C.6 Discrimination training
Discrimination training teaches a client to respond one way when a particular stimulus is present and not to respond that way (or to respond differently) when other stimuli are present. The RBT runs it by reinforcing the target response in the presence of the discriminative stimulus (SD) and withholding reinforcement in the presence of other stimuli (S-deltas), exactly as the supervisor's plan describes.
What the outline says
Task C.6 in content area C, Behavior Acquisition (19 of the 75 scored questions): "Implement discrimination training."
The key terms
A discriminative stimulus (SD) is a stimulus in whose presence a response has been reinforced. Because of that history, the response becomes more likely when the SD is present. An SD does not force a response; it signals that reinforcement is available for it.
An S-delta (SΔ) is a stimulus in whose presence the same response has not been reinforced. The response becomes less likely when the S-delta is present. An S-delta is an antecedent, not a punisher: it comes before the response, and the response simply does not pay off in its presence.
Stimulus control is the result you are aiming for: the response occurs more often in the presence of the SD than in its absence. When a client says "cat" for a cat and not for a dog, saying "cat" is under the stimulus control of cats. Stimulus discrimination is the name for this pattern of responding differently to different stimuli.
How discrimination training works
The core procedure is differential reinforcement across stimuli. A typical receptive identification trial looks like this:
- The RBT sets out the materials as the plan directs, for example three pictures.
- The RBT gives the instruction once ("Touch cat").
- If the client touches the cat, the RBT delivers the planned reinforcer right away.
- If the client touches another picture, the RBT does not deliver the reinforcer and uses the error correction written in the plan.
- The RBT records the response, varies the arrangement, and starts the next trial.
Over many trials, the instruction and the cat picture together become the SD for touching the cat, and other pictures become S-deltas for that response. If reinforcement were delivered no matter which picture the client touched, no stimulus would predict reinforcement better than another, and no discrimination would form.
Comparison stimuli matter. When the target picture is the only item on the table, any touch is correct, so a 100% score shows nothing about discrimination. Adding pictures to the array makes a correct choice meaningful. It also lowers the chance of being right by guessing: about 50% with two pictures and about 33% with three. A client who scores near those levels over many sessions is probably not yet discriminating, and the RBT should share those data with the supervisor.
Simple and conditional discriminations
In a simple discrimination, the same stimulus is always correct for a response, such as saying "red" when the red card is shown. In a conditional discrimination, which stimulus is correct depends on another stimulus. In receptive identification, the same three pictures can be on the table, but the correct one changes with the instruction: "Touch cup" on one trial, "Touch ball" on the next. Matching-to-sample, where the client matches a sample picture to the identical picture in an array, is another conditional discrimination.
This is why a client who always touches the cup, whatever the RBT says, has not learned the skill even if half of the trials are scored correct. The instruction is not yet controlling the choice.
Keeping the right stimulus in control
Clients learn from whatever reliably predicts reinforcement, including things the RBT never meant to teach. Common unplanned cues include:
- Position: the target is always in the middle or always on the right.
- Order: targets alternate in a fixed pattern (dog, cat, dog, cat), or the target card is always placed last.
- The RBT's behavior: a glance, a lean or a change in voice toward the correct choice.
- Features of the materials: a bent corner, a stain or a different card size on the target card.
- Timing: reinforcing a touch that happens before the instruction is finished.
Follow the plan's instructions for rotating positions and mixing targets in an unpredictable order, keep your face and body neutral while the client chooses, and give the instruction fully before accepting a response. If accuracy is high with you but low with other staff, or the client makes the same error with materials that share an odd feature, tell your supervisor what you noticed.
Discrimination and generalization
Stimulus generalization is responding the same way to stimuli that are similar to the one used in teaching. Discrimination is responding differently to stimuli that differ. Good teaching needs both. A toddler who says "doggy" for dogs, cats and cows is overgeneralizing and needs discrimination training. A client who says "apple" only for the one apple photo used in teaching has the opposite problem: generalization to other apples has not occurred yet (see C.8).
Discrimination training is not limited to tabletop programs. A teen learning to use a quiet voice in the library and a louder voice on the sports field, an adult learning to knock only when a door sign says "Open," and a child learning to turn when their own name is called are all discriminations. In each case the setting, the sign or the name is the SD, and responding pays off only when it is present.
Your role as an RBT
The supervisor selects targets, SDs, the array size, the error-correction procedure and the mastery criterion. The RBT implements those steps consistently, records each trial accurately, and reports patterns in the data. If the client makes many errors after a new comparison stimulus is added, keep following the plan, record the data and tell the supervisor; do not remove the comparison or switch targets on your own. Discrimination training builds on C.3 (discrete-trial teaching), uses prompts and prompt fading from C.7, and connects to C.8 when skills need to extend to new examples.
Examples and non-examples
| Example | Non-example |
|---|---|
| In a color program, the RBT reinforces "red" when the red card is shown and does not reinforce "red" when the blue card is shown, using the plan's error correction instead. | The RBT praises the client for saying "red" to every card to keep the client motivated. |
| The RBT rotates the position of the target picture across trials and presents "Touch dog" and "Touch cat" in an unpredictable order. | The RBT always puts the target picture in the middle and alternates dog, cat, dog, cat. |
| At a vocational site, putting an item in the left bin earns points only when the item is a bolt, so holding a bolt becomes the SD for that response. | Points are given whenever the client puts anything in either bin, so the item in hand does not matter. |
Common mistakes
- Reinforcing incorrect choices to keep a client happy. Reinforcement in the presence of an S-delta weakens the discrimination; follow the plan's error correction instead.
- Treating a high score with only one item on the table as mastery. With one item any touch is correct; discrimination shows only when comparison stimuli are present.
- Giving unplanned cues, such as looking at the correct card or always placing it in the same spot. Keep a neutral face and vary positions and order as the plan directs.
- Calling an S-delta a punisher. An S-delta comes before the response and signals that reinforcement is not available; nothing is added after the response.
- Confusing an SD with a motivating operation. An SD signals that reinforcement is available; a motivating operation changes how valuable the reinforcer is.
How it's tested
Expect scenarios that ask you to name the SD or S-delta in an everyday situation, to pick what should happen after a correct or incorrect choice, or to spot an unplanned cue such as position, order or the RBT's glance. Watch for distractors that swap SD with motivating operation, reinforcer or prompt, and for answers in which the RBT changes the program instead of following it and telling the supervisor. Questions may also contrast discrimination with generalization.
Practice questions for task C.6
10 of our 34 questions on this task. The rest are in the Behavior Acquisition practice sets and the mock exams.
-
Question 1What does discrimination training involve?
Show answer
B: Reinforcing a response when one stimulus is present and not when other stimuli are present. Discrimination training is differential reinforcement across stimuli: the response is reinforced when the SD is present and not reinforced (extinction) when an S-delta is present. Over time the response occurs mainly in the presence of the SD.
- A: That describes shaping, which changes the form of a response rather than which stimuli it occurs in the presence of.
- C: That describes forward chaining, which links steps of a behavior chain.
- D: That describes programming for generalization, which spreads a response to new stimuli instead of narrowing which stimuli control it.
Task C.6 · 40-hour curriculum section E
-
Question 2In discrimination training, an S-delta (SΔ) is a stimulus in whose presence:
Show answer
D: a response has not been reinforced. An S-delta is a stimulus in whose presence a given response has not been reinforced. Because responding does not pay off when it is present, the response becomes less likely in its presence.
- A: That defines a discriminative stimulus (SD), the stimulus that signals reinforcement is available.
- B: A prompt is added help from the teacher; an S-delta signals that reinforcement is not available.
- C: That describes an establishing operation, a kind of motivating operation, not an S-delta.
Task C.6 · 40-hour curriculum section E
-
Question 3In a clinic, a client now touches the cat picture when told "Touch cat" and does not touch the other pictures in the array. Touching the cat picture after that instruction is reinforced; touching the other pictures is not. The client's pattern of responding shows:
Show answer
C: Stimulus control. When a response reliably occurs in the presence of a particular stimulus and not in its absence, the response is under stimulus control. Differential reinforcement in discrimination training is what produces it.
- A: Response generalization is the appearance of new, untrained responses similar to the trained one; here one response reliably occurs in the presence of one stimulus.
- B: No prompt is involved; the client responds correctly to the instruction and picture alone.
- D: Spontaneous recovery is the return of a behavior after extinction, often at the start of a later session; it does not describe accurate responding.
Task C.6 · 40-hour curriculum section E
-
Question 4In a school sight-word program, a student is nearly always correct with the RBT but much less accurate when a coworker runs the same program. The RBT realizes that after giving the instruction, they tend to look at the correct card while the coworker keeps their eyes on the student. What is the most likely explanation?
Show answer
C: The RBT's glance is partly controlling the student's choices. An unplanned cue such as a glance can become an SD for the correct choice. When the glance is absent, accuracy drops because the printed word alone does not yet control the response. The RBT should keep a neutral gaze and tell the supervisor what they noticed.
- A: A mastery criterion guides decisions about when a target is learned; it does not change the student's accuracy within a session.
- B: An extinction burst is a temporary increase in behavior when reinforcement stops; nothing suggests the coworker stopped reinforcing correct responses.
- D: Lower accuracy with the coworker shows the skill has not carried over; full generalization would mean similar accuracy with both.
Task C.6 · 40-hour curriculum section E
-
Question 5Which statement best describes the difference between stimulus generalization and stimulus discrimination?
Show answer
B: Generalization is responding the same way to similar stimuli; discrimination is responding differently to different stimuli. In stimulus generalization, a response learned with one stimulus also occurs with similar stimuli. In stimulus discrimination, the response occurs in the presence of some stimuli and not others. Teaching usually needs both: enough generalization to new examples and enough discrimination to avoid non-examples.
- A: Both are defined by how responding relates to stimuli, and both can occur in any setting.
- C: Discrimination is built through differential reinforcement (reinforcement in the presence of the SD, none in the presence of the S-delta), not punishment.
- D: A change in response form describes response generalization; stimulus discrimination is about which stimuli occasion the response.
Task C.6 · 40-hour curriculum section E
-
Question 6A client is learning to tell red from blue. The plan reinforces saying "red" when the red card is shown and says not to reinforce "red" when the blue card is shown. The client says "red" to the blue card. In discrimination training, what should happen after that response?
Show answer
A: It is not reinforced, and the plan's error correction is used. Discrimination training reinforces the response only when the SD is present. A response in the presence of the S-delta is not reinforced, and the RBT then uses the error correction written in the plan.
- B: Reinforcing "red" in the presence of the blue card teaches that "red" pays off with any card.
- C: Any reinforcement in the presence of the S-delta weakens the discrimination, and the plan says not to reinforce it.
- D: Removing tokens is response cost, a punishment procedure that is not in this plan; discrimination training relies on withholding reinforcement.
Task C.6 · 40-hour curriculum section E
-
Question 7A client has a cup, a shoe and a ball on the table. On some trials the RBT says "Touch cup" and on others "Touch ball," so which item is correct depends on what the RBT says. What kind of discrimination is this?
Show answer
A: A conditional discrimination. In a conditional discrimination, which stimulus is correct depends on another stimulus, here the spoken instruction. Receptive identification with an array of items is a common conditional discrimination.
- B: In a simple discrimination the same stimulus is always correct; here the correct item changes with the instruction.
- C: Generalization is responding in the same way to similar stimuli, not choosing an item based on an instruction.
- D: A behavior chain is a sequence of linked responses; here each trial calls for one response.
Task C.6 · 40-hour curriculum section E
-
Question 8In a two-target program, an RBT presents "Touch dog" and "Touch cat" in strict alternation: dog, cat, dog, cat. The client is correct on nearly every trial. What is the main concern?
Show answer
C: The client may be responding to the pattern, not the instruction. A predictable order lets the client respond to the sequence rather than to what the RBT says. Plans present targets in a varied, unpredictable order so that only the instruction predicts which response will be reinforced.
- A: Being correct on nearly every trial means reinforcement is frequent.
- B: No prompts are being used; prompt dependence means waiting for help before responding.
- D: Clear differences between targets make a discrimination easier, not harder.
Task C.6 · 40-hour curriculum section E
-
Question 9In a discrimination program, a client has 20 trials and responds correctly on 16 of them. What percentage of trials were correct?
Show answer
B: 80%. Percentage correct = correct trials ÷ total trials × 100. Here 16 ÷ 20 = 0.80, and 0.80 × 100 = 80%.
- A: 75% would be 15 correct out of 20.
- C: 16 is the number of correct trials, not the percentage; it must be divided by the total.
- D: 20% is the percentage of incorrect trials (4 out of 20).
Task C.6 · 40-hour curriculum section E
-
Question 10A plan says to add a second picture to a client's "Touch shoe" target once the client reliably touches the shoe picture when it is alone. After the second picture is added, the client makes several errors in a row. What should the RBT do?
Show answer
B: Follow the plan and its error correction, record data and tell the supervisor. Errors are common when a comparison stimulus is first added. The RBT implements the plan as written, including its error correction, records accurate data, and lets the supervisor decide whether to change anything.
- A: Going back to an earlier step is a change to the plan that the supervisor decides, not the RBT.
- C: Reinforcing errors weakens the discrimination the program is trying to build.
- D: Dropping a program is a plan change; the RBT should follow the plan and report the data.
Task C.6 · 40-hour curriculum section E · RBT Ethics Code 2.02
Answer every question to see your score.
Key terms
| Discrimination training | A procedure that reinforces a response in the presence of one stimulus (the SD) and withholds reinforcement for that response in the presence of other stimuli (S-deltas). |
|---|---|
| Discriminative stimulus | A stimulus in whose presence a behavior has been reinforced, and in whose absence it has not, so the behavior becomes more likely when the stimulus is present (SD). |
| S-delta | A stimulus in whose presence a particular response has not been reinforced, so that response becomes less likely when the stimulus is present (written SΔ). |
| Stimulus control | When a behavior reliably occurs (or occurs differently) in the presence of a particular antecedent stimulus because of a history of reinforcement in its presence. |
| Stimulus discrimination | Responding differently in the presence of different stimuli, for example making a response when the SD is present but not when another stimulus is present. |
| Stimulus generalization | When a behavior that was reinforced in the presence of one stimulus also occurs in the presence of other, similar stimuli that were not part of training. |
| Error correction | A planned procedure the RBT follows after an incorrect or missing response, usually prompting the correct response and then giving the learner another chance to respond. |
| Motivating operation | An environmental variable that temporarily changes how reinforcing a stimulus is and changes how often behavior that has produced that stimulus occurs. |
| Percentage | A percentage is a ratio expressed as parts per 100: the number of occurrences divided by the total, multiplied by 100. |
| Trend | Trend is the overall direction in which data points on a graph move over time: increasing, decreasing, or showing no clear change (zero trend). |