For RBTs, Behavior Technicians, Educators & Parents implementing teaching programs
Practitioner provides complete hand-over-hand guidance through the entire target response.
Example: guiding a learner's hand completely through placing a puzzle piece.
Light physical guidance — a touch, nudge, or partial assist — without completing the movement for the learner.
Example: a gentle tap on the elbow to initiate reaching for an object.
Demonstrate the target response for the learner to imitate. Can be full or partial modeling.
Example: picking up the spoon and putting it in the bowl, then handing it back.
A physical gesture (point, nod, gaze shift) directs the learner without physical contact or modeling.
Example: pointing to the correct picture card on the table.
Environmental cue: a picture schedule, color code, written cue, or object placement that guides responding.
Example: placing the correct item closer to the learner as a positional cue.
A spoken instruction, hint, or partial verbal prompt (first sound, partial phrase) delivered without gestures or physical assist.
Example: saying "What do you want?" or providing the first sound "Wa—".
Define your fading criteria BEFORE the session. Decide how many correct responses at what level trigger moving to the next step.
Record the prompt level used AND whether the response was correct. You cannot fade what you do not measure.
Deliver the highest-magnitude reinforcement for prompted-free (independent) responses. Differentially reinforce lower prompt levels.
Pair prompts with the natural SD so the natural cue gains stimulus control. The prompt should become unnecessary — not a permanent fixture.
| Situation | Strategy | Why | Watch out for |
|---|---|---|---|
| Brand-new skill with zero prior responding | Most-to-Least (MTL) | Prevents error practice; builds fluency from a supported correct response | Fading too slowly → prompt dependency |
| Skill the learner is almost mastering | Least-to-Most (LTM) | Gives learner a chance to be independent first; avoids unnecessary support | Waiting too long between prompts → frustration |
| Safety skill (e.g., stop at curb) | MTL with slow fade | Errors in safety contexts carry real risk; errorless learning is non-negotiable | Fading before fluency is established |
| Maintenance / generalization probe | No prompt — natural SD only | Probes measure true stimulus control; prompting invalidates the probe | Accidentally prompting during probes (even gesturally) |
| Learner showing prompt dependency | LTM with extended wait | Extending latency re-establishes independent responding as the first behavior | Giving in to challenging behavior by prompting prematurely |
If an error occurs, implement your program's error-correction procedure (e.g., model → test → distractor → retest) rather than simply re-delivering the trial with a higher prompt. Error correction targets the specific incorrect response; prompting prevents errors. Know which procedure your program calls for before every session.
Want more? Explore free and full training resources in the Special Learning library.