Levels of Processing

The Levels of Processing view was put forward by Craik and Lockhart in 1972, as an alternative to thinking about memory purely in terms of separate stores. Although the view has been refined several times since it was first proposed, its central idea has stayed the same: how well any new information is eventually retained depends on the way it is perceived, analysed, and understood at the time it is processed — that is, on how deeply it is processed, not on which 'store' it happens to sit in.

Craik and Lockhart proposed that incoming information can be analysed at more than one level. At the shallowest level, a person may attend only to the physical or structural features of the material — for example, noticing only the shape of the letters in the word 'cat', regardless of whether it is written in capitals or small letters, or what colour ink was used. At an intermediate level, a person may instead attend to the phonetic sounds attached to the letters, so that the structural features get converted into at least one meaningful spoken word — the word 'cat' understood as a specific sequence of sounds rather than just a shape. Both of these levels — structural and phonetic — produce memory that is comparatively fragile and decays fairly quickly.

There is a third and deeper level still. To make sure information is retained over a longer period, it needs to be analysed and understood in terms of its meaning — for instance, thinking of a cat as an animal that has fur, four legs, and a tail, and belongs to the mammal group, or forming a mental image of a cat and connecting that image to one's own experiences. Analysing information in terms of its structural or phonetic features is shallower processing; encoding it in terms of the meaning it carries — semantic encoding — is the deepest level of processing, and it is this depth that produces memory that resists forgetting considerably better than shallow processing does. …