This cluster shows how to teach learners to pay attention to the right parts of a task. It tells you why seeing the cue, getting the right amount of rewards, and using clear differences between choices matter. These studies help BCBAs build lessons where the learner always knows what to do and does not guess. Good stimulus control means fewer errors and faster learning for any skill.
Stimulus control means the learner responds to the right cue at the right time — reliably. This cluster looks at the practical side of stimulus control in applied settings: how to set up discrimination tasks so learners notice the right features, and what goes wrong when they do not.
One important finding is overselectivity. Some learners — especially those with autism — may focus on one small feature of a stimulus and ignore the rest. If a learner can only identify a red card with a picture of a ball by the color, not the object, you have a stimulus control problem. Research shows that overselectivity can be reduced by making the learner observe the whole stimulus before responding and by starting with simpler compound stimuli.
Another finding is that non-critical features accidentally correlated with reinforcement will be learned as if they are the critical ones. If the correct card happens to always be on the left, the learner learns 'left,' not 'correct answer.' Studies show this is a common, silent failure in teaching programs. You need to carefully randomize everything that is not the skill you are teaching.
Research also shows that stimulus control can split across dimensions — a learner might respond based on location sometimes and on color other times. This divided control can be detected through careful probing, and it often resolves when reinforcement contingencies are made consistent. Using eye-tracking or careful observational data during sessions can reveal what the learner is actually attending to.
Common questions from BCBAs and RBTs
Overselectivity is when a learner attends to only one small feature of a stimulus and ignores the rest. Research shows you can reduce it by requiring the learner to observe the whole stimulus before responding and by starting discrimination training with simpler, less confusing compound stimuli. Building in an explicit observing response requirement — even a few extra looks — helps.
Check your materials for accidental correlations. If color, position, or any other non-target feature consistently predicts the correct answer, the learner will attend to it instead. Randomize all non-critical features across trials and probe the target dimension in isolation to confirm what is actually controlling the response.
It means the learner's response is controlled by two different features at different times — for example, sometimes by color and sometimes by position. Research shows this is detectable through probing and usually resolves when reinforcement is made consistently contingent on only one dimension. Run probes, identify which dimensions are competing, and tighten your contingencies.
Prior reinforcement history can silently stay active and reassert stimulus control when conditions change. Research shows that previous learning does not disappear — it competes with new learning. When behavior shifts in a way you did not expect, trace the reinforcement history to identify what was previously reinforced in similar situations.
Research shows that when a conditional cue is weak or hard to detect, learners may learn to treat it as partially reinforced rather than learning true occasion setting. Make your conditional cues highly salient and easily distinguishable from the background. Probe for occasion setting by testing whether the cue modulates responding differently from a standard discriminative stimulus.