Differential reinforcement changes behavior by reinforcing something, not by punishing.
Why it matters
Most behavior-reduction plans include some form of differential reinforcement, and families and staff usually find it acceptable. But each variation answers a different question. Pick DRO when you need DRL, or forget to plan for extinction, and the plan can stall or make things worse.
Practice vignettes often describe a contingency and ask you to name it, or describe a goal and ask you to pick the best procedure. Pay attention to two things in each stem: what earns reinforcement, and what happens after the problem behavior.
Key ideas
- Differential reinforcement has two parts. You reinforce one response class (or the absence of a behavior, or a certain rate of it), and you withhold reinforcement for the other responses. In its classic form, the problem behavior is placed on extinction.
- DRA (alternative behavior): you reinforce a specific, useful behavior that can stand in for the problem behavior, ideally with the same reinforcer that maintains the problem behavior. Functional communication training, where a learner asks for a break instead of screaming, is a DRA variation.
- DRI (incompatible behavior) is a type of DRA in which the alternative cannot physically happen at the same time as the problem behavior, such as hands in pockets instead of hand-biting.
- DRO (other behavior) delivers reinforcement when the problem behavior has not occurred for a set interval (interval DRO) or is not occurring at a set moment (momentary DRO). No specific replacement behavior is named. Start with an interval a little shorter than the average time between responses in baseline, so the learner contacts reinforcement often, then lengthen it gradually.
- DRL (low rates) reinforces a behavior when it occurs at or below a set rate. Use it when the behavior is fine in moderation, such as asking questions or requesting help. Variations include full-session DRL, interval DRL, and spaced-responding DRL, which reinforces a response only if enough time has passed since the last one.
- DRH (high rates) reinforces a behavior when it occurs at or above a set rate. Use it to increase a behavior that happens too slowly or too rarely, such as completed math problems.
- With or without extinction. Adding extinction usually produces faster, larger reductions. When extinction is unsafe or not feasible (for example, a caregiver cannot ignore elopement near a street), you can still use DRA without extinction by making reinforcement for the alternative richer than reinforcement for the problem behavior: more immediate, more frequent, bigger, or higher in quality.
Definitions
- Differential reinforcement: reinforcing some responses (or a particular rate, or the absence of a response) while withholding reinforcement for others.
- DRA: reinforcing a specified alternative behavior while the problem behavior produces little or no reinforcement.
- DRO: delivering reinforcement contingent on the absence of the target behavior during an interval or at a specific moment.
- DRL: reinforcing a behavior only when its rate stays at or below a criterion.
- DRH: reinforcing a behavior only when its rate meets or exceeds a criterion.
- Extinction: discontinuing the reinforcement that has maintained a behavior, which reduces the behavior over time.
Examples from a session
Four procedures, four goals
- DRA in a clinic: Mateo, 4, screams to get his therapist's attention. His BCBA teaches him to tap her arm and say "Look!" Each tap gets enthusiastic attention right away; screaming gets none.
- DRO at home: Aisha, 12, picks at her skin while watching TV. Her parent sets a 3-minute timer. If no picking occurs before it ends, Aisha earns a point toward extra screen time. If picking occurs, the timer restarts (a resetting DRO).
- DRL in a classroom: Liam, 8, calls out answers about 20 times per class. His teacher wants participation, not silence, so he earns a preferred job if he calls out 5 times or fewer (full-session DRL).
- DRH at a day program: Grace, 34, packs 6 kits per hour at her job training site. She earns a break with a coworker when she packs 9 or more.
Common mistakes / exam traps
- DRL is for behavior you want to keep. If the goal is zero occurrences, DRO or DRA fits better. If the goal is "less often but still sometimes," choose DRL.
- DRH increases behavior. It is not a reduction procedure, even though it appears next to DRL.
- DRO can reinforce other problem behavior. Because DRO only checks for the absence of one behavior, a reinforcer delivered while the learner is yelling may strengthen yelling. Watch what is happening at the moment of delivery.
- Extinction side effects. When DR is paired with extinction, expect a possible extinction burst early on. Keep the plan consistent rather than reading the burst as failure. Plan for resurgence if the alternative behavior later stops being reinforced.
- Mismatched reinforcers. A DRA that delivers stickers for a behavior maintained by escape may compete poorly. Match the reinforcer to the function when you can.
Exam trap
The trap: DRO vs. fixed-time (noncontingent) reinforcement. Both deliver a reinforcer on a clock, so ask whether the problem behavior can cancel or delay it. In DRO, an occurrence resets the interval or forfeits that delivery. On a fixed-time schedule, the reinforcer arrives on time no matter what the learner is doing.
Exam trap
The trap: DRO vs. DRL vs. DRH vs. DRA. When two answers look close, check the goal stated in the stem. Reduce toward zero without naming a replacement: DRO. Reduce but keep: DRL. Increase: DRH. Teach a specific replacement: DRA (or DRI if it is physically incompatible).
Quick check
Rosa, 10, earns a token at the end of each 2-minute interval with no spitting. Spitting drops, but staff notice she is often yelling at the moment the token arrives, and yelling is increasing. What best explains this?
Related topics
- Positive and negative reinforcement procedures (G.1)
- Time-based reinforcement schedules, such as fixed-time (G.3)
- Operant vs. respondent extinction, as procedure and as process (B.11)
- Choosing socially valid replacement behavior (H.3)
- Planning for unwanted effects of reinforcement, extinction, and punishment (H.4)