Importance-Based Chunking

Importance-based chunking is grouping the pieces of a topic by why each one matters: the role it plays, and what it changes about the rest. The order the pieces arrived in, and the headings they arrived under, are not the basis. The comparison that settles which reason matters is where the material gets encoded. A structure copied from the headings the material arrived under rearranges the same pieces without doing that comparison.

A chunk is a decision about the pattern a set of ideas will sit under. Meaningfulness is how familiar that pattern already is. The decision holds when the grouping survives the feeling that it is not right yet. Importance-based chunking is part of Deep Processing, the processing that decides how much has to be memorised later.

Core takeaways

• A group is made by asking why its members matter. The first answer is usually not the deepest one; asking what each reason is in turn for gives a broader group that more of the topic can attach to.
• A good grouping is relevant to a goal that can be stated, connected to what else is being learned now, and intuitive because it uses a pattern already held. Seeing one member should let the next members name themselves.
• More than four branches off one point means the grouping is not finished, and a label whose membership has to be memorised is doing no work.
• A good grouping is found only after a period of confusion. Confusion ends in one of two ways: the material was organised so it makes sense, or the thinking stopped. A grouping that feels obvious without the confusion first is usually the second.
• How well someone groups is set by two things: how comfortable they are sitting in confusion, and how good the questions they make from it are. Prior knowledge does not set it.
• The first twenty to thirty minutes on a topic go to its three to six main groups and how they relate. Rote memorising waits until connecting has been tried and has failed.

What makes a group important

The question being asked is why is this important? It is a different question from what do these share? Shared features are found by analysis. Importance is settled by evaluation. The structure that evaluation is aiming at is the one that would make the topic easiest to understand, retrieve, and use.

The basis of that importance changes by subject. Function is the right basis for tissue types. A group of chemical reactions often matters because they share a mechanism or a reactivity, and function is beside the point there. A literature topic can turn on the writer’s position or background. The same piece can sit in more than one group depending on the filter: function, mechanism, failure mode, application context, importance. The choice between two valid bases is which structure needs less memorizing for the person holding it.

How material is presented decides how it gets grouped. The grouping a textbook or a lecture hands over is usually sound and intuitive, and that is why it feels like the only one. That given structure is well above grouping by first letter, and still below what a later pass can reach.

Weak:

Epithelium arrives as three headed lists: cell types (flat, cube-shaped, tall),
cell arrangements (single-layered, multi-layered, one that looks multi-layered
and is not), and important linings, each with a function and a site.

Strong:

The linings are combinations of a type and an arrangement, so the third list
falls out of the first two. Grouping by why the distinctions exist puts the
functions on the combinations instead of beside them: protection where surfaces
abrade, exchange where material must cross, movement where material is swept.

Rearranging the given categories is not the same as answering why they exist. A map built from type against arrangement is already better than the three headings, and it still tangles, because the grouping is a rearrangement rather than an answer.

What makes one grouping better than another

The words picked for a group are the choice of schema. Naming groups is higher-order work rather than clerical work: each label files the set under one pattern and leaves the others unused. Without a pattern, material fades in days to weeks and survives only by being called up again, which is what repetition-only study is doing.

The best pattern to form is three things at once. It is relevant to a real goal that can be stated: the exam, the work, the project. It is internally connected to the other things being learned in this same domain right now. It is intuitive, meaning already inside what is held, a way of thinking already familiar. Meaningfulness describes the person doing the filing. It is not a property of the topic. A pattern becomes meaningful in two ways. Either it has been worked with often enough to feel familiar, or it is a direct extension of something already held, so the new item slots in without a stop.

A mnemonic is a forced pattern and sits below that trio; it is better than nothing. Pure repetition is not a schema at all. No connection forms, so it sits beneath the scale rather than at the bottom of it. Holding the top of the rubric every time is not realistic. The work moves down the scale deliberately when it must.

Seeing one piece of a group should let the next pieces name themselves.

That is the derivability test. It is run between candidate structures, and it is not run against one structure in isolation. Name a group “before” and the next names itself; nobody would name the next group “plastic”. Drivers or triggers point to actions or responses. Mechanisms and drivers point to prerequisites, then to responses. Name a group “factors” and the next name is already constrained.

A mnemonic fails the same test. An acronym over a five-item emergency checklist, spelling an ordinary word that has nothing to do with the situation, whose items do not connect to each other, has one merit: it is pronounceable. The label tells nothing about what the next item must be.

Technical terms are names, and they are not groups. Group by the underlying idea and memorise the term as the name of that idea. Letting the terms be the groups raises the memorising load, adds retrieval failure points, and is usually slower. All three follow from the same cut. Sibling branches have to match on how intuitive they are. If one branch of a split is ordinary language and its sibling is in jargon, the split is wrong. The mismatch is visible with no subject knowledge at all.

Two candidate structures can hold the same content and make different things memorable. Two branches converging on one outcome emphasises that both routes end there. A backbone with branches off the side emphasises the process. There is often no right answer, and deciding between them is the work. Derivability is what makes that comparison operable.

Similarity is not enough to form the group. The group forms on a judgement that the similarity is important enough to build on. Importance decides which similarity to group on. Meaningfulness decides whether the resulting label is one that person can hold. Done properly, grouping by importance produces intuitive groups, so checking whether the groups feel intuitive is how the importance work gets checked. One reason a size ceiling exists: three tags across a thousand items returns three hundred per tag, which is no better than not filing at all. Groups have to be specific enough to separate things.

Immediate reasons and root reasons

The usual stall comes after the importance question has been asked: the first answer gets accepted. A root reason is the fundamental reason something matters, the way a root cause is the fundamental reason something happens. Each time a reason arrives, the next question is what that in turn is for. The chain of questions stops when the question stops making sense.

Genetic diversity matters because it produces variation.
A chunk built there is called variation.
Why does variation matter? Adaptability, and the long-run survival of a population.
The chunk that gets built is survival.

The deeper chunk is broader, so it absorbs more sub-concepts, and it connects outward more easily, so the rest of the topic has somewhere to attach. The deeper the reason, the less the group depends on the case that produced it. That claim is reasoning from the breadth, and it is not a measured finding.

The first structure is almost never the simplest available. The better one usually comes from a perspective other than the one the topic was taught in, and finding that perspective reliably feels difficult before it feels obvious.

What the structure buys

A chunk, once the decision is made, is a group of ideas held together by one reason they matter. The backbone is the few largest of those groups and how they relate. Under them sit mid-level groups, and under those the smaller ones holding concepts and details.

Each level reduces what has to be held at once and increases what can be unpacked, and unpacking runs top-down. That load account lives in full on Cognitive Load & What Mental Effort Is Trying to Cue. A teaching image for the same idea: a hundred files into twenty folders into five parent folders, so any file is two steps from the top. The image is an image. What working memory actually holds, on current estimates, is nearer four items in the focus of attention, which is why the working ceiling below is four branches rather than a hundred items.

The load reduction and the retrieval gain are only there because the group was judged rather than inherited. Grouping can reduce load, and not every grouping does. A low-quality grouping can look organized while leaving the understanding where it was.

Reading your own structure

More than four branches off a single point means the grouping is not finished.

The repair is to ask why each of those items matters. The reasons cluster, and the clusters are the missing sub-chunks. The same signal shows without branches: where relationships are becoming chaotic, those items want to move closer or into the same group; where one group carries a high density of connections, its contents want grouping further.

A map that only mirrors the headings the material arrived under, or groups by surface similarity while hiding function, is the given structure from earlier, left unjudged. Spiderwebbing, too many arrows between the leaves because the groups were fixed before the relationships were judged, is the next stage of the same sequence. Copying a hierarchy gives connections that all run outward, with nothing lateral. Adding the lateral relationships afterwards, on a structure already fixed, is what produces the tangle. The fault is the ordering. The order that works is relationships found first, judged, and only then turned into groups.

The materials-free repair is to put the items into a flat list before grouping any of them, so the headings stop deciding. Groups appear that were in none of the headings. A concept can leave the group the textbook put it in and sit in one the textbook never had.

If the only way to remember what is in a group is to memorize the membership, the label is doing no work. Terminology-heavy labels are the sign. Relabel in ordinary words, what against how, mechanism against action, and the relabelling forces the comparison that was skipped. Relabelling is itself the higher-order act.

A group that is working explains why its members belong together, reduces what has to be held, makes later details easy to place, and creates retrieval prompts. Two tests have to be survived: it can be rebuilt from memory, and it still holds on a case the material never mentioned. Groups that form through evaluation rather than surface similarity are the ones that pass.

A map that looks complete without being easier to retrieve, or that cannot support a useful question, or on which every relationship carries equal weight, has not finished that evaluation.

False obviousness, and the two ways confusion ends

The standard above says which grouping is better. It does not say how the good one is found. A group that feels obvious without first passing through confusion is usually a simplification that skipped the encoding. That failure has a name: false obviousness. Real obviousness arrives after confusion. Obviousness that arrives instead of confusion is the false kind.

False obviousness has a testable form. A pair of group names that cannot derive each other looks like a structure and is not one:

A pair that will not derive each other, "things to think about before grouping"
and "things to think about while eating breakfast on the 17th of May",
is not a weak structure. It is no structure. The only way to hold the
two groups is to memorise them.

Change one name and the others have to move with it. Names that can
each be swapped without disturbing the rest were never a structure.

Any action that ends confusion is one of two things: the material was organised so it makes sense, or the thinking stopped.

There is no third. When the confusion has gone, the self-check is which one happened. If it was the first, the understanding is noticeably deeper and the material feels more retainable, and both of those can be told from inside. If neither is true, name the action that ended it. Saying “I don’t know” and stopping is itself an action taken to end the confusion. It feels like honesty, and its effect is escape from the confusion.

The response that replaces escape is to interrogate the feeling. What specifically does not make sense, gotten very specific, is the question worth asking. Questions generated without first feeling the confusion come out generic and recycled and move nothing. The habit of ending the feeling instead of reading it is invisible to the people who have it. Hearing it described does not produce recognition, and un-training it takes a long time; a single session does not do it.

Thinking on Paper does two jobs. Working non-linearly moves the working-memory half of overwhelm out of the head and onto the page. It also produces the productive half: deciding where a thing goes and which arrows reach it is a decision, and making it produces confusion, which is what prompts the questions. Overwhelm and confusion are different loads.

On paper?	What the load actually is	What to do
No, still held in the head	Too many elements, not knowing how they connect, working-memory overflow	Externalise. This is not confusion.
Yes, items already on the page	A missing item (a definition) or a missing connection	Sit with it. This is productive.

Cognitive Load & What Mental Effort Is Trying to Cue is what the load is actually reporting. Habitual work on paper means nearly all the overwhelm still felt is confusion.

Two colours mark the stop-and-answer cycle, if they go on the page at all. Red is a deliberate stop: sit with the confusion and generate questions out of it. Green is to go and answer them. Then sit with the answers and ask what is still unclear; when nothing is, the topic is learned. The whole skill is staying with the first of the two endings and catching the slip into the second. That slip is The Shortcut Problem as it appears in chunking: the version of the method that feels cheaper is the one that fails.

What actually sets the level

How good a grouping comes out is set by two things: how comfortable confusion is to sit in, and how good the questions made from it are. There is no third. Prior knowledge does not set the level, and how far along a course has gone does not set it either. Prior knowledge is what makes a group intuitive, because it is the stock of patterns available to be tried. A pattern picked up years ago in an unrelated field, applied to new material, is what ordinarily gets called an analogy. The two claims are made together: prior knowledge supplies the candidate patterns; the skill decides whether they get searched and whether the search stops at the first one that fits.

A specialist with no tolerance for confusion still groups badly. A relative beginner who interrogates confusion groups well. That is possible because some structural cues need no subject knowledge at all. The four-branch ceiling is one. A node that leads only to one other node is asking to be collapsed. A dead end on such a chain merges into its parent.

Spending a long time searching for a pattern without finding one diagnoses one of three things: it was not known that patterns were what to look for, thinking in patterns is unfamiliar, or not enough patterns have been seen yet. The repair is the same in all three cases because the alternative does not exist. There is no route to durable, usable knowledge that skips the pattern, so the decision reduces to persist or give up.

How the skill grows, and where it stalls

Every skill grows the same way: make a prediction, get data on whether it was right, iterate the prediction. A stall at any stage is a failure in one of those three ingredients.

The skill has four stages. The first is low comfort with confusion and little practice generating questions. Groups come out superficial or rote, and the discomfort is escaped quickly. The second is taking action: forcing a grouping even if it is wrong. What the group says does not matter yet. Producing one does. The third is self-correction loops: willing to re-group as understanding grows, and then actually doing it. The fourth is intuitive and accurate on the first pass. “It could be this, it could be that” runs live while first learning.

Above the fourth, the operation does not change. A practised grouping holds two or three candidates at once, all of which could work, and four or five can be tried and discarded in thirty to forty seconds. That is a practitioner’s description of speed. It is not a new capability, and it is not a measurement.

The benefit from the first stage to the second feels small, from the second to the third moderate, from the third to the fourth large. Those sizes are felt, and unmeasured. The two stalls matter more than the gradient. The stall from the second stage to the third is being afraid to correct a grouping and trying to be right the first time. The stall from the third to the fourth is not being used to doing the hard thinking right at the start. The shape matches the ordinary competence ladder, from not knowing the skill is weak, to knowing, to doing it without watching. That is a practitioner’s description, and it is not a trial result.

Running a pass, and reading it

A healthy pass looks like this. The items are encountered and get onto paper non-linearly. A candidate pattern is proposed and run against derivability. The specific gap, a missing item or a missing connection, is sat with and interrogated. Something relevant, connected, and intuitive is committed to. A click of fit arrives; that click is a feeling, and it is not a finding about this technique. The grouping is then encoded, retrievable, usable.

The failure pass takes a different middle. The items are encountered. Confusion and overwhelm arrive. The first arrangement is grabbed, or a mnemonic, or rote. The confusion lifts because the thinking stopped. No pattern forms. The material decays in days to weeks.

The repair starts by noticing the lift was too easy, and asking whether the understanding deepened or whether the thinking just stopped. The gap is re-opened and named: missing item or missing connection. Questions are built out of that specific confusion (red) and answered (green). The grouping is made again with derivability and the four-branch ceiling in mind. Later, when the first structure has become so obvious it is no longer teaching anything about the topic, it is rebuilt. Restructuring from memory a week later is itself a retrieval session, which is the concrete tie to Spaced Interleaved Retrieval, the sessions that show which groups have stopped earning their place.

Pause when “this is not making sense any more” arrives. The method’s own defaults: one or two concepts if the material is genuinely hard, four to seven if it is simple, and past five to seven new things almost regardless of how familiar the topic is. Early on, grouping only happens at those pauses. With practice it happens as the material arrives and the pauses become rarer.

The first twenty to thirty minutes have one objective: the three to six main groups of the whole topic and how they relate. No definitions, no detail. How fast something is learned depends on how new it is, so a fact that takes fifteen minutes at the start of a session can take thirty seconds once the backbone exists. Leaving the detail behind feels unsafe. It is returned to shortly, and it does not matter whether the early groups are accurate, because they will change. A textbook’s or teacher’s grouping beats an arbitrary one, and roughly 60–70% of the time, on the method’s own estimate, it still needs rearranging to fit. A textbook’s structure is a starting position. It is not the answer.

A long arrow between two related items is usually a group that has not been made. Where a structure jumps from a state straight to two specific items, the missing logical parent is usually the fix: one intermediate category, and one memorising demand gone.

A thin first pass over the material and the main study pass use the same mental operation. The only difference is how much is taken. The thin pass stays close to what is already held, because five items connect more easily than a hundred. A guess at a structure is made and left in place, which sets up a prediction error later. It is the easier of the two because being wrong and walking away is allowed. Its product is prior knowledge, which is what makes the main pass’s groups intuitive. The thin pass builds the prior knowledge the main pass then uses, which is why Prestudy is a mechanism rather than a cross-reference: the thin pass at full length, allowed to be wrong, building what the main pass then leans on. The main pass takes more at once and corrects as it goes. The skill is practised inside Bear Hunter System, the encoding routine whose restructuring pass is this work.

Rote comes last. It is not skipped.

Most of what gets decided “must be memorised” does not have to be. An item cannot be known to be arbitrary until the effort of connecting it has been spent. There is no advantage to doing rote repetition early, because that memory decays fast either way, so all of it defers to the end, and the early effort goes to pattern-finding. That decay claim is not absolute: spaced retrieval of genuinely arbitrary items does last. The sequencing still holds: connecting first, rote last. Applying knowledge and solving problems, including getting them wrong, is what gives any frame of reference for judging a pattern at all. A wrong answer says the pattern reached for was not the right one, which is exactly the comparison this work needs. Problem-solving supports the technique and never substitutes for it.

The click of a structure fitting, and the later rebuild of a structure rather than an extension of one, is what Schema Construction, Assimilation, and Reorganization covers underneath. When to stop, when to re-group, and when the structure has expired sit with Self-Regulation.

Pause when the material stops making sense: one or two hard concepts, four to seven simple ones.
The first twenty to thirty minutes are the three to six main groups, no detail.
A textbook's grouping is a starting position; most of the time it still needs rearranging.
Rote waits until connecting has been tried and has failed.
Even a grouping that is not redrawn was practice: the thinking is the product.

Working looks like this. The material gets easier to understand and retention rises. Complicated groups start feeling logical. Large volumes feel less overwhelming and take less time. Someone who has grouped a topic well can collapse it to a few top groups, explain it to a child, and expand any group on demand. Someone who has not grouped it well cannot do those three things.

Not working looks like this. No clear backbone, points scattered. Branches over the four-branch ceiling. Too many relationships to keep track of. Retention dropping. Study time rising. The amount that can be covered in one sitting falling sharply, which is the most visible sign and the one most often ignored.

Doing this properly feels overwhelming, and fluency feels lower than it was, because the standard just rose. That is the expected experience, and it is not a sign of failure. It eases with practice. Even when nothing is redrawn, the effort of working out the better structure was the practice.

A structure built well is one that gets outgrown. After a month or two of using a topic, the groups made for it become so obvious they are no longer teaching anything about it. That obviousness is the signal to build the next ones. It differs from false obviousness in when it arrived: after the work, and after a month or two of use.

What it costs, and where it fails

The method costs a second pass, and the cost comes after the map already looks finished. The first structure is a draft, and the regrouping is where the encoding happens.

Material with no importance gradient has no “why it matters” to group by: a vocabulary list, a table of constants, an alphabetized reference. Forcing a reason onto it manufactures a structure that will not survive the next item. For genuinely arbitrary material there is nothing to find, and the pattern search is a real cost paid for nothing. The other limit is the habit described above of ending confusion instead of reading it: it is invisible to the people who have it, and un-training it takes a long time.

Two people with different prior knowledge can group the same topic differently and both be right, because the working test is which structure needs the least memorizing for the person holding it. A group borrowed from someone else’s map, however good, carries none of the encoding.

If the regrouped structure cannot be rebuilt from memory a day later, the reasons were labels rather than judgments, and the repair is upstream at the questions rather than on the map.

The criterion before the map

An inadequate structure almost always traces back to importance questions that were never asked when the topic was opened. The criterion is applied before a map exists, which is why its failures are caused upstream of where they show.

The Bear Hunter System is the four-pass loop the criterion runs inside. The first pass hypothesizes what matters. The middle pass tests the hypothesis against the material. The last pass prunes what the structure could not earn. Aim, Shoot, and Skin are those three jobs: questions, working map, prune.

How to practice this

1. Before grouping anything, put every item from the material into a flat list. Notice groups appearing that were under none of the original headings.
2. For each group, ask why it matters, then ask what that reason is in turn for. Notice the group getting broader as the reason gets deeper.
3. Look at one piece of a group and see whether the next pieces name themselves. If they do not, the label is doing no work and wants relabelling in ordinary words.
4. Notice where more than four branches leave one point; that grouping is not finished. Ask why each of those items matters, and the answers cluster into the missing sub-groups.
5. When confusion lifts, ask which of the two endings happened. Deeper understanding and more retainable material mean it was organised; neither means the thinking stopped.
6. Spend the first twenty to thirty minutes on the three to six main groups only. Notice a fact that took fifteen minutes at the start taking thirty seconds once the backbone exists.

Related pages

• Deep Processing: the parent dimension, what deep processing is and where this criterion sits inside it; the processing that decides how much has to be memorised later.
• Bear Hunter System: the four-pass loop the criterion runs inside, hypothesized in the first pass, tested in the middle, pruned in the last; the encoding routine this skill is practised inside, and where in it the restructuring pass falls.
• Spaced Interleaved Retrieval: the test of whether the groups can actually be unpacked from memory; the sessions that show which groups have stopped earning their place, and the pass where restructuring from memory happens.
• Self-Regulation: the control layer that decides when a structure gets rebuilt rather than defended, when to stop, when to re-group, and when the structure has expired.
• Higher-Order Learning: rating importance as a practised behaviour rather than a one-off judgment.

6 more related pages

• Cognitive Load & What Mental Effort Is Trying to Cue: why a hierarchy lowers what has to be held at once, what the effort is signalling when it does not, and how overwhelm gets told apart from confusion.
• Knowledge Mastery: From Recognition to Usable Knowledge: the level at which judging which relations matter becomes the thing being practised.
• Thinking on Paper - Mindmaps and Other Techniques: getting the grouping out of a head where it can be rearranged; the offload that makes the remaining difficulty legible as confusion rather than overload.
• Schema: what a group is a choice about, the structure the decision builds.
• The Shortcut Problem: why the cheaper-feeling version of a method is the one that fails; the failure pass is this problem as it appears in chunking.
• Prestudy: the thin pass at full length, allowed to be wrong, building the prior knowledge the main pass then leans on.

Open questions

• Does the size ceiling hold at every level of abstraction?
• What does false obviousness look like in the material being grouped right now?

Sources

• Chi, M. T. H., Feltovich, P. J., & Glaser, R. (1981). Categorization and representation of physics problems by experts and novices. Cognitive Science, 5(2). Experts group by function and principle where novices group by surface features. With Chase and Simon 1973, the bound on the claim that the level is not set by how much is already known: expertise changes what can be grouped, so prior knowledge is not irrelevant.
• 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). Semantic and evaluative processing encodes more durably than surface processing; repetition is shallow relative to connected encoding.
• Anderson, J. R., & Reder, L. M. (1979). An elaborative processing explanation of depth of processing. In Levels of Processing in Human Memory.
• Miller, G. A. (1956). The magical number seven, plus or minus two. Psychological Review, 63(2). The chunk as the unit of span: the lasting contribution, not the number.
• Cowan, N. (2001). The magical number 4 in short-term memory. Behavioral and Brain Sciences, 24(1). The focus of attention holds about four chunks.
• Chase, W. G., & Simon, H. A. (1973). Perception in chess. Cognitive Psychology, 4(1). Hierarchical chunk structures in expert memory.
• Ericsson, K. A., & Kintsch, W. (1995). Long-term working memory. Psychological Review, 102(2). Retrieval structures increase what can be unpacked from long-term memory; the retrieval structures behind the derivability test.
• Sweller, J. (1988). Cognitive load during problem solving. Cognitive Science, 12(2). Schema formation and automation reduce load. Overwhelm as element interactivity plus working-memory overflow.
• Kapur, productive failure (2008, 2016). Attempting a structure before it is right.
• VanLehn, impasse-driven learning (1988). Sitting in the gap as the condition that produces the question.
• Bjork, desirable difficulties (1994). Difficulty that is met improves later performance.
• Dunlosky, Rawson, Marsh, Nathan & Willingham (2013). Keyword mnemonics rate low-to-moderate with limited transfer; practice testing of arbitrary items does last. The hedge on rote.
• Bjork, Dunlosky & Kornell (2013). Fluency is not competence: the field cousin of false obviousness.
• Risko & Gilbert (2016). Cognitive offloading: paper removes the working-memory half.