The Multi-Store Model of Memory
The Multi-Store Model
Where does a memory go after you first notice something? The multi-store model of memory proposes that information passes through three separate stores: the sensory register, short-term memory and long-term memory. Each store is different in how it codes information, how much it can hold (capacity), and how long it lasts (duration).
The multi-store model has three stores: the sensory register (huge capacity, near-instant duration), short-term memory (about 7 items, up to around 30 seconds) and long-term memory (potentially unlimited capacity, potentially permanent duration).
The Sensory Register
The sensory register receives information from every sense at once. Its capacity is huge — it can take in a massive amount of raw sensory data — but its duration is extremely brief, lasting only fractions of a second. Almost all of it fades before you ever notice it. Only the small amount you actually attend to moves on into short-term memory.
Coding in the sensory register matches whichever sense picked the information up: visual information is coded visually (iconic memory) and sound is coded acoustically (echoic memory). The brief flash you still ‘see’ after a camera flash goes off is your iconic sensory register at work.
Short-Term Memory
Short-term memory (STM) holds information you are using right now, such as a phone number you have just read. Its capacity is limited to around 7 items, plus or minus 2, and without rehearsal information lasts only about 15–30 seconds before it fades. Coding in STM is mainly acoustic — we tend to hold information as sounds in our head, even words we have only read silently.
Rehearsal is the mechanism the model uses to explain how information moves from STM into LTM. Repeating information keeps it active in STM, and enough repetition strengthens the trace so it can be stored long-term. Without rehearsal, the STM trace decays and disappears — which is why you can lose a phone number almost instantly if someone interrupts you before you dial it.
Memory depends on specific brain structures, and damage to them produces different patterns of forgetting. Damage to the hippocampus is linked to anterograde amnesia — difficulty forming any new lasting memories from that point onward. Damage to the frontal lobe is linked to retrograde amnesia — difficulty recalling memories from before the injury. In contrast, the cerebellum supports procedural memory, such as physical skills, which can survive largely intact even after severe damage elsewhere in the brain.
Long-Term Memory
Long-term memory (LTM) is the relatively permanent store for information that has passed through STM. Its capacity is thought to be essentially unlimited, and duration can last a lifetime. Coding in LTM is mainly semantic — information is stored by its meaning rather than its sound or appearance.
STM codes mainly acoustically, holds around 7 items (±2), and lasts up to about 30 seconds without rehearsal. LTM codes mainly semantically, has potentially unlimited capacity, and can last a lifetime.
Background — In 1985, musician Clive Wearing developed a severe viral brain infection that damaged his hippocampus, leaving him with profound amnesia. Wilson, Kopelman and Kapur (2008) studied his case in detail to understand how the brain damage had changed his memory.
Method and results — This was a longitudinal case study of a single participant. The researchers used neuropsychological memory tests, clinical observation over many years, and analysis of the diary Clive kept, alongside information from his wife. Clive could not form new lasting episodic memories — he repeatedly believed he had only just woken up, writing near-identical diary entries again and again and crossing out what he had written minutes earlier. Yet his procedural memory remained intact: he could still play the piano and conduct a choir fluently.
Conclusions and criticisms — The case supports the idea that different types of long-term memory depend on different brain structures: hippocampal damage disrupted his ability to form new episodic memories and to recall some earlier ones, while his procedural memory, linked to the cerebellum, survived. As a single case study, the findings cannot be generalised to memory in the wider population and cause and effect is hard to establish with certainty, but the detailed, real-world, longitudinal data would be impossible to gather in a lab experiment.
A weakness of the model is that it may overstate the importance of pure repetition. Rehearsal predicts that any information repeated enough times should eventually reach LTM, but thinking about meaning (elaboration) is generally far more effective at building lasting memories than mechanical repetition alone. The Wilson, Kopelman and Kapur (2008) study also challenges the idea that damage affects memory as one single system: Clive Wearing’s brain damage disrupted his ability to form new episodic memories while leaving his procedural memory intact, suggesting long-term memory is not a single unitary store as the model implies.
How Information Moves Between Stores
Information reaches the sensory register automatically from every active sense. At any moment it holds a vast amount of raw data — every sound, colour and sensation around you — almost all of which is never consciously noticed and fades within a fraction of a second. The gateway from the sensory register into STM is attention: when you selectively attend to something, such as hearing your name in a noisy room, it transfers into STM and becomes available for conscious processing.
Once information is in STM, rehearsal decides whether it passes into LTM. Maintenance rehearsal — simple repetition — keeps information circulating in STM, and with enough repetition the trace can be stored more permanently in LTM. Without rehearsal, the STM trace decays within about 30 seconds and is lost.
Not everything reaches LTM only through deliberate repetition. Elaborative rehearsal — linking new information to meaning and to things you already know — is more effective than simple repetition, and emotionally significant events can enter LTM rapidly even without conscious rehearsal at all.