Schedules, Extinction, Stimulus Control, Discrimination, and Generalization

Schedules of Reinforcement

A schedule of reinforcement specifies the response requirement that must be met for reinforcement. Continuous reinforcement (CRF) reinforces every response and is ideal for acquisition (teaching new behavior). Intermittent reinforcement reinforces only some responses and builds persistence and resistance to extinction, which is why thinning to intermittent schedules is the goal for maintenance.

Two dimensions define the four basic intermittent schedules:

• Ratio vs. interval: ratio schedules count responses; interval schedules require the first response after a period of time has elapsed.
• Fixed vs. variable: fixed schedules use a constant requirement; variable schedules use an average that changes trial to trial.

Crossing them yields FR, VR, FI, VI. A critical distinction: interval schedules still require a response after the time passes — they are NOT the same as noncontingent reinforcement (NCR), which delivers a stimulus on a time schedule regardless of behavior.

Characteristic Response Patterns

Memorize the signature pattern each schedule produces on a cumulative record — this is directly tested.

Memory hooks: Ratio schedules drive high rates because more responding earns more reinforcers; interval schedules produce moderate rates because extra responding within the interval does not speed up reinforcement. Variable schedules eliminate pausing because the next reinforcer is unpredictable. A scalloped record is the unmistakable fingerprint of FI; a near-vertical, steady climb is VR.

Extinction and Its Side Effects

Extinction is the discontinuation of the reinforcer that has maintained a behavior, producing a gradual decrease in responding. Extinction is function-specific: you must withhold the maintaining reinforcer.

• Attention-maintained behavior → planned ignoring (withhold attention).
• Escape-maintained behavior → escape extinction (do NOT allow escape; the demand continues).
• Tangible-maintained behavior → withhold access to the item.
• Automatically maintained behavior → sensory extinction (mask or remove the sensory product, e.g., a padded surface).

'Ignoring' is extinction only when attention is the maintaining reinforcer. Ignoring escape-maintained behavior actually delivers the escape and strengthens it. Predictable extinction effects include:

• Extinction burst — a temporary increase in frequency, duration, magnitude, or novel topographies right after extinction begins.
• Extinction-induced variability and emotional/aggressive responding.
• Spontaneous recovery — the behavior reappears after time has passed even though the reinforcer is still withheld.

These effects mean extinction is working as expected, not failing. The exam asks whether the maintaining reinforcer is being withheld and whether the behavior decreases across opportunities.

Stimulus Control, Discrimination, and Generalization

Stimulus control exists when the rate or form of a response changes in the presence of an antecedent stimulus because of a reinforcement history. A discriminative stimulus (SD) signals that a response will be reinforced and evokes it. An S-delta (SΔ) signals reinforcement is not available, and responding drops in its presence. Discrimination training reinforces a response in the SD and withholds it in the S-delta, narrowing responding to the relevant cue.

Generalization is the opposite spread:

• Stimulus generalization — the same response occurs in the presence of new but similar stimuli (untrained). A generalization gradient describes how responding decreases as stimuli become less similar to the training SD; a flat gradient means broad generalization, a steep gradient means tight discrimination.
• Response generalization — new, untrained responses with the same function appear after training one response.
• Maintenance (response maintenance) — performance continues after the intervention is partly or fully removed (generalization across time).

Putting the Concepts to Work

These principles guide real intervention decisions, and the exam tests the applications as often as the definitions.

Choosing a schedule across phases. Start a new skill on CRF so every correct response is reinforced and acquisition is fast. Once the skill is established, thin the schedule to intermittent (often VR or VI) so the behavior becomes durable and resistant to extinction. Thin gradually; abrupt thinning can produce ratio strain, a breakdown in responding when the requirement jumps too high too fast.

Programming for generalization (Stokes & Baer tactics, high-yield in Domain B and beyond):

• Train loosely and use multiple exemplars so responding is not tied to one narrow SD.
• Program common stimuli — embed cues from the natural environment into training.
• Recruit natural communities of reinforcement so the behavior keeps paying off after training ends.
• Train sufficient exemplars to flatten the generalization gradient appropriately.

Extinction safety. Because of the extinction burst and possible aggression, extinction is usually combined with reinforcement of an alternative (DRA) rather than run alone, and it must be implemented consistently — intermittent 'giving in' converts the schedule into an intermittent reinforcement schedule that makes the behavior more persistent, the opposite of the goal.

Discrimination vs. generalization at a glance: discrimination training builds a tight, context-appropriate response (answer in the SD, withhold in the S-delta); generalization spreads the gain across new stimuli (stimulus), new forms (response), or time (maintenance). A well-designed program does both — it narrows responding to the right occasions while ensuring it appears across all relevant settings and people.