Most-to-least prompting

Teaching begins with effective assistance

The clinician first identifies a controlling or highly effective prompt that is likely to occasion the response. The prompt occurs with the natural cue according to the plan. After stable correct responding, assistance reduces through prewritten levels.

For opening a transit app, the initial prompt might be a full video model. Later levels could be a partial model, a visual arrow, and the ordinary app screen alone. Physical guidance is one possible prompt form, not a required first level.

Fading requires observable criteria

“Fade when ready” creates uneven decisions. A plan can require a defined number or percentage of correct prompted responses, low latency, stable performance across sessions, or another measurable criterion before changing levels.

It should also state when to return to more support, pause, or revise the procedure. A prompt that stops working, rising distress, equipment change, or repeated errors may show that the current step is too large or the cue needs redesign.

Independent responding needs its own measurement

Correct responses during strong prompts show that the prompt can produce the response. They do not show that the natural cue controls it. Insert planned delay trials or separate unprompted probes when clinically appropriate, and keep those data distinct from prompted teaching.

A delayed version combines early error prevention with an opportunity to respond before the prompt. The delay interval and what happens after an error need clear definitions.

The procedure can reduce errors

Strong early prompting can be useful when repeated errors carry safety, effort, or learning costs. It may also reduce contact with independent attempts if fading is slow. Staff can become part of the cue, and the learner may wait for assistance.

Review prompt salience, position, timing, teacher proximity, and response latency. Prompt dependence is an instructional signal that calls for reassessment, not a character trait of the learner.

Comparative research supports individualization

Libby and colleagues taught autistic children to build Lego play structures. The most-to-least condition produced acquisition for all participants and fewer errors than least-to-most, while three participants learned faster with least-to-most. A delayed version produced fewer errors than least-to-most and nearly comparable speed in the second study.

A study of positional prompts illustrates another most-to-least arrangement for receptive identification. Its narrow participants, stimuli, and prompt forms cannot determine another learner’s plan. Together, these papers show why prompt type and fading require direct measurement.

Clinical authority and consent remain visible

The BACB BCBA Test Content Outline, 6th edition covers prompting, fading, stimulus control, observable goals, contextual fit, integrity, and evaluation. It is examination content, not a clinical protocol.

For covered behavior analysts, the current BACB Ethics Code addresses competence, client involvement, consent and assent when applicable, risk, assessment, data, and ongoing evaluation. Any physical prompt needs explicit justification, competent delivery, applicable consent and assent, health and trauma review, and an accessible way to stop.

AAC, communication, breaks, mobility, food, water, bathroom access, prescribed care, and emergency help remain available throughout teaching.

A fictional fading record

Elliot practices attaching a bike light before an evening ride. Across 12 teaching opportunities, the plan starts with a model and fades to a gesture. Elliot completes 9 of 12 correctly at the assigned prompt level. In four separate natural-cue probes, Elliot responds independently in 2 of 4.

Teaching performance is 9/12 and independent probe performance is 2/4. Combining them would blur prompt effectiveness and stimulus-control transfer. The record also preserves latency, errors, prompt changes, pauses, and Elliot’s view of the routine.

Review learning, burden, and transfer

Track correct responses at each prompt level, trials to a fading criterion, independent probe performance, errors, latency, integrity, duration, and generalization. Define eligible opportunities and exclusions before data collection.

If independence lags, review the prompt itself, fading size, natural cue, delay, reinforcement, task design, and goal fit. Continuing a controlling prompt indefinitely is not a fading plan.

Fading decisions should be reversible when evidence changes. If a weaker prompt produces repeated errors or distress, the clinician can restore effective support, examine the step size, and test another route. Record every level change with its date and basis so later reviewers can distinguish planned fading from staff inconsistency. Across people and settings, compare the same response definition while allowing ordinary supports that the learner will continue to use.

Review those records together.