Levels of processing
Levels of processing is the idea, proposed by Craik and Lockhart in 1972, that how well you remember something depends on how deeply you process it when you meet it. Thinking about meaning (deep) builds a stronger memory than noticing looks or sounds (shallow), however long you stare at it.
In more detail
Craik and Lockhart offered the idea as an alternative to the older view of separate memory stores. Their claim was that memory is a by-product of processing. Work with the meaning of an item and the trace lasts. Touch only its surface and the trace fades quickly. They described a continuum, and most experiments use three levels: structural (what it looks like), phonemic (what it sounds like) and semantic (what it means).
The best-known test is Craik and Tulving (1975). People saw words, each paired with a yes or no question: is the word in capital letters (structural), does it rhyme with weight (phonemic), would it fit into a given sentence (semantic). In a surprise recognition test afterwards, words that had been processed for meaning were recognised far more often than words processed for letter case, although everyone had seen each word for the same short time. Self-reference goes deeper still: Rogers, Kuiper and Kirker (1977) found that asking whether a trait word describes you gave better recall than asking what it means.
The model has limits. Eysenck (1978) criticised it as circular, because depth was never measured on its own: deep processing was simply whatever produced better memory. Transfer-appropriate processing (Morris, Bransford and Franks, 1977) adds that the best way to encode depends on the test, and on a rhyming test sound-based encoding can beat meaning-based encoding. For studying the core advice survives. If your exam asks you to explain, apply and connect, encode the material in exactly that way.
Levels of processing from shallow to deep
| Level | What you attend to | Example study task | How well it sticks |
|---|---|---|---|
| Structural (shallow) | Letters, layout, fonts | Highlighting key words, copying a slide word for word | Weakest |
| Phonemic (middle) | Sound, rhyme, rhythm | Repeating a term aloud, making up a rhyme | Middle |
| Semantic (deep) | Meaning and links to what you know | Explaining in your own words, giving an example, asking why | Strong |
| Self-referential (deep) | How the idea relates to you | Tying a concept to a case from your own life or course | Among the strongest |
Example
Mia studies opportunity cost. Version one: she copies the definition from the slide and highlights it in yellow, twice. Version two: she closes the slides, explains the idea out loud with her own example (giving up a Saturday shift to study means losing the wages), and writes down how it differs from sunk cost. Both versions take about five minutes. A week later a quiz asks her to apply opportunity cost to a new case. Version one left her with a few words. Version two left her with an idea she can use.
How to use it
- 1
Cover the source and explain the idea in your own words before you check the slide.
- 2
Ask why and how: why is this true, how does it work, what would change if it were not?
- 3
Give one example from the course and one from your own life.
- 4
Link the idea to a concept you already know and write down how the two differ.
- 5
Match the format of the exam: if it asks you to apply, practise applying, not only explaining.
How StudyPDF does this
StudyPDF builds deep processing into the quiz format. Short answer, matching and fill-in-the-blank questions make you produce the answer instead of recognising it, every question comes with an explanation and the source page from your lecture, and you can ask Bo why an answer is right or how two concepts differ, with the answer drawn from your own material.
See also
Put it to work on your own course
Upload your lectures and ask Bo. Cited answers, plus flashcards, exams, and more built from your material.
Common questions
What are the three levels of processing?
Structural processing looks at how something appears, such as letters or layout. Phonemic processing looks at how it sounds. Semantic processing looks at what it means. They run from shallow to deep, and in the classic experiments memory got better the deeper the level.
How do you use levels of processing to study?
Move from copying to explaining. Put ideas in your own words, ask why and how, give examples and connect new material to what you already know. These tasks force semantic processing, which is why they beat highlighting or rereading.