Much of what we teach is not a brand-new behavior but the right behavior at the right time. A learner may already say "cup," but we want them to say it when they see a cup and not when they see a bowl. Teaching a learner to respond differently to different stimuli is called discrimination training, and it is part of almost every program you will run.
Why it matters
Labeling colors, pointing to the named picture, reading sight words, answering "What's your name?" correctly, and waiting for the walk signal all depend on discrimination. If a learner "knows" a target only because it is always in the same spot or because your eyes drift toward it, the skill will fall apart in the real world.
On the exam, expect scenarios that ask you to name the procedure, identify an SD or S-delta, spot an accidental cue, or choose how to arrange trials so the learner must attend to the right stimulus.
Key ideas
- Reinforce in the presence of one stimulus, not the other. In discrimination training, a response is reinforced when the SD is present and is not reinforced when the S-delta is present. Over time, the learner responds mostly when the SD is there.
- The goal is stimulus control. When a behavior reliably occurs more often in the presence of one stimulus than in its absence, that stimulus has gained control over the behavior.
- Some discriminations are simple, some are conditional. In a simple discrimination, one stimulus signals the response (seeing a dog, the learner says "dog"). In a conditional discrimination, the correct choice depends on another stimulus. With a cup, spoon, and plate on the table, which item is correct depends on whether you said "Touch cup" or "Touch spoon." Receptive identification and matching tasks are conditional discriminations.
- Arrange trials so only the relevant cue predicts reinforcement. Rotate where targets sit in the array, vary the order of instructions, and keep your body, voice, and gaze neutral.
- Discrimination and generalization are two sides of one coin. Generalization means responding the same way to similar stimuli; discrimination means responding differently to different stimuli. Good teaching builds both: "dog" for every kind of dog, but not for cats.
- You follow the written plan. Your supervisor chooses the targets, array size, prompting and error-correction steps, and mastery criteria. Your job is to run trials precisely and record data accurately.
Definitions
- Discriminative stimulus (SD): a stimulus in whose presence a particular response has been reinforced, so the response becomes more likely when it is present.
- S-delta (SΔ): a stimulus in whose presence that response has not been reinforced, so the response becomes less likely when it is present. Both terms always name a response: one picture can be an SD for one response and an S-delta for another.
- Discrimination training: reinforcing a response in the presence of the SD and withholding reinforcement for that response in the presence of the S-delta.
- Stimulus control: the outcome of discrimination training; the response occurs more often in the presence of the SD than in its absence.
- Simple discrimination: one stimulus alone tells the learner which response will be reinforced.
- Conditional discrimination: the correct response to a stimulus depends on (is conditional on) another stimulus, such as the spoken instruction.
- Faulty stimulus control: responding is controlled by an irrelevant feature, such as an item's position, the order of trials, or the instructor's glance, instead of the intended stimulus.
Examples from a session
Three learners, three discriminations
- Clinic, receptive identification: Amara, an RBT, places pictures of a car, a ball, and a cat in a row and says "Touch ball" to Theo, 4. When Theo touches the ball, she praises him and gives a quick turn with a spinning toy. If he touches the car, no reinforcer follows and she uses the error-correction step in his plan. On the next trial she shuffles the pictures and asks for a different item.
- Home, sight words: Lina, 9, is learning to read "push" and "pull" on door signs. Her RBT holds up a card. When Lina reads the word correctly, she earns a token; reading "pull" for "push" earns nothing. The two cards appear in an unpredictable order.
- School, safety skill: Devon, 15, practices crossing at a crosswalk. Stepping off the curb when the walk symbol shows is praised; the "don't walk" hand is the S-delta for stepping off, so waiting is what gets praised then.
In each case, reinforcement depends on the stimulus that is present, and the RBT makes sure nothing else (position, order, or a hint in their face) predicts the right answer.
Common mistakes / exam traps
- Leaving the target in the same spot. If the correct picture is always on the left, the learner may learn "touch left." That is a position bias, a form of faulty stimulus control. Rotate positions across trials.
- Asking for the same target over and over. Ten "Touch ball" trials in a row can be a useful early step if the plan calls for it, but they do not show discrimination. Mixing in other instructions shows whether the learner is listening to what you said.
- Giving accidental cues. Glancing at the answer, leaning toward it, saying the target word louder, or placing it closer to the learner can all control responding. Keep items evenly spaced and look at the learner, not the target.
- Reinforcing errors to keep things positive. Praising a response made in the presence of the S-delta weakens the discrimination. Follow the plan's error-correction and prompting procedures instead.
Exam trap
If a scenario says the learner is accurate only when the target sits in one spot, or only when one instructor (who may be giving subtle cues) runs the program, suspect faulty stimulus control before assuming mastery.
Exam trap
Discrimination vs. generalization: a child who calls every four-legged animal "dog" is generalizing too broadly. The fix is discrimination training, not more generalization practice.
Quick check
During a color program, an RBT gives a sticker when a learner says 'green' to a green block and gives no sticker when the learner says 'green' to a yellow block. In this program, what is the yellow block?