Insights from the H5P Toronto Community Event

Learning and Development News You Can Understand | H5P: A Click Is Not a Thought, Turning Passive Training Into Active Competence
H5P: A Click Is Not a Thought. Cover image for ldnucu.com, Learning and Development News You Can Understand

Listen to the podcast on Spotify

The Problem

Escape the Seat-Time Trap

For decades, workplace training and higher education have operated inside a persistent structural trap: prioritizing the distribution of information over the verification of skill. Organizations build static slide decks, record long talking-head videos, and mandate compliance clickthroughs under the quiet assumption that visual exposure translates into behavioral change. In practice, passive exposure rarely produces operational readiness. Moving beyond that legacy model requires designing for genuine learner agency and scenario-based decision-making, which forces teams to rethink what they actually measure in adult environments.

At ldnucu.com, Learning and Development News You Can Understand, we examine how interactive architectures perform when tested in real classrooms and workplaces, drawing on practitioner research and case studies presented at the H5P Toronto Community Event, also known as the H5P Toronto Meetup 2026, held at the Hilton Toronto on November 11, 2026. When instructional teams replace passive navigation with intentional decision checkpoints, they alter both cognitive engagement and diagnostic feedback. Instead of logging passive seat time, organizations begin verifying active competence. Long-term retention is neurologically difficult to build when a person's only physical action is clicking a navigation button every three minutes.

Two loops, two kinds of evidence
Passive clickthrough loop compared with an active competence loop The passive pattern runs in one line from Watch to Click Next to Move on while a progress bar fills. The active pattern is a circle: Observe, Decide, Receive diagnostic feedback, Recalibrate. Watch Click Next Move on The progress bar fills. Nothing asks the learner to decide. Passive seat-time loop Observe Decide Receive diagnostic feedback Recalibrate Active competence loop Decisions replace seat time

Passive clickthrough. Select a step below to see what each stage does and does not verify.

Passive seat-time loop compared with the active competence loop
PatternSequenceWhat it verifies
Passive seat-time loopWatch, click Next, move onSeat time. No decision is requested, so skill is not verified.
Active competence loopObserve, decide, receive diagnostic feedback, recalibrateActive competence, through decisions that expose and correct misconceptions.

Consider a hypothetical illustration, not a reported study. A regional utility is onboarding 40 junior plant operators on emergency valve-isolation protocols. Under a legacy compliance model, the operators watch a 28-minute narrated slideshow and pass a five-question recall quiz at the end. Three weeks later, during a simulated pressure spike, most of them hesitate, because they memorized terminology rather than practicing sequential judgment. Now imagine the utility redesigns the onboarding around three decision checkpoints that pause mid-procedure and ask operators to select the next isolation step before the pressure gauge responds. Operators confront their own misconceptions immediately, in a zero-risk environment.

Your team is asked to convert a 45-minute safety compliance PDF into an interactive digital asset. A stakeholder suggests adding "Click to Reveal" accordion tabs on every page so learners have to click before advancing. Does this satisfy active competence design?

Choose an answer to see the reasoning.

Practice question: a stakeholder proposes "Click to Reveal" accordion tabs on every page of a 45-minute safety compliance PDF. Does that satisfy active competence design? No. As Eric Young of Good Roads emphasized, a click is not a thought. A checkpoint must ask the reader to predict, diagnose or apply before the explanation appears.

Cognitive Mechanics

The Protege Effect and Multimedia Alignment

Standard unsegmented video often underperforms text-based study because it encourages mind wandering and superficial visual consumption. To overcome this, University of British Columbia (UBC) Associate Professors of Teaching Kayli Johnson (Chemistry) and Simon Lolliot (Psychology) ground their interactive video architecture in cognitive psychology, specifically combining the protege effect with Richard Mayer's Cognitive Theory of Multimedia Learning.

The protege effect describes the cognitive advantage gained when a learner teaches or guides someone else, forcing the brain to organize, retrieve, and articulate knowledge rather than passively recognizing it. In UBC's open-source textbook videos, Johnson and Lolliot created an apprentice persona named Chris Preston, a well-meaning but inexperienced 16-year-old lab assistant. At critical moments, the video pauses and Chris asks the viewer what to do next. By shifting the viewer from passive student to active mentor, the design triggers deeper generative processing. It mirrors the difference between riding as a passenger and driving a vehicle: a passenger can watch scenery pass for an hour without learning the route, whereas a driver must process every navigational choice to build a durable mental model.

Richard Mayer's embodiment principle reinforces this mechanism. Filming a physical procedure from a first-person point of view lets learners mentally rehearse the exact hand movements they will perform. Pairing that visual perspective with Mayer's temporal contiguity principle (delivering the spoken explanation at the exact moment the visual action occurs) and personalization principle (using conversational human narration) strengthens encoding during feedback.

Move the camera, change the mental rehearsal
Wide lecture shot compared with a first-person laboratory view An illustrated laboratory bench. In the wide shot an instructor stands far from the camera and the bench items are tiny. In the first-person view two forearms hold a spatula over a weigh boat, an apprentice speech bubble asks what to do next, and a sound wave stays aligned with the action. Wide shot APPRENTICE What should I do next? Spoken explanation arrives with the action First-person view
Wide lecture shot First-person view

Wide lecture shot. From across the room the learner watches someone else's hands. Slide toward the first-person view to see what changes.

Wide lecture shot compared with first-person framing
PrincipleWide shot from across the roomFirst-person view with an apprentice prompt
EmbodimentThe learner cannot rehearse the hand movements from their own angle.The learner mentally rehearses the exact movements they will perform.
Temporal contiguityCorrections arrive later, away from the moment of misconception.Spoken explanation arrives at the moment the action happens.
PersonalizationFormal narration toward a wall-mounted camera.Conversational narration, with an apprentice who asks the viewer what to do next.

Here is a second hypothetical illustration, again not a reported study. A hospital sterile-processing department wants new technicians to master autoclave loading rules. Instead of a presenter lecturing toward a wall-mounted camera, the video uses a first-person camera mounted at eye level over the loading cart. A newly hired apprentice character calls in on an audio prompt, holding an overpacked instrument tray sideways, and asks whether to run the cycle now or re-rack the instruments. The viewer must diagnose the steam-circulation error and choose the corrective action, which activates the protege effect while mentally rehearsing the exact tray orientation.

A video editor films a chemistry titration from across the lecture hall so the instructor's full body is visible, and places all corrective text feedback in a separate PDF handout downloaded after the video ends. Which two multimedia principles are being violated?

Choose an answer to see the reasoning.

Practice question: a titration filmed from across the lecture hall, with corrective feedback placed in a PDF handout after the video. The setup violates Mayer's embodiment principle (no first-person rehearsal of the hand manipulation) and temporal contiguity principle (feedback is separated from the moment of misconception).

The Learner's Mind

Metacognitive Calibration and Affective Scaffolding

One of the greatest hidden risks in video-based training is poor metacognitive calibration, the gap between how well people think they understand a topic and how well they actually perform. Because polished video feels effortless to watch, viewers mistake visual fluency for mastery. They recognize the material being presented and confuse that recognition with competence, which creates an illusion of mastery.

When Johnson and Lolliot tested branching interactive videos at UBC in demanding subjects such as IUPAC chemistry nomenclature, the standardized system for naming chemical compounds, where a single misplaced syllable alters the compound being named, pausing the video for consequential choices immediately exposed gaps in understanding. Students who experienced that real-time calibration demonstrated faster relearning two weeks later, compared with peers who watched non-interactive video.

Equally decisive is affective scaffolding, the emotional safety built into practice environments. High-stakes graded exams trigger stress that impairs memory retrieval and encourages ego defense. By contrast, ungraded formative sandboxes with unlimited attempts allow learners to fail privately and iterate without penalty. At the H5P Toronto Community Event, H5P Senior Director of Sales Nils Viken cited a PhD study by Theresa Huff showing that 60 percent of participants reported lower test anxiety with H5P interactive design, while Eric Young of Good Roads noted that ungraded practice removes severe exam-day panic among municipal field workers who carry negative memories of traditional schooling. Learners stop defending their ego and start focusing on understanding the system.

Where does a confident learner actually stand?

Hypothetical case. Technicians rate their own confidence at 92 percent after a training video, yet fail 45 percent of practical troubleshooting tasks in the field. Select the quadrant that describes this team, then explore the others.

High confidence
Low confidence
Incorrect performance
Correct performance

The token marks where the hypothetical team sits today. Select a quadrant to read what it means and what moves a learner out of it.

The four calibration states
StateConfidence and performanceWhat helps
Illusion of masteryHigh confidence, poor performanceUngraded, consequential branching checkpoints inside the video expose the gap before the field does.
Well-calibrated masteryHigh confidence, good performanceThe goal: branching aligns subjective confidence with objective performance.
Recognized gapLow confidence, poor performanceRoute wrong answers to descriptive feedback segments and allow repeated attempts.
Under-confident competenceLow confidence, good performanceUngraded practice removes exam anxiety so learners can see what they can do.

In the hypothetical case, technicians rate their confidence at 92 percent but fail 45 percent of field troubleshooting tasks. That is an illusion of mastery. Smooth video delivery creates superficial recognition without retrieval practice, and ungraded branching checkpoints fix it without triggering exam anxiety.

Choosing the Tool

Purpose-First Design and the Activity Selection Matrix

Moving from cognitive theory to daily production requires disciplined tool selection. At the H5P Toronto Meetup 2026, George Brown Polytechnic Educational Technology and Digital Content Specialist Evelyn Chan, author of AI in Education: Ethics, Literacy, and Interactive Design (eCampusOntario Pressbooks), stressed that effective design begins with the learning purpose, not the software widget. Designers must first ask what they want learners to think, do, or experience before selecting an interaction type.

Without that discipline, teams facing 54 distinct H5P interaction types often suffer decision paralysis or fall back on endless multiple-choice quizzes. To solve this authoring bottleneck, Eric Young and the team at Good Roads built a two-dimensional Activity Selection Matrix that cross-references the six levels of Bloom's Taxonomy (Remember, Understand, Apply, Analyze, Evaluate, Create) against five content typologies: facts, concepts, processes, procedures, and principles.

Here is a hypothetical illustration of the matrix at work, not a reported project. A pharmaceutical logistics team needs warehouse staff to master two distinct objectives. First, executing an exact five-step thermal-pack sealing sequence, which sits at Apply and Procedures. Second, deciding whether to quarantine or release a pallet after a 40-minute dock temperature excursion, which sits at Evaluate and Principles. Using the matrix, the designer pairs the five-step sealing sequence with an Image Sequencing activity rather than a word search or recall quiz, and pairs the quarantine judgment call with a Branching Scenario in which each decision triggers realistic spoilage and regulatory consequences.

Read the matrix like a designer
Bloom's level
Content typology

Pick a level and a typology. The two gold cells mark the pairings from the cold-chain illustration above.

The Good Roads matrix crosses six Bloom's levels (Remember, Understand, Apply, Analyze, Evaluate, Create) with five content typologies (facts, concepts, processes, procedures, principles).

The two pairings from the cold-chain illustration
ObjectiveMatrix positionPairing
Execute a five-step thermal-pack sealing sequenceApply and ProceduresImage Sequencing, not a word search or recall quiz
Quarantine or release a pallet after a 40-minute excursionEvaluate and PrinciplesBranching Scenario with realistic spoilage and regulatory consequences

The matrix also clarifies an important distinction between structural containers and choice questions. Containers such as Interactive Video, Interactive Book, Branching Scenario, and Virtual Tour (360) are shells that hold multiple activities, and their cognitive rigor depends entirely on the prompts placed inside them. Likewise, a multiple-choice question can test simple recall or complex judgment depending on how the scenario is framed. Finally, high-impact feedback loops must replace binary red X flags with descriptive guidance, routing incorrect answers to targeted animation or spoken explanation that resolves the underlying misconception.

The same wrong answer, two kinds of feedback
Binary flag
Waiting for an answer.
Descriptive routing
Waiting for an answer.

A red X signals an error without correcting the misconception. A branching interactive video can instead route an incorrect answer to a dedicated segment with synchronized animation and narration, which turns the mistake into targeted instruction.

Under Production Pressure

Three Builds in Higher Education, Hospitality and Project Management

Three real-world case studies from the H5P Toronto Community Event illustrate how purpose-first architectures hold up under production constraints.

First, at UBC, Kayli Johnson and Simon Lolliot implemented a pre-lab inversion strategy, replacing 30-minute in-person pre-lab lectures with interactive videos that test qualitative intuition, such as asking students to visually estimate 10 milligrams of a chemical on a spatula before entering the lab. Across their classroom evaluations, comprehension effect sizes nearly doubled, rising from a Cohen's d of 0.4 for in-person talks to 0.7 for interactive videos. Crucially, those gains depended on releasing the interactive videos a full week in advance to support spaced repetition, rather than the day before. They also discovered a vital technical rule: editors must leave a full two-second silence buffer before interaction markers in the timeline, so students watching at 2x speed do not experience jarring audio cutoffs while the interaction loads.

Second, at Centennial College, Professor Hoshedar Batliwalla built a 360-degree virtual reality orientation for the college's 20,000-square-foot hospitality Event Centre, using an explore, discover, and prepare sequence. Walking through the front desk, barista station, and back-of-house kitchen beforehand eliminates first-shift disorientation. Production required placing the 360-degree camera at chest height rather than tall tripod height to preserve natural human spatial proportions, alongside managing glass reflections and mobile responsiveness.

Third, Tariq Butt, Founder of TAliB Learning Inc., engineered The Last Sealift, an asynchronous project management branching scenario set in the remote northern community of North Reach. Learners act as project manager for a replacement health center when drift ice blocks the final supply ship, weighing a $1.19 million airlift against a one-year deferral. Building that realistic complexity required 90 distinct H5P nodes, 144 possible choice sequences, and 12 outcome screens, all held together by a strict file and asset naming taxonomy so that no branching path collapses into a dead end.

Inspect each build
UBC effect size and the two-second buffer A bar chart shows a Cohen's d of 0.4 for in-person pre-lab talks and 0.7 for interactive videos. A timeline shows speech, a two-second silence buffer, then an interaction marker before the next sentence. 0.4 0.7 In-person talk Interactive video Cohen's d, comprehension Timeline at 2x playback Speech 2 s silence Next line Interaction marker The buffer keeps the next sentence from cutting off while the interaction takes real clock time to load.
  • Replaced 30-minute in-person pre-lab talks with interactive videos that test qualitative intuition.
  • Videos released a full week in advance, not the day before, to support spaced repetition.
  • A full two-second silence buffer before every interaction marker.
Three builds under production pressure
BuildKey numbersProduction rule
UBC pre-lab inversionCohen's d 0.4 for in-person talks, 0.7 for interactive videosRelease videos a full week ahead and leave a two-second silence buffer before interaction markers.
Centennial College 360 orientation20,000 square foot Event CentreCamera at chest height, with reflections and mobile responsiveness managed.
The Last Sealift90 H5P nodes, 144 choice sequences, 12 outcome screens, $1.19 million airliftA strict file and asset naming taxonomy so no path becomes a dead end.

Infrastructure and Risk

Platform Convergence, Generative AI, and Governance

Complex interactive architectures require reliable enterprise infrastructure to scale. As D2L Founder and CEO John Baker and H5P's Nils Viken discussed at the H5P Toronto Community Event, D2L's acquisition of H5P accelerates the shift from fragile HTML embed codes toward native Learning Tools Interoperability (LTI) integration. Embed codes frequently break gradebook synchronization when browser updates or network hiccups occur, whereas LTI provides a secure, deep data pipe between the platform and the interactive content.

Meanwhile, generative AI tools such as H5P Smart Import and D2L Lumi, demonstrated in workshops by H5P's Aaron Fitzgibbon and Andrew Furgal and echoed in practitioner results from Scott Hallman at the University of Otago, are inverting authoring economics. Drafting assessment items shifts from 15 to 20 minutes per question down to curating and refining auto-generated drafts in seconds, operating at a 10x to 100x scale. Subject matter experts move from blank-page drafting to curating, refining, and validating.

Yet speed without governance introduces serious risk. Enterprise deployments must enforce strict data hygiene, with customer inputs excluded from large language model training sets and erased within 30 days, to protect proprietary manuals and meet privacy requirements. Designers must also guard against AI hallucinations in unvetted drafts and against cognitive offloading, where learners use AI to bypass the productive struggle needed to build mental models. Human-in-the-loop oversight and Universal Design for Learning (UDL) accessibility, including closed captioning, keyboard navigation, and screen-reader compatibility, remain essential.

Audit an AI authoring workflow
AI authoring pipeline with five safeguard gates A pipeline runs from a dense manual to an AI draft, expert review, delivery through the learning platform, and learners. Five numbered gates sit along it and turn green when the matching safeguard is switched on. Dense manual AI draft Smart Import, Lumi Expert review Platform delivery Learners 1 2 3 4 5

4 of 5 safeguards missing

    Before launching an AI authoring workflow, confirm five safeguards:

    • Customer inputs are excluded from large language model training sets.
    • Inputs are erased within 30 days.
    • Subject matter experts review every AI draft, which guards against hallucinations in unvetted content.
    • Native LTI integration replaces fragile embed codes, which prevents gradebook synchronization failures.
    • Universal Design for Learning accessibility is in place: closed captioning, keyboard navigation and screen-reader compatibility.

    Looking Ahead

    The Future of Modular Learner Agency

    Intentional, modular interactive design redefines instructional credibility across corporate and academic settings. The value of these architectures lies in verifying active competence and diagnostic decision-making rather than logging passive seat time. When organizations replace passive seat-time tracking with ungraded formative sandboxes, purpose-first activity selection, and diagnostic feedback loops, learners who predict outcomes and apply principles build resilient mental models that transfer directly to real-world performance.

    Looking ahead, learners will increasingly use AI tools under clear pedagogical guardrails to convert dense technical manuals into personalized, scenario-based practice sandboxes calibrated to their own knowledge gaps, transforming static reference material into adaptive rehearsal and placing high-fidelity cognitive rehearsal directly in the hands of the end user.

    From dense manual to personal practice sandbox
    Three ways a learner can use a dense technical manual A stack of manual pages on the left connects to a learner on the right. With a static manual the learner rereads. With unguarded AI the learner hands the thinking to the tool. With guardrails the manual becomes a practice loop of attempt, feedback and retry, with expert oversight. Dense technical manual Reread and recognize AI does the thinking Unvetted drafts, no productive struggle Attempt Feedback, retry Human oversight The learner

    Static manual. The reference stays dense and static. The learner can reread it and recognize it, which is the passive pattern this whole piece began with.

    Three ways to use a dense technical manual
    ApproachWhat happens
    Static manualThe learner rereads and recognizes static reference material.
    AI without guardrailsRisk of hallucinations in unvetted content and of cognitive offloading, where the learner bypasses the productive struggle that builds mental models.
    AI with pedagogical guardrailsThe manual becomes a personalized, scenario-based practice sandbox calibrated to the learner's own knowledge gaps, with human oversight and UDL accessibility.

    Grounded in purpose-first instructional design, these interactive frameworks put the tools for high-fidelity cognitive rehearsal directly into the hands of the learner. Thanks for learning with us at ldnucu.com, Learning and Development News You Can Understand.

    Watch the podcast on YouTube

    Previous Post Next Post

    نموذج الاتصال