Skip to content

Learning-for-Use (LfU) Model

Pattern · All patterns
Evidence · 16 claims (8 for, 8 mixed) · 26 studies (7 causal, 7 quant-synthesis, 6 review, 3 theoretical, 1 associational, 1 qualitative, 1 design), q1–q4 · 6 of 26 report an effect size · 11 claims rest on one study

Description and scope

A reusable policy for a unit of inquiry in which learners first meet a task they cannot yet do and that makes the target knowledge necessary → they build the knowledge through guided investigation, with explicit instruction that connects their findings to the canonical idea → they use the knowledge in a consequential task that demands it → they revisit it on a new case later. Edelson (2001) named the three phases motivate, construct and refine. The intended change is in conceptual-understanding and in using the idea on a case not yet seen (near-transfer, far-transfer), not in recall of facts or speed with a procedure. The goal is usually a concept or principle inside a complex-skill, at unit grain over days-weeks.

No claim in this wiki tests Learning-for-Use as such: no recorded study compares a motivate-construct-refine unit with the same content taught another way, and no claim page names the model. What the wiki does hold are claims about each phase, carried here by extrapolation:

  • Motivate (meet the task before the idea). Letting learners attempt problems the concept solves before it is taught, then teaching it from their attempts, improved conceptual knowledge and transfer and not procedural skill, with the effect reversed for young children and for domain-general skills (productive failure [+S]). That is the nearest tested relative of "create a need to know", but it measures learning, not the motivation the phase is named for. No claim recorded here measures whether a need-to-know opening raises interest; two descriptive reports from a storyline curriculum and one teacher's account are the closest.
  • Construct (guided investigation). Guided inquiry in school science is associated with better achievement than the usual instruction, and teacher-led inquiry with larger effects than student-led (inquiry-based teaching [+S]); unassisted discovery learns less than explicit instruction for novice learners (minimal guidance [~S]).
  • Refine (use in a consequential task). That practising whole tasks improves transfer is a training-design argument that no study recorded here tests (whole-task practice [+W]), and for absolute novices with high-element-interactivity tasks, practising parts first reduced load (part-task practice [~M]).

So the sequence below is the earlier page's, with each step's evidence status stated; the response-dependent branches are untested design proposals.

How the converted siblings sit beside it. Problem-based learning is the more general pattern for problem-centred units and owns how much guidance novices get and whether a problem-first phase comes before instruction; LfU applies those decisions and adds a fixed arc (need, construction, use, revisit) whose last phase must require the knowledge built. Anchored instruction owns building one shared scenario; an LfU unit may use an anchor as its motivating task. 4C/ID owns whole-task training for complex skills, which is the nearest relative of the refine phase. The principles it draws on, inquiry-based learning, productive failure (teach first or attempt first), scaffolding and fading and transfer of learning, each own their relationship; this page does not repeat their models.

Inputs: establish the design brief

Field Record or elicit
Learner and starting observation Before the unit, give each learner the motivating task, or a small version of it, and ask what they would do and what they would need to know. Record whether they can begin at all, what ideas they bring (including wrong ones), and whether they can generate more than one approach. Preserve unknowns rather than labelling learners novice. Record age (child, adolescent, adult) and setting (pre-k-12, postsecondary, adult-workplace).
The target knowledge Whether it is a concept or principle that a task can make necessary, a procedure or domain-general skill (for which teaching first did better), or facts to recall. How high its element interactivity is in the whole task.
The motivating and consequential tasks What task creates the need, and what later task requires using the knowledge (a decision, a design, an explanation, a product for someone). Whether the second task can be done without the target knowledge: if it can, it does not refine it.
Context and activities classroom, online-self-paced, online-instructor-led, workplace-clinical or mixed; single-session or days-weeks; group size; how much teacher or facilitator time there is to check in; which tools, data or simulations are available for investigation.
Learner-valued goal Ask what the learner wants from the unit (to solve the local problem, to pass the course, to be able to do the job). Record where it differs from the designer objective; the motivating task works only if learners see it as theirs.
Designer objective and outcome Which outcome the assessment weighs: conceptual explanation, use on a new case (near-transfer, far-transfer), procedure fluency, or factual recall; immediate or delayed. Name the one the unit is judged by, since the phase evidence differs by outcome.

A request to "design an LfU unit on X" does not specify these. Ask what task makes X necessary, what later task requires it, who the learners are and how the unit will be judged before choosing the problem or the inquiry.

Sequence and conditional policy

The default sequence is the earlier page's four steps, with doses and progression added. Steps 1 and 4 have no claim that tests them as specified; steps 2 and 3 carry neighbouring claims by extrapolation. Doses are untested proposals unless a claim gives them.

  1. Motivate: meet the task first (one session; 20–40 minutes of a first lesson). Present the motivating task (a local problem, a design challenge, a decision) and let learners attempt it briefly, alone or in pairs, before any teaching. Ask each to write what they tried and what they would need to know; collect the questions publicly. Evidence status: attempting problems before instruction improved conceptual knowledge and transfer when instruction followed and built on the attempts (productive failure [+S]; invention tasks [+M] found the benefit only when a learning resource was available later). The "need to know" as a motivational mechanism is untested; a teacher's account that opening on learners' existing ideas stimulates interest is one author's experience (inquiry first [+W]).
  2. Construct: guided investigation with explicit consolidation (3–8 sessions of 40–60 minutes, proposed). Learners investigate with data, a simulation, readings or experiments chosen so the target idea is what explains the results. Guidance is planned, not optional: a driving question per session, prompts, a structured procedure, and feedback. Each session ends with a teacher-led synthesis (10–15 minutes) that compares learners' findings and attempts with the canonical idea and states it. A reading placed after an investigation where results disagree can serve the same need (reading with a need to know, a strategy page, untested). Evidence status: guided inquiry raises science achievement and teacher-led inquiry more than student-led (inquiry-based teaching [+S]); guidance improved post-inquiry learning outcomes (guidance in inquiry [+M]); attempts help only when instruction builds on learners' solutions or uses contrasting cases (PS-I design features [~M]).
  3. Refine: use the knowledge in a consequential task (2–4 sessions, proposed). Learners apply what they built to the motivating task, now in full, or to a new task that cannot be done without the idea (a recommendation to a real audience, a design that must work, an explanation of a new case). Prompt them to explain why their solution follows from the idea. Evidence status: prompted self-explanation improved understanding and problem solving on immediate tests (self-explanation [+S]); whole-task practice for transfer is an untested argument (whole-task practice [+W]).
  4. Revisit on a new case (one 20–30 minute task, 1–2 weeks after the unit, proposed). Give a case with different surface features and ask learners to use and explain the idea without the unit's materials. Evidence status: untested; from the earlier page. It doubles as the unit's delayed outcome.

Support and progression across sessions (proposals). In the first unit of a course, the designer supplies the motivating task, the investigation procedure and the data, and the teacher leads every synthesis. By the second or third LfU unit, learners pose some of the sub-questions, choose among two or three investigation routes and lead part of the synthesis; the canonical statement is still checked by the teacher. Prompts in the refine phase move from sentence starters to a single "why does this follow?" question. Contingent support, raised when a learner fails and lowered when they succeed, is the form the evidence favours, but almost all of it comes from one-to-one tutoring (contingent scaffolding [~M]).

When a learner moves on (proposals). From construct to refine: the learner can explain the target idea in their own words and use it correctly on two of three short new cases at the end of a synthesis. From refine to the next unit: the consequential task uses the idea, and the learner can say why. A learner who meets neither after the planned sessions gets explicit teaching of the idea with a worked example, not another round of open investigation.

Observation Discriminating probe Proposed action, with its uncertainty
Learners produce nothing in the motivating attempt ("we don't know where to start") Ask for any approach, however rough, and check they can read the task's parts (the map, the graph, the terms). If they cannot read the task, teach those parts first and keep the attempt short; if they can but generate nothing, offer two contrasting cases to compare. The productive-failure evidence assumes learners generate solutions; not tested for this case.
Interest is high at the start and fades over the unit Ask privately what they are now trying to find out and why. If learners cannot connect the day's activity to the task, open each session by restating the driving question and what is still unknown; shorten the construct phase. Teachers in a storyline curriculum reported this fading (OpenSciEd enactment [~M]); the remedy is untested.
Learners finish the investigation with findings but no idea ("the water near the farm had more nitrates") Ask them to predict a new case from their finding. If they cannot, the synthesis has not connected findings to the idea: state it, and compare their findings with it. Unassisted discovery of the target idea is the condition where learning drops (minimal guidance [~S]).
The refine task is done well without using the idea Ask which part of their solution depended on the idea. If none did, the task does not require the knowledge: change the task, not the learners. Untested proposal.
The class "agrees" on the idea but individuals cannot use it Collect individual written explanations before discussion; score individual use on the revisit case. If individuals fail, add individual practice with feedback before the refine task; do not read group agreement as each learner's understanding. Untested proposal.
Learners have no idea what to investigate next Ask what question the last result raised. Supply the next question, and say it is supplied; nearly half of students in one curriculum's field test had no ideas about what to investigate next (OpenSciEd enactment [~M]). The same report describes pseudoagency: students seeing a teacher's attempt to open sensemaking as disingenuous, so a supplied question is better named as supplied.

Fitting the design to a situation

The three facts that most change the decision:

  • Whether learners know enough to attempt the motivating task. The problem-first advantage depends on learners generating solutions, reversed for second to fifth graders, and unassisted discovery learns less than explicit instruction for novices. If unstated, ask: given the task cold, can they describe at least one way to start?
  • Whether the target is a concept or principle, or a procedure or fact set. Attempting first helped conceptual knowledge and transfer, not procedural knowledge; for domain-general skills teaching first did better; problem-based formats may cost basic knowledge. If unstated, ask: what will the final assessment ask learners to do?
  • How much time there is, and whether it runs over days. The full arc needs several sessions, and in one meta-regression the benefit of a faithful problem-first design tended to shrink for interventions spanning only a few hours (time span [~M], a marginal interaction). If unstated, ask: how many sessions, of what length, and over how many weeks?
If the brief says… Then change… Basis
Learners are novice in the domain (cannot read the task's data or terms) Teach the task's vocabulary and representations before the motivating attempt; keep the attempt to 10–15 minutes; give a structured investigation procedure and a teacher-led synthesis every session; fade only after the second unit. Minimal guidance [~S]; part-task practice [~M] for high-element-interactivity tasks
Learners are advanced or already hold the idea Skip most of the construct phase: state the idea, and spend the time on harder consequential tasks and new cases; let learners design their own investigation. Problem-first effects rose with grade level [~M] (school and university grades, not expertise as such); otherwise untested
Learners are child (second to fifth grade) Teach the idea first with concrete examples, then use the motivating task as the application; keep the "need to know" opening short and teacher-led. Problem-first effects were negative for 2nd–5th graders [~M]
The target is a procedure or domain-general skill (e.g. designing a fair test), or factual recall Teach the procedure or facts explicitly first; use the LfU arc only for the concept the procedure serves, and keep a separate check of basic knowledge. Productive failure [~S] (no procedural advantage; reversal for domain-general skills); PBL and basic knowledge [~M]
adult-workplace learners, the task is their own work problem Use a real case from their work as the motivating task and a real decision as the refine task; run the construct phase as short guided analyses with a facilitator synthesis. Untested in workplaces; problem-first effects rose with grade level [~M], g = 1.03 for postgraduates and professionals (education settings, not workplace training)
single-session only (one 60–90 minute block) Keep one arc: 10-minute attempt, 30–40 minutes of guided investigation with a 10-minute synthesis, 15–20 minutes on one consequential task, 5 minutes to write the idea in their own words. Do not expect transfer evidence from one session. Untested as a format; time span [~M]: the fidelity benefit of problem-first designs tended to shrink for interventions of a few hours (marginal, p = .055)
A fixed short time budget across sessions (e.g. four 30-minute slots) Give one slot each to motivate, construct and refine, and the fourth to the revisit a week later; cut the number of investigations, not the synthesis. Untested proposal
online-self-paced, no teacher The system shows the task, collects a typed attempt and questions, runs a guided simulation or data set with prompts and immediate feedback, then gives a worked statement of the idea that refers back to typical attempts, then the applied task with a self-explanation prompt. Guidance in inquiry [+M] (the studies' settings are not reported on the claim page); otherwise untested
Mixed modes: some learners remote, some in the room, or phone-only Put the motivating task, the driving question and the class's collected questions in one shared document all can see; use data or readings that open on a phone; take attempts and explanations as typed or voice replies; run the synthesis so remote learners' findings are named. Untested proposal
One teacher for many learners, or scarce facilitator time (volunteers, a short check-in per group) Spend facilitator time on the syntheses, not the investigation; give groups written prompts and a checklist for the investigation; at each check-in ask one question ("what does your result tell you about the driving question?"). Contingent scaffolding [~M]: its evidence is mostly one-to-one tutoring, and the claim page reports contingency is hard across a whole class
Learners with low confidence or literacy Make the motivating attempt low stakes (unscored, in pairs, oral or drawn); use visual data and short readings; give sentence frames for explanations. Untested proposal
A unit running several weeks, with interest fading Re-state the driving question and what is still unknown at the start of each session; keep the unit short enough that the refine task arrives within two to three weeks. OpenSciEd enactment [~M] (teacher reports from one curriculum's field test); remedy untested
The refine task is a product or performance for a real audience Make the audience's question the motivating task, so the need to know is theirs; check the product uses the idea and not only presentation; keep the whole performance whole rather than splitting it into items. Untested; whole-task practice [+W] is an argument, not a test
The brief names its own routine, rubric or categories (its own investigation template, scoring rubric or job aid) Keep it; change only what the evidence says matters: add a synthesis that states the idea after the investigation, and one "why does this follow?" prompt in the applied task. If the rubric rewards tool use, keep that criterion and add an unaided use of the idea beside it. Inquiry-based teaching [+S] (teacher-led inquiry); self-explanation [+S] on immediate tests
The motivating material is private or sensitive (learners' own health, family or workplace data) Let learners choose whether to use their own case or a supplied one, and who sees their attempt; collect questions anonymously. Untested proposal
High stakes, or assessment on new cases weeks later Judge the unit by use and explanation of the idea on a new case at the delay that matters, plus a separate check of basic knowledge; do not judge it by the refine product alone. Productive failure [~S] (the benefit is in conceptual and transfer outcomes); the delay is untested

Twelve of the sixteen rows cite a claim, and each is carried beyond the learners or settings its claim page reports, as its Basis cell says (the real-audience row cites only an untested argument); the other four (fixed short budget, mixed modes, low confidence or literacy, private material) are untested proposals.

Choosing configurations from evidence

  • Attempt first, then teach from the attempts. Productive Failure Improves Conceptual Learning [+S]: across 53 studies and 166 comparisons, problem solving followed by instruction beat instruction followed by problem solving on conceptual knowledge and transfer (g = 0.36) with no difference on procedural knowledge (g = −0.03); larger with higher fidelity to productive-failure design, and reversed for second to fifth graders and for domain-general skills. The claim's table gathers 8 distinct studies. Invention tasks [+M]: ninth-graders who tried to invent a statistical procedure before instruction transferred better than tell-and-practice students only when a worked-example resource was present at test (two classroom experiments, n = 95 and 102). PS-I design features [~M]: a review concludes the benefit appears only with contrasting cases or instruction built on student solutions. Grade level [~M]: the pooled effect was negative for second to fifth graders (g = −0.09) and largest for postgraduates and professionals (g = 1.03). Carry these to LfU's motivate phase as "an attempt at the task before teaching can prepare learners to learn from the teaching, if the teaching uses the attempts", not as evidence that a motivating problem raises interest.
  • Guide the investigation. Inquiry Based Teaching Improves Science Achievement [+S]: a meta-analysis of 37 studies found g = .50 for inquiry-based science teaching, with teacher-led inquiry about .40 larger than student-led, and a synthesis of 138 K–12 studies found a positive trend for conceptual understanding without a pooled effect; the claim's table gathers 8 distinct studies, most quasi-experimental. Guidance in inquiry [+M]: one meta-analysis, d = 0.50 on post-inquiry learning outcomes. Minimal guidance [~S]: unassisted discovery learned less than explicit instruction (580 comparisons, d = 0.38 favouring explicit instruction), while enhanced discovery beat other instruction (d = 0.30); 5 distinct studies in its table. This is the boundary of the construct phase: an LfU unit with driving questions, prompts and teacher-led synthesis is a guided form; the same unit left to "explore" is not.
  • Make the applied phase generative. Self-explanation [+S]: prompting learners to explain improved understanding and problem solving by a moderate average amount (g = .55 across 69 effect sizes), on immediate tests, with little evidence on delayed or classroom gains; 9 distinct studies in its table. It supports a "why does this follow?" prompt in the refine task, not the refine task itself.
  • Whole task or parts. Whole-task practice [+W] is a training-design argument (van Merriënboer, Kester & Paas 2006); its table gathers 5 distinct studies (3 theoretical, 2 causal), none with an effect size coded. Part-task practice [~M]: for novices with high-element-interactivity material, isolated parts first helped (one experiment). Use the consequential task whole for learners who can manage it, and isolate parts first for absolute novices.
  • Contingent support. Contingent scaffolding [~M]: fully contingent one-to-one tutoring of long division beat fixed or no support at immediate and one-month tests (8 per condition), and interactive tutoring helped transfer; a meta-analysis found explicit strategy training beat scaffolding. Its table gathers 25 distinct studies across many narrower claims. Mostly one-to-one; a qualification for class-wide inquiry.
  • What the storyline curriculum reports. OpenSciEd relevance and coherence [+M]: in field tests of a storyline science curriculum that opens units on an anchoring phenomenon, more than 90% of students reported instruction relevant and 87% said they understood how the lesson tied to the bigger picture (descriptive survey percentages from one white paper, no comparison). OpenSciEd enactment [~M]: teachers reported declining interest in the anchoring phenomenon, 20% of those interviewed struggled with unit length and pacing, and nearly half of students had no ideas about what to investigate next. Neither claim page calls the curriculum LfU.

Do not rank LfU against direct instruction or other inquiry designs from these claims: none compares them, and none measures the time the arc costs. A precise abstention names the missing comparison (a motivate-construct-refine unit against the same content and consequential task taught explanation-first, on use of the idea in a new case at a delay) and the local observation that would stand in for it: individual use and explanation on the revisit case.

The evidence by question

Further evidence, not yet read against this model

Claims bearing on the page that are not used as core evidence above.

Illustrative design instance and observation record

Invented illustration, not a tested unit. A grade 8 science class (adolescent, classroom, days-weeks) of 28 learners has three weeks, four 50-minute lessons a week, for a unit on water quality. The town has asked residents whether a stream is safe for children to paddle in. Motivate (lesson 1): pairs get last year's readings for three sites and are asked to give an answer in 15 minutes; each writes what they would need to know; the class lists 20 questions. Construct (lessons 2–8): each lesson opens on one listed question; groups test samples or work with the town's data using a structured sheet, and the teacher closes each lesson with a 12-minute synthesis comparing groups' results and stating the idea (dissolved oxygen, nitrate sources, indicator species). A reading on nitrate runoff comes after two groups' results disagree. At lesson 8, learners explain the idea and predict two new sites; those who cannot get a worked example and a re-teach. Refine (lessons 9–11): groups write a recommendation to the town council and must explain which reading supports it and why. Revisit (two weeks later, 25 minutes): each learner judges a different river from new data, alone, without the unit's sheets. A learner says she wants to know whether her younger brother can paddle; the designer's objective is explaining the idea on a new case. They agree to score the recommendation and the individual revisit separately.

The table changed the default in three places: the vocabulary of the readings was taught before lesson 1 because the starting attempts showed learners could not read the data table; the synthesis was kept every lesson because these were novice learners; and the unit was held to three weeks because of the reported fading of interest in long units. These are local choices to observe, not settled ones.

Record: motivating task and what learners attempted, with their questions → each session's driving question, investigation and guidance given → what each synthesis stated and which attempts it used → individual explanation and prediction at the construct-to-refine check → the consequential task and whether the idea was used, and why → individual use on the revisit case at the stated delay → candidate interpretations (idea built, task done without the idea, group agreement only, could not read the data) and their basis. Preserve the learner's stated purpose and any change in it. An agent may summarise this record but must not fabricate missing inputs or assign numerical probabilities of understanding without a calibrated model.

Elements and limits

Problem presentation, scaffolded inquiry, invention, scaffolding, self-explanation, application, transfer tasks, part-task practice, practice and reflection. Principles the pattern draws on: Problem-based Learning, Inquiry-based Learning, Productive Failure, Scaffolding and Fading, Transfer of Learning, Authentic Audiences & Purposes and Real-world Math.

This pattern is scoped to units in which a task can make a concept or principle necessary and a later task can require it. Procedures and domain-general skills, fact sets, and very young learners are better served by teaching first (direct instruction). The policy supports observation and design reasoning; whether a motivate-construct-refine unit produces better use of an idea on new cases than the same content taught explanation-first, for which learners, at what time cost and at what delay, remains to be tested.

Examples

  • Science learners exploring environmental data because they need it to explain a local problem.
  • Engineering learners investigating a principle so they can apply it in a design challenge.
  • Business learners constructing concepts through inquiry and then using them in a decision scenario.

Key Sources

Page metadata
Field Value
type pattern
id learning-for-use-lfu-model
title Learning-for-Use (LfU) Model
description A reusable unit policy in which learners first meet a task that makes a concept necessary, build it through guided inquiry with explicit synthesis, then use it in a consequential task and revisit it later; no claim in this wiki tests the model as a whole, and its phases rest on neighbouring claims (problem-first teaching helps conceptual learning and transfer when instruction builds on the attempts, guided inquiry beats unguided discovery for novices).
status review
grain_size unit
author Daniel Edelson
canonical True
generated by claude/unspecified on 2026-10-15
trust unverified — no one has confirmed this page's content

Frontmatter is the page's machine-readable record — see the Schema & Guide.