Constructivism Learning Theory: Types, Principles, Examples

Go from storyboard to an interactive course in minutes

Start Free

Picture a new accounts payable officer, in a hypothetical company, reading an email on a Tuesday morning. The usual contact at a long-standing supplier asks for its bank account to be changed "for all payments from this week", with new details on letterhead attached. Her induction last month covered invoice entry and the payment run, and she passed its quiz. She updates the record, because in her experience a request from the usual contact is safe. The quiz checked that she could recall what she was told. It never put a request like this in front of her.

I'm an instructional designer at Coursebox, where we make instructional design software that turns a brief into a course outline, so I have a stake in how courses get planned. Constructivism offers a way to understand the officer's mistake: people fit new experience to what they already believe, and can revise that belief through evidence, explanation and reflection. This guide covers the theory in its originators' words, its types and principles, the evidence on guidance, four design methods worked through one finance course, and two free templates.

What is constructivism learning theory?

Constructivism learning theory is the view that people build knowledge by fitting new experience to what they already know and changing their understanding when it no longer fits, rather than receiving it ready-made from a teacher, a book or a screen.

View the full-size diagram

Ernst von Glasersfeld, the philosopher most associated with radical constructivism, summed it up in a 1989 encyclopedia entry as two principles:

• "knowledge is not passively received but actively built up by the cognizing subject"
• "the function of cognition is adaptive and serves the organization of the experiential world, not the discovery of ontological reality"

The second principle says that what we know is a working model that helps us act, so knowledge is judged by whether it works. Constructivism is a theory of how people come to know things, not a teaching method, though designers draw methods from it. What it changes for a designer is where to start: from what learners already believe, and from a problem that shows them where that belief stops working.

The theory has several strands:

A workplace program built on these ideas shows what guided constructivist practice looks like: realistic problems, the learner planning the approach, and a coach on call.

Case study: Avionics troubleshooting practice in the US Air Force

In US Air Force avionics shops in the late 1980s, the first priority was "to keep planes flying" rather than to train new technicians, so the hardest faults went to the most experienced technicians and trainees rarely got to work on them. The Air Force Human Resources Laboratory and the University of Pittsburgh built Sherlock, a computer-based practice environment for diagnosing faults in the test equipment used on F-15 avionics.

• What they did: Sherlock gave trainees 34 troubleshooting problems in order of difficulty, presented the way a real fault would appear in the shop. In most of them the standard fix in the technical orders did not solve the problem, so trainees had to plan their own way of isolating the fault. A built-in coach, described as "a coach or master technician" for "apprenticeship learning", offered up to five levels of hints that the trainee chose, from a recap of their own actions to detailed help with the next step.
• Evidence and result: In a field test at two F-15 wings, 32 technicians were put into matched pairs and one of each pair was randomly assigned to Sherlock, for an average of 20 hours over about three weeks. The others continued normal on-the-job training. On a verbal troubleshooting test scored by researchers who did not know who had been tutored, the tutored group averaged 79.0 against 58.9, an effect size of 1.02 in Valerie Shute's 1991 Air Force review. The tutored technicians, with about 28 months' experience, scored close to technicians averaging 114 months' experience, according to an Air Force laboratory paper in NASA's proceedings of the 1989 SOAR workshop, and the gap was still about 15 points five to six months later.
• Limitation: This was a test, not shop records, with 32 technicians at two bases. The comparison group got no alternative course, so Sherlock also added 20 hours of focused practice. I could not read the primary evaluation report, so these figures come from two Air Force laboratory documents that summarise it. Sherlock was also an intelligent tutor built from an analysis of expert troubleshooting, so the credit belongs to coached practice on realistic problems, not to constructivism alone.
• Lesson: When the job cannot give novices the hard problems, build realistic ones where the standard procedure fails, let learners plan their own approach, and keep graded hints on call that the learner chooses. Then measure with a problem-solving test, not a recall quiz.

Where constructivism came from, and who founded it

No single person founded constructivism. Von Glasersfeld credited the "first explicit formulation of a constructivist theory of knowledge" to the Italian philosopher Giambattista Vico in 1710, and named James Mark Baldwin and Jean Piaget as the first to make it "a major component of developmental theory". Piaget is the figure most often called its father.

If you read one original, read von Glasersfeld's two-page encyclopedia entry on constructivism in education, free on his official site. Then read D. C. Phillips's 1995 article "The good, the bad, and the ugly" for how the versions differ and where they overreach.

The worked example: a supplier bank-detail change request

I'll carry one hypothetical example through the rest of this guide. A mid-size manufacturing company is building a short self-paced course for accounts payable (AP) officers in their first three months. They can enter invoices and run the weekly payment run, but they have not handled a request to change a supplier's bank details or read the verification procedure, so the course cannot assume that knowledge from the job title.

The case: on a Tuesday, two days before the weekly payment run, the usual contact at a long-standing packaging supplier emails AP to update the bank account "for all payments from this week", because the supplier has "moved banks", with a letter on letterhead giving the new details. A $48,600 invoice from that supplier is due in Thursday's run, and the email's domain differs from the usual one by one letter. The course supplies a case pack with everything the decision needs:

• the email and the letter;
• the supplier's record in the vendor master: bank details, the phone number on file and a change history showing no changes in four years;
• the payment run schedule;
• the procedure: any bank-detail change must be confirmed by calling the supplier on the phone number already on file, never a number in the request, and approved by a second officer before the next payment.

Officers should leave able to choose hold, check or pay against the record and the procedure, justify it in writing, and build their own check for next time. The example suits constructivism because new officers arrive with a plausible but wrong rule, "a request from the usual contact is safe", which the case contradicts, and because a wrong answer has a real cost, so the design has to be guided.

Types of constructivism

There is no official list of types; Phillips sorted the versions along lines rather than into boxes, such as whether the building of knowledge is individual or social. In instructional design, the three types most often named are cognitive, social and radical constructivism; Bruner's modes of representation and Papert's constructionism are close relatives. Each type follows the same pattern: the core claim, what it changes in a course, and what it looks like in the AP course.

1. Cognitive constructivism (Piaget)

Cognitive constructivism is about how each person builds knowledge in their own mind. Piaget put it plainly in 1964: "Knowledge is not a copy of reality." And: "To know an object is to act on it."

Assimilation fits new experience into the structures a person already has, and accommodation changes those structures when experience will not fit. In The Construction of Reality in the Child he wrote that "assimilation is conservative", while "accommodation is the source of changes", and that "all knowledge is simultaneously accommodation to the object and assimilation to the subject". The two are not steps that take turns. In his 1964 paper he called the balancing process between them equilibration, "a process of self-regulation".

Piaget's four stages, ending when "the child reaches the level of what I call formal or hypothetic-deductive operations", describe how children's thinking develops. They are not a guide to how adults learn at work, where what learners already know about the task is the useful variable.

• Core claim: People make sense of new experience through what they already believe, and can revise those beliefs as they work through evidence, explanation and reflection.
• What it changes in a course: Find out what learners currently believe before you explain, and design a case that will not fit the wrong belief.
• In the AP course: The officer's first answer shows her existing rule. The one-letter domain and the "all payments from this week" deadline do not fit "the usual contact is safe", and that conflict makes her change the rule to "check the request against the vendor record, whoever sends it".

2. Social constructivism (Vygotsky)

Social constructivism is about how thinking develops between people before it develops within one person. In Mind in Society (1978, p. 57), Vygotsky wrote that every function in development "appears twice: first, on the social level, and later, on the individual level". His zone of proximal development is the gap between what someone can do alone and what they can do "under adult guidance or in collaboration with more capable peers" (p. 86). Later writers call that helper the "more knowledgeable other", but the phrase is not in the book.

The label "social constructivism" was applied to Vygotsky later. Two Adelaide researchers, Liu and Matthews, wrote in a 2005 review of Vygotsky's philosophy that "whether Vygotsky ever considered himself as a social constructivist is not clear to us", although he remains the strand's main figure. The word scaffolding is also often credited to him, but it comes from David Wood, Jerome Bruner and Gail Ross (1976), who used it for the help a tutor gives so that a learner can complete a task they could not yet manage alone.

• Core claim: Ways of thinking start in shared activity and conversation, then become the individual's own.
• What it changes in a course: Plan who learners will think with, including someone more capable, such as an expert or a stronger peer.
• In the AP course: Officers read two colleagues' decisions on the same request and a senior officer's reasoning on an earlier one. The senior officer is the more capable peer.

3. Radical constructivism (von Glasersfeld)

Radical constructivism accepts both of von Glasersfeld's principles. In his 1989 entry he wrote, "To accept only the first principle is considered trivial Constructivism by those who accept both". The test he set for knowledge is that it "only has to be 'viable'", meaning it fits within the limits that experience sets on what we can do and think.

It is often described as saying that reality does not exist or that all knowledge is subjective. Von Glasersfeld rejected that reading in a 1995 chapter on constructivist teaching: "I have never denied an 'absolute' reality, I only claim, as the sceptics do, that we have no way of knowing it." In a 1990 exposition he added that "all action, be it physical or conceptual, is subject to constraints", which is why people cannot construct any reality they like.

• Core claim: Knowledge is a model that works or does not work in experience, and it is revised when it stops working.
• What it changes in a course: Treat learners' errors as information about the model they are using, and give them cases where that model stops working.
• In the AP course: The rule "the usual contact is safe" worked for months of routine invoices and stops working on the Tuesday request; the aim is a rule that works on more cases, which the procedure supplies. That does not mean any answer will do: paying the invoice is still wrong.

4. Bruner's modes of representation

Jerome Bruner is often listed with the constructivists for discovery learning (1961) and the spiral curriculum, his 1960 proposal that a course should return to the same big ideas at increasing depth. His most usable idea for training is in a 1964 paper on cognitive growth, where he named "three modes of representation": "enactive representation, iconic representation, and symbolic representation". Enactive knowledge is held in action, such as tying knots; iconic knowledge in images; symbolic knowledge in language and other symbols. Unlike Piaget's stages, the three modes remain "more or less intact throughout life", so adults use all three.

• Core claim: People know things in three ways, by doing, by picturing and by stating, and each supports the others.
• What it changes in a course: Give learners the task to do, a picture of it and the words for it, rather than words alone.
• In the AP course: Officers check the vendor master record themselves (enactive), follow a timeline of the request and the payment run (iconic) and read the procedure (symbolic).

5. Constructionism (Papert)

Constructionism is a relative of constructivism, not another name for it. In their 1991 chapter "Situating constructionism", Seymour Papert and Idit Harel wrote that constructionism "shares constructivism's connotation of learning as 'building knowledge structures'" and "adds the idea that this happens especially felicitously in a context where the learner is consciously engaged in constructing a public entity". Their short version is "learning-by-making". The chapter adds, in Papert's first person, "I do not mean to imply that construction kits see instruction as bad. That would be silly."

• Core claim: People learn especially well when they build something that other people can see and use.
• What it changes in a course: Finish with learners making and sharing a real artefact, such as a job aid or a checklist.
• In the AP course: Each officer writes and shares a personal verification checklist for the next bank-detail change request.

Principles of constructivism

The types share six principles, drawn from the originators and from the designers who turned the theory into methods.

• Puzzlement drives learning. Savery and Duffy's second proposition is that "Cognitive conflict or puzzlement is the stimulus for learning". Janet Kolodner said the same of failure: when expectations fail, "a need to learn arises".
• Knowledge is tied to where it is used. John Seely Brown, Allan Collins and Paul Duguid argued in 1989 that knowledge learnt out of context tends to "lie inert". Our guide to situated learning theory covers the related idea of learning through a community of practice.
• Learners need guidance. Richard Mayer, in a 2004 article on discovery learning, wrote that learners need "enough freedom to become cognitively active" and "enough guidance so that their cognitive activity results in the construction of useful knowledge".

Matching content to visual, auditory or kinaesthetic learning styles is not a constructivist principle either, and a 2008 review by Pashler and colleagues concluded that "there is no adequate evidence base to justify incorporating learning-styles assessments into general educational practice."

Does constructivist teaching work? The evidence on guidance

The evidence favours guided versions of constructivist teaching. The sharpest criticism targets minimal guidance.

The criticism came from Paul Kirschner, John Sweller and Richard Clark in a 2006 article on minimal guidance. They argued that minimally guided instruction "is less effective and less efficient than instructional approaches that place a strong emphasis on guidance of the student learning process", and that the advantage of guidance recedes "only when learners have sufficiently high prior knowledge to provide 'internal' guidance" (p. 75). Mayer called the leap from active learning to unguided activity the "constructivist teaching fallacy" (p. 15). A 2011 meta-analysis by Alfieri and colleagues found that unassisted discovery taught less than explicit instruction, while discovery with feedback, worked examples or scaffolding did better than other methods.

The reply from Cindy Hmelo-Silver, Ravit Duncan and Clark Chinn, in a 2007 response, was that problem-based and inquiry learning "are highly scaffolded" (p. 99), so they are not minimal guidance at all, and that direct instruction can be given "on a just-in-time basis" (p. 100). Two meta-analyses since then test guided designs directly. I could read only their published abstracts, so their figures here are abstract-level.

• Guidance in inquiry. Ard Lazonder and Ruth Harmsen pooled 72 studies of inquiry-based learning in 2016. Guidance improved learning outcomes, with an average effect of d = 0.50.
• Problem first, then instruction. Tanmay Sinha and Manu Kapur compared the two orders across 53 studies in 2021.

The effect was stronger when designs followed the principles of productive failure closely, and nothing in the abstract covers workplace training. Instruction is still part of the winning design: the order changes, but the teaching stays.

How much guidance to give depends on what learners already know. Slava Kalyuga, Paul Ayres, Paul Chandler and John Sweller called this the expertise reversal effect: "Instructional techniques that are highly effective with inexperienced learners can lose their effectiveness and even have negative consequences when used with more experienced learners" (2003, p. 23). New AP officers need the procedure, a worked example and hints; an officer with ten years in the job needs harder cases and less help. For this AP course, use a brief simulated first attempt to reveal learners' reasoning, then teach and reduce support as expertise grows. If learners cannot make a useful start, give a worked example or explanation first. Problem-first teaching is a design choice, not a requirement of constructivism.

Four constructivist design methods, worked through one course

Four constructivist methods suit workplace courses well. For each I give what it needs, when it fits, and its part in the AP course, where new officers decide whether to hold, check or pay after the Tuesday request to change a packaging supplier's bank details.

1. Problem-based learning

Problem-based learning (PBL) organises learning around a problem that learners work on before they are taught. In his 2006 overview, John Savery named two conditions for success: "ill-structured problems" and "a tutor who guides the learning process and conducts a thorough debriefing". He added that "learners who are new to PBL require significant instructional scaffolding".

• What it needs: A realistic problem from the job, resources to research it, a facilitator who guides without giving the answer, and a debrief.
• When it fits: Learners know enough to start, and the job throws up problems that need research and judgement.
• In the AP course: The Tuesday request organises the whole course, and officers meet it before any explanation. Because the procedure gives a right answer, the course borrows PBL's order (problem first, research, debrief) more than its open-endedness.

Case study: Problem-based training for occupational doctors in the Netherlands

Physicians training as occupational health specialists in the Netherlands attend a school of occupational medicine one day a week during four years of work. In 1999, researchers at all four Dutch schools tested whether the teaching method changed how trainees who still had to learn the national guidelines for workers with mental health problems applied them in practice.

• What they did: One group took a four-day program "based on the principles of problem-based learning", in groups of about nine with a facilitator. Each cycle started from a case in a participant's own practice, moved to agreed learning questions and private study, and ended with reporting back to the group. The other group took a four-day lecture-based program with "the same objectives and content".
• Evidence and result: In the randomised trial published in 2003, 118 physicians were randomised and 78 could be compared on practice: 34 in the PBL group and 44 in the lecture group. Their performance on 12 guideline indicators, scored from their own case records, rose from 74% to 81% in the PBL group and from 76% to 79% in the lecture group, a significantly larger gain for PBL. Knowledge scores were the same. The PBL group rated the course 6.0 out of 10 against 6.9, and the program cost about 15% more.
• Limitation: The groups did not differ significantly after training; only the size of the gain did. Performance came from the physicians' own case forms, not observation, and more of the PBL group dropped out of the practice comparison. These were doctors in specialist training, so this is health care, not general workplace training.
• Lesson: PBL built on learners' own cases nudged practice up slightly more than lectures with the same content, but people liked it less and it cost more. Satisfaction scores would have pointed the wrong way, so judge a constructivist course on what people do afterwards.

2. Case-based learning

Case-based learning uses worked or partly worked cases from real practice, so that learners reason from earlier examples to a new one. Janet Kolodner's 1997 article on case-based reasoning opens with workplace examples, such as phone company staff learning to sell Yellow Pages ads with simulated clients, and argues that learners must "reflect appropriately on their experiences". Cases and PBL sit at different points in a course: Savery and Duffy noted that most case strategies present the case "after the topic is covered in order to help test understanding and support synthesis", while in PBL "all of the learning arises out of consideration of the problem".

• What it needs: Cases with enough detail to reason from, including what an expert noticed and why.
• When it fits: You have real cases, and the skill is applying judgement from one case to the next.
• In the AP course: After their first attempt, officers watch a senior officer work through an earlier change request from a different supplier, saying what she checks and why: a worked example of the reasoning, not the answer to the Tuesday case.

Case study: Fault-finding training for apprentices at a German technology company

Skilled troubleshooting in industrial automation has high economic value, because it cuts the cost of production downtime. A "globally-acting German technology company", unnamed in the study, trained its third-year mechatronics apprentices on a computer simulation of a real automotive press, built on cognitive apprenticeship, with videos of an experienced maintenance worker diagnosing faults at the original plant.

• What they did: In a four-day in-house course, half the apprentices also worked video cases on two other machines in groups of four to six. The video paused when an expert noticed a failure, learners noted what they would check next and answered questions such as "Which type of fault do you assume?", then watched the expert's solution and compared it with their own reasoning. The other half got worksheets on troubleshooting instead.
• Evidence and result: In Hochholdinger and Schaper's 2013 study, 42 apprentices were split into two matched groups of 21 by pretest score. On simulator faults similar to the training, the video-case group found the fault 56% of the time against 30% (d = 0.92). On a real production machine, an automatic sorter, the figures were 47% against 38%, a difference that was not significant.
• Limitation: The groups were matched, not randomised, and small. Both groups already used the constructivist simulator, so the study tests the extra video cases, not constructivism against a lecture. There was no follow-up on the job.
• Lesson: Expert-solved cases with a "what would you check next?" pause helped apprentices on similar faults but did not reliably carry over to a different real machine. If the goal is transfer to new equipment, give learners practice on varied equipment as well as richer cases.

3. Inquiry-based learning

Inquiry-based learning asks learners to investigate a question by examining information themselves. Savery separated it from PBL by the tutor's role: in inquiry the tutor is also "a provider of information", while in PBL the tutor "does not provide information related to the problem". The abstract of Lazonder and Harmsen's meta-analysis of 72 studies reports that inquiry works better with guidance.

• What it needs: Information or data to examine, a clear question, and prompts or explanations for learners who get stuck.
• When it fits: The job requires people to check facts and records for themselves rather than take them on trust.
• In the AP course: Officers check the case pack themselves: the sender's domain against the vendor master, the change history, the phone number on file and the payment date. Optional hints point to where to look, not to the answer.

4. Cognitive apprenticeship

Cognitive apprenticeship makes an expert's thinking visible and then hands the work over to the learner in stages. Allan Collins, John Seely Brown and Susan Newman set it out in 1989, and Collins, Brown and Ann Holum's 1991 article "Making thinking visible" describes six methods:

1. Modelling: an expert performs the task so learners can build a model of the process.
2. Coaching: the expert watches learners work and offers "hints, scaffolding, feedback, modeling, reminders, and new tasks".
3. Scaffolding and fading: supports for the task, then "the gradual removal of supports until students are on their own".
4. Articulation: learners put their knowledge and reasoning into words.
5. Reflection: learners compare their problem solving "with those of an expert, another student, and ultimately, an internal cognitive model of expertise".
6. Exploration: learners solve problems on their own, "the natural culmination of the fading of supports".

Scaffolding, the authors add, requires "accurate diagnosis of the student's current skill level or difficulty". In practice, ask one open question about a similar situation before you set the problem: if most learners cannot start, add a worked example first; if most already have the right rule, set a harder case.

• What it needs: An expert who can think aloud, hints and feedback, and tasks that grow harder as support falls away.
• When it fits: The skill depends on expert thinking that learners cannot see by watching the result.
• In the AP course: The senior officer models her checks, hints coach officers through the Tuesday case, officers articulate their decision in writing and compare it with colleagues', and the new case is the exploration, with the hints gone. Our guide to the cognitive apprenticeship model has more examples of each method.

Which method fits which training need

Most constructivist courses combine methods. This table is my synthesis of the sources in this guide, not a ranking.

The AP course, step by step

Together, the methods give the AP course this sequence:

The new case switches from email to phone, with a different supplier and warning signs, so officers must apply the reasoning rather than repeat the Tuesday answer.

How to assess constructivist learning

Assess constructivist learning with a new case that learners have not seen, judged against criteria you publish in advance. In a 1990 ERIC digest on authentic assessment, Grant Wiggins wrote that such assessment "achieves validity and reliability by emphasizing and standardizing the appropriate criteria for scoring", and that the debate with traditional tests "is not 'either-or'".

For the AP course, that means:

• A written decision on the new phone case, judged on three criteria. The officer checked the request against the vendor master record, named the verification step from the procedure, and stated the next action for the payment run.
• A record of changed reasoning. Keep each officer's first answer to the Tuesday case beside their final one; the difference shows what changed.
• A short factual check for the procedure only. Multiple choice can check both facts and case decisions if the choices demand reasoning. Here, a written response lets the reviewer inspect the officer's explanation as well as the decision.

Treat the first answer and the comparison as formative assessment, and use a structured debrief such as Gibbs' reflective cycle when a team reviews a real request. None of this shows that officers verify requests at work. That needs a check on the job a few weeks later, such as a review of the change requests each officer handled, which is level 3 in Kirkpatrick's model.

What course requests ask for

We read course requests typed into Coursebox in the 90 days to 7 October 2026: one per person, from English-language browsers, with no uploaded file. Of 266 prompts, 98 were course requests. Two readers split them and classified each by hand against written definitions, quoting the words behind every positive call, and I re-read each problem and scenario call. With no independent second rater, treat the figures as classification estimates. Re-reading positive calls does not establish an upper bound or detect all missed requests. This pass identified 98 course requests; neighbouring articles used different coding passes on the same 266 prompts.

About 14 of the 98 mentioned a realistic problem, case, scenario or project, and in only 6 did learners work on it; in the rest it was content to read or a quiz item. None asked for the problem before the explanation, 6 explicitly asked for the explanation first, and none named a constructivist method.

Five of those 14 paired the scenario with a recall knowledge check. These are requests, not finished courses, but when a brief mentions a problem it rarely says where it goes or what help learners get. For a constructivist module, write the problem, the order, the supports and an open-text assessment into the brief yourself.

Constructivism and neighbouring theories

For behaviourism and cognitivism, our three-theory comparison covers which theory to use when. This table compares three other neighbours, as my synthesis of how designers use them, not a ranking.

Our guides to Kolb's experiential learning theory, social cognitive theory and connectivism cover each in depth.

Limitations of constructivism

• It describes learning better than it prescribes teaching. Kirschner, Sweller and Clark put it this way: "The constructivist description of learning is accurate, but the instructional consequences suggested by constructivists do not necessarily follow" (2006, p. 78). The theory does not say how much to explain.
• Knowledge and skills can move in different directions. A 2003 meta-analysis of 43 studies of PBL by Dochy and colleagues, mostly in medical education, found a positive effect on skills (0.46) and a small negative effect on knowledge that largely disappeared without two outlier studies. A 2009 synthesis by Strobel and van Barneveld found traditional teaching better for short-term retention on board exams.
• It costs more, and learners may like it less. Problem-based courses take longer to design and run, and learners can rate them lower, as physicians did in a Dutch randomised trial of problem-based training (6.0 against 6.9 out of 10).
• Transfer is not automatic. An authentic task in training does not guarantee use on different equipment, as a study of German mechatronics apprentices found: the gain on simulator faults did not reliably carry over to a real machine. Plan practice across varied cases.
• It is harder to assess. Open-text decisions need written criteria and someone to apply them consistently.
• The workplace evidence is thin. I could not find a controlled study outside health care that measured performance on the job; the non-health cases here measured tests and simulators.
• The philosophy can be overstated. Phillips called constructivism "something akin to a secular religion" in education, and named as its bad side "the tendency within many forms of constructivist epistemology ... towards relativism".

Using constructivism with an AI course tool

On 8 October 2026 I gave Coursebox a brief for the AP course, "Hold, check or pay? Handling supplier bank-detail change requests". Coursebox is our product, so weigh my account with that in mind. I wrote the brief problem-first. It included the full Tuesday case and case pack, the sections in order from "Before you start" to a wrap-up, open-text answers for every decision about the case, and a final assessment on a new phone request judged against the three criteria. I asked for about 40 minutes.

Most of the order came back as I had briefed it. The outline put the case and a "What would you do?" page before any explanation, in the first content section after the course map. Next came the senior officer's worked example, as pages called "A senior officer starts" and "Following the evidence trail", and later a page where learners write their own verification checklist.

Two parts did not match the brief. The hinted case work came back as a five-question quiz rather than work on the case itself, and the "Compare your reasoning" step was missing.

I sent one follow-up message asking for those two changes. The revised outline showed "Work the case with hints" as a case analysis assignment, a "Compare your reasoning" page, the new-case assessment, and then a separate wrap-up page, "Raising a held payment". It had grown to about 63 minutes against the 40 I asked for, across 6 sections and 12 pages.

I checked the revised outline only, not the rendered pages or the hints, and two short quizzes remain (a one-question first-decision reflection and a four-question check on the worked example), so the next step before use is to open each page and trim the course back toward 40 minutes.

The question I get most often in this area is how to put a realistic scenario into a course: 15 of the 26 support threads I found over the past year on scenarios, open answers, hints and discussion were about scenarios, though part of that followed the launch of AI scenarios in May 2026. For open decisions like the ones in this course, Coursebox has a scenario feature, short-answer quiz questions that the AI marks against the answers you provide, and assignments graded against a rubric.

Putting constructivism to work

Constructivism explains why a course can teach the rule and still leave people making the old mistake: learners interpret new information through what they already believe, and may need explanation, practice and feedback to revise it. Use it for jobs that need judgement on new cases, such as verifying a change request or diagnosing a fault, and design it as guided practice: a realistic problem first, then a worked example and hints, then comparison and a new case on their own. For routines with one right response, or learners who cannot yet make a start, explanation and practice come first.

Pick one decision your learners get wrong at work and build the case pack for it: the documents they would see, the record they should check and the procedure that settles it. Choose whether a brief first attempt or a worked example should come first, based on what learners already know. Write the open question they will answer and the criteria you will judge a new case against, then try it with a few learners and compare their first and final answers to see whether the case changed their thinking.