Processing quality is the thinking a session actually produces, and strategies are the workflows for producing it. Meta-strategies sit above both: the response decided in advance for the moment a method turns expensive. The three names are one lens with three questions in it, and in practice a session rarely separates cleanly into thinking that was weak and a workflow that was weak. The lens earns its overhead at one moment in particular. A method that is working starts to cost, the cost gets read as evidence that nothing is going in, and something cheaper gets picked up instead. Learners who experienced a strategy as more effortful rated it less effective and were less likely to choose it again, in conditions where that strategy was producing better retention. A method gets dropped on how it feels while it runs rather than on the results it produced. All of this sits inside the ICS System, which holds the five trainable capabilities and the encoding-and-retrieval loop they run in. The three levels decide which of those capabilities a failed session actually needed.

The Retreat Moment

The retreat has a recognisable shape. A workflow starts producing — comparisons get made, the map gets messy, holding the whole thing takes real attention — and the load climbs past comfort. What happens next arrives without feeling like a decision. The page gets tidier, the highlighting resumes, the video goes back on, and the session ends looking productive. Those cheap familiar moves are the lower-order shortcuts: re-reading, copying out, highlighting, watching it again. They are also what students name most often when surveyed about their own study, and they sit near the bottom of the utility rankings once the same set of techniques is compared on delayed tests.

The response to the retreat is written down before the session, because the moment it fires is the moment deliberation has already failed. The line gets decided while nothing is costing anything: if I feel the pull to re-read, then I close the source and write the question. Plans in that if-then shape carry an average effect of d = .65 across 94 independent tests, measured on top of already holding the goal. A diagnosis asks for reasoning at the exact point reasoning has gone missing, while a line written in advance only has to be recognised.

The Shortcut Problem covers the substitution itself, from why the cheaper move gets chosen to what it costs later. Cognitive Load & What Mental Effort Is Trying to Cue reads the effort as information about how much is being carried at once, and that reading is what keeps a method running through its expensive stretch.

What the Session Produced

Processing quality names the kind of cognitive work actually happening during a study activity. That covers comparing, judging importance, chunking, spotting relationships, following causes into effects, reasoning about conditions, reconstructing, explaining, and carrying a rule to a case that was not in the source. None of it is visible from outside. Flashcards, a mapping session, a summary, a highlighter pass, a set of questions: every one of them is a container, and the same container holds unrelated kinds of work on two different days. That is why highlighting and re-reading rate low on delayed tests while explanation rates moderately well — prompting a learner to explain material to themselves moves performance by about half a standard deviation — and why deeper encoding beats surface encoding regardless of how long the surface work took. Adding techniques on top of weak thinking makes a session noisier without making it better, because the count of methods is not the variable that moves outcomes.

Deep Processing is the capability underneath all of it, the one that thinks about material while encoding it instead of moving it from one page to another, and it names the end state too: Knowledge Mastery is material that can be recalled cold, related to its neighbours, judged for importance, and applied to a case that never appeared in the source. Importance-Based Chunking handles one sub-move of that, grouping information by what each piece does and what it changes in place of the order the source happened to use. Deep Processing Practice is an index of practice moves, a short list to pull from on a session that needs one. And the line between work and the appearance of work is the whole subject of Are You Thinking, or Just Consuming?, which is the fastest check available on a session that looked busy.

Workflows and Layers

A workflow creates the conditions for that thinking and guarantees none of it. The form of the activity decides the outcome, and The Technique Is Only as Good as the Thinking It Produces makes that case at full length. Three workflows carry most of the load. Prestudy is a short, deliberately shallow pass before the main session, run to build a crude frame and generate questions rather than to understand anything. The Bear Hunter System is the encoding workflow proper: three passes that turn material into a structure holding up with the source closed. Spaced Interleaved Retrieval is the review schedule, and it runs on three rules — wait until recall is effortful, mix topics inside a session instead of blocking them, and treat every gap a recall exposes as the next thing to study. Prestudy, BHS, and SIR runs all three together as one loop, end to end.

The three encoding passes are layers rather than stages: each one crosses the whole topic and corrects the one before it. Aim writes down the questions the material has to answer before the source is open, so attention has a target to fill. Shoot reads against those questions and builds a rough map of how the answers connect, on the understanding that parts of it will be wrong. Skin cuts that map down to the structure that survives without the source open. A first layer that turns out wrong has still done its job, which is to give later detail somewhere to sit. What a crude frame buys is relevance, so the load of the next pass goes on structure instead of on facts with nowhere to go, and each pass exposes what the one before it missed.

The stack has a boundary condition. On material a frame already exists for, the prestudy pass comes out and retrieval starts immediately, because scaffolding that helps a novice measurably hurts a learner who already holds the schema — the support becomes extra material to process. That boundary also prices the stack. A prestudy pass adds a session before the session, and three encoding passes mean touching one topic three times before any of it is retrievable, so the full stack belongs on material that will still be needed months from now rather than on a reading due tomorrow. Interleaving costs something inside the session too: mixing topics depresses performance while practising and improves retention at a delay, so more errors in the room and faster re-entry a week later is the expected shape rather than a sign the schedule has gone wrong.

Questions Before Material

Sorting problems by deep structure runs downstream of knowledge. Novices group physics problems by surface features precisely because they lack the schemas that would let them group by principle, and thinking skills of that kind do not detach from the domain they were learned in. What transfers early is the question. Asking what this depends on and what would have to be true for it to fail points attention at structure, and structure is what schemas get built out of. The pattern of questions improves before the knowledge does, and new information then has somewhere to land.

The same order is the case against consuming first and organising later. Without rehearsal, recall of a three-letter string falls under 10% by 18 seconds. The loss is now read as interference rather than pure decay, and rehearsal or chunking stretches the window considerably, but the order of magnitude holds. Material arriving at a mind that is not primed to catch it is largely gone before there is anything to organise, which is the work a prestudy pass and a set of pre-written questions are doing.

From Overwhelm to a Question

Overwhelm tracks one variable: how many elements have to be held at the same time because they interact. A hundred unrelated facts sit lighter than six that only make sense together. That makes the feeling readable as a description of the material rather than a verdict on the person holding it, which is what turns it into a cue.

The weak responses all shrink the problem instead of the confusion — dropping the relationship and studying the parts separately, copying an explanation across, asking for the answer before a guess exists, simplifying until the structure that mattered has gone. The useful response converts the feeling into one question precise enough to be answered.

"I'm confused about this chapter."        nothing to act on
"I don't understand why A leads to B."    closer
"I don't understand why A leads to B
 under condition C, when B follows on
 its own everywhere else."                one question, one next move

Guessing before checking earns its place on the evidence. Unsuccessful retrieval attempts improve later learning of the answer, and errors held with high confidence get corrected more readily than low-confidence ones. All of that is conditional on the correction landing. An open guess is a debt, closed in the same session or the error is what gets encoded — a high-confidence error comes back if the correction is later forgotten, so the check is the part that cannot be skipped.

Noticing the shift at all is trainable, and Building the Radar is about catching, while it is happening, that attention has gone passive or a method has quietly turned mechanical. What to do once it is caught belongs to Self-Regulation, the capability of steering a drifting session back, as distinct from choosing the right method at the start. Programmes that train it move strategy use by .72 in primary and .88 in secondary students, and metacognitive prompts issued during a task move self-regulated activity by about half a standard deviation. The trainable part is the noticing and the return, which is what those numbers are measuring.

The outcome names the thinking it requires.
The thinking names the workflow that forces it.
The workflow generates overwhelm as a by-product of holding relationships.
The overwhelm converts into one written question.
The question closes in the same session, and its answer corrects the structure.

Retrieval Beside Encoding

Retrieval runs alongside the three levels. Encoding is the selective half, filtering what gets stored before it is stored, and retrieval tests and strengthens whatever made it through. Neither substitutes for the other: retrieval is bounded by what was encoded in the first place, and encoding without retrieval decays unnoticed because nothing checks it.

Retrieval has the better cost-to-yield ratio. Practice testing rates high utility in the largest comparison of study techniques, above nearly everything on the encoding side, and it needs no setup beyond closing the source. Repeated restudy wins at five minutes and loses at a week, which is why it keeps getting chosen and why the comparison has to be made at a delay. That reversal is the argument for building the retrieval half early, while the encoding workflow is still clumsy: it is the cheapest available path into high-quality thinking.

Checks That Catch Self-Deception

The easy metrics all count the same kind of thing. Pages covered, cards completed, notes written, hours logged, questions answered — every one of them measures activity, and activity is not the variable. Repeated re-reading raises confidence while lowering recall a week later, so volume metrics track the exact thing that misleads.

Three checks work, and the trade between them is why all three are needed. The subjective read is fast and free, a daily sense of whether the session felt connected and organised, and it is the least trustworthy of the three. The uncalibrated objective check is real output judged by a self-set standard: questions written and answered, a topic taught to someone, something built. The calibrated objective check is output judged by the standard that will do the judging in the end — a past paper with the official mark scheme, a rubric, a real performance — and it is slow enough that it cannot carry the daily load.

A session that felt smooth and connected is the one whose objective check should come sooner. Judgments of learning inflate in a predictable direction: they run high when the answer is present during study and absent at test, and the smoothest sessions are the ones where the answer sat in view throughout. Without a stated frequency the three checks stay a list, and a list does not get run. Even an arbitrary default starts it — subjective at the end of every session, uncalibrated once a week, calibrated at every real opportunity that exists. Whether a technique is producing the thinking it was chosen for is a separate check again, and Are You Learning, or Just Using Techniques is where that test is written out.

Problem Maps Over Labels

Complaints like “I procrastinate”, “I have bad time management”, “I am inconsistent”, or “this topic is confusing” name a symptom and stop. A problem map is the written breakdown underneath one: the conditions it shows up under, what sets it off, which feeling arrives with it, what gets done next, which variables move together, and which one gets changed first. The writing is where the relief comes from, since putting the structure of a problem outside the head measurably lowers the demand of holding it, the same offloading that makes a written question cheaper than a remembered one.

Seeing the problem clearly does not finish it. Holding a clear goal does not by itself produce the behaviour, and the distance between intending and doing is the best-documented failure in this area: closing it takes the same if-then line that closes the retreat, worth d = .65 across 94 tests on top of the goal already being held. The last line of a problem map is that sentence, one trigger and one response, written before the situation arrives.

Diagnosing a Failed Session

What the session felt like	Which question it raises	What changes next session
Two hours in and nothing joined up.	Processing quality	One comparison, one chunking pass, or one out-loud explanation before any new material.
I know what good work looks like here and have no way of producing it.	Strategy	The prestudy pass and the three encoding passes, run end to end on one topic.
The method works and I stop using it on the hard days.	The retreat	The if-then line, written before the next session, naming the exact pull.
The notes look right and nothing can be done with them.	Processing quality and strategy together	The map rebuilt around relationships, then retrieved from with the source closed.
More material keeps going in instead of getting worked on.	The retreat	The next hour of intake converted into five written questions, answered cold.

The Case Against

The three levels are not three mechanisms. Meta-strategies reduce to Self-Regulation pointed at a strategy, with no separate machinery underneath. Running the diagnosis on every session costs more than it returns, because the attention goes on classifying the session instead of on the material, and the classification changes nothing on a session that was already working. The five-capability view in Dimensions of Learning carries the same risk and the same repair, since a dimension is one of the separately trainable capabilities that set learning performance, where the weakest one caps the rest, and naming the weakest earns its keep only when the next move changes as a result.

No study sets a threshold here, so the working rule of thumb is two sessions: if naming the level has not changed what the next session does, twice running, the diagnosis stops and the next concrete piece of work starts instead. The table comes back when the same failure repeats a third time.

How technique failure distributes across the three levels is unmeasured, and nobody has counted it. What is documented is narrower and sharper — effort gets read as evidence of poor learning, and that reading is where an otherwise-working method dies. That single moment is what the three levels are worth running for, and the sentence written before the session is what carries a method through it.

Open Questions

• Which method in current use costs the most before any of it pays back?
• Which cheaper move gets picked up first when the retreat happens?
• What would name the confusion while there is still session left to use it?
• Which recurring complaint has never been written out with its conditions and triggers?

Sources

• Kirk-Johnson, A., Galla, B. M., & Fraundorf, S. H. (2019). Perceiving effort as poor learning: the misinterpreted-effort hypothesis. Cognitive Psychology, 115. pubmed.ncbi.nlm.nih.gov/31470194
• Dunlosky, J., Rawson, K. A., Marsh, E. J., Nathan, M. J., & Willingham, D. T. (2013). Improving students’ learning with effective learning techniques. Psychological Science in the Public Interest, 14(1), 4–58. journals.sagepub.com
• Roediger, H. L., & Karpicke, J. D. (2006). Test-enhanced learning: taking memory tests improves long-term retention. Psychological Science, 17(3), 249–255. pubmed.ncbi.nlm.nih.gov/16507066
• Bjork, E. L., & Bjork, R. A. (2011). Making things hard on yourself, but in a good way: creating desirable difficulties to enhance learning. unh.edu (PDF)
• Gollwitzer, P. M., & Sheeran, P. (2006). Implementation intentions and goal achievement: a meta-analysis of effects and processes. Advances in Experimental Social Psychology, 38, 69–119. sciencedirect.com
• Kornell, N., Hays, M. J., & Bjork, R. A. (2009). Unsuccessful retrieval attempts enhance subsequent learning. JEP: LMC, 35(4), 989–998. web.williams.edu (PDF)
• Richland, L. E., Kornell, N., & Kao, L. S. (2009). The pretesting effect: do unsuccessful retrieval attempts enhance learning? learninglab.uchicago.edu (PDF)
• Metcalfe, J. (2017). Learning from errors. Annual Review of Psychology, 68, 465–489. annualreviews.org · free copy (PDF)
• Koriat, A., & Bjork, R. A. (2005). Illusions of competence in monitoring one’s knowledge during study. JEP: LMC, 31(2), 187–194. bjorklab.psych.ucla.edu (PDF)
• Peterson, L. R., & Peterson, M. J. (1959). Short-term retention of individual verbal items. JEP, 58(3), 193–198. summary carrying the 3/6/18-second figures
• Chi, M. T. H., Feltovich, P. J., & Glaser, R. (1981). Categorization and representation of physics problems by experts and novices. Cognitive Science, 5(2), 121–152. onlinelibrary.wiley.com
• Willingham, D. T. (2007). Critical thinking: why is it so hard to teach? American Educator. aft.org
• Tricot, A., & Sweller, J. (2014). Domain-specific knowledge and why teaching generic skills does not work. Educational Psychology Review, 26, 265–283. link.springer.com
• Sweller, J., van Merriënboer, J. J. G., & Paas, F. (2019). Cognitive architecture and instructional design: 20 years later. Educational Psychology Review, 31, 261–292. link.springer.com
• 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. doi.org
• Bisra, K., Liu, Q., Nesbit, J. C., Salimi, F., & Winne, P. H. (2018). Inducing self-explanation: a meta-analysis. Educational Psychology Review, 30, 703–725. link.springer.com
• Guo, L. (2022). Using metacognitive prompts to enhance self-regulated learning and learning outcomes: a meta-analysis. Journal of Computer Assisted Learning. onlinelibrary.wiley.com
• Dignath, C., & Büttner, G. (2008). Components of fostering self-regulated learning among students: a meta-analysis. Metacognition and Learning, 3, 231–264. link.springer.com
• Risko, E. F., & Gilbert, S. J. (2016). Cognitive offloading. Trends in Cognitive Sciences, 20(9), 676–688. cell.com