What counts as a prompt
A prompt is supplementary help delivered alongside the instruction, and it is meant to be temporary. The goal of every prompting procedure is the same: transfer control of the response from the prompt to the instruction or natural cue, so the learner eventually responds without the help.
Prompts come in two broad kinds. Response prompts act on the learner's behavior directly - physical guidance, a model to imitate, a gesture, or a verbal hint. Stimulus prompts change the materials instead - placing the correct card closer, making it larger or brighter, or adding a picture cue - and are faded by gradually returning the materials to normal. Most prompt hierarchies describe response prompts, with visual prompts included where a clinic uses them.
The prompt levels, from most to least intrusive
The table below is a common hierarchy, ordered from the most support to the least. "Intrusive" here means how much the prompt controls the response: full physical guidance produces the response almost entirely, while a visual cue only points the way.
| Prompt level | What the clinician does | Example: teaching "touch your nose" |
|---|---|---|
| Full physical | Guides the learner through the entire response, hand over hand. | Gently guides the learner's hand all the way to their nose. |
| Partial physical | Uses a light touch or nudge to start the response, and the learner completes it. | Lightly lifts the learner's elbow toward their face. |
| Model | Demonstrates the response for the learner to imitate. | Touches their own nose. |
| Gestural | Points, nods, or looks toward the correct response without performing it. | Points toward the learner's nose. |
| Verbal | Adds words beyond the instruction: a hint, part of the answer, or the whole answer. | Says, "It's on your face." |
| Visual | Adds a picture, written word, or other visual cue that indicates the response. | Holds up a picture of a child touching their nose. |
| Independent | Gives the instruction alone, with no added help. | The learner touches their nose after hearing "touch your nose." |
Why the order is not universal
There is no single correct ordering. Verbal prompts are physically unobtrusive, but they are often among the hardest to fade, because learners come to wait for the words; for that reason some clinics place verbal prompts above gestural ones, or treat verbal and visual prompts as separate tracks rather than rungs on one ladder. A hierarchy also depends on the skill: a model is meaningless for a receptive task the learner cannot see performed, and physical prompts rarely fit vocal responses.
What matters is that the hierarchy is written down for each program and used the same way by everyone who runs it. A "gestural" mark means nothing if one clinician uses it for a point and another for a glance, and that kind of inconsistency is a common cause of low interobserver agreement on prompted targets.
Most-to-least and least-to-most prompting
Most-to-least prompting starts with the level of help that guarantees a correct response and fades to less help as the learner succeeds, usually across trials or sessions. It keeps errors rare, which suits new skills, learners whose errors tend to repeat, and situations where an error is unsafe. Its risk is prompt dependence: if fading is too slow, the learner learns to wait for help.
Least-to-most prompting, also called the system of least prompts, does the reverse within each trial. The learner gets a chance to respond independently, and the clinician steps up through the hierarchy only as far as needed. It suits skills partly in the learner's repertoire and the steps of a task analysis the learner already does on some days. Its cost is more errors and longer trials while help escalates.
Either way, write the fading rule into the program: when the learner moves to a less intrusive level, and when they move back. A rule such as "move down one level after two consecutive sessions at 80% or more at the current level" keeps fading from drifting with whoever happens to run the session.
Time delay and other fading strategies
Prompt delay fades help by changing when it arrives rather than what it is. In constant time delay, early trials deliver the prompt immediately after the instruction - a zero-second delay - and later trials wait a fixed interval, often a few seconds, so the learner can respond before the prompt comes. Progressive time delay lengthens the wait gradually. Both keep errors low while giving independent responding room to appear.
Graduated guidance adjusts physical help moment to moment, giving only as much as the learner needs and shadowing their hands as help is withdrawn. Stimulus fading reduces a stimulus prompt, such as an enlarged or highlighted card, until the materials look as they will in everyday use. All of these share one principle: help is planned to disappear, and the schedule for removing it is part of the program.
Prompt dependence and how to spot it
Prompt dependence is the point where a learner responds correctly only with help. In the data it looks like high percent correct with flat percent independent, responding that stalls at one prompt level session after session, or a learner who waits after the instruction rather than attempting a response.
It is far easier to see when the prompt level is recorded on every trial than when trials are scored only as prompted or unprompted. Common remedies are fading faster, adding a time delay, and reinforcing independent responses more richly than prompted ones, so that responding without help is clearly worth more to the learner.
Recording prompt levels in the data
Score the level that was actually used, on each trial or each step. Under least-to-most prompting, that is the lowest level at which the learner responded correctly. Count only unprompted correct responses toward percent independent; a prompted correct response is progress, but it is not independence, and mixing the two is how a graph shows mastery a learner has not reached. The discrete trial training guide covers how percent correct and percent independent differ.
Keep the hierarchy itself with the data. If a program's levels change partway through, older sessions should still be read against the hierarchy they were scored under, or trend lines will compare different definitions of the same mark.
How Cadence handles prompt levels
Cadence scores prompted trials against a prompt hierarchy each clinic can configure per target, and falls back to a seven-level default - full physical, partial physical, model, gestural, verbal, visual, and independent - when none is set. A trial counts toward a target's percent independent only when it was correct and unprompted, so the graph and the mastery criterion measure the same thing.
The same prompt levels apply to each step of a task analysis, and they are what Cadence's capture shows on screen, whether a clinician taps the level or says it aloud. All 8 measurement systems are native, so prompt data sit inside the target they belong to rather than in a notes field.