levels of processing
preference for incidental orienting tasks
When researchers tell people “try to remember this” (intentional instructions), they lose control because everyone uses different strategies, and that affects memory results.
If researchers give a specific task like “count the vowels” (orienting task), they have more control because they know what mental process is happening and can check it.
Incidental instructions are tasks that feel like everyday thinking, so they make experiments more realistic and useful for understanding real-life memory.
The goal is to design experiments that are both controlled and similar to real-world remembering, so the findings actually matter outside the lab.
critical assessments of levels of processing
Reg. explaination
Conceptual Criticisms
Circular reasoning risk: Depth of processing was defined by memory outcome, so well-remembered items were assumed to be “deeply processed” without an independent measure.
No independent index of depth: This made the theory hard to test objectively.
Baddeley’s critique:
Questioned the usefulness of broad functional principles; preferred focusing on mechanisms tied to brain processes.
Challenged the idea of a single linear sequence of processing levels and noted lack of evidence for more levels within semantic processing.
General criticisms:
Theory was not falsifiable, making it unscientific.
Ideas were too vague—critics argued clarity was missing.
Empirical Criticisms
Doubts about two types of rehearsal: Maintenance processing might still strengthen memory.
Evidence showed sensory information can persist longer than Craik & Lockhart claimed.
Neuropsychological findings conflicted with the theory:
Amnesic patients can process deeply (understand conversations) but still fail to remember, contradicting the depth-memory link.
High schooler explaination
the first big problem with the theory is circular reasoning: it says “deep processing makes memory better,” but then assumes anything remembered well must have been deeply processed. That’s like saying “good players score more points” and then deciding anyone who scores a lot must be a good player—without another way to measure skill.
There’s no independent way to measure depth, which makes the theory hard to test scientifically.
Some psychologists, like Baddeley, argued we should focus on specific brain mechanisms instead of vague ideas about “levels.” He also said it’s unrealistic to think there’s one straight line of processing levels and pointed out there’s no proof of extra levels beyond meaning (semantic).
Other critics said the theory was too vague and not falsifiable, meaning you couldn’t design an experiment to prove it wrong—so it wasn’t truly scientific.
Experiments raised doubts too:
Rehearsing information (even simple repetition) might help memory more than the theory claimed.
Sensory information sometimes lasts longer than expected.
Brain studies showed contradictions: for example, amnesic patients can understand conversations (deep processing) but still forget them, which breaks the link between depth and memory
cirularity and the need for an independent index of depth
Passage: “Rather, the accusation of circularity arises in relation to within-domain differences such as those obtained between items that warrant "yes" and "no" answers to orienting questions, both of which obviously entail semantic-level analysis, or to differences between qualitatively distinct orienting tasks, say between evaluating synonymity of adjectives and evaluating those same adjectives as self-descriptors. In these cases, there appears to be no index of depth other than through the post hoc observation of their differential effects on memory performance.“
Explaination: The criticism is about circular reasoning when comparing tasks that both involve deep (semantic) processing but still produce different memory results.
For example:
One task asks, “Does this word mean the same as another word?” (synonym check).
Another asks, “Does this word describe you?” (self-description).
Both tasks require thinking about meaning (semantic level), yet people remember words differently depending on the task.
The problem: There’s no independent way to measure which task is “deeper” except by looking at memory performance afterward. That makes the theory circular—depth is judged by memory results, not by an objective measure.
Passage: “A measure of the depth of processing involved in performing some task would be the numerical output from some function applied to an explicit model of the processes involved in performing that task. For example, a numerical index of the depth of processing involved in deciding that the word dog refers to an animal could only be obtained from an explicit model of the operations on semantic memory needed to answer questions of category membership. Hence a complete definition and quantification of terms such as depth or level is appropriately thought of as the end product of research, not the starting point. In this sense, our original use of these terms was a tentative label lo characterize processes that require further explication and specification through ongoing research.“
Explaination: The authors are saying: if you want a real measure of “depth of processing”, you’d need a mathematical model of how the brain processes information during a task.
For example, if you ask, “Does the word dog mean an animal?” you’d need a model of all the mental steps involved in checking category membership in your memory. Then you could calculate a number that represents “depth.”
Because we don’t have such detailed models yet, we can’t fully define or measure depth right now.
So when Craik and Lockhart first used terms like “depth” or “level,” they were just rough labels, not precise scientific measures. The goal was to study and refine these ideas through research over time.
In short: Depth of processing isn’t something you can measure directly yet—it’s a concept that needs more research and modeling before it becomes a precise scientific term.
Passage: “There are many examples of this ongoing development and refinement, some of which will be discussed below. For present purposes the ease of self-reference. orienting tasks provides a good example of the sense in which depth of processing is a general concept needing to be explicated, rather than a well-defined explanatory construct. The self-reference phenomenon first reported by Rogers, Kuiper, and Kirker (1977) is interesting for several reasons. In the first place the high level of performance produced by this incidental orienting task, relative either to intentional instructions or to other semantic orienting tasks, illustrates the point that the relation between memory performance and various forms of semantic-level processing is not a simple one. It clearly represents little advancement in explanatory understanding to stale that the superiority of the self-reference orienting task is a consequence of it requiring deeper levels of processing; it is scarcely better than saying lhat it yields a stronger trace or that it increases the probability of the item entering a particular memory store.”
Explaination: The authors are saying that self-reference tasks (like asking “Does this word describe you?”) show why “depth of processing” is still a broad idea, not a precise scientific term.
These tasks lead to very strong memory performance, even better than other semantic tasks or intentional memorization.
But here’s the problem: simply saying “it works because it’s deeper processing” doesn’t really explain anything. That’s like saying “it works because it’s better”—it’s vague and doesn’t tell us why.
The point is: the relationship between memory and different kinds of meaning-based processing is complex, and calling one task “deeper” isn’t enough to give a real scientific explanation.
In short: Self-reference tasks show that depth of processing is a useful idea, but not yet a clear, detailed theory. Just saying “deeper = better memory” doesn’t explain why self-reference works so well.
Passage: “Moreover, notice that it was levels of processing as a research framework that did much to stimulate the collection of such data in ihe first place. The self-reference phenomenon comprises precisely the kind of data needed to build a comprehensive theory of memory encoding; and it was the levels framework that stressed that its explanation was to be found in an analysis of the particular meaning-extracting opera- tions involved in performing the orienting task. Obviously such an analysis, if it is to isolate those principles that arc important for remembering, may require con- siderable research into the structure of the self-concept, as well as into how such judgements are made; but, when completed, the achievement will have been to provide an explication of what was initially meant by depth. Such an analysis is well under way (e.g., Bellezza, 1984; S.B. Klein & Kihlstrom, 1986). It may well be that self-reference is effective because performing the comparison judgement effectively structures the elements into an organized set, as S.B. Klein and Kihlstrom have argued; but as a research framework, levels of processing is entirely neutral on the matter. Finally, notice how such a research programme provides yet another example of the ongoing integration of memory research into the broad context of cognition.”
Explaination: The authors are saying that the levels of processing framework was important because it encouraged researchers to collect data like the self-reference effect.
The self-reference phenomenon (remembering words better when you relate them to yourself) is exactly the kind of finding needed to build a full theory of how memory encoding works.
The framework suggests that the explanation lies in analyzing the specific mental steps used in the task—like how we judge whether a word describes us.
To do this properly, researchers need to study things like the structure of the self-concept and how those judgments are made. When that research is complete, it will clarify what “depth” really means.
Some researchers (e.g., Klein & Kihlstrom) think self-reference works because it organizes information into a structured set, but the levels framework doesn’t take sides—it’s just a starting point for research.
Finally, this kind of work shows how memory research is becoming part of the bigger picture of cognition (how we think and process information).
In short: The levels of processing idea wasn’t a final answer—it was a research roadmap. Self-reference tasks are helping scientists figure out what “depth” really means and how memory connects to broader thinking processes.
types of rehearsal
Passage: “The evidence for an increase in recognition memory is clcarcut. One possibility here (discussed by Greene, 1987, and by Navch-Benjamin & Jonides, 1984) is that recall depends primarily on intcritem elaboration, whereas recognition is incremented by intraitem integration (Mandler, 1979); if maintenance rehearsal leads to little intcritem processing, the observed result would be expected.”
Explaination: The passage is talking about two types of memory:
Recall: When you have to bring information back from memory without any hints (like answering an essay question).
Recognition: When you just need to recognize something you’ve seen before (like picking the right answer on a multiple-choice test).
Researchers noticed that recognition memory tends to improve more than recall in certain situations. Why?
One theory is:
Recall works best when you make connections between different items you’re learning (this is called interitem elaboration).
Recognition improves when you combine details within each individual item (this is called intraitem integration).
If you only repeat the information over and over (called maintenance rehearsal), you’re not really making connections between items. So recall doesn’t improve much—but recognition does, because repeating helps you integrate details within each item.
Interitem Elaboration
This means making connections between different items you’re learning.
Example:
You’re studying vocabulary words: “photosynthesis” and “chlorophyll.”
Instead of just memorizing each word, you think:
“Photosynthesis uses chlorophyll to capture sunlight.”
You’ve linked the two items together, which helps recall because you built a network of relationships.
Other examples:
Linking historical events: “The Industrial Revolution led to urbanization, which influenced labor movements.”
Connecting math concepts: “The Pythagorean theorem relates to distance formulas in coordinate geometry.”
Intraitem Integration
This means combining details within one item to make it more meaningful.
Example:
You’re learning the word “photosynthesis.”
You break it down: “Photo = light, synthesis = putting together.”
Now the word itself feels more integrated and easier to recognize later.
Other examples:
For a chemical formula like H₂O, you think: “H means hydrogen, O means oxygen, and together they make water.”
For a historical date: “1776 = 1 Declaration of Independence, 7 letters in ‘freedom,’ 76 trombones in the parade.” (a mnemonic inside the item).
long term retention of sensory features
Passage: “It seems clear at this point that our original suggestions were again too simple. Surface information is often lost rapidly, but there arc also cases in which a record of surface form clearly persists to affect later performance over long retention intervals. We can offer a few comments from the perspective of levels of processing. First, some of the cases involving the dramatically long-term retention of surface form do not involve the explicit retrieval of surface information, but involve the implicit use of such information to facilitate current performance. Kolcrs's (1979) demonstrations of long-lasting information regarding transformed typographies fall into this category. In general, it now seems that many perceptual memory (priming) tasks are data-driven and are sensitive to modality-specific information (e.g., Roediger, Wcldon, & Challis, 1989). However, even if such modality-specific information cannot be explicitly retrieved, it must be represented in some manner to affect performance on implicit memory tasks, and such findings are incompatible with our original statement.”
Explaination: The authors are revising their earlier idea because it turned out to be too simple. Originally, they thought surface information (like the exact wording or appearance of something) disappears quickly from memory. That’s often true—but there are exceptions where surface details stick around and influence performance even after a long time.
They explain this using the levels of processing perspective:
In some cases, people don’t consciously recall surface details, but those details still help them perform tasks without realizing it. This is called implicit memory.
For example, Kolers (1979) showed that people could remember unusual text formats (like transformed typography) for a long time, even if they weren’t aware of it.
Many priming tasks (where past exposure helps you respond faster later) rely on these hidden surface details. These tasks are data-driven and depend on specific sensory features (like the way something looked or sounded).
So even if you can’t explicitly recall those details, they must still be stored in some way because they affect performance. This finding contradicts the authors’ original claim that surface information is always lost quickly.
Transfer Appropriate Processing: Transfer-appropriate processing (TAP) is a psychological theory stating that memory performance is best when the cognitive processes used during retrieval match the processes used during encoding. This means that the way you study should mirror the way you will be tested to improve recall. For example, if you are studying for a multiple-choice test, you should practice with multiple-choice questions rather than just reading the material.
levels of processing and retrieval
Passage: “One is that relative to shallow encoding, deep processing dccontextualizcs retrieval. That is, granted that retrieval is never totally independent of the retrieval context, deep processing sustains retrieval over a wider range of retrieval conditions and contexts: Retrieval becomes (relatively) robust against changes in the context and form of the retrieval. If this is true, then an interesting research question is to establish the degree to which various encoding operations render retrieval contextually robust. Certain operations may make the retrieval highly dependent on the recapitulation of the context, others may make retrieval independent of context. Such differences in contextual robustness may exist quite apart from the level of performance: For some encoding operations, retrieval may be uniformly low, regardless of the cuing conditions for retrieval, or it may be uniformly high; for other operations, it may be high if certain retrieval conditions hold, but poor otherwise.
The second possibility is that deep processing boosts retrieval, but such boosting is highly dependent on retrieval context so that levels effects are greatly reduced if retrieval conditions differ too markedly from what is appropriate relative to the encoding. Put differently, transfer appropriate processing is correct; but if the form of processing is deep, the increment in performance when it is also appropriate is greater than the increment for shallow processing, even when it too is paired with appropriate retrieval conditions. Fisher and Craik's (1977) finding that semantic processing tested with semantic cues yields higher recall than rhyme processing tested with rhyme cues supports this second possibility, although as suggested above, some deep encoding operations may serve to make retrieval relatively context independent.”
Explaination:
First Paragraph
Main idea: Deep processing (thinking about meaning, making connections) makes retrieval more context-independent compared to shallow processing (surface-level features like sound or appearance).
Why? Because retrieval is never completely free from context, but deep processing tends to make memory more robust across different contexts.
Implication: Researchers want to know how different encoding strategies affect this robustness. Some strategies might make recall very dependent on the original context, while others make it less so.
Example: If you learn a word by its meaning (deep processing), you might recall it even if the test environment changes. If you learn it by its sound (shallow processing), you might only recall it well if the test also focuses on sound.
Second Paragraph
Main idea: Deep processing improves retrieval, but only when the retrieval context matches the encoding context—this is called transfer-appropriate processing.
Key point: If retrieval conditions differ too much from encoding, the advantage of deep processing shrinks.
Example: Fisher & Craik (1977) found that if you encode words semantically and then get semantic cues at test, recall is high. If you encode by rhyme and get rhyme cues, recall is also good. But if the cue type doesn’t match the encoding type, performance drops.
Nuance: Even with deep processing, context still matters—though sometimes deep encoding can make retrieval less context-dependent.
In short:
First possibility: Deep processing makes memory strong across many contexts (context-independent).
Second possibility: Deep processing helps most when retrieval matches encoding (context-dependent), but the boost is bigger than for shallow processing when matched.