4 Differential Reinforcement

How the procedures differ

Differential reinforcement arranges reinforcement to strengthen a useful behavior, reduce a behavior, or both. The procedures differ in what earns reinforcement: DRA and DRI reinforce a behavior that occurs, DRO reinforces the nonoccurrence of a target behavior, and DRL reinforces a lower rate of a behavior. These labels describe the reinforcement contingency, not a guarantee of outcome.

Selecting a procedure begins with clearly defining the behavior, considering what consequence may be maintaining it, and choosing a safe, achievable response or criterion. Reinforcement should match the learner’s needs and, when reducing problem behavior, the function of that behavior.

Choosing what earns reinforcement

Use DRA when there is a useful replacement behavior to teach. The specified, appropriate alternative may serve the same need as the target behavior, but it does not have to be physically incompatible with it. For example, if a learner calls out to get attention, teacher attention can be provided for raising a hand or using an agreed-upon signal.

Use DRI when a safe, practical response is available that cannot physically occur at the same time as the target behavior. DRI is a specific type of DRA. For example, reinforcing hands on a book or desk can be an alternative to grabbing materials.

Use DRO when reinforcement depends on the target behavior being absent during a specified interval or at a check-in moment. DRO does not specify which behavior must occur instead. It may be useful when reinforcing a named replacement is not practical or as one component of a broader plan; because it does not teach a particular replacement, pair it with skill teaching when needed. For example, a learner could earn a token if no calling out occurs during a two-minute interval.

Use DRL when a behavior is acceptable or useful at some level but occurs too often. Reinforcement is earned only when the behavior occurs infrequently enough to meet a preset lower-rate criterion, rather than requiring that it stop. For example, a learner could earn a reward for making no more than three appropriate requests during a class period.

Reinforcement schedules

A schedule of reinforcement specifies when reinforcement is delivered. Choose a schedule the learner can contact successfully and the implementer can follow consistently.

• Continuous reinforcement (FR 1): Reinforce every occurrence of the selected response. Initially reinforcing each appropriate request for help is one example, and this schedule is often useful while teaching a new response.

• Intermittent reinforcement: Reinforce according to a response-count or time-based rule, such as every second eligible response or at planned checks. Once a response is established, reinforcement may be made less frequent gradually. Monitor whether the skill and behavior change remain stable as reinforcement is thinned.

Intervals and rate criteria

DRO plans need a clearly specified interval and check. In whole-interval DRO, reinforcement is provided only if the target behavior did not occur at any point during the interval. In momentary DRO, check whether the behavior is absent at the end of the interval and reinforce if it is. Some plans reset the interval after the target behavior occurs; the written plan should state whether to reset or continue timing.

DRL criteria can be arranged in three ways:

• Spaced-responding DRL: Reinforce a response only after a specified minimum time has passed since the previous response.

• Interval DRL: Reinforce if the response count stays at or below a criterion within each interval.

• Full-session DRL: Reinforce if the total count for a session meets the criterion.

Criteria can be adjusted gradually as data support change. For example, if a learner requests attention every minute but appropriate requests are acceptable, an interval DRL might begin with no more than four requests in a 20-minute session. If the goal is to increase the spacing between requests, spaced-responding DRL may fit better. Base the exact starting criterion on observation and an authorized plan, not guesswork.

Planning a procedure

Use an observable definition that specifies what counts as an instance and how the behavior will be measured, such as by frequency, duration, or intervals. Avoid vague labels such as “being disruptive.”

Consider the context and function of the behavior. Use assessment and team guidance to identify what the behavior may access or help the learner avoid. Choose a replacement that can reasonably obtain the same outcome; for example, teaching a break request when behavior is associated with escaping a difficult task.

Then choose what to reinforce: a useful alternative for DRA, an alternative physically incompatible with the target behavior for DRI, the absence of the target behavior for DRO, or a lower rate for DRL. Set a clear schedule and criterion, including the response or interval, how the learner knows reinforcement is available, what reinforcer is delivered, and what happens when the criterion is or is not met.

Plan for teaching and fading. Teach and prompt a replacement response if needed, begin with a feasible reinforcement schedule, and thin it gradually only when data show success. Collect data on both the target behavior and the replacement response or criterion.

Applying the procedures

Suppose observation and team assessment indicate that calling out during a lesson is a way to obtain teacher attention. The learner can raise a hand but does so inconsistently.

• Possible DRA plan: Teach and prompt hand-raising, initially provide teacher attention each time the learner raises a hand appropriately, and record both hand-raises and call-outs. Consider thinning reinforcement gradually when the response is reliable.

• Possible DRI plan: Reinforce a response such as keeping a hand raised while waiting, if it is physically incompatible with calling out in the defined situation and practical for the learner.

• Possible DRO plan: Reinforce intervals in which no calling out occurs. Define the interval and whether a call-out resets it; also teach an appropriate way to get attention so the learner is not left without a replacement skill.

• Possible DRL plan: If occasional verbal contributions are appropriate and the concern is excessive frequency, reinforce meeting a defined lower-rate criterion rather than requiring zero call-outs.

For another choice example, teaching a learner to request a break instead of pushing work away is closest to DRA because it reinforces a specified replacement response. Reducing frequent but acceptable questions may fit DRL because the goal is a lower rate, not necessarily zero questions.

Implementation and safety

Follow the written plan and the supervisor’s direction. Use reinforcement that is meaningful to the learner, deliver it promptly and consistently when the criterion is met, and record the data specified by the plan. If the learner does not yet have the selected skill, teach it rather than assuming it is already available. Review results with the responsible team and adjust criteria based on data.

Withholding reinforcement for problem behavior may be included alongside differential reinforcement, but it is not automatic and should not be improvised. Extinction-based components can produce temporary increases or changes in behavior and may raise safety concerns. Any such component requires an appropriate, individualized plan, training, and monitoring; follow applicable safety procedures and seek supervisory guidance.

• Extinction-based component