Memory

Memory is a very complex topic, much researched and at the heart of the �cognitive revolution�: what follows is therefore even less reliable than usual, but it has been filtered and distilled with the needs of teachers in mind: please go elsewhere for a synoptic view, such as Gross (1996) ch 12; Rose (1993)

Memory is of course central to learning, which could not happen without it: indeed �memorising� is a synonym for the lowest levels of rote learning.

The diagram to the right illustrates schematically the current view of memory, based on the model of Atkinson and Shiffrin (1968). Input of sensory information starts at the top, goes through �sensory memory� or a �sensory buffer� into short-term memory (STM), and hence to long-term memory (LTM).

Two points are worth mentioning for our purposes:

• the process is selective and �lossy�. At each stage information is filtered, selected and even altered (represented by the grey arrows which stand for information lost). 
• Progression to long-term memory (if it is reached) may well take less than a minute. It is drastically simplistic, but still salutary, to consider that from presenting something you have one minute to get it into a student's mind (LTM)!

The sensory buffer is only technically a part of memory: it might make more sense to regard it as part of the perceptual system. Information may stay there for about 1/15 of a second, while the brain assembles it to �make sense�. We are familiar with the illusion by which a succession of still pictures presented rapidly enough appear to be moving: it is the basis of all cinematography. Once the frame rate drops below about 16 frames per second, however, we may well become conscious of the flicker or jumps from one still image to another. Similarly, we do not hear a succession of phonemes (or speech sounds), but complete words or phrases: it is as if the brain waits to assemble a meaningful sound before passing it on to the next stage.

Which is Short-term Memory (if the information makes it that far). The research suggests that STM deals best with sounds rather than visual stimuli, but that may perhaps be a reflection of the problem of devising experiments which do justice to the fact that visual stimuli are taken in all at once, whereas sounds are processed in a linear fashion�over time. In any event, the STM:

holds material for about 15-30 seconds, although this can be expanded by practice. This is much shorter than most of us think�a lot of people seem to think that it lasts for ten minutes or so. You can see, however, how disabling damage to the STM is, as in the case of dementia.   
has a capacity of seven items (plus or minus two). However, �items� are defined by meaning rather than size, so it may be difficult to remember telephone numbers of more than seven digits, but if �01234� is remembered as �Bedford STD code� it becomes just one item, and remembering �7,9,3,1,5,6� after it becomes simpler. If that too is �chunked� as �my work 'phone number� it is even easier (it's not my number any longer incidentally). This of course assumes that a label for the "chunk" already exists in long-term memory.

The feedback arrow on the exit from STM represents what Atkinson and Schiffrin called �rehearsal�: the process of continually recalling material into the STM in order to work on it�memorise it or digest it. I do violence to the technicalities if I say [STM + rehearsal + retrieved LTM] = �Working memory�, but it makes good pragmatic sense. The process of digestion must (IMO) be important: it is where processes such as Piaget's assimilation and accommodation come into play, as part of the process of �coding� material for long-term storage. It is another stage at which detail may be lost, and/or of course changed. Memory is fallible and malleable

And so to Long-term Memory. Theoretically, LTM has infinite capacity and lasts for the rest of your life.

Tulving (1985) suggested the useful distinction between three components of LTM:

• Semantic memory stores concepts and ideas 
• Episodic (sometimes referred to as �autobiographical� or �narrative�) memory contains memories of events.
• Procedural memory concerns skills and �know-how� rather than �know-that� knowledge.

A few moments' thought will make it clear that these are fundamentally different. People with amnesia, for example, typically lose episodic memory, but other memories may be relatively intact. Episodic and semantic memory are more prone to distortion than procedural memory, which is more robust: a skill lost through lack of practice typically comes back rapidly when called upon, and without significant degradation. However, semantic and episodic memories are more amenable to linguistic description and communication.

It is a common-place that LTM �plays tricks� and of course that we forget things. That is too complex a matter to go into here, but it is worth mentioning that the �tricks� are principally in the direction of simplification and minimising discrepancies within the memory store (see cognitive dissonance), and forgetting (apart from the kind associated with neurological damage) was described�in a specific context�very well a couple of thousand years ago:

�A sower went forth to sow, bearing precious seed;

Some fell by the wayside,

Some memories never even make it to LTM.

Some fell upon stony ground.

Some are not consolidated, perhaps because of conflict with existing �knowledge�

Some fell among thorns,

Known as �interference�, competing in context with other more pressing concerns, or emotional factors leading to repression.

But other fell upon good ground.

A sower went forth to sow, bearing precious seed,
A sower went forth to sow�let all who hear take heed!�

(Verse by Bruce Reed, set to music by David Silk, 1962: based on Mark 4:3-20)

The only things not mentioned in the parable are �trace decay� from non-use, and �cue-dependence� where memories are so tied in to a setting�an episodic memory, probably�that they cannot be recalled without it.

A variant of cue-dependence, however, is much used in mnemonics�tricks to facilitate memorising.