Identify Examples of Stimulus Discrimination: A Practical Guide for ABA Clinicians
Stimulus discrimination sounds technical, but it describes something you see every day in your clinic and in the real world: a learner responding to one cue but not to similar ones. If your student follows the “raise your hand” rule in a quiet classroom but forgets it when the room gets loud, or a client performs a safety skill perfectly at the clinic but never at home, you’re likely looking at a discrimination issue.
Understanding how to spot, measure, and teach stimulus discrimination is essential to designing interventions that actually work where your clients need them most.
This article is for practicing BCBAs, clinic owners, senior therapists, and caregivers who want to move beyond the textbook definition and learn how discrimination shapes real behavior. We’ll walk through what stimulus discrimination is, how to recognize it, when to teach it, and the ethical considerations that matter when deciding how narrow or broad your client’s responding should be.
One-Paragraph Summary
Stimulus discrimination is the ability to respond to one specific stimulus while not responding to similar stimuli. You spot it by watching whether a behavior happens reliably in one context (when the discriminative stimulus is present) and rarely or never in a similar context (when it’s absent).
In ABA, a child might label a Labrador as “dog” when shown that picture but not when shown a poodle. In everyday life, a driver stops at a red light but proceeds at green—same road, different cues, different responses.
Recognizing discrimination matters because it tells you whether a trained skill will actually work in the settings where it needs to. Used ethically, stimulus discrimination helps clients become independent and respond to relevant environmental cues rather than depending on artificial supports.
Clear Explanation of the Topic
What Stimulus Discrimination Actually Is
Stimulus discrimination happens when a learner responds differently to two or more stimuli because of their different histories of reinforcement. The learner has learned that one cue (the discriminative stimulus or SD) predicts reinforcement, while a similar or different cue (called S-delta or SΔ) does not.
When this learning is solid, the behavior shows up reliably when the SD is present and rarely or not at all when only the SΔ is present.
Think of it as the learner asking, “Will I get reinforced right now?” The SD says yes. The SΔ says no. Over time, the learner’s behavior reflects that difference.
Discrimination Versus Generalization
These two concepts often get confused because they sound opposite—and they are.
Stimulus generalization is when a response spreads to stimuli similar to the one that was trained. A child trained to label a Labrador as “dog” might generalize and call all dogs “dog,” including poodles and golden retrievers. That’s generalization—responding across stimuli.
Stimulus discrimination is the opposite direction. Responding narrows down to only the specific stimulus (or a very specific set of stimuli) that has been reinforced. The same child learns to label only Labradors as “Labrador” and not to use that label for poodles. That’s discrimination—responding selectively to one stimulus while not responding to similar ones.
Both are important. The key is knowing which one your client needs in each situation.
Key Terms You Need to Know
A discriminative stimulus (SD) is a cue that signals reinforcement is available. When the SD is present and the learner makes the right response, reinforcement happens. An SD might be a teacher’s raised hand, a green traffic light, a certain tone of voice, or a specific visual sign.
An S-delta (SΔ) signals that reinforcement is not available. When the SΔ is present, the same response does not lead to reinforcement. Examples include a teacher’s hand down, a red light, or silence when your client expects a verbal cue.
Stimulus control is the broad term for how environmental cues influence behavior. Discrimination is one type of stimulus control—it means the environment is controlling which behavior happens and when.
A response class is a group of similar behaviors that all have the same function or lead to the same consequence. If a child can raise their hand, shout out, or stand up to get the teacher’s attention, all three are in the same response class. Discrimination can apply to the stimulus that sets up the response class, not just individual behaviors.
How Discrimination Gets Learned
Discrimination is learned through differential reinforcement. The SD is followed by reinforcement when the target response happens. The SΔ is not followed by reinforcement—that’s extinction.
Over many trials, the learner’s brain registers the pattern: “This cue means reinforcement is coming; that cue means it’s not.” The behavior follows suit.
For example, if you reinforce a child for saying “sit” when the therapist faces them (SD) but do not reinforce “sit” when the therapist faces away (SΔ), the child gradually learns to say “sit” mainly when facing the therapist. The reinforcement history shapes which stimulus controls the behavior.
What Good Discrimination Looks Like
When discrimination is working well, you’ll see reliable responding in the presence of the SD and low or absent responding in the presence of the SΔ.
If you run ten trials with the SD present and the learner responds correctly nine or ten times, that’s strong SD control. If you run ten trials with the SΔ present and the learner responds incorrectly zero to two times, that’s good SΔ control. Together, they show discrimination.
You can also measure the strength of discrimination by comparing response rates or accuracy in SD trials versus SΔ trials. The bigger the gap, the sharper the discrimination.
Why This Matters in Practice
Discrimination Determines Real-World Success
A skill trained in the clinic is only useful if your client uses it in the real world. Stimulus discrimination is often the missing link.
A teenager might be perfectly polite to the therapist (strong SD control of politeness to that adult) but completely rude to teachers (weak generalization). Or a child might follow safety rules with their mother but ignore them with their father. These gaps usually mean the stimulus control is too narrow—the behavior is locked to one specific cue rather than the broader, more functional cue you actually want.
Understanding discrimination helps you diagnose why trained skills fail in new settings. When a skill doesn’t transfer, it’s often because the SD in the clinic is very different from the SD in the real world. Recognizing this lets you adjust your training instead of just reteaching the same way.
Discrimination Shapes How You Teach
Some skills need tight discrimination. A child learning to cross the street safely should respond to the walk signal (SD) and not respond to the don’t walk signal (SΔ). You want that discrimination to be absolute. You might teach using prompts, high reinforcement, and careful, repeated practice with both the SD and SΔ to build strong control.
Other skills need to be more flexible. If you’re teaching a child to follow instructions, you probably want them to respond to instructions from teachers, parents, therapists, and coaches—not just one person. In that case, you’d use multiple exemplar training: practice the skill with many different people and in many different settings so the learner generalizes rather than narrows their responding.
The difference matters. Teaching for discrimination and teaching for generalization use different strategies, and mixing them up can waste time or create exactly the wrong outcome.
The Risk of Narrow Stimulus Control
One common pitfall is creating stimulus control that is too narrow. You might unintentionally teach a client to respond only to a specific prompt, therapist, location, or material. This is sometimes called prompt dependence—the behavior depends on the prompt rather than the natural cue.
A child might label pictures only when the therapist points at them but refuse when asked to label freely. That’s not useful; it’s limiting.
Narrow stimulus control also creates barriers to independence and community inclusion. If a skill only works at the clinic, the client misses opportunities to use it at school, home, or in the community. That’s not just a training issue; it’s an ethical issue.
Key Features and Defining Characteristics
Stimulus discrimination has a few hallmarks that help you recognize and test for it.
Differential responding is the most obvious: the learner responds one way in the presence of the SD and a different way (usually no response or an error) in the presence of the SΔ. This difference is reliable—it happens trial after trial, not just once.
That differential responding is linked to a history of differential reinforcement. You can’t just observe the response; you have to know (or be able to test) whether the SD was actually reinforced and the SΔ was not.
Discrimination is also observable and testable. You can present the SD and SΔ in separate trials and measure the difference.
The stimulus control can be specific and narrow, or it can be broader. A learner might respond only to a Labrador picture (very narrow) or to any dog picture (broader). Both are discrimination, but the range of stimuli that control the behavior differs.
Discrimination can also be partial or absolute. Absolute discrimination means the response is correct on the SD every time and absent on the SΔ every time. Partial discrimination means the response is more likely on the SD than the SΔ, but not perfect.
One key boundary: discrimination is not the same as stimulus fading. Fading is a *procedure*—you gradually change the stimulus to shift control. Discrimination is the *outcome*—the learner responds differently depending on the stimulus.
When You Would Use This in Practice
Common Clinical Moments
Assessment is a big one. When a skill fails in a new setting, your first move should often be to assess for stimulus discrimination. Does the behavior happen at the clinic but not at school? That suggests the SD at the clinic is very different from the SD at school. Running a quick discrimination test—presenting the SD and SΔ in trials at both locations—can tell you whether this is the issue.
Teaching precise, safety-related responses is another major use. If you’re teaching a client when it’s safe to cross the street, you want very sharp discrimination between a walk signal (SD) and a don’t walk signal (SΔ). You’d deliberately teach discrimination using clear contrasts, high reinforcement, and multiple practice trials so the control is rock-solid.
Designing prompts and fading is a third place where discrimination planning matters. If you decide to use a prompt, you need a plan to fade it and transfer control to the natural SD. That plan is based on understanding how discrimination works—you’re aiming to get the learner to respond to the natural cue, not the prompt.
Decision Points
Before you design teaching, ask yourself: “Does this skill need to be highly discriminated to a specific cue, or should it generalize across many similar cues?”
A child learning to sit at a desk might need to sit whenever a teacher asks (generalized to many teachers) but only during instruction time, not recess (discriminated from play time).
A client learning to ask for a break might need to ask the therapist, teacher, and parent (generalized across people) but only when they’re genuinely overwhelmed (discriminated from attention-seeking).
Your answer shapes your teaching plan.
Examples in ABA
Example 1: The Picture Labeling Task
A therapist shows a child a picture of a Labrador and says, “What dog is this?” When the child says “Labrador,” they get praise and a small reward. This happens many times with Labrador pictures—dozens of trials over days and weeks.
Then the therapist shows a picture of a poodle and asks the same question. The child says “Labrador,” but nothing happens. No praise, no reward. This is extinction on the poodle picture.
Over time, the child learns the pattern. They say “Labrador” reliably when shown a Labrador (SD). They do not say “Labrador” when shown a poodle (SΔ). This is stimulus discrimination.
Why it works: The Labrador picture is the SD—it signals reinforcement. The poodle picture is the SΔ—it signals no reinforcement. The child’s different responses reflect that learned history.
Trial data might look like this:
- Trial 1: Labrador shown (SD) → Child says “Labrador” → Praise given → Correct
- Trial 2: Poodle shown (SΔ) → Child stays silent → No reinforcement → Correct
- Trial 3: Labrador shown (SD) → Child says “Labrador” → Praise given → Correct
- Trial 4: Poodle shown (SΔ) → Child tries to say “Labrador” → No reinforcement → Error
Over many trials, errors on SΔ trials decrease as discrimination sharpens.
Example 2: The Classroom Hand-Raising Rule
A teacher sets up a rule: students raise their hands in a quiet classroom when the teacher is ready for questions. Raising your hand gets attention and praise. This is the SD.
In a noisy classroom—during group work, transitions, or when the teacher is managing behavior—raising your hand does not get attention. The teacher might even redirect the student to focus on their peers. This is the SΔ.
A student who has learned this discrimination raises their hand reliably during quiet, structured instruction time and rarely raises their hand during chaotic group work. The behavior is controlled by the quiet-classroom cue, not just by the presence of the teacher.
Why it’s discrimination: The student’s responding is shaped by the context because that context has been paired with different consequences.
Examples Outside of ABA
Example 1: The Traffic Light
A driver stops at a red light (SΔ) because stopping is the learned response when red is present—not stopping leads to accidents, tickets, or worse. The driver goes at a green light (SD) because that’s when driving is reinforced by reaching the destination safely.
The color of the light controls the driver’s behavior because of differential reinforcement: stopping when red leads to safety; going when green leads to reaching the goal.
Example 2: The Store’s Open Sign
A shopper enters a store with an “Open” sign (SD) and receives service—that’s reinforcing. A shopper sees a “Closed” sign (SΔ) and does not enter; approaching the store is not reinforced when closed.
Over time, the sign controls the behavior. The shopper discriminates: they approach when the sign says “Open” and not when it says “Closed.”
These everyday examples show that stimulus discrimination is not an ABA invention; it’s a principle of how humans and animals learn to match their behavior to the environment.
Common Mistakes and Misconceptions
One frequent error is calling any weak response a “discrimination problem” without actually comparing SD and SΔ performance. A child who rarely follows instructions might not have a discrimination problem; they might have a skill deficit or a motivation problem. To call it discrimination, you need to show that the child responds differently in the presence of an SD versus an SΔ.
Another mistake is assuming that discrimination is always good. Sometimes you want broad generalization, not narrow discrimination. A student who responds to instructions only from a specific teacher has learned discrimination, but it’s not functional—the student needs to follow instructions from many adults.
A third pitfall is confusing discrimination with stimulus fading. Fading is a procedure; discrimination is an outcome. You can fade a prompt in a way that builds discrimination or in a way that teaches generalization. Know which one you’re doing.
Prompt dependence is easy to create by accident. If you always prompt a response and then reinforce it, the prompt becomes the SD, not the natural cue. The learner responds to the prompt but not to the natural cue when the prompt is gone. Preventing this means fading the prompt on a schedule and rewarding unprompted responding more heavily.
Finally, some people conflate simple discrimination with conditional discrimination. Simple discrimination is responding to one stimulus and not another. Conditional discrimination is more complex: the “correct” response depends on a relationship among multiple stimuli. Both are discrimination, but they’re structurally different.
Ethical Considerations
When you’re teaching stimulus discrimination, you’re shaping which cues control your client’s behavior. That’s powerful, and it comes with ethical weight.
Narrow stimulus control can limit independence and community participation. If you teach a skill that only works in the clinic or only with one therapist, you’ve created a barrier to the client using that skill in real life. Before you design narrow discrimination, ask: “Is there a good reason why this behavior should be locked to this specific cue?”
Safety is one good reason. A client learning to cross at walk signals should have tight discrimination to that signal. But teaching a client to speak only to the therapist, not to peers, is usually not a good reason.
Equity and access matter too. If a discriminative stimulus requires a resource the client doesn’t have access to—a specific device, a particular setting, a trained person who isn’t available at home—you’ve set up a barrier. Consider whether the SD you’re relying on is fair and available in all the places where the skill needs to happen.
Documentation and informed consent are non-negotiables. Your treatment goals should spell out what stimulus control you’re aiming for and why. Caregivers should understand whether the goal is to narrow responding to one cue or to generalize across many cues.
Ethical practice means balancing precision with client autonomy. Sometimes precision is necessary. Often, broader generalization is more functional. Use the least intrusive, least restrictive approach.
Assessment and Intervention Planning
Go back to the scenario where a skill fails to transfer. Your client can perform a safety task perfectly at the clinic but never does it at home.
Before reteaching, assess for stimulus discrimination. Watch whether the behavior occurs reliably in the clinic and rarely at home. This tells you the problem is not the skill itself; it’s the transfer of stimulus control.
If you’re designing a new intervention for a skill that must be precise—like crossing the street or responding to emergency signals—build in deliberate discrimination training from the start. Use clear contrasts between the SD and SΔ, reinforce heavily on the SD, and run extinction trials on the SΔ so the learner’s responding sharpens quickly.
When prompting, always plan how you’ll transfer stimulus control from the prompt to the natural cue. Will you fade the prompt gradually? Will you use prompt delays? Will you practice unprompted responding and reinforce that more heavily? The plan depends on what stimulus control you want to end up with.
How to Measure Stimulus Discrimination
The Simplest Test
The simplest way to test discrimination is to run alternating SD and SΔ trials and calculate the percent correct on each.
Percent correct = (Number of correct responses on SD trials / Total SD trials) × 100
If a learner is correct on eight out of ten SD trials, that’s 80% correct on the SD. If they make zero errors on ten SΔ trials (meaning they didn’t respond when they shouldn’t have), that’s 100% correct avoidance of the SΔ. Compare the two, and you have a sense of discrimination strength.
What Good Numbers Look Like
In ABA, we often aim for 80–100% accuracy on the SD as a mastery criterion. For the SΔ, you’re looking for low error rates—ideally 0–20%, meaning the learner mostly doesn’t respond when the SΔ is present.
The bigger the gap between SD accuracy and SΔ error rate, the sharper the discrimination.
100% accuracy is not always realistic or necessary. A learner might reliably hit 85–95% accuracy on the SD and 5–15% errors on the SΔ, and that can be perfectly functional. For safety-critical tasks, you’ll want higher accuracy. For social or academic skills, a bit less precision might be acceptable.
Tracking Over Time
Plot your data on a graph with sessions on the x-axis and percent correct on the y-axis. You should see SD accuracy going up and SΔ error rate going down as discrimination is learned.
The curve shows whether learning is happening and helps you spot plateaus or regressions. If SD accuracy is stuck at 60% after many sessions, you might need to change your reinforcement strategy. If SΔ errors start creeping back up, you might need a refresher on extinction trials.
Related Concepts to Know
Stimulus control is the umbrella term. It means that a stimulus influences whether a behavior happens. Discrimination is one type—selective responding.
Generalization is discrimination’s opposite direction. When a learner responds to many similar stimuli, they’re generalizing. Both can be functional or dysfunctional depending on the context.
Prompt fading is the process of gradually removing an artificial cue so that the natural environmental cue takes over. It’s a key tool for preventing prompt dependence and building control to the right SD.
Matching to sample is a more complex discrimination task where the learner has to pick the stimulus that matches a sample. The same principles of differential reinforcement and SD/SΔ control apply.
FAQ: Common Questions Clinicians Ask
What’s the fastest way to test whether a learner has learned a discrimination?
Run a quick alternating trial session: present five SD trials and five SΔ trials in random order. If the learner is correct on most SD trials and makes few or no errors on SΔ trials, discrimination is solid. If they’re scattered, you’re still building it. This takes 10–15 minutes and gives you usable data.
How do I know if my client needs discrimination training or generalization training?
Ask what the real-world goal is. If the behavior should happen in many contexts and with many people, aim for generalization. If it should happen only in specific, safe situations, aim for discrimination.
A greeting should generalize (work with anyone); a street-crossing rule should discriminate (work only at the walk signal). Your answer determines your teaching strategy.
Can a learner unlearn a discrimination?
Yes. If you stop differentially reinforcing—if you start reinforcing the SΔ as much as the SD, or if you stop reinforcing the SD altogether—the discrimination will weaken.
You can also deliberately teach generalization using multiple exemplar training: practice the skill across many different SDs until the learner’s responding broadens.
What should I do if prompts are making my client dependent?
First, stop reinforcing prompted responses as much as unprompted responses. Start reinforcing independence more.
Second, fade the prompt on a schedule: maybe use a full prompt for trials 1–5, a half prompt for trials 6–10, a gesture for trials 11–15, and no prompt for trials 16 onward.
Third, insert prompt delays: wait 3–5 seconds after the SD before prompting, so the learner has a chance to respond without the prompt.
These three strategies together usually transfer control away from the prompt to the natural SD.
Should I worry about discrimination if my client is only just learning the skill?
Not initially. When a skill is brand new, the learner is just figuring out what response to make. You can use prompts to help them succeed.
But as soon as the response is reliable with prompts, start planning for discrimination: introduce SΔ trials, fade the prompts, and reinforce unprompted responding. Don’t wait until the skill is “perfect” to start thinking about control transfer—start early.
Key Takeaways
Stimulus discrimination is about teaching a learner to respond reliably to one specific cue (the discriminative stimulus) and not to similar cues (S-delta). It’s measured by comparing accuracy on SD trials to errors on SΔ trials. The outcome depends on the reinforcement history: SD gets reinforced, SΔ does not.
Whether you need discrimination or generalization depends on your goal. Safety-critical behaviors usually need tight discrimination. Social and academic skills usually benefit from generalization. Know which one you’re aiming for, and design your teaching accordingly.
Discrimination is powerful but can create unintended barriers. Narrow stimulus control—especially to a single therapist, setting, or prompt—limits real-world use and independence. Ethical practice means balancing precision with the client’s autonomy, community participation, and access to natural cues.
Finally, test discrimination early and often. Running alternating SD/SΔ trials gives you clear data about whether control is where you want it to be. If a skill fails to transfer, assess discrimination before you reteach. Understanding stimulus control is how you move from clinic success to real-world success.
