–30 seconds without rehearsal, supporting the multi-store model’s idea of STM as temporary storage.

3. Baddeley (1966a, 1966b) – Encoding in Short-Term and Long-Term Memory

• Aim: To investigate how information is encoded in short-term and long-term memory.

• Method: Participants learned word lists that were acoustically similar (e.g., caught, short, taut, nought), semantically similar (e.g., huge, great, big, wide), or dissimilar controls. They recalled the list order either immediately (testing STM) or after a delay filled with an interference task (testing LTM).

• Results: On the immediate (STM) test, acoustically similar words were recalled markedly worse, while semantically similar words showed only a small deficit (about 64% vs 71% correct). On the delayed (LTM) test the pattern reversed: semantically similar words were now recalled worst.

• Conclusion: This supports MSM by showing short-term memory is primarily acoustic, whereas long-term memory encodes mainly based on meaning.

4. Miller (1956) – Capacity of Short-Term Memory

• Aim: To explore how many items short-term memory can hold.

• Method: Reviewed various studies where participants remembered lists of letters, numbers, or words.

• Results: Most participants could accurately remember about 7 items, give or take 2 (known as Miller’s magic number, 7±2).

• Conclusion: Demonstrates short-term memory has limited storage capacity, reinforcing the MSM’s claim that STM can hold only a small amount of information at once.

5. Jacobs (1887) – Digit Span Test

• Aim: To test the capacity of short-term memory for numbers and letters.

• Method: Participants repeated sequences of digits or letters in the correct order; sequences increased in length until mistakes occurred.

• Results: Participants recalled on average 9 digits and about 7 letters.

• Conclusion: Demonstrates the limited capacity of short-term memory, supporting MSM’s idea of STM’s restricted storage.

6. Bahrick et al. (1975) – Duration of Long-Term Memory

• Aim: To investigate how long memories last in real-life settings.

• Method: Participants (aged 17 to 74) identified former classmates from yearbook photos and recalled their names years after leaving school.

• Results: Recognition remained strong (around 90%) even after 15 years, and substantial (70–80%) even after nearly 50 years.

• Conclusion: Indicates long-term memory has a very large capacity and can retain information over decades, supporting MSM’s distinction of LTM as long-lasting.

7. Conrad (1964) – Acoustic Encoding in STM

• Aim: To examine how short-term memory encodes information.

• Method: Participants saw lists of letters briefly and recalled them immediately. Letters sounded either similar (e.g., B, C, D) or different (e.g., F, H, Q).

• Results: Recall mistakes often involved mixing up letters that sounded alike, rather than those that looked alike.

• Conclusion: Short-term memory primarily encodes information acoustically (by sound), reinforcing MSM’s explanation of STM encoding.

8. HM (Henry Molaison) – Amnesia and Memory Dissociation

• Aim: To find out whether short-term and long-term memory are separate stores, using the case of a patient whose hippocampus was surgically damaged.

• Method: Surgeons removed most of HM’s hippocampus and amygdala in 1953 to relieve severe epilepsy. Researchers then studied his memory for decades, using immediate memory-span tasks and new-skill tests such as mirror-drawing.

• Results: HM’s short-term memory stayed normal, but he could never form new long-term memories. He never learned his parents had died, yet could still learn new motor skills like mirror-drawing without remembering practicing them.

• Conclusion: Intact short-term memory alongside an inability to form new long-term memories is exactly the dissociation the multi-store model predicts, showing STM and LTM are separate stores.

9. Shallice and Warrington (1970) – KF and the Reverse Dissociation

• Aim: To test whether short-term and long-term memory are separate stores, using the case of a patient (KF) with a different pattern of brain damage than HM.

• Method: Researchers assessed KF’s short-term store using auditory-verbal digit span and immediate recall of spoken letters and digits, then compared this with his long-term learning ability.

• Results: KF’s short-term memory was severely impaired, with a digit span of only about two items, yet his long-term learning and comprehension stayed essentially intact. The deficit was specific to spoken material; his visual short-term memory was relatively spared.

• Conclusion: KF is the mirror image of HM, completing a double dissociation between STM and LTM. His intact long-term learning despite a near-abolished STM is puzzling for the model. It also shows STM is not one unitary store, motivating the working memory model.

Critical Evaluation

Strengths

1. Clear Structure and Influential

One strength is that MSM provides a clear, easy-to-understand structure of memory.

This clear structure allows researchers to build upon and test its validity.

Its influence has led to extensive research on memory, increasing our understanding of the processes involved.

2. Evidence from Research

Numerous studies support MSM by demonstrating clear distinctions between Short-Term Memory (STM) and Long-Term Memory (LTM). For instance:

• Glanzer & Cunitz demonstrated the primacy and recency effects, showing differences in how STM and LTM function.

• Case study of HM: his intact short-term memory alongside an inability to form new long-term memories (see Key Study 8) is exactly the dissociation the multi-store model predicts.

Weaknesses

1. Oversimplification of Short-Term Memory

The MSM views STM and LTM as singular, uniform stores, but research indicates that memory is more complex and involves multiple components.

Working memory (Baddeley & Hitch, 1974) showed that short term memory is more than just one simple unitary store and comprises different components.

• Central Executive: Directs attention and allocates resources.

• Phonological Loop: Manages verbal and auditory information.

• Visuospatial Sketchpad: Handles visual and spatial data.

Baddeley and Hitch (1974) argue that the picture of short-term memory (STM) provided by the Multi-Store Model is far too simple.

Fig 2. The Working Memory Model Components (Baddeley and Hitch, 1974).

The model proposes that every component of working memory has a limited capacity, and also that the components are relatively independent of each other.

Long-term memory is unlikely to be a single store either. It seems improbable that knowing how to play a computer game, the rules of subtraction, and what you did yesterday are all stored in the same way.

Indeed different types of long-term memory have been identified, namely episodic (memories of events), procedural (knowledge of how to do things) and semantic (general knowledge).

2. Neglects Different Types of LTM

MSM treats LTM as a single, uniform store, but research indicates multiple distinct long-term memory systems:

• Episodic memory: Personal experiences and events (e.g., your birthday party).

• Procedural memory: Skills and actions (e.g., riding a bike).

• Semantic memory: General knowledge and facts (e.g., capital cities).

3. Overemphasis on Rehearsal

Rehearsal is considered a too simple explanation to account for the transfer of information from STM to LTM.

For instance, the multi-store model ignores factors such as motivation, mood and strategy (e.g. mnemonics) which underpin learning.

Also, rehearsal is not essential to transfer information into LTM.

For example, we can recall skills we never rehearsed, such as swimming. Yet we often forget material we did rehearse, such as revision notes.

Rehearsal therefore plays a smaller role in transferring information from STM to LTM than Atkinson and Shiffrin (1968) claimed.

4. Structure-Focused, Ignoring Detailed Processes

MSM emphasizes memory structures but overlooks the cognitive processes involved in encoding and retrieval.

It highlights attention and maintenance rehearsal. But it fails to explain the cognitive depth long-term retention actually requires.

Later models take a different approach. The levels of processing model (Craik & Lockhart, 1972) argues that depth and meaningfulness of processing, not rehearsal itself, determines retention.

Elaborative rehearsal analyzes information deeply: linking it to existing knowledge, building vivid imagery, or forming meaningful associations. This produces more durable memories than simple repetition.

5. One-Directional Flow Is Too Simple

The MSM shows information flowing only one way: from sensory memory to STM, then to LTM.

But chunking and pattern recognition both draw on LTM. Recognising a chunk, or matching new input against a stored pattern, needs information to flow back from LTM too.

The flow is not strictly one-way (Craik & Tulving, 1975).

6. Over-Reliance on Artificial Lab Tasks and Rare Cases

Much of the evidence for the MSM comes from lists of unrelated words or letters recalled in a lab, which has low ecological validity.

The rest often rests on a handful of rare, non-generalisable amnesic patients, such as HM, KF and Clive Wearing, each with idiosyncratic brain damage.

Findings from these unusual cases may not generalise well to everyday memory in the wider population.

Contemporary Research

Since the 1970s, many researchers have moved away from picturing short-term memory as a separate box. Instead, they treat it as the temporarily activated portion of long-term memory, held in the current focus of attention (Cowan, 2017).

Even so, a causal test still finds short-term and long-term memory can be separately targeted.

• Aim: Grover et al. (2022) tested whether working memory and long-term memory depend on separate brain systems that can each be independently boosted, using brain stimulation rather than case studies.

• Method: Adults aged 65-88 received four days of brain stimulation during a memory task. One protocol targeted the parietal cortex; another targeted the prefrontal cortex. Both memory types were tested across the four days and again one month later.

• Results: Parietal stimulation improved working memory, while prefrontal stimulation improved long-term memory. Both improvements were still present a month later.

• Conclusion: Working memory and long-term memory rely on distinct, separately targetable brain systems – a modern echo of the multi-store model’s original STM-LTM split.

Beyond Rehearsal: Activation and Consolidation

This state-based view has direct neural support. Using pattern analysis to decode brain activity, Rose et al. (2016) found that the active neural signal for a temporarily unattended item fades to baseline.

A single pulse can briefly bring it back, though. This “activity-silent” finding suggests the item was retained in rapidly modifiable synaptic weights, not by ongoing firing in a dedicated store.

Retention looks different too. The MSM pictures LTM storage as rehearsal simply “copying” material across from STM.

Contemporary work instead treats retention as active, time-extended consolidation, much of it happening offline during sleep. A large meta-analysis found a reliable sleep benefit for later recall (Berres & Erdfelder, 2021).

Memories keep stabilising well after encoding, independently of further rehearsal.

Applications

The multi-store model does more than describe memory in the abstract. Because it explains why and how information is lost or kept, it also grounds several practical applications, from how students should revise to why eyewitness accounts can go wrong.

Education and Revision

The distinction between maintenance and elaborative rehearsal is directly useful for study technique. Rote repetition alone is weak.

Deep, meaningful processing works far better: self-testing, elaborating on material, connecting new information to things you already know, and using imagery or mnemonics.

A student who just re-reads notes is using maintenance rehearsal. That alone rarely works.

Explaining the material in your own words, or linking it to something you already understand, works far better – that is elaborative rehearsal in action.

Chunking long strings of information into meaningful units, and spacing out revision sessions, both help work within STM’s 7±2 limit and consolidate material into LTM.

Middle-of-session material suffers most from the serial-position effect. So teachers often front-load key points and revisit them later.

Memory Disorders

The STM-versus-LTM framework underpins how amnesia is clinically described.

It distinguishes a general loss of consciousness from a specific failure of one memory process.

Anterograde amnesia, seen in patients like HM and Clive Wearing, is understood as a failure to transfer or consolidate information from STM into LTM, with STM itself left intact.

Clive Wearing’s case is striking. A brain infection in 1985 left him able to hold a memory for only seconds at a time.

Yet he can still read music and conduct a choir, because that procedural skill survived even though his conscious memory for events did not.

For example, a patient with anterograde amnesia can improve at a new motor skill through repeated practice. But they may have no memory of ever practicing it.

This understanding guides rehabilitation strategies, such as exploiting a patient’s intact procedural learning, using errorless learning techniques, and relying on external memory aids.

Eyewitness Memory

STM’s limited capacity and short duration help explain why eyewitness testimony can be fragile.

Retrieval from LTM is also reconstructive and cue-dependent, which adds a further source of distortion.

A witness’s memory of a brief, stressful event passes through the same STM bottleneck. So does any other information.

Any detail they never attended to is simply unavailable later, however confident the witness sounds.

This is not lying. It is how memory works.

The pattern is part of the foundation for research into misleading information and the reliability of witness testimony.

References

Atkinson, R. C., & Shiffrin, R. M. (1968). Chapter: Human memory: A proposed system and its control processes. In Spence, K. W., & Spence, J. T. The psychology of learning and motivation (Volume 2). New York: Academic Press. pp. 89–195.

Baddeley, A. D. (1966a). The influence of acoustic and semantic similarity on long-term memory for word sequences. Quarterly Journal of Experimental Psychology, 18(4), 302–309. https://doi.org/10.1080/14640746608400047

Baddeley, A. D. (1966b). Short-term memory for word sequences as a function of acoustic, semantic and formal similarity. Quarterly Journal of Experimental Psychology, 18(4), 362–365.

Baddeley, A .D., & Hitch, G. (1974). Working memory. In G.H. Bower (Ed.), The psychology of learning and motivation: Advances in research and theory (Vol. 8, pp. 47–89). New York: Academic Press.

Bahrick, H. P., Bahrick, P. O., & Wittlinger, R. P. (1975). Fifty years of memory for names and faces: A cross-sectional approach. Journal of Experimental Psychology: General, 104(1), 54–75. https://doi.org/10.1037/0096-3445.104.1.54

Conrad, R. (1964). Acoustic confusions in immediate memory. British Journal of Psychology, 55(1), 75–84. https://doi.org/10.1111/j.2044-8295.1964.tb00899.x

Cowan, N. (2017). The many faces of working memory and short-term storage. Psychonomic Bulletin & Review, 24(4), 1158–1170. https://doi.org/10.3758/s13423-016-1191-6

Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11, 671-684.

Glanzer, M., & Cunitz, A. R. (1966). Two storage mechanisms in free recall. Journal of Verbal Learning and Verbal Behavior, 5(4), 351–360. https://doi.org/10.1016/S0022-5371(66)80044-0

Grover, S., Wen, W., Viswanathan, V., Gill, C. T., & Reinhart, R. M. G. (2022). Long-lasting, dissociable improvements in working memory and long-term memory in older adults with repetitive neuromodulation. Nature Neuroscience, 25(9), 1237–1246. https://doi.org/10.1038/s41593-022-01132-3

Jacobs, J. (1887). Experiments on “prehension”. Mind, os-12(45), 75–79. https://doi.org/10.1093/mind/os-12.45.75

Miller, G. A. (1956). The magical number seven, plus or minus two: Some limits on our capacity for processing information. Psychological Review, 63(2), 81–97. https://doi.org/10.1037/h0043158

Peterson, L. R., & Peterson, M. J. (1959). Short-term retention of individual verbal items. Journal of Experimental Psychology, 58(3), 193–198. https://doi.org/10.1037/h0049234

Raaijmakers, J.G.W. & Shiffrin, R.M. (2003). Models versus descriptions: Real differences and langiage differences. behavioral and Brain Sciences, 26, 753.

Scoville, W. B., & Milner, B. (1957). Loss of recent memory after bilateral hippocampal lesions. Journal of Neurology, Neurosurgery, and Psychiatry, 20(1), 11–21.

Shallice, T., & Warrington, E. K. (1970). Independent functioning of verbal memory stores: A neuropsychological study. Quarterly Journal of Experimental Psychology, 22(2), 261–273. https://doi.org/10.1080/00335557043000203

Sperling, G. (1960). The information available in brief visual presentations. Psychological Monographs: General and Applied, 74(11), 1–29. https://doi.org/10.1037/h0093759