Differential reinforcement is a family of procedures in Applied Behavior Analysis (ABA) that reduce unwanted behavior by systematically reinforcing other, more desirable responses while withholding or minimizing reinforcement for the problem behavior. Rather than relying on punishment or simply ignoring a behavior and hoping it goes away, differential reinforcement reshapes what a person does by making better alternatives more rewarding. It is one of the most widely used and researched strategies in ABA, with applications ranging from reducing self-injury in individuals with developmental disabilities to curbing substance use in adults.
How Differential Reinforcement Works
The core logic is straightforward: if two behaviors compete for the same person’s time and energy, the one that pays off more reliably will eventually win. Differential reinforcement tips the scales by increasing the payoff for a preferred behavior while reducing or eliminating the payoff for the problem behavior. A child who screams for attention, for instance, gets no response when screaming but immediate attention when raising a hand. Over time, hand-raising replaces screaming because it is the more efficient route to what the child wants.
What distinguishes differential reinforcement from simply rewarding good behavior is the deliberate contrast. The reinforcement is “differential” because one response class gets more reinforcement than another along at least one dimension, whether that is how often, how quickly, or how richly the reinforcer is delivered. Recent thinking in the field has formalized this idea. Some researchers have proposed that differential reinforcement of alternative behavior should be defined as providing greater reinforcement along at least one dimension for the desired response while minimizing reinforcement for the problem response, rather than strictly requiring that the problem behavior receive no reinforcement at all.1PubMed Central / Wiley Online Library. On the definition of differential reinforcement of alternative behavior That distinction matters in practice, because completely cutting off reinforcement for a problem behavior is not always safe or feasible.
The Main Types
ABA practitioners use several variations of differential reinforcement, each suited to different situations. The names sound technical, but the differences are practical.
Differential Reinforcement of Alternative Behavior (DRA)
DRA reinforces a specific replacement behavior that serves the same function as the problem behavior. If a child bites to get a toy, you teach and reinforce asking for the toy instead. The replacement has to actually work for the person, meaning it needs to produce the same kind of outcome the problem behavior was producing. A large review of the research found that DRA has been effective across a wide range of problems, from mild communication difficulties to life-threatening behaviors like food refusal, and is considered well-established for treating destructive behavior in individuals with developmental disabilities.2PubMed. A review of empirical support for differential reinforcement of alternative behavior A systematic review of differential reinforcement for individuals with autism found that DRA produced some of the strongest outcomes among the different subtypes.3Advances in Neurodevelopmental Disorders. Differential Reinforcement in Applied Settings for Individuals with Autism: A Systematic Literature Review
Differential Reinforcement of Incompatible Behavior (DRI)
DRI is a narrower version of DRA. Instead of reinforcing any appropriate alternative, you reinforce a behavior that physically cannot happen at the same time as the problem behavior. If an adolescent flaps his fingers repetitively, you might reinforce scribbling with crayons, since that requires keeping hands on the table and occupied.4PubMed Central. Effectiveness of Differential Reinforcement of Incompatible Behavior (Dri) in Managing Problem Behavior of Adolescents with Intellectual Disability with Autism: A Single Subject Design In a nursing home case, a resident’s disruptive behaviors were reduced by reinforcing attendance at structured recreational programs, activities that were incompatible with wandering and disrupting others.5Clinical Case Studies. Using a Differential Reinforcement of Incompatible Behavior (DRI) Schedule to Reduce Maladaptive Behaviors in a Nursing Home Resident The strength of DRI is that the replacement behavior physically blocks the problem, but the weakness is that truly incompatible alternatives are not always easy to identify.
Differential Reinforcement of Other Behavior (DRO)
DRO reinforces the absence of the problem behavior over a set interval. If a child goes five minutes without hitting, reinforcement is delivered. The person does not need to do anything specific; they just need to not do the target behavior. DRO is widely used, but it comes with a quirk that practitioners need to watch for: because reinforcement is delivered at the end of the interval regardless of what the person is doing (as long as the target behavior did not occur), whatever the person happens to be doing right before the reinforcer arrives can get accidentally strengthened. A study with college students found that all 13 participants showed increased rates of untargeted “other” behaviors during DRO schedules compared to simple extinction, suggesting that DRO can inadvertently reinforce behaviors you never intended to encourage.6PubMed. Differential reinforcement of other behavior increases untargeted behavior One early study of DRO for self-injurious behavior also observed a side effect: crying episodes appeared when reinforcement was not delivered at the end of an interval.7PubMed Central. Effects and side effects of DRO as treatment for self-injurious behavior
Differential Reinforcement of Low Rates (DRL)
DRL is the gentlest form. It does not aim to eliminate a behavior, just reduce it to a manageable frequency. This is ideal when a behavior is perfectly appropriate in moderation but happens too often. A student raising her hand in class is fine; raising it 30 times an hour disrupts learning. One study found that a DRL schedule significantly reduced primary school children’s excessive requests for teacher attention while still allowing them to ask for help at a reasonable rate, and the teacher rated the intervention as highly useful and acceptable.8PubMed Central. Using differential reinforcement of low rates to reduce children’s requests for teacher attention The same approach has been used with adults at day-training centers to reduce excessive question-asking.9PubMed. Reducing excessive questions in adults at adult-day training centers using differential-reinforcement-of-low rates
Why the Behavior’s Function Matters
Differential reinforcement works best when it is matched to the reason the problem behavior exists in the first place. Behavior analysts do not just observe that a child hits; they figure out why the child hits. A child who hits to escape a difficult task needs a different replacement behavior than a child who hits to get a parent’s attention. When practitioners conduct a functional analysis first and then design the differential reinforcement procedure around the identified function, outcomes are consistently better. One study demonstrated this directly with children who had a genetic syndrome associated with challenging behavior: after identifying the specific function of each child’s behavior through functional analysis, a treatment combining differential reinforcement and extinction reduced challenging behavior for both children.10PubMed. Using a Functional Analysis Followed by Differential Reinforcement and Extinction to Reduce Challenging Behaviors in Children With Smith-Magenis Syndrome
Getting this wrong can make things worse. If a child’s aggression is maintained by escape from demands and you reinforce an alternative behavior with attention, you have not addressed the actual motivation. The child still has no good way to escape, so the aggression is likely to persist or intensify. This is why functional assessment is considered a prerequisite to differential reinforcement, not an optional add-on.
The Extinction Debate
Traditionally, differential reinforcement has been paired with extinction, meaning the problem behavior no longer produces any reinforcement at all. If a child used to get attention for tantrums, extinction means tantrums now get zero attention. This combination tends to produce faster and more complete behavior reduction. Research confirms that when extinction is included, the target behavior drops to lower levels during treatment compared to when it is not.11PubMed Central. Resurgence following differential reinforcement of alternative behavior implemented with and without extinction
But extinction has real drawbacks. It can produce extinction bursts, temporary spikes in the intensity or frequency of the problem behavior as the person “tries harder” with what used to work. For dangerous behaviors like severe self-injury or aggression, those bursts can be unsafe. Extinction can also be practically impossible in some settings; a teacher cannot completely prevent classmates from reacting to a student’s outburst. For these reasons, researchers have increasingly explored differential reinforcement without extinction, which involves making the alternative behavior much more attractive than the problem behavior without completely eliminating the payoff for the problem behavior. One approach manipulates the quality, speed, or amount of reinforcement so that the alternative simply becomes the better deal. Studies have found that manipulating reinforcer quality, immediacy, and magnitude for the alternative response all reduced problem behavior even when the problem behavior still produced some reinforcement, with the strongest effects seen when all three dimensions were improved together or when reinforcer quality alone was enhanced.12PubMed Central. Varying reinforcer dimensions during differential reinforcement without extinction: A translational model
Resurgence and Why Gains Can Slip
Even when differential reinforcement works well initially, there is a real risk that the old problem behavior comes back if the replacement behavior stops paying off. This phenomenon, called resurgence, is one of the most important practical concerns in ABA. Imagine a child has learned to ask for breaks instead of throwing objects. If the classroom stops honoring break requests, throwing objects may reappear because it is still in the child’s behavioral history.
Research has shown a counterintuitive pattern: the more effective the initial treatment, the bigger the resurgence risk can be. High rates of alternative reinforcement are better at suppressing problem behavior during treatment, but when that alternative reinforcement is discontinued, the old behavior bounces back more. Low or gradually thinning rates of reinforcement are slower to suppress the problem behavior initially, but when those reinforcement schedules end, the old behavior is less likely to resurge.13PubMed. Effects of high, low, and thinning rates of alternative reinforcement on response elimination and resurgence The practical takeaway is that clinicians should not flood a person with reinforcement for the alternative behavior and then abruptly pull back. Gradual schedule thinning, where the reinforcement for the alternative behavior is slowly reduced over time to something sustainable, is critical for maintaining gains long term.14PubMed Central. Reinforcement schedule thinning following functional communication training: review and recommendations
What Happens When Implementation Is Inconsistent
In a research lab, differential reinforcement can be implemented precisely. In a home, a classroom, or a group home, consistency is much harder. Parents get distracted. Teachers have 25 other students. Staff rotate shifts. This inconsistency, referred to as treatment integrity failures, can seriously undermine effectiveness.
Not all errors are equally damaging. Research has shown that accidentally reinforcing the problem behavior is far more harmful than occasionally failing to reinforce the appropriate behavior. In other words, if you sometimes forget to praise the replacement behavior, the procedure takes a hit but can still work. But if you sometimes give in to the problem behavior (reinforcing it even once), the damage to the treatment is disproportionately large.15PubMed Central. Effects of treatment integrity failures during differential reinforcement of alternative behavior: a translational model The order of these errors matters too: if a period of inconsistency follows a period of good implementation, the effects are different than if inconsistency comes first.
A study examining DRO specifically found that the procedure remained effective when implemented with about 80% or 60% fidelity, but fell apart for most participants when fidelity dropped to 40% or lower.16PubMed Central. Parametric evaluation of procedural fidelity level with combined errors during differential reinforcement of other behavior That gives practitioners a rough benchmark: you do not need perfection, but you need to be right more often than you are wrong, and you especially need to avoid reinforcing the behavior you are trying to reduce.
Differential Reinforcement Outside the Clinic
While most of the research on differential reinforcement comes from ABA work with individuals with developmental disabilities, the underlying principle shows up far beyond clinical settings. Contingency management for substance use disorders is essentially differential reinforcement applied to drug and alcohol use. Individuals receive tangible rewards (vouchers, prizes, or money) for demonstrating abstinence (confirmed by drug tests), while continued use produces no reward. Meta-analyses have found that contingency management is among the more effective approaches to promoting abstinence, improving clients’ ability to stay clean long enough to benefit from other treatment components.17PubMed. Contingency management for treatment of substance use disorders: a meta-analysis This approach has even been adapted for delivery via mobile phone, where it has outperformed control conditions in reducing tobacco and alcohol use among adults not in formal treatment programs.18PubMed Central. Mobile telephone-delivered contingency management interventions promoting behaviour change in individuals with substance use disorders: a meta-analysis
The matching law, a principle from behavioral research, helps explain why differential reinforcement works across such different contexts. The idea is that when people have two options, they tend to distribute their behavior roughly in proportion to the relative payoff of each option. One study tested this with children exhibiting both problem behavior and appropriate behavior that served the same function: when reinforcer rates were manipulated on competing schedules, relative response rates closely tracked relative reinforcement rates, just as the matching law predicts.19PubMed Central. Concurrent reinforcement schedules for problem behavior and appropriate behavior: experimental applications of the matching law The same principle has been demonstrated in completely unrelated settings, like college basketball players allocating their shots between two-point and three-point range in proportion to their success rates at each distance.20PubMed Central. An application of the matching law to evaluate the allocation of two- and three-point shots by college basketball players These findings suggest that differential reinforcement taps into something fundamental about how organisms allocate effort, not something specific to clinical populations.
Ethical Complications in Practice
Differential reinforcement is generally considered one of the least restrictive behavioral interventions available, which is part of why it is so widely recommended. But “least restrictive” does not mean free of ethical concerns. One published case report described a program for an individual with intellectual disabilities whose ritualistic behavior led to severe social isolation. The differential reinforcement program successfully reduced the rituals and improved social engagement. However, care staff had concerns about the specific reinforcer being used and about whether the individual had meaningfully consented to the intervention. Those staff concerns led to inconsistent application of the program and its eventual termination. A follow-up assessment found that the individual’s behavior had returned to pre-treatment levels.21PubMed Central. Ethical issues involved in the implementation of a differential reinforcement of inappropriate behaviour programme for the treatment of social isolation and ritualistic behaviour in an individual with intellectual disabilities
That case illustrates a tension that recurs across ABA: an intervention can be technically effective but practically unsustainable if the people implementing it do not buy into it or if the person receiving it has not had a meaningful say. Differential reinforcement requires everyone involved, from parents to teachers to direct care staff, to follow through consistently. If caregivers find the reinforcer objectionable, or if the replacement behavior feels artificial or burdensome to the individual, the program tends to erode. Social validity, the idea that interventions should be acceptable and meaningful to everyone involved, is not just a nice ethical aspiration; it is a practical requirement for durability.
Choosing Among the Subtypes
With several variations available, practitioners have to match the right subtype to the situation. A few practical guidelines shape those decisions. DRA is the default for most cases where a clear functional replacement exists. If the person can learn a new way to get what they want, DRA is the most natural fit and the best-supported option. DRI works well when a physical incompatibility makes the solution obvious, but it is less flexible because truly incompatible behaviors are not always available. DRO is useful when you cannot identify a specific replacement, when the problem behavior is dangerous enough that you simply need it to stop, or when you are targeting multiple problem behaviors at once. DRL is the right choice when the behavior is appropriate in moderation and you do not want to eliminate it entirely.
The systematic review of differential reinforcement for individuals with autism found that DRA and DRL produced larger effect sizes than DRO and DRI, though the review also noted that study quality varied widely, with only three studies meeting all quality indicators.22Advances in Neurodevelopmental Disorders. Differential Reinforcement in Applied Settings for Individuals with Autism: A Systematic Literature Review These differences do not mean DRO and DRI are poor choices. They may reflect the different clinical situations in which each subtype tends to be used. DRO, for example, is often reserved for the most severe or complex cases, which makes it look worse on paper even if it is the right tool for those specific situations.
Making It Work Over the Long Term
The hardest part of differential reinforcement is not the initial behavior change. It is keeping those gains once the structured intervention fades. In real life, no one can deliver reinforcement on a dense schedule forever. A child who initially earns a sticker after every single appropriate request eventually needs to sustain that behavior with less frequent and less predictable rewards. Schedule thinning, the gradual shift from rich reinforcement to something more natural, is where many programs succeed or fail. Thin too fast and the replacement behavior drops out, leaving room for the old problem to resurge. Thin too slowly and the intervention becomes unsustainable for the people implementing it.
Most experts recommend small, incremental steps with clear criteria for when to move to the next step. If the person is stable at the current schedule for a set number of sessions, you thin slightly. If the problem behavior starts creeping back, you briefly return to a richer schedule before trying again. The goal is a smooth landing at a level of reinforcement that the natural environment can realistically maintain, something that looks less like a clinical procedure and more like the kind of feedback and recognition that happens in everyday life.