Multi-store model (Atkinson & Shiffrin, 1968)

Capacity is how much a store holds; duration is how long. Candidates define one and label it the other, or give a general definition without tying it to the scenario. Say which feature you mean — "STM is small" earns nothing — and where a situation is given, attach it: "the digits were gone before she reached the keypad".

The named error here: candidates say attention is needed to enter the sensory register, or that rehearsal puts information there — both reverse the model. The sensory register is the first store, entered automatically by everything the senses pick up. Attention is the arrow out of it (into STM); rehearsal is one arrow further (STM to LTM).

Candidates often draw the three stores accurately but never touch the scenario — that is AO1 only, and since AO3 needs AO2 an unapplied answer can lose the evaluation marks too. Not application: the character''s name, repeating the stem, copying a line. Application reads a specific detail through the model, e.g. the six-digit code held just long enough to type.

"A weakness is that it is too simple" names a point and stops. A creditworthy evaluation has three parts: the point, why it follows from the model, and the consequence. E.g. free-recall studies find the first and last list items recalled best, which fits separate stores, so the model makes a testable prediction. Identifying is only the first part.

A tautology repeats the term instead of defining it — "retrieval is when you retrieve something", or "capacity is how much capacity a store has". It earns nothing: no mechanism is shown. Every definition must carry the mechanism — for example, retrieval brings material back from LTM into STM so it can be used, and capacity is the number of items a store can hold at once.

"Describe the difference" and "compare" set the ideas against each other, not two definitions side by side, which leaves the contrast implicit — one mark at most. A connective (whereas) earns the second mark: the multi-store model uses fixed stores linked by attention and rehearsal, whereas reconstructive memory rebuilds fragments through schemas.

Two distortions. Rehearsal has two jobs: maintenance (keeping material in STM past ~18-30s) and transfer into LTM — dropping the first loses a mark. "Rehearse it and you''ll remember it" is the model''s weak point: we recall unrehearsed shocking moments yet forget repeated material, so meaning and emotion affect transfer, which it explains by repetition alone.

Two errors. The MSM is not reconstructive memory: it is a structure of stores linked by attention and rehearsal, whereas reconstructive memory rebuilds fragments through schemas (Bartlett). "There is evidence, so it must be right": the lab studies give good control and reliability but used word lists, so their ecological validity is limited.