Last week we looked at stimulus vs. response generalization, and one section kept pulling in a neighbor: discrimination. Generalization and discrimination can coexist, but that only makes sense once you are clear on what discrimination is, and how it relates to a term that sounds almost identical: stimulus control.
Most candidates can define both. Stimulus control: behavior changes depending on an antecedent stimulus. Stimulus discrimination: responding differently in the presence of different stimuli. Put those side by side and they read like the same sentence. The TCO lists them as separate tasks (B.12 and B.13), so candidates need to recognize what counts as an example of each.
Stimulus discrimination is differential responding across antecedent conditions. That differential responding is evidence that behavior is under stimulus control.
The two terms overlap heavily, and that is the point. Often the same data demonstrate both. The skill is knowing which one a question is asking you to name.
| Stimulus Control | Stimulus Discrimination | |
|---|---|---|
| What it describes | The systematic relation between antecedent stimulus conditions and responding | Differential responding across those conditions |
| Evidence | Responding varies reliably with the relevant stimulus conditions | Rate, latency, duration, magnitude, or form of responding differs across those conditions |
| Where it comes from | Often established through differential consequences across stimulus conditions | Often the same differential-consequence history; discrimination training arranges it on purpose |
| Question to ask | "Does responding reliably vary with this antecedent condition?" | "How does responding differ across these conditions?" |
Drivers brake at a red light and continue through a green one. Braking varies reliably with the light. That is stimulus control: a systematic relation between antecedent conditions and behavior, which can show up as a change in frequency, latency, duration, or magnitude.
Such control often reflects a history of differential consequences, and that history is worth recognizing when a stem describes it. But it is not part of the definition. Control can also be established or transferred in other ways, such as prompting and fading, so you will not always be handed an explicit SD and S-delta arrangement. Identify control from the reliable variation in responding.
Two things are worth noticing. First, stimulus control describes a relation. It is not a procedure, and it is not a single response. Second, seeing a stimulus appear and then seeing behavior occur is not enough. Temporal pairing alone does not demonstrate control. You need evidence that responding varies systematically with the relevant antecedent condition, or a clearly described learning history that would produce that variation. The contrast can be presence versus absence of the stimulus. You do not literally need two different objects, but you need more than "this happened before that."
"Control" does not mean the stimulus makes the behavior happen. It means responding varies with that stimulus condition. Often that reflects what has happened in its presence before. Candidates who read "control" as "forces" get pulled toward answers about compliance or obedience.
Now change the example. A learner requests a snack at a much higher rate when the pantry is open and a parent is in the kitchen than when the pantry is locked. There are two stimulus conditions, and responding differs across them. That contrast is stimulus discrimination, and it is also evidence that requesting is under stimulus control of the pantry conditions.
The word doing the work is differential. If a stem describes the same responding across new stimuli, that points toward generalization, although the two can coexist. If it describes different responding across stimulus conditions, that points toward discrimination.
The beginner version is "responds here, doesn't respond there." That works for clean examples, so keep it. Just do not turn it into the definition. Discrimination can show up as differences in rate, latency, duration, magnitude, or form, not only an on/off split. And differential responding does not have to be perfect: some responding in the less-reinforced condition does not erase control.
Suppose a learner completes math worksheets at a steady pace when the teacher is nearby and much more slowly when she is not. Responding occurs in both conditions. The rate differs, and that difference is the discrimination.
Do not wait for an all-or-nothing split. If the stem shows that responding differs across stimulus conditions in how much, how fast, how long, how loud, or in what form, you are still looking at discrimination.
The stimulus in whose presence a particular response has not been reinforced is the S-delta. That is specific to the response and its consequence. It is not always "nothing": sometimes it is a different stimulus, and sometimes it is the absence of the SD, like a light that is off. And it is not defined as a punisher. What defines it is that the response has not produced reinforcement there.
Here is the chain that keeps these straight:
Do not pick the procedure when the stem describes the outcome. If the learner has already been trained and now responds differentially, the answer is not "discrimination training." Ask whether the stem is describing what someone did (procedure) or what the learner does (pattern or relation).
Stimulus control does not automatically mean appropriate stimulus control. Suppose a learner is taught to select a picture of a dog from an array, but the dog picture is always placed on the left. The learner selects the dog every time during teaching. Then the dog picture is moved to the right, and the learner still selects the left-hand picture.
Responding is under stimulus control. The problem is that position, not the relevant features of the dog, is controlling the response.
This is why correct responding during teaching does not automatically prove that the intended stimulus has acquired control. Ask what feature actually predicts the learner's responding. Position, color, size, and other irrelevant features can acquire control just as relevant features can. So can prompts: if a learner answers correctly only when the teacher glances at the right choice, the glance, not the target, may be controlling the response.
"The learner answers correctly" is not enough to tell you what controls the response. Change the supposedly irrelevant feature and see whether performance changes with it.
A generalization gradient shows how responding changes across values of a stimulus dimension, such as wavelength or pitch. Its shape is not good or bad by itself. It reflects a learning history, and discrimination histories can change that shape.
This is the link back to last week. Generalization and discrimination are not opposites you have to choose between. Both describe how responding changes across stimulus conditions. Effective behavior often needs responding to extend across the relevant conditions and differ across others.
The TCO also asks you to distinguish motivating operations from stimulus control (B.17). Here is a first-pass shortcut:
Keep the shortcut, but do not treat it as the whole story. An SD does not just announce that something physically exists. It rests on a history of differential consequences, and the mere presence of a stimulus, or elevated responding, is not enough to call it one.
Take a learner who asks for water. If she asks when the water cooler is present and not when it is missing, and asking has been reinforced only when the cooler is there, that is stimulus control by the cooler, an SD. If she asks after running in the heat and rarely after a drink break, the value of water changed, and that is an MO. Both can operate at the same time. The exam will ask which one the stem is describing. If that split is the part you want to drill, we covered it in An Antecedent Is Not Automatically an SD: Value vs. Availability on the BCBA Exam.
| Pattern | Best Concept | Why |
|---|---|---|
| Responding reliably varies with an antecedent condition | Stimulus control | A systematic antecedent–behavior relation, often built through differential consequences |
| Responding differs across stimulus conditions | Stimulus discrimination | Differential responding, which is also evidence of stimulus control |
| Correct responding tracks a feature that is not the intended stimulus | Stimulus control by an irrelevant feature | Responding is controlled, but not by the stimulus the teacher intended |
| A trained response occurs with untrained stimuli | Stimulus generalization | Established responding extends across stimulus conditions |
| Consequences are arranged differently across stimulus conditions | Discrimination training | The procedure that produces the pattern |
| Behavior increases after a long time without access to the reinforcer | Motivating operation | The effectiveness of the consequence changed |
| The skill is still present three months later in the same setting | Response maintenance | Persistence across time |
During class discussions, hand-raising reliably produces teacher attention when the discussion light is green but not when it is red. Over time, hand-raising occurs at a much higher rate during green-light periods. Stimulus control and stimulus discrimination. Hand-raising is under stimulus control of the discussion-light condition, and the differential responding demonstrates stimulus discrimination.
A learner completes worksheets at a steady pace when the teacher is nearby and much more slowly when she is not. Stimulus discrimination. Responding occurs in both conditions, but the rate differs across them.
A learner says "red" to every color card. Not yet discrimination. Responding does not vary with color, so the behavior is not yet under the control of that dimension.
A learner asks for water only after outdoor play in the heat. Motivating operation. The effectiveness of water changed, and nothing in the stem shows a stimulus condition tied to different consequences.
A learner taught to say "dog" to one picture now says "dog" to pictures of dogs never used in teaching. Stimulus generalization. Established responding extended to untrained stimuli.
Before reading the explanation, pick the one you would choose on exam day.
Which scenario is the BEST example of stimulus discrimination?
The child's requesting differs across two stimulus conditions: it occurs often with the blue cup, where juice follows, and rarely with the red cup, where it does not. That differential responding is stimulus discrimination, and it is also evidence that requesting is under stimulus control of the cups. The thank-you example is stimulus generalization, because established responding extended to people who were not part of the teaching. The snack example differs across conditions too, but what changed is how effective food is, which points to a motivating operation rather than stimulus conditions tied to different consequences. The handwashing example is response maintenance, because the behavior persisted across time.
The hard part is rarely memorizing the definitions. It is noticing what the stem is asking you to name. Ask what evidence shows that responding actually varies with the antecedent condition. If the focus is the systematic relation between the antecedent and the behavior, you are being asked about stimulus control. If the focus is differential responding across conditions, you are being asked about stimulus discrimination. And when a stem reads like both, the distractors are usually the neighbors: generalization, an MO, or maintenance.
Treat "responds here, not there" as a first-pass heuristic, not a law. Real responding can differ in rate, latency, duration, magnitude, or form, and the same data can support both terms. Your job is to answer the relation the stem actually asks you to name.
Stimulus control: responding reliably varies with antecedent stimulus conditions. Such control often reflects a history of differential consequences. Stimulus discrimination: responding differs across those conditions. Differential responding is evidence of stimulus control. Ask what evidence shows responding varies with the antecedent, decide which relation the stem is focused on, and check the neighbors before you commit.
Knowing the definitions is only the starting point. BA Prep questions are designed to make you separate stimulus control, stimulus discrimination, generalization, and motivating operations when they show up in the same scenario. That is the kind of distinction the BCBA 6th Edition Test Content Outline expects.
Start Free TrialBehavior Analyst Certification Board. (2022). BCBA test content outline (6th ed.). Behavior Analyst Certification Board. Updated September 2024. The current outline addresses stimulus control (B.12), stimulus discrimination (B.13), and the distinction between motivating operations and stimulus control (B.17), along with discriminative stimuli in behavior-change procedures (G.5) and simple and conditional discriminations (G.6). https://www.bacb.com/wp-content/bcba-outline-6thEd/
Cooper, J. O., Heron, T. E., & Heward, W. L. (2020). Applied behavior analysis (3rd ed.). Pearson.
Halbur, M. E., Caldwell, R. K., & Kodak, T. (2021). Stimulus control research and practice: Considerations of stimulus disparity and salience for discrimination training. Behavior Analysis in Practice, 14(1), 272–282. https://doi.org/10.1007/s40617-020-00509-9