The levels of processing (LOP) framework, proposed by Fergus Craik and Robert Lockhart in 1972, challenged the dominant multi-store model of memory by arguing that memory is not determined by which store information reaches but by how deeply it is processed during encoding. Shallow processing (analyzing physical features) produces weak, short-lived memories, while deep processing (analyzing meaning) produces strong, durable memories.
The Framework
Craik and Lockhart proposed a continuum of processing depth. Structural processing (the shallowest level) involves analyzing physical features: Is the word in uppercase or lowercase? Phonological processing (an intermediate level) involves analyzing sound: Does the word rhyme with "train"? Semantic processing (the deepest level) involves analyzing meaning: Does the word fit the sentence "The ___ was tasty"? Each deeper level of processing engages more elaborate, more distinctive, and more meaningful analysis.
Phonological: "Does the word rhyme with TRAIN?" → ~35% recall
Semantic: "Does the word fit: 'He sat on the ___'?" → ~65% recall
Deeper processing → Better incidental memory (even without intention to learn)
Key Findings
The LOP effect is remarkably robust. Deeper processing improves memory even when participants are not trying to memorize (incidental learning conditions). The effect holds across different types of material, different age groups, and different retrieval tests. Semantic processing produces better memory than phonological processing, which produces better memory than structural processing — a consistent ordering found in hundreds of studies.
One of the most powerful encoding strategies is self-referential processing — relating information to oneself. Rogers, Kuiper, and Kirker (1977) showed that judging whether a word describes oneself produces better memory than semantic processing of the same word. This self-reference effect suggests that the self-schema provides a rich, well-organized knowledge structure that facilitates deep, distinctive encoding. It has practical implications for education: relating material to personal experience enhances learning.
Criticisms and Refinements
The LOP framework has been criticized on several grounds. The definition of "depth" is circular — deep processing is defined as whatever produces better memory. The framework does not adequately explain why some elaboration improves memory while irrelevant elaboration does not. And it initially underemphasized the importance of retrieval conditions: transfer-appropriate processing theory (Morris, Bransford, and Franks, 1977) showed that phonological encoding can produce better memory than semantic encoding when the retrieval test requires phonological information.
Despite these criticisms, the core insight — that how information is processed matters more than how long it is rehearsed — has been thoroughly validated and has transformed educational practice. The distinction between maintenance rehearsal (rote repetition, which contributes little to LTM) and elaborative rehearsal (processing for meaning, which strengthens LTM) directly reflects the LOP principle.