Prompting, Prompt Fading, Modeling, Instructions, and Rules
Key Takeaways
- A prompt is a supplementary antecedent that increases correct responding; the goal is always transfer of stimulus control to the natural SD, never permanent prompting.
- Response prompts act on the learner's behavior (verbal, model, physical/gestural); stimulus prompts act on the teaching materials (highlighting, position, exaggerating a relevant feature).
- Most-to-least and errorless approaches minimize errors for new/high-risk skills; least-to-most promotes independence for skills partly in repertoire but allows initial errors.
- Progressive/constant time delay transfers control by inserting a delay between the SD and the prompt; graduated guidance fades physical support within and across trials.
- Prompt dependence is a control failure - the response is under the prompt's control, not the natural cue - and is prevented by systematic fading plus reinforcement for unprompted responding.
Prompting Is Stimulus-Control Engineering
A prompt is a supplementary antecedent stimulus that increases the likelihood of a correct response before the response is reliably under the control of the natural SD. Prompts are temporary scaffolding. The terminal goal is transfer of stimulus control from the prompt to the relevant natural cue. A plan that prompts without a fading mechanism risks prompt dependence.
Prompts split into two families:
- Response prompts act on the learner's behavior: verbal instructions, modeling (showing), and physical/gestural guidance. These vary in intrusiveness, roughly from gestural (least) up through verbal, model, partial physical, and full physical (most).
- Stimulus prompts act on the teaching materials/antecedent: highlighting the correct choice, changing its position or size, or exaggerating a relevant feature (then fading that exaggeration). Stimulus prompts split into within-stimulus prompts (altering the relevant SD itself) and extra-stimulus prompts (adding a separate cue, such as a pointing arrow).
Prompt-Fading Hierarchies
The two response-prompt directions differ in error tolerance:
- Most-to-least (MTL): start with the most intrusive prompt that guarantees success, then systematically fade to less intrusive prompts. Best for new or difficult skills and high error cost (errorless-style teaching). Risk: independence is slow if fading lags.
- Least-to-most (LTM): start with the least intrusive prompt (or none) and increase support only as needed. Best for skills partly in repertoire; promotes independence but allows initial errors, which can be costly for safety skills.
| Strategy | Best fit | Risk to monitor |
|---|---|---|
| Most-to-least | New/difficult skills, high error cost | Slow independence if fading lags |
| Least-to-most | Skills partly in repertoire | Repeated errors before stronger prompts |
| Constant time delay | Transfer from prompt to natural cue | Errors if delay introduced too fast |
| Progressive time delay | Gradual transfer of control | Slower; requires careful step sizes |
| Graduated guidance | Motor/physical responses | Intrusive support if not faded promptly |
| Stimulus fading/shaping | Discrimination training | Learner attends to the irrelevant cue |
Time delay inserts a delay between the natural SD and the prompt. In constant time delay, the delay jumps from 0 s to a fixed value (often 3-5 s); in progressive time delay, it grows in small steps. The learner learns to respond to the SD before the prompt arrives, transferring control. Graduated guidance fades the intensity of physical support within and across trials, often shadowing the learner's movement.
Response Versus Stimulus Prompts
Know the dividing line cold: if you change what you do to the learner (say it, show it, guide the hand), it is a response prompt. If you change the materials (make the correct card bigger, brighter, or closer), it is a stimulus prompt. Stimulus fading gradually removes an exaggerated material feature; stimulus shaping gradually changes the form of the stimulus itself. Both transfer control to the untouched natural stimulus array. A classic error is calling a positional cue (moving the correct answer closer) a response prompt — it is a stimulus prompt.
Modeling, Instructions, and Rules
Modeling works when the learner can imitate the modeled action and the action then contacts reinforcement. It is efficient for social, motor, vocational, and daily-living skills. It is insufficient when generalized imitation is weak or the model is not salient. A video model can increase salience and consistency and is reusable across staff. Modeling is itself a response prompt, so it too must be faded to transfer control to the natural cue.
Instructions and rules are verbal descriptions of contingencies. They can change behavior rapidly without the learner contacting every contingency directly (rule-governed behavior). The catch: rule-governed behavior can be insensitive to changing consequences — a learner may keep following a rule even after the contingency shifts. So you must verify with data that the rule still produces the intended behavior in the natural context, and that behavior comes under contingency-shaped control where durability matters.
Instructions also require listener (receptive) skills: an instruction is useless if the words do not yet function as stimuli for the learner.
Selection Logic and Preventing Prompt Dependence
Ask three questions: (1) Are errors harmful? If yes, lean MTL/errorless. (2) Does the learner have prerequisite imitation/listener skills? If not, modeling or verbal prompts will fail. (3) Can staff implement the sequence with integrity?
Always pair prompts with reinforcement for correct responding and, crucially, reinforce unprompted (independent) responses more richly than prompted ones — a differential reinforcement of prompted vs. unprompted arrangement that keeps independence the most worthwhile option. Prompt dependence means the response is under the prompt's control rather than the natural cue's — diagnosed when the learner responds only after an added cue. Causes include fading that is too slow, weak reinforcement for independence, unclear SDs, or criteria advanced too quickly.
A related concept is the transfer-of-stimulus-control trial: prompt the response, then immediately re-present the SD with a reduced or absent prompt and reinforce the now-unprompted response. Done systematically, this is how every prompting strategy hands control back to the natural environment. The exam's correct answer almost always includes a fading plan; an option that prompts forever, however effective in the moment, is a trap.
A behavior analyst is teaching a safety skill (crossing at a signal) where an error could be dangerous. Which prompting approach is MOST defensible?
During discrimination training, a teacher places the correct picture card noticeably closer to the learner and slightly larger, then gradually moves it back to the array and equalizes its size across trials. This is BEST described as:
A learner reliably completes a task only after an adult repeats the instruction a second time; without the repeat, the learner does nothing. This pattern most likely indicates:
Which statement about rule-governed behavior is the MOST important caution for a behavior analyst?