Levels of Processing: Why Deeper Encoding Can Help Memory
Author: Angela Bogdanova · Published: October 12, 2026 · Editorial Policy
Research cluster: Memory & Forgetting · Written by Artificial Sapiens
Levels of processing is a framework in cognitive psychology proposing that what people do with information during encoding has important consequences for later remembering. Processing a word for its meaning, relationships, and implications often produces better subsequent memory than attending only to its typography or sound. The classic finding is powerful, but its modern interpretation requires an important qualification: the value of an encoding operation depends on the information it preserves and the demands of the eventual memory test.
The framework was introduced by Fergus I. M. Craik and Robert S. Lockhart in 1972 and developed through experimental research by Craik and Endel Tulving. It remains a foundational way to explain why meaningful engagement can support retention, even when people have no explicit intention to memorize. Contemporary memory science also emphasizes elaboration, distinctiveness, organization, retrieval cues, and the fit between encoding and retrieval. These mechanisms explain both the successes and the limits of the original account. See the foundational papers by Craik and Lockhart (1972), Craik and Tulving (1975), and the later reassessment by Craik (2002).
What Are Levels of Processing in Memory?
Levels of processing refers to differences in the kinds of cognitive analysis applied to information while it is encountered. A perceptual judgment might register the size or shape of a written word. A phonological judgment might register how it sounds. A semantic judgment connects the word with meaning, concepts, knowledge, and relationships. In the traditional account, semantic analysis is described as deeper processing.
A concise definition is useful: the levels-of-processing framework explains variation in subsequent retention partly through the qualitative character of encoding operations, with meaning-oriented processing commonly producing more memorable representations than surface-oriented processing. The framework describes a regularity in experimental memory tasks; it does not establish a single numerical scale of depth or an unchanging hierarchy for every material and retrieval goal.
In practical terms, someone who asks whether the word “harbor” appears in uppercase letters has attended to a visible feature. Someone who asks whether “harbor” rhymes with “barber” attends to its sound. Someone who asks how a harbor supports ships, trade, and shelter constructs a meaningful connection. The last operation supplies more potential ways to recover the concept later, although a test specifically asking about rhyme could favor the phonological encoding task.
Levels of processing is sometimes called the depth-of-processing approach or levels-of-processing theory. These names describe the same research tradition. The label “deep learning” is also used informally for meaningful study, but it is not a precise synonym and should not be confused with the machine-learning term.
The Three Familiar Encoding Conditions
Structural or visual processing
Structural processing concerns how information appears. Researchers may ask whether a word is printed in capital letters, whether it contains a certain letter, or whether it is shown in a particular font. Participants can answer by examining graphic features without engaging much with the word's meaning.
For the target word “ORCHARD,” a structural question would be, “Is this word written entirely in capitals?” A person could answer correctly while paying little attention to trees, fruit, agriculture, or any personal association with an orchard. Such processing can be appropriate when the eventual goal is to remember formatting or distinguish visually similar stimuli.
The designation shallow is task-relative. Looking at the physical characteristics of a handwritten signature can be exactly the operation needed to recognize that signature later. The traditional prediction of weaker retention concerns many subsequent tests of word identity or meaning, not an inability to remember visual details.
Phonemic or phonological processing
Phonological processing concerns the sound structure of language, including pronunciation, syllables, and rhyme. A common orienting question asks, “Does this word rhyme with ‘tree’?” Another asks whether two words begin with the same sound.
For “ORCHARD,” the learner could examine pronunciation or count syllables without retrieving a rich semantic interpretation. Such sound-based operations can establish information that a semantic judgment might omit. This distinction matters in language learning, poetry, pronunciation practice, and tasks that later test auditory or rhyme features.
Phonological encoding is conventionally placed between graphic and semantic encoding in textbook diagrams. Modern evidence does not justify treating it as universally inferior. The decisive question is which information was encoded and how it will later be accessed.
Semantic processing
Semantic processing concerns meaning. An experiment may ask whether a word belongs to a category, whether it fits a sentence, or whether it has a particular conceptual property. For “ORCHARD,” the prompt “Could someone harvest fruit here?” demands access to meaning.
Semantic judgments can recruit existing knowledge, connect a target to other concepts, and differentiate it from similar targets. Someone who imagines a family visiting an orchard, recalls the smell of apples, and explains why orchards require cultivation has constructed several related cues. These connections often support later recall and recognition.
Semantic activity varies in quality. A quick category judgment, a richly elaborated explanation, and an unrelated chain of associations are all meaningful in a broad sense, but they need not produce equal memory outcomes. The richness and distinctiveness of the resulting representation matter.
Where the Framework Came From
The 1972 proposal and the multistore debate
In the late 1960s and early 1970s, influential memory models characterized information as passing through different stores, often distinguishing a short-term store from a long-term store. Craik and Lockhart argued that duration in a store and mere repetition could not fully explain which experiences endured. They proposed that a central explanatory variable was the processing performed on an item, ranging from analyses of sensory features to increasingly meaningful operations.
Their 1972 framework paper was a theoretical proposal intended to redirect experimental questions. It did not demonstrate that short-term and long-term memory systems were fictional, nor did it supply a complete neural mechanism. Subsequent research has shown the continuing importance of working-memory processes, long-term memory systems, and consolidation. The framework's lasting contribution is its insistence that encoding quality deserves independent investigation.
The proposal also challenged a simple equation between rehearsal time and durable learning. Repeating a telephone number to keep it briefly available is a different operation from connecting that number to a meaningful structure. The contrast motivated research on maintenance rehearsal and elaborative processing.
Evidence that preceded the theory
Work by Hyde and Jenkins helped establish that the kind of incidental task performed during exposure to words could affect later memory. Participants who made meaning-oriented judgments showed memory advantages over people performing more superficial judgments, even without a straightforward intentional-memorization requirement. Their 1969 study helped set the experimental stage for the levels-of-processing account.
These incidental-learning designs are important because they separate deliberate study intentions from the operations actually performed. People can form lasting memories without deciding in advance to memorize, while deliberate efforts can be inefficient if they involve only repetitive or uninformative operations.
The Craik and Tulving Experiments: What Was Actually Tested?
In a landmark series published in 1975, Craik and Tulving reported ten experiments examining depth of processing and episodic retention. Participants answered orienting questions about words and were subsequently tested on their memory, often unexpectedly. Some questions directed attention to physical features, others to sounds, and others to meaning. The general pattern favored semantic encoding in conventional later memory tests. The original report is available through the American Psychological Association DOI record.
Consider a simplified illustration of their approach. A person sees “CANDLE” after an uppercase-letter question, a rhyme question, or a sentence-completion question. The task requires an answer but may give no warning of the subsequent test. Later, when asked to identify previously seen words, the participant often performs better after a semantic question.
The result is informative because the target word need not change. What changes is the mental operation invited by the question. This experimental structure made it possible to study encoding independently of whether people thought they were memorizing.
The findings require more than the slogan “deep is better.” The experiments also probed congruity, elaboration, and how the processing question related to the target. An appropriate and meaningful sentence context can integrate a word more effectively than a strained or mismatched context. Thus, a binary semantic judgment is not an automatic guarantee of rich or distinctive encoding.
The experiments established a robust and influential laboratory pattern across a range of tasks. They did not prove a universal, context-free law that every semantic judgment beats every perceptual or phonological judgment on every possible test. Later work was designed precisely to test that stronger claim.
Incidental learning versus intentional learning
Intentional learning occurs when people expect a memory test and deliberately try to retain information. Incidental learning is measured when the later memory test is not an announced goal of the original task. The levels-of-processing studies demonstrated that orienting activities can produce significant differences in incidental retention.
This does not mean intention is irrelevant. A learner expecting an exam may choose more effective encoding operations, schedule retrieval practice, or allocate attention differently. The experimental point is narrower: favorable processing can support retention without an explicit memorization instruction.
Does a semantic task simply take longer?
A possible explanation for the original results is that semantic questions require additional seconds or mental effort, thereby allowing more processing. Yet time alone does not determine what information is stored. A long series of repeated visual checks may still furnish fewer conceptual retrieval routes than a brief but meaningful comparison.
The 1975 work and later theoretical critiques made it essential to distinguish elapsed time, task difficulty, effort, and qualitative encoding. These variables can correlate, but they are not interchangeable. A difficult, confusing task can consume substantial effort without creating a useful representation.
Why Meaningful Encoding Often Helps
Elaboration builds relationships around the target
Elaboration adds meaningful relationships that can be used later. Instead of reading “photosynthesis” repeatedly, a learner explains how light energy contributes to chemical processes in plants, contrasts photosynthesis with respiration, and constructs an example of a leaf receiving sunlight.
The memory advantage need not arise because the learner has created a permanently deeper layer. A more concrete explanation is that additional relationships furnish multiple routes for retrieval. Some associations will prove more useful than others, and irrelevant embellishment can distract from the central concept.
Craik's 2002 assessment revisits how levels of representation, encoding-retrieval relationships, and consolidation complicate any single explanatory account.
Distinctiveness helps separate similar experiences
Distinctiveness means that a representation includes features useful for discriminating one episode or item from others. A generic sentence about a river may be less helpful than remembering the specific river, the bridge crossing it, and why that crossing mattered to a historical decision.
Semantic elaboration can create distinctive features, but it can also emphasize common category meaning at the expense of individual surface details. If twenty words are all processed simply as “types of food,” that shared relation may be insufficient to distinguish which particular foods appeared in the list.
This is why effective encoding often combines meaningful organization with item-specific detail. Learners should ask what makes the current example different from nearby examples rather than simply adding more broadly related facts.
Organization connects items into a recoverable structure
People remember collections of information more effectively when relationships among items are accessible. A list of disconnected concepts can be reorganized into causes, consequences, categories, and contrasts. This structure reduces the need to search for each item independently.
Organization and elaboration partly overlap, but the distinction is useful. Elaboration adds associations to an item; organization clarifies relationships across items. An especially effective study explanation may do both.
Research on the self-reference effect provides a related example. In a meta-analysis by Symons and Johnson (1997), self-referential encoding was associated with a memory advantage, and the authors argued that elaboration and organization contributed to it. The result supports a mechanism richer than a simple one-dimensional “depth meter.”
Existing knowledge offers retrieval routes
A familiar concept has a larger web of previously learned associations than an unfamiliar one. Explaining an unfamiliar physical law through an accurate, well-understood everyday analogy may help because the analogy activates relevant prior knowledge.
Prior knowledge can also introduce errors. If someone assumes how an unfamiliar process works based on a familiar but misleading analogy, meaningful processing can reinforce a mistaken explanation. Better encoding requires the meaning itself to be accurate.
A practical implication follows: before elaborating, establish what the new claim actually says. Then connect it to verified concepts and check where the analogy breaks down.
Maintenance Rehearsal and Elaborative Rehearsal
Maintenance rehearsal repeats information to keep it momentarily accessible. Saying a room number silently until reaching the door is an everyday example. Elaborative rehearsal introduces meaning, relationships, or transformations intended to support later retrieval.
Craik and Watkins examined rehearsal in short-term memory and showed why sheer repetition should not automatically be equated with superior long-term retention. Their 1973 paper helped challenge a straightforward “more repetitions equals stronger memory” interpretation.
Still, maintenance rehearsal is useful for its actual purpose. Keeping a code accessible for a few seconds is a legitimate cognitive achievement. It would be misleading to describe this operation as useless simply because it often produces weaker durable retention than elaborative learning.
Repetition also has different effects depending on its form. Re-reading, explaining, recalling from memory, and revisiting information across days are separate operations. Evidence for distributed practice or retrieval practice cannot be assigned automatically to passive repetition. The distinction is covered in our articles on retrieval practice and spaced repetition.
Transfer-Appropriate Processing: The Essential Limitation
When sound-based encoding beats meaning-based encoding
In 1977, Morris, Bransford, and Franks compared different encoding conditions with different recognition tests. When the test asked for ordinary word recognition, semantic processing outperformed rhyme-oriented processing. When the test instead emphasized rhyme, the advantage reversed: rhyme-oriented encoding supported better performance.
The researchers replicated the general pattern with immediate and delayed tests and examined whether repetition of sound features could explain it. Their findings favored transfer-appropriate processing: memory performance depends in part on the match between operations at encoding and those required at retrieval. The experiments are documented in Morris, Bransford, and Franks (1977).
This result is one of the clearest reasons not to turn levels of processing into the unconditional rule “semantic always wins.” Different tests ask for different information. An analysis that preserves meaning may excel on a conceptual test, while one that preserves sound may excel on a phonological test.
How encoding and retrieval interact
Transfer-appropriate processing does not deny the benefits of semantic processing. It adds the question “appropriate for what?” A student preparing to explain a biological mechanism should practice causal explanations and applications. A student preparing for a pronunciation test should also attend to phonological detail. Someone identifying a face may need distinctive visual features that ordinary semantic labeling overlooks.
The related principle of encoding specificity emphasizes that what can be retrieved depends on how information was encoded and the cues available at the time of remembering. Tulving and Thomson's 1973 work helped establish this approach.
These ideas are compatible when stated precisely: many semantic activities support a broad range of useful retrieval cues, but the usefulness of a cue depends on the task and the circumstances in which it is sought.
Encoding depth and test format
Recognition, free recall, cued recall, source judgments, and perceptual-identification tasks need not rank encoding strategies identically. Even when two tests use the same words, one may require recognition of identity while another requires precise memory for sound, font, temporal order, or where an item appeared.
A well-designed study strategy therefore begins with the kind of knowledge the learner needs to retrieve. Meaningful conceptual understanding is central for explanation and transfer; exact spelling, terminology, pronunciation, and item-specific details may require complementary encoding operations.
Is Depth of Processing a Measurable Psychological Quantity?
The problem of circular definitions
One of the major theoretical criticisms concerns the meaning of depth itself. If a researcher calls an encoding task deep because it produces strong memory, then uses the strong memory to prove that deep processing works, the explanation becomes circular.
Michael Eysenck's 1978 critique raised important concerns about specification and explanatory power. Later assessments, including Craik (2002), explicitly discuss measurement and whether there are discrete levels at all.
Modern experimental work therefore gains precision by describing the operation that participants performed—semantic classification, imagery, generation, relational encoding, surface judgment—and measuring its consequences on defined tests. The phrase “deeper encoding” is most informative when the task, comparison, and outcome are stated.
A framework rather than a complete theory of memory
The original paper called itself a framework. It organized questions about the cognitive operations preceding successful remembering. A complete account of memory must also explain distinct memory systems, neural plasticity, consolidation, interference, retrieval, reconstruction, and the way knowledge changes with practice.
Levels of processing is therefore best treated as a useful set of experimentally developed propositions concerning encoding, embedded within broader cognitive and neuroscientific theories. A single study showing that semantic judgments improve recognition supports a local prediction; it does not demonstrate every proposed claim about how all memories are stored.
Does greater effort always mean greater depth?
Effort, duration, complexity, and successful encoding can vary independently. Trying unsuccessfully to make sense of a confusing passage may demand significant effort without leaving a coherent representation. A rapid but accurate explanation to someone who already knows the subject may generate effective cues with little felt strain.
Effort can matter when it encourages constructive processing, correction, and retrieval. It becomes misleading when treated as a substitute for specifying what the learner actually accomplished.
Levels of Processing and Other Memory Concepts
Encoding versus consolidation
Encoding refers to how information is initially processed and represented. Consolidation concerns stabilization and reorganization over time. A meaningful encoding operation can improve the initial representation available to later processes, but it is not itself the whole process of consolidation.
The distinction is developed in Memory Encoding: How Experiences Become Memories and Memory Consolidation: How Memories Become Stable. The present article addresses depth, elaboration, and experimental encoding tasks as its primary topic.
Encoding versus retrieval
An item may be encoded successfully yet difficult to retrieve with the available cue. Conversely, a highly familiar concept may be recognized even when the person cannot reconstruct the original encounter with it. Remembering involves both what was retained and what the test makes accessible.
Our dedicated account, Memory Retrieval: Recall, Recognition, and Cues, covers retrieval mechanisms more broadly. The levels-of-processing framework adds insight into which properties are likely to be recoverable after different initial operations.
Levels of processing and working memory
Working memory maintains and manipulates information for ongoing tasks. Deep-versus-shallow encoding effects can be examined while a person actively holds information, but immediate maintenance and delayed long-term retention are not identical outcomes. Research by Rose and Craik (2012) examined levels-of-processing manipulations in working-memory-span contexts and illustrates the importance of task and retention interval.
A strong encoding result in a delayed recognition task should not be extrapolated automatically to increased working-memory capacity, processing speed, or general intelligence.
Meaningful encoding and accurate remembering
Semantic processing can enhance later familiarity with a theme or meaning while leaving precise contextual details weaker. Memory often preserves gist in addition to item-specific information, and both can contribute to judgments about prior experience.
Some experimental work links deep encoding to effects on false recognition and monitoring strategies. The direction of a false-memory effect depends on materials, distinctiveness, and test demands; it is not valid to say either that semantic elaboration always prevents false memories or that it always creates them. Gallo and colleagues (2008) investigated relevant interactions between deep processing and recognition errors.
Meaningful study should therefore include checks on factual accuracy, original sources, and distinguishing details whenever exact recollection matters.
What Brain Research Adds
Neuroscientists can examine neural activity while participants encode material and compare activity for items remembered versus forgotten on a later test. This subsequent-memory approach identifies patterns associated with successful encoding under particular experimental conditions.
A review by Paller and Wagner (2002) described how cognitive neuroscience began to connect encoding operations with later episodic remembering. A seminal study by Wagner and colleagues (1998) examined neural activity during verbal encoding and subsequent memory outcomes.
Semantic and relational processing can engage regions involved in language, control, and conceptual analysis, while medial temporal structures play important roles in building durable episodic representations. The precise pattern depends on stimuli, processing demands, task timing, and the method used to analyze brain activity.
A 2024 neuroimaging meta-analysis by Kim synthesized subsequent-memory findings and proposed a model of attentional and memory-related dynamics. Its proposed explanatory model is a contemporary interpretation of accumulated imaging evidence, not a universal biological scale of encoding depth.
There is no established single “deep-processing center” that can be read from a brain scan to determine whether a learner has truly understood material. Functional imaging findings provide converging clues about involved networks; they do not replace experimental comparisons of what people did and later remembered.
Individual Differences, Development, and Aging
The usefulness of semantic encoding depends partly on a person's knowledge of the material, vocabulary, strategy use, available attention, and familiarity with the task. If a concept is unfamiliar, a request to “think about its meaning” can fail because the learner has no stable meaning yet. Teaching the concept accurately must come before elaborate application.
Developmental and aging research examines changes in spontaneous strategy selection, cognitive control, available knowledge, and recollection. An individual who does not naturally use an elaborative strategy may benefit from explicit instruction, while someone with extensive domain knowledge may construct useful associations readily.
A review of aging and encoding by Craik and Rose (2012) discusses relationships among processing resources, encoding, and age-related memory performance. Patterns observed at a group level do not allow a diagnosis or a reliable inference about an individual from their study habits.
Everyday forgetting has many possible contributors. If a person experiences persistent or worsening changes that interfere with daily functioning, appropriate professional assessment considers history, context, health, medications, and functional change. A levels-of-processing explanation by itself cannot evaluate a clinical memory concern.
How to Use Deeper Encoding in Everyday Learning
The practical goal is to create accurate, discriminable, retrievable representations. A meaning-oriented approach is often a productive starting point, especially for material that must be explained, applied, or connected to other concepts. It should be combined with retrieval and scheduling methods supported by separate experimental literatures.
1. Explain the meaning in precise language
After reading a paragraph, state its central claim in your own words. Identify what causes what, what evidence supports the claim, and which conditions might change it. An explanation must preserve the original meaning; fluent paraphrasing of an incorrect understanding can strengthen an error.
For example, instead of repeating “negative reinforcement,” explain that removing an unpleasant stimulus following a behavior can increase the future likelihood of that behavior. Then distinguish this mechanism from punishment and from merely feeling relieved.
2. Compare close concepts and identify diagnostic differences
Ask how one term differs from a neighboring term and build a concrete example of each. Distinguishing retrieval from encoding, or recognition from recall, provides relational organization and item-specific cues.
This is more useful than generating many loosely related associations. Good comparisons identify differences that matter for future questions and decisions.
3. Generate an example and a counterexample
Apply a principle to a new situation. Then identify a similar-looking case where the principle does not apply. This encourages active use of meaning and helps expose misunderstanding.
For levels of processing itself, a semantic example is explaining why a harbor shelters vessels. A counterexample to “semantic always wins” is a rhyme-recognition test in which sound-focused encoding is the better preparation.
4. Build retrieval cues appropriate to the goal
Decide what you will need to remember. For an oral presentation, practice explaining a causal argument without notes. For exact vocabulary, practice definition and spelling. For identifying a diagram, include spatial and visual information as well as conceptual meaning.
Matching encoding to the anticipated retrieval goal follows the logic of transfer-appropriate processing. Because real-world demands are sometimes unpredictable, combining several relevant kinds of encoding is a reasonable practical strategy.
5. Retrieve later instead of only processing again
After meaningfully studying a topic, close the source and attempt to recall the explanation, examples, and distinctions. Check the attempt against accurate feedback and revisit missed points after a delay.
This additional step is supported by an evidence base distinct from the original levels-of-processing theory. The review by Dunlosky and colleagues (2013) evaluates multiple learning techniques, while Roediger and Karpicke (2006) document lasting benefits of retrieval practice under their tested conditions. The overall strategy is to use meaningful encoding to construct good representations and retrieval practice to make access more reliable.
Practical Examples by Task
Remembering a scientific explanation
A student reads that a chemical reaction speeds up with temperature. A surface approach copies the sentence several times. A meaning-oriented approach explains particle collisions, identifies the limits of the explanation, draws a relevant graph, and compares the prediction with an actual experiment. The student then tries to reconstruct the explanation from memory.
The meaning-based activities may support comprehension and recall because they create explanatory structure. Retrieval later offers a separate test of whether that structure is accessible without the source.
Learning a new person's name
The learner attends to the spoken name, repeats it accurately, connects it with a specific encounter, and recalls it later when the person is not visible. Semantic associations may help if they are natural and appropriate, but names are sometimes arbitrary labels. Accurate sound encoding and retrieval practice are especially important.
This case shows why meaningful elaboration is one component of a larger memory strategy rather than a universal replacement for phonological learning.
Reading a legal, historical, or technical document
A reader needs both the meaning of a claim and exact qualifications. Explaining the argument supports a conceptual representation; checking dates, names, negations, citations, and scope preserves precise details. In a context where wording matters, semantic gist alone can be insufficient.
The appropriate encoding strategy includes whichever surface and semantic features will distinguish the relevant claim from a nearby but different one.
Understanding unfamiliar medical information
A person reading health information can ask what a term means and how it connects to their clinician's explanation. This may improve understanding of the material, but memory techniques do not establish whether a treatment is appropriate or replace medical interpretation. Recording accurate instructions and requesting clarification are important when the consequences of a memory error are significant.
What the Evidence Does and Does Not Establish
The strongest evidence for a levels-of-processing effect comes from controlled experimental contrasts, especially in verbal materials, in which encoding questions elicit different operations and subsequent memory is measured. Semantic processing commonly benefits conventional word memory outcomes. Historical experimental findings and later theoretical work support this pattern.
Evidence is also clear that encoding-retrieval compatibility can change which operation is advantageous. Morris and colleagues' rhyme-test result is a direct demonstration rather than a minor exception. The original framework is most useful when integrated with retrieval demands, elaboration, distinctiveness, and the nature of the information.
The applied conclusion is more limited than the familiar educational slogan. A study showing improved recall of a trained list does not by itself show lasting improvements in grades, occupational functioning, or generalized memory capacity. Practical recommendations must distinguish performance on the practiced task, near transfer to similar material, far transfer to different tasks, and demonstrable benefits in daily functioning.
Research reviews of learning techniques provide better evidence for some combinations, especially retrieval practice and spacing, than for a blanket promise that “think deeply” will improve every outcome. Deep semantic engagement can be valuable, but its effects depend on accuracy, test type, knowledge, and the availability of later retrieval.
Common Misinterpretations
“Deep processing always produces stronger memories”
Semantic encoding often improves word recognition and recall, but an operation directed at sound can outperform it when the test requires rhyme information. The outcome depends on what must be recovered. The transfer-appropriate processing experiments directly falsify a universal ranking.
“More effort or longer study time means deeper encoding”
Time and effort can support useful operations, but they are not direct measures of encoding quality. Complex confusion may yield fewer useful cues than a brief accurate explanation. Researchers should specify and test the operations rather than infer depth from effort alone.
“Repetition is worthless”
Maintenance rehearsal can serve immediate retention, and retrieval spaced across time can support durable learning. These are different types of repetition with different functions. The failure of simple repetition to guarantee durable retention does not make every repeated encounter unhelpful.
“Meaningful memory is necessarily accurate memory”
A coherent interpretation can improve gist memory while leaving details or sources uncertain. Elaborating a misunderstanding can reinforce the misunderstanding. Accuracy checks, item-specific distinctions, and reliable source material remain essential.
“The theory explains every memory problem”
The framework addresses an important part of encoding in experimental and everyday settings. It is not a diagnosis, an account of every memory system, a complete model of neural consolidation, or evidence that someone with memory concerns simply needs to try harder.
“Deep processing is a brain-training program”
Levels of processing is an experimental framework, not a standardized commercial intervention. It does not validate broad claims about brain games, supplements, permanent increases in memory capacity, or transfer to unrelated cognitive functions.
Frequently Asked Questions
Who proposed the levels-of-processing theory?
Fergus I. M. Craik and Robert S. Lockhart introduced the levels-of-processing framework in 1972. Craik and Endel Tulving subsequently conducted influential experiments published in 1975 that tested physical, phonological, and semantic encoding tasks.
What are the three levels of processing?
The traditional teaching distinction is structural or visual processing, phonemic or sound-based processing, and semantic or meaning-based processing. These are useful examples of encoding operations, rather than universally separable brain compartments or a validated numeric scale.
Which level is usually best for remembering words?
Semantic encoding usually produces better later word recognition or recall in classic experimental comparisons. Which operation is best for a particular goal depends on the retrieval test. If the test asks whether a word rhymes with another word, phonological encoding may provide the advantage.
What is deep processing in simple terms?
Deep processing is an informal description of engaging with meaning and relationships while encoding information. Explaining what a concept means, connecting it to relevant prior knowledge, and distinguishing it from similar concepts are common examples.
What is a real-life example of shallow processing?
Noticing that a term is printed in blue, has eight letters, or begins with a capital letter is surface-focused processing when the goal is simply to inspect appearance. That inspection may be useful if the later task is to recognize the typography rather than to explain the term's meaning.
Is semantic encoding the same as elaborative rehearsal?
They overlap but are not identical. Semantic encoding concerns processing meaning. Elaborative rehearsal involves actively making connections that help retain information. A brief semantic classification can occur without extensive elaboration, and elaborate connections can vary in relevance and quality.
Does levels of processing explain why I forget what I read?
It offers one possible explanation: reading without meaningful engagement can leave few effective retrieval cues. Forgetting also depends on attention, prior knowledge, interference, retention interval, and retrieval conditions. The framework does not determine the cause of any individual's forgetting.
How does transfer-appropriate processing differ from levels of processing?
Levels of processing emphasizes how encoding operations affect later memory, with semantic encoding commonly advantageous in standard tasks. Transfer-appropriate processing emphasizes whether the operations at study match those required at test. The 1977 Morris, Bransford, and Franks experiments showed that different recognition tests can reverse the apparent advantage.
Does self-reference improve memory?
Self-referential encoding has shown a memory advantage in a meta-analysis. Organization and elaboration explain important parts of this effect. Personal relevance can supply strong cues, but the outcome depends on the comparison task, the material, and what must be recalled.
Does deeper processing increase working-memory capacity?
Evidence for better delayed retention after a particular encoding operation does not establish an increase in working-memory capacity. Immediate maintenance, complex span, and long-term retention are separate outcome measures. Claims about capacity require direct tests of capacity.
Is processing depth more important than studying for longer?
Qualitatively useful encoding operations can matter greatly, and extra time devoted only to superficial repetition may have limited long-term value. The comparison is not an absolute rule. More time can be useful when it supports correct elaboration, feedback, distributed practice, and retrieval.
Can deep processing create false memories?
Meaning-based processing may strengthen gist and shared associations, while item-specific detail and monitoring can help distinguish actual encounters. False-recognition findings vary by design. Meaningful learning should include checks of exact claims and their sources whenever detail matters.
Does the theory apply to pictures, faces, and sounds?
Researchers have investigated meaning-oriented and feature-oriented processing beyond simple word lists, including facial recognition. The most relevant operations and outcomes depend on the material. A face-identification task may require distinctive visual information; an auditory task may require sound-based features.
What is the main criticism of levels-of-processing theory?
A central criticism is that “depth” can be difficult to define and measure independently of memory performance. Other challenges include retrieval-task dependence and the possibility that elaboration, distinctiveness, and organization explain particular effects better than a single depth continuum.
Is levels of processing still accepted in modern psychology?
The classic differences between encoding operations remain influential empirical phenomena. The original framework is now interpreted alongside models of memory systems, encoding-retrieval interactions, organization, distinctiveness, and neuroscience. It is better regarded as a productive research framework than a complete standalone explanation of memory.
What is the most evidence-based way to use this idea while studying?
First understand the material accurately, then explain it, compare related ideas, and develop task-relevant examples. Follow that encoding with retrieval practice, corrective feedback, and spaced revisiting. These steps draw on separate evidence bases and should be chosen for the specific learning goal.
Related Articles
For the broader account of how memory works, explore the full sequence of encoding, retention, and retrieval. Our article on memory encoding covers the initial formation of memory representations, and memory retrieval examines the role of cues and tests. For practical methods with their own evidence base, see how to improve memory, retrieval practice, and spaced repetition.
Conclusion
The levels-of-processing framework helped establish that remembering depends strongly on what a person does with information during encoding. Meaningful engagement frequently makes words and concepts more memorable because it can support elaboration, organization, distinctiveness, and useful retrieval cues.
Its scientific significance becomes clearer when the limits are retained. Encoding operations must be evaluated against the memory outcome being tested; meaning is often useful, and sound or visual detail may sometimes be decisive. The most defensible practical approach combines accurate understanding with task-relevant representation, later retrieval, feedback, and spacing. It treats memory as a coordinated cognitive process whose success depends on both the information established during learning and the conditions under which that information must be recovered.
References
Adesope, O. O., Trevisan, D. A., & Sundararajan, N. (2017). Rethinking the use of tests: A meta-analysis of practice testing. Review of Educational Research, 87(3), 659–701. https://doi.org/10.3102/0034654316689306
Bower, G. H., & Karlin, M. B. (1974). Depth of processing pictures of faces and recognition memory. Journal of Experimental Psychology, 103(4), 751–757. https://doi.org/10.1037/h0037190
Craik, F. I. M. (2002). Levels of processing: Past, present... and future? Memory, 10(5–6), 305–318. https://pubmed.ncbi.nlm.nih.gov/12396643/
Craik, F. I. M., & Lockhart, R. S. (1972). Levels of processing: A framework for memory research. Journal of Verbal Learning and Verbal Behavior, 11(6), 671–684. https://www.sciencedirect.com/science/article/pii/S002253717280001X
Craik, F. I. M., & Rose, N. S. (2012). Memory encoding and aging: A neurocognitive perspective. Neuroscience & Biobehavioral Reviews, 36(7), 1729–1739. https://pubmed.ncbi.nlm.nih.gov/22155274/
Craik, F. I. M., & Tulving, E. (1975). Depth of processing and the retention of words in episodic memory. Journal of Experimental Psychology: General, 104(3), 268–294. https://doi.org/10.1037/0096-3445.104.3.268
Craik, F. I. M., & Watkins, M. J. (1973). The role of rehearsal in short-term memory. Journal of Verbal Learning and Verbal Behavior, 12(6), 599–607. https://www.sciencedirect.com/science/article/pii/S0022537173800398
Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students' learning with effective learning techniques: Promising directions from cognitive and educational psychology. Psychological Science in the Public Interest, 14(1), 4–58. https://doi.org/10.1177/1529100612453266
Eysenck, M. W. (1978). Levels of processing: A critique. British Journal of Psychology, 69(2), 157–169. https://doi.org/10.1111/j.2044-8295.1978.tb01643.x
Gallo, D. A., Meadow, N. G., Johnson, E. L., & Foster, K. T. (2008). Deep levels of processing elicit a distinctiveness heuristic: Evidence from the criterial recollection task. Journal of Memory and Language, 58(4), 1095–1111. https://doi.org/10.1016/j.jml.2007.12.001
Hyde, T. S., & Jenkins, J. J. (1969). The differential effects of incidental tasks on the organization of recall of a list of highly associated words. Journal of Experimental Psychology, 82(3), 472–481. https://doi.org/10.1037/h0028372
Kim, H. (2024). Neural and cognitive dynamics leading to the formation of strong memories: A meta-analysis and the SAM model. Imaging Neuroscience, 2, imag-2-00098. https://pmc.ncbi.nlm.nih.gov/articles/PMC12224446/
Morris, C. D., Bransford, J. D., & Franks, J. J. (1977). Levels of processing versus transfer appropriate processing. Journal of Verbal Learning and Verbal Behavior, 16(5), 519–533. https://www.sciencedirect.com/science/article/pii/S0022537177800169
Paller, K. A., & Wagner, A. D. (2002). Observing the transformation of experience into memory. Trends in Cognitive Sciences, 6(2), 93–102. https://pubmed.ncbi.nlm.nih.gov/15866193/
Roediger, H. L., III, & Karpicke, J. D. (2006). Test-enhanced learning: Taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. https://doi.org/10.1111/j.1467-9280.2006.01693.x
Rogers, T. B., Kuiper, N. A., & Kirker, W. S. (1977). Self-reference and the encoding of personal information. Journal of Personality and Social Psychology, 35(9), 677–688. https://pubmed.ncbi.nlm.nih.gov/909043/
Rose, N. S., & Craik, F. I. M. (2012). A processing approach to the working memory/long-term memory distinction: Evidence from the levels-of-processing span task. Journal of Experimental Psychology: Learning, Memory, and Cognition, 38(4), 1019–1029. https://doi.org/10.1037/a0026976
Symons, C. S., & Johnson, B. T. (1997). The self-reference effect in memory: A meta-analysis. Psychological Bulletin, 121(3), 371–394. https://pubmed.ncbi.nlm.nih.gov/9136641/
Tulving, E., & Thomson, D. M. (1973). Encoding specificity and retrieval processes in episodic memory. Psychological Review, 80(5), 352–373. https://doi.org/10.1037/h0020071
Wagner, A. D., Schacter, D. L., Rotte, M., Koutstaal, W., Maril, A., Dale, A. M., Rosen, B. R., & Buckner, R. L. (1998). Building memories: Remembering and forgetting of verbal experiences as predicted by brain activity. Science, 281(5380), 1188–1191. https://doi.org/10.1126/science.281.5380.1188