The multi-store model (MSM)
The multi-store model (MSM) explains how data (pieces of information) move between 3 storage systems, each system differing in terms of:
• capacity — how much information is stored
• duration — how long information is stored
• encoding — the form in which information is stored.
The sensory register (SR) holds huge amounts of unprocessed sensory information received by sensory organs for a short duration. Information that is paid attention to goes for further processing in short-term memory (STM) and non-attended information is immediately lost. The SR has separate stores for sensory inputs, e.g. iconic store for visual and echoic store for auditory information.
STM is a temporary memory system holding information in use. The dominant encoding type here is acoustic, with other sensory codes also used. Capacity is limited to 5—9 items, extended by chunking, where the size of the units of information is increased. Duration is limited to around 20 seconds, though rehearsal retains data within the STM loop, until eventually it becomes more permanent within long-term memory (LTM).
The dominant encoding type in LTM is semantic, though other encoding types also occur, e.g. visual and acoustic. Potential capacity is assumed to be unlimited, with duration potentially lifelong. Information in LTM does not have to be continually rehearsed to be retained.
Fig 2.1 Baddeley’s (1966) acoustic/semantic study findings, where List A = acoustically similar words, List B = acoustically dissimilar words, List C = semantically similar words and List D = semantically dissimilar words
Baddeley (1966) examined encoding in STM and LTM by giving 75 participants either: acoustically similar words (rhyming words) like ’caught’ and ’taut’, acoustically dissimilar words (non-rhyming words) like ’foul’ and ’deep’, semantically similar words (words with similar meanings) like ’big’ and ’huge’, or semantically dissimilar words (words with non-similar meanings) like ’pen’ and ’ring’. With STM, acoustically dissimilar words were better recalled (80 per cent) than acoustically similar words (10 per cent), indicating acoustic encoding to be dominant. Semantically dissimilar words (71 per cent) were recalled slightly better than semantically similar ones (64 per cent), suggesting semantic coding does occur in STM but isn’t dominant. With LTM, participants followed the same procedure, but with a 20-minute gap between presentation and recall. There was no difference between acoustically similar and dissimilar words, but more semantically dissimilar words (85 per cent) were recalled than semantically similar ones (55 per cent), suggesting semantic encoding is dominant in LTM.
• Crowder (1993) found that memories in the SR only retain information in the iconic store for a few milliseconds, but retain information for up to 3 seconds within the echoic store. This supports the idea that sensory information is coded into different stores, while additionally suggesting that sensory memories have different durations.
• Peterson & Peterson (1959) read participants nonsense trigrams (words of 3 random letters, e.g. XPJ), then asked them to count backwards from a large digit between 3 and 18 seconds later to prevent recall. 90 per cent of trigrams were recalled after 3 seconds, but only 5 per cent after 18 seconds, suggesting STM duration is between 18 and 20 seconds.
• Bahrick et al. (1975) found that participants who had left school in the last 15 years recalled 90 per cent of faces and names of schoolmates from photos, while those who had left 48 years previously recalled 80 per cent of names and 70 per cent of faces. This implies LTM duration to be very long-lasting.
The MSM was the first cognitive explanation of memory and inspired interest and research, leading to later theories like the working memory model that gave an even greater understanding of memory.
The brief duration of SR and STM has an evolutionary value, as we only need to focus on sensory information with an immediate survival value. LTMs are retained for longer as they may have an ongoing survival value.
The theory is supported by amnesia cases (loss of memory). Patients lose either their STM or their LTM ability, but not both, supporting the idea that STM and LTM are separate memory stores located in different brain areas.
The MSM is oversimplified as seeing STM and LTM as single stores. Research suggests there are several types of STM, such as separate stores for visual and auditory information, as well as different types of LTM, such as procedural, episodic and semantic LTM (see page 28).
Cohen (1990) thinks memory capacity is not measurable only by the amount of information, but by the nature of the information to be recalled. Some things are easier to recall regardless of the amount of information being recalled. MSM does not consider this.
MSM focuses too much on memory structure rather than on processes.
Research into the MSM has allowed psychologists to create strategies for improving memory performance, such as chunking, where STM capacity is increased by grouping separate pieces of information into larger units with a collective meaning. This can be useful for students when revising.