Skip to main content

3. Plan the support, then fade it

Decide: how much help do learners get at the start, and how does it go away?

Help is like scaffolding on a building: essential at first, in the way once the building stands. The goal is a learner who no longer needs the lesson.

3.1 Show it done before learners do it​

Do

  • Let beginners see the whole task done once, the way an expert does it, before they try it themselves.
  • Show not just what the expert does, but why: what they check, what they look out for.
  • When you record a demonstration, say out loud what you are checking and what beginners usually get wrong at that moment.

In ALF

The learner experiences: they've seen it done once, on the same machine they are about to use, and know what they are aiming for.

Avoid: letting beginners figure out the task by trial and error.

Why: beginners learn more from studying worked examples than from solving problems on their own. Showing the expert's thinking turns imitation into understanding. Worked-example effect (Sweller), Cognitive Apprenticeship (Collins, Brown & Newman).

3.2 Remove the help from the end first​

Do

  • Run 1: every step has a full instruction.
  • Run 2: the last step only has a short cue, for example just its name.
  • Run 3: the last two steps only have cues, and so on, until learners do the whole task alone.

In ALF

  • Make each run its own scene, or copy the scenario and shorten the panels: Write the instructions.
  • Put the runs in order in the main menu.
  • With a Motion demonstration, make the ghost hands more see-through in later runs (Ghost Hand Opacity), or leave the playback out once learners know the task.
  • In AR, lower the model's opacity from run to run, until the learner works on the real object alone: Set transparency.

The learner experiences: they do the end of the task alone early, and know the start well when they get there.

Avoid: removing the help everywhere at once, or from the start of the task first.

Why: taking away worked steps from the end works better than from the beginning, because learners still have the full example for the parts they haven't mastered. Backward fading (Renkl & Atkinson).

Not built in yet

ALF doesn't fade help automatically depending on how a learner performs. Build the runs as separate scenes for now.

3.3 Give each stage less help than the one before​

StageTypical help
OrientOverview, background, the words learners will need.
Practise in VROne instruction per step, next to the part. Then shorter in later runs.
Transfer in ARLabels and key cues on the real object. Few words.
Perform for realNone from the system. A checklist for the learner or a supervisor.

In ALF: the instruction panel's position and size are set per step: Place the instruction panel. Labels on parts: Add a label.

Why: support should decrease while realism and learner independence increase. 4C/ID, Fitts & Posner's stages of motor learning.

3.4 Keep background knowledge out of the steps​

Do

  • Explain how things work (parts, principles, reasons) at the start, in the orientation, or in an exploded view.
  • In the steps, say only what to do now.

In ALF: an exploded view in the orientation for the background; short step instructions for the procedure.

Avoid: long explanations in the middle of a procedure, which break the flow and are forgotten by the next step.

Why: background knowledge is best studied beforehand, step-by-step instructions are best given just in time. 4C/ID: supportive and procedural information.

3.5 Let learners choose more or less help​

Do

  • If your group is mixed, offer the lesson at two levels, for example With guidance and On my own.
  • Tell learners they can go back to the guided version at any time, and that going back a step is normal.

In ALF

Avoid: forcing experienced learners through every guided step.

Why: help that beginners need gets in the way of learners who already know more. Learners who plan and check their own learning learn more. Expertise reversal (Kalyuga), self-regulated learning (Zimmerman).

Not built in yet

ALF can't yet adapt the help to how a learner performs. Learner choice is the workaround. Pair it with a short self-check, so learners judge their level well.