Discrimination, Prompting and Fading

What You Will Be Able to Do

Teaching a new skill often means helping a learner respond correctly, then gradually removing that help. This lesson shows how to tell whether a learner is actually responding to the right cue, and how help is added and faded.

By the end, you should be able to:

• Explain why rotating card positions shows discrimination rather than position guessing.
• Record each trial as independent or prompted without adding interpretation.
• Classify a prompt as a stimulus prompt or a response prompt.
• Distinguish stimulus fading, least-to-most prompting, simultaneous prompting and time delay.
• Follow a written prompting plan, know what to do if the learner protests or shows distress, and know when to check with your supervisor.

No background is needed. Terms are explained as they appear.

Discrimination: Responding to the Right Cue

Discrimination means the learner responds differently depending on which relevant cue is present. This happens because those responses have been followed by different consequences. Touching the named card is reinforced. Touching the other card is not. Reinforcement is a consequence that makes a response more likely in the future.

Here is an example. Rae, a learner, sees two cards: a cup and a shoe. You say, "Touch shoe."

• Trial 1: shoe on the left. Rae touches left. This is correct.
• Trial 2: shoe on the right. Rae touches left. This is an error.
• Trial 3: shoe on the right. Rae touches right. This is correct.

Why rotate the positions? Suppose the shoe were always on the left. Rae could touch the left card every time and score perfectly while ignoring the pictures. When you rotate positions, only responding to the picture is reliably correct. Sometimes asking "Touch cup" instead does the same job for the instruction. Neither one side nor one card should always be right.

Check 1: A learner is correct on 5 of 5 trials. The target card was on the right every time. Did they show discrimination?

Answer: Check 1

You cannot tell yet. Always touching the right side would produce the same score. A high score is not evidence of discrimination unless position has been rotated. Run trials with rotated positions, or report the problem with the data sheet to your supervisor.

Recording What You Saw

Record each trial as one of these: independent correct, prompted correct, independent error, prompted error or no response. Write down what you observed, such as "touched shoe after model." Do not write an interpretation, such as "Rae understands shoes." An interpretation is not an observation.

Suppose 10 trials produced 6 independent correct, 3 prompted correct and 1 independent error.

• Percent independent correct = 6 ÷ 10 = 0.6 = 60%.
• Percent correct including prompted = (6 + 3) ÷ 10 = 9 ÷ 10 = 90%.

These numbers measure different things. Always say what you counted and what you divided by.

Also say when and under what conditions the data were collected. Two percentages can look alike and still describe different things. For example, 60% independent correct in one 10-trial session is not the same measure as 60% across a whole week of sessions. A percentage collected while the prompt came immediately is not the same as one collected while you waited 4 seconds before prompting. Before comparing numbers, check that the observation period and conditions match.

A trial that was never run is not an error and not a zero. Do not invent a score for it. Follow your data sheet and your supervisor's rules for how unrun trials are marked and counted.

Two Families of Prompts

A prompt is extra help that occurs before or during a response, so the learner responds correctly while the natural cue is present. There are two families.

• Stimulus prompts change the materials or cue so the correct choice stands out. Examples include placing the correct card closer (position), adding a feature such as a bright border (redundancy), or wiggling the correct card (movement).
• Response prompts act on the learner's response. Examples include spoken or signed instructions (verbal), demonstrating the response yourself (model), and guiding the learner's hand (physical guidance).

A quick test: Did I change what the learner is looking at, or did I help what the learner does?

Fading: Moving Control to the Natural Cue

The natural cue is what should eventually signal the response. Here, that is "Touch shoe" with ordinary cards. If help never fades, the learner may develop prompt dependence, which means waiting for help instead of responding to the natural cue. Fading is planned from the start, but transfer is gradual. Prompts are often needed at first, even when the natural cue is present.

• Stimulus fading: The added feature shrinks step by step. For example, a thick yellow border becomes thin, then disappears. You watch whether responses stay correct as it shrinks.

• Least-to-most prompting: The learner first gets a chance to respond with no help. If no correct response comes within the plan's interval, you give the next prompt in the plan, increasing help until the controlling prompt. The controlling prompt is the prompt that reliably produces a correct response. The AFIRM procedure calls for at least three levels: independent, at least one intermediate prompt, and controlling. For example: independent → verbal → model. Your written plan sets the actual levels and their order. There is no single universal order.

• Simultaneous prompting (an errorless approach): "Errorless" means arranging help so that errors rarely happen, instead of correcting errors afterward. In teaching sessions, you give the controlling prompt immediately with the instruction, every trial. The prompt is not faded during teaching trials. Instead, separate probe sessions use no prompt and test whether the learner responds independently.

• Time delay: In the AFIRM procedure, teaching begins at a 0-second delay, meaning the controlling prompt comes immediately, for at least two trials. Then the wait increases in one of two ways:

  • Constant time delay uses a fixed wait every trial. AFIRM describes 3–5 seconds as typical.
  • Progressive time delay increases the wait gradually.

  If the learner responds correctly before the prompt, you reinforce the response. Your plan sets the actual delay and when to move past 0 seconds.

The key contrast is what happens to the help. Least-to-most starts with no help and adds prompts only as needed. Simultaneous prompting keeps the controlling prompt immediate in every teaching trial and checks independence in separate unprompted probe sessions. Time delay also starts with the controlling prompt immediate, then systematically lengthens the wait before it, so the learner gets a growing chance to respond first.

Your Role

Follow the authorized, individualized plan from your supervisor. The plan specifies which prompts to use, their order, the delay and the fading rules. Do not add physical guidance or require eye contact unless the plan includes it. Use only procedures you have been trained in and confirmed competent to perform.

A plan never overrides the learner's safety, the learner's rights, the law or your workplace's requirements. If the learner pulls away, protests or shows distress, treat that as important information, not something to push through. Do not improvise a new procedure or add prompts to get the response. Use the plan's procedure for this situation if it has one. If there is a risk of harm or a concern about the learner's rights, take the necessary steps to protect the learner and follow your organization's policies. Tell your supervisor promptly and document what happened as required.

Report data problems to your supervisor promptly.

Practice on Your Own

1. A data sheet places the target card on the left for all 10 trials. What is the problem?
2. You print the correct card larger than the other card. Is this a stimulus prompt or a response prompt?
3. During teaching sessions, you model the response right after every instruction. In separate probe sessions, you give no prompt. Which procedure is this?
4. A plan lists: independent → gesture → model. The plan states that the model reliably produces a correct response for this learner. Which procedure is this, and which level is the controlling prompt?
5. Over sessions, you wait 1 second, then 2, then 3 before prompting. Is this constant or progressive time delay?
6. Over sessions, a red outline around the correct card becomes fainter. Which procedure is this?
7. After the instruction, you say, "It's the one with laces." Is this a stimulus prompt or a response prompt?
8. A coworker says, "Rae was 70% independent correct today and 70% last month, so nothing has changed." Today's number came from one 10-trial session. Last month's came from all sessions that month. What should you point out?

Answer: Practice on Your Own

1. Touching the left card would always be correct, so the data could not show discrimination. Positions need rotating. Raise this with your supervisor.
2. It is a stimulus prompt (redundancy). It changes the material, not the learner's response.
3. This is simultaneous prompting. The prompt is immediate in every teaching trial and is not faded there. Unprompted probe sessions check independent responding.
4. This is least-to-most prompting. Model is the controlling prompt because the plan says it reliably produces a correct response. Being listed last is not what makes a prompt controlling.
5. It is progressive, because the wait grows. Constant would use the same wait every trial after the 0-second trials.
6. This is stimulus fading. An added feature of the material is gradually removed.
7. It is a response prompt (verbal). It gives the learner spoken help with the response; the cards themselves do not change.
8. The two percentages cover different observation periods, so they are not automatically comparable. Matching numbers do not show that performance is the same.

Apply a Written Plan

Plan: Rae touches the named card from two (cup, shoe). Positions follow the data sheet's rotation. The procedure is constant time delay. The controlling prompt is a model, where you touch the correct card. Trials 1–2 use a 0-second delay. Later trials use a 4-second delay.

• A correct response before the prompt earns praise plus a sticker and is recorded as independent correct.
• A correct response after the model earns brief praise and is recorded as prompted correct.
• An error or no response within 4 seconds is followed by the model.

Answer these:

a) Trial 3: You say "Touch cup." Rae touches the cup at 2 seconds.

b) Trial 5: Rae touches the shoe at 1 second when asked for the cup.

c) A coworker suggests hand-over-hand guidance because it is faster.

d) Trial 7: After the model, Rae pushes the cards away, cries and turns away from the table.

Answer: Apply a Written Plan

a) Give praise plus a sticker, and record independent correct. Rae responded before the 4-second prompt.

b) Record independent error and give the model. Rae made an error before any prompt, and the plan says to model after an error.

c) Do not switch. Physical guidance is not in the plan. Bring the suggestion to your supervisor.

d) Do not push through by repeating the instruction, adding prompts or guiding Rae's hand. This plan says nothing about distress, so do not invent a procedure. Protect Rae if there is any risk of harm, follow your organization's policies, tell your supervisor promptly and document what you observed.

Recap

• Discrimination means responding to the relevant cue. Rotate positions so a position habit cannot look like skill.
• Record observations, not interpretations. State your denominator, and compare numbers only when the observation period and conditions match. Unrun trials are not zeros; follow your data sheet's rules for them.
• Stimulus prompts change the materials. Response prompts help the learner's response.
• Least-to-most starts with no help and adds prompts as needed. Simultaneous prompting gives the controlling prompt immediately in every teaching trial and checks independence in separate probe sessions. Time delay starts with an immediate prompt, then lengthens the wait before it.
• Follow your individualized plan and stay within your competence. A plan never overrides safety, client rights, law or workplace requirements. Report and document concerns, including protest or distress, to your supervisor.

Sources

• Sam, A., & AFIRM Team (2024). Prompting and Time Delay, updated. UNC Frank Porter Graham Child Development Institute.
• Kansas TASN. Stimulus and Response Prompts.
• Grow, L., & LeBlanc, L. (2013). Teaching receptive language skills: Recommendations for instructors. Behavior Analysis in Practice, 6(1), 56–75.
• Fisher, W. W., et al. (2019). Establishing initial auditory-visual conditional discriminations and emergence of initial tacts in young children with autism spectrum disorder. Journal of Applied Behavior Analysis, 52, article starting at 1089. Article. © 2019 The Authors. CC BY 4.0. Concepts paraphrased; course examples are original.
• BACB (2023). RBT Test Content Outline (3rd ed.); BACB (2021). RBT Ethics Code (2.0).

Independent study resource

This study resource is not affiliated with, sponsored by, or endorsed by the Behavior Analyst Certification Board (BACB). BACB, Registered Behavior Technician and RBT are registered trademarks of the Behavior Analyst Certification Board. Source citations do not imply endorsement by their authors or publishers.