Discriminative stimulus

History gives the stimulus its function

A light, word, person, location, internal sensation, or combination can function as an SD. Its physical appearance does not make it discriminative. The relevant history is that a defined response has contacted a consequence differently when the stimulus is present.

Suppose a library checkout light turns green when the station is operating. Scanning a book while it is green usually produces a checkout confirmation. Scanning while the screen says “closed” does not. The green display may acquire an SD function for scanning.

An SD is more than something that comes first

Temporal order alone is weak evidence. A teacher may say “start” before every trial, while item position or a gesture actually controls responding. A logo may be visible whenever a reward occurs without changing behavior.

Lattal’s contingency review explains why response-consequence relations require more than close timing. To assess an SD, define the response, stimulus condition, consequence, comparison condition, and learning history.

Availability and value answer different questions

An SD concerns whether a consequence is differentially available for a response. A motivating operation concerns how effective that consequence is now and changes behavior that has produced it.

The open checkout station may signal that scanning will work. A person who has no books to borrow has little reason to scan. Conversely, wanting to borrow a book does not make a closed station available. The motivating-operation tutorial by Langthorne and McGill describes this value-versus-availability distinction.

S-delta identifies a comparison condition

Behavior analysts often use S-delta for a condition in which the response has not produced the same reinforcer. In the library example, a clear “out of service” display might be the comparison.

These labels are functional. A red light is not inherently an S-delta, and a green light is not inherently an SD. Different histories, contexts, and responses can give the same stimulus different functions.

Stimulus control can be intended or accidental

Instruction may target the printed destination on a transit sign while learners respond to color. A clinician may intend a spoken instruction to guide a selection while eye gaze signals the answer.

Balance positions, vary irrelevant features, remove prompts, and include relevant comparison conditions. Probe with new examples that retain the defining feature. Report when a response follows the intended stimulus and when another feature could explain it.

A fictional discrimination record

Elena is learning whether a self-checkout station is available. Across eight naturally occurring opportunities with the operating light on, she scans a selected book in 7 of 8. Across eight opportunities with the “closed” display, she begins scanning in 1 of 8.

This pattern is consistent with differential responding in the observed conditions. It does not prove the light alone controlled behavior. Staff presence, screen content, prior instructions, book choice, and time of day also differed. The team checks performance after balancing those features and keeps ordinary assistance available.

Clinical teaching begins with a meaningful response

Choose an outcome the person values, a response form they can use, and stimuli they can perceive. Speech, sign, gesture, writing, and AAC can all be valid. Eye contact is never required to demonstrate discrimination.

Preserve communication, food, water, bathroom use, mobility, rest, prescribed care, and emergency help regardless of performance. Use the least intrusive effective prompts and fade them according to data.

The ABAI basic-principles page and BACB BCBA Test Content Outline, 6th edition include stimulus control and discrimination in training content. Neither source prescribes a plan for one person.

Measure the relation you claim

Report responses divided by eligible opportunities in SD and comparison conditions. Also record prompts, response latency, missed opportunities, reinforcer delivery, procedural integrity, and changes in motivation or access.

A high SD rate is easier to interpret when comparison responding is low and both conditions are otherwise similar. Generalization probes should vary irrelevant features without changing consequence availability. Maintenance probes show whether the pattern continues after teaching; they do not replace tests of stimulus control.

Plan transfer without erasing support

When the everyday SD differs from the teaching cue, program a clear bridge. A practice card can gradually incorporate the real symbol, location, wording, or communication partner. Test new examples without removing AAC, visual schedules, or safety supports that belong in the natural setting. Generalization means the relevant relation works across planned variation; it does not require unsupported performance.