Differential reinforcement is a technique used in Applied Behavior Analysis (ABA) that works on two tracks at once: reinforcing a desired behavior while withholding reinforcement for an undesired one. Instead of relying on punishment to reduce problem behaviors, it shifts the balance so that appropriate behaviors “pay off” and problematic ones gradually fade. It’s one of the most widely used strategies in ABA, particularly with children on the autism spectrum, because it builds skills at the same time it reduces challenging behavior.
How the Two Processes Work Together
Reinforcement, in behavioral terms, is anything that follows a behavior and makes it more likely to happen again. A child who asks for a break and gets one is more likely to ask again next time. Differential reinforcement pairs that process with extinction, which is the gradual weakening of a behavior when the reinforcement that was maintaining it stops. If a child has been getting attention every time they throw materials off a desk, and that attention is now consistently withheld, the throwing loses its payoff over time.
The power of differential reinforcement is that these two processes run simultaneously. You’re not just ignoring a problem behavior and hoping for the best. You’re actively making a better behavior more rewarding so the learner has a clear path to getting what they need. The replacement behavior fills the gap left by the one you’re trying to reduce.
The Main Types of Differential Reinforcement
Alternative Behavior (DRA)
DRA involves picking a replacement behavior and reinforcing it whenever it occurs. The key detail: the replacement and the problem behavior can physically happen at the same time, but the reinforcement only follows the replacement. A classic example is reinforcing a child for asking for a break verbally instead of swiping materials off the table. Both behaviors could technically happen in the same moment, but only the verbal request earns the reinforcement. DRA works best when the replacement behavior serves the same function as the problem behavior. If a child throws things to escape a difficult task, the alternative needs to also provide escape, like a request for help or a short break.
Incompatible Behavior (DRI)
DRI is a narrower version of DRA. The replacement behavior you choose is one that physically cannot occur at the same time as the problem behavior. Reinforcing a child for sitting in their chair, for instance, is incompatible with standing up and wandering the room. You can’t do both at once. This built-in incompatibility makes DRI particularly effective because success with one behavior automatically prevents the other.
Other Behavior (DRO)
DRO takes a different approach. Instead of reinforcing a specific replacement behavior, it reinforces the absence of the problem behavior over a set time interval. If a child goes the full interval without engaging in the target behavior, they earn reinforcement. If the behavior occurs, the timer resets.
The interval length matters a lot. Practitioners typically set the initial interval well below the baseline rate of the behavior to ensure early success. If a child currently hits every 5 minutes on average, the first DRO interval might be set at 2 or 3 minutes. Once the child consistently earns reinforcement at that level, the interval gradually increases. There are two main variations: whole-interval DRO, where the timer resets every time the behavior occurs, which tends to reduce behavior more quickly; and momentary DRO, where you only check whether the behavior is happening at the moment the interval ends, which is easier to run in busy classrooms or group settings.
Low Rates of Behavior (DRL)
DRL is designed for behaviors that aren’t necessarily problems on their own but happen too frequently. A child who raises their hand constantly during class discussion, for example, doesn’t need to stop raising their hand altogether. They just need to do it less. DRL reinforces reduced rates of the behavior rather than eliminating it entirely.
There are a few ways to structure this. Spaced-responding DRL reinforces the child only when a minimum amount of time passes between instances of the behavior, gradually shaping longer gaps. Interval DRL divides a session into equal time blocks and delivers reinforcement if the behavior stays below a set number within each block. Full-session DRL looks at the total count across an entire session. In all cases, the starting criterion is set just below baseline so the learner can succeed early, then expectations increase gradually.
High Rates of Behavior (DRH)
DRH is the flip side of DRL. It’s used when a learner is already displaying a desired behavior but not often enough. Reinforcement is delivered for incremental increases in how frequently the behavior occurs. A child who eats very slowly at lunch, for example, could be reinforced for taking bites at a slightly faster rate. DRH is less commonly discussed than the other types because it’s building frequency rather than reducing problem behavior, but it fills an important role when the goal is “more of this,” not “less of that.”
Setting Up a Differential Reinforcement Plan
The first step is defining the problem behavior clearly enough that anyone observing could agree on whether it’s happening. “Being disruptive” is too vague. “Throwing objects off the desk” is specific and measurable.
Next comes figuring out why the behavior is happening. This is the function of the behavior, and it’s the most critical piece. A child who screams during math might be doing it to escape the work, to get attention from the teacher, or because the noise in the room is overwhelming. The function determines what kind of replacement behavior will actually work. If the function is escape, the replacement needs to provide escape. If it’s attention, the replacement needs to earn attention. A mismatch here is one of the most common reasons differential reinforcement plans fail.
Before any intervention starts, you collect baseline data on how often the problem behavior is occurring. This gives you a starting point for comparison and helps you set realistic reinforcement schedules. If a child is engaging in the behavior 20 times per hour, you know the replacement behavior needs to be reinforced frequently enough to compete.
Then you choose your reinforcement schedule. Depending on the behavior and the setting, reinforcement can be delivered every time the desired behavior occurs, every third time, or on a time-based schedule like every two minutes. Early on, more frequent reinforcement is better. You want the learner to contact the reinforcement quickly and often so the connection between the new behavior and the reward is strong.
Mistakes That Undermine the Process
The single most common error is assuming you know what’s reinforcing to a particular learner. Many people find stickers motivating, but not everyone does. What functions as a reinforcer varies enormously from person to person. A preference assessment, where you systematically test what the learner actually gravitates toward, is essential before building any plan around a specific reward.
Fading reinforcement too quickly is another frequent problem. A child’s behavior might improve dramatically in the first few days of a new plan, and the adults around them declare the problem solved and drop the reinforcement schedule. Almost inevitably, the problem behavior returns. Reinforcement needs to be thinned gradually over time, not removed the moment things look better.
Inconsistency across people and settings can also sink a plan. If a child’s classroom teacher is running a DRO but the recess aides don’t know about it, the problem behavior continues to get reinforced in those gaps. Everyone who interacts with the learner needs to be on the same page about what’s being reinforced and what’s being withheld.
Timing matters more than most people realize. If you’re reinforcing appropriate language and a child happens to curse right as you’re handing over a token, you’ve accidentally reinforced the cursing. The reinforcement needs to land as close to the desired behavior as possible, with no problem behavior sneaking in between.
Finally, there needs to be a long-term plan for transitioning from artificial reinforcers (tokens, stickers, special privileges) to the natural reinforcers that exist in everyday life. A child who learns to raise their hand should eventually be maintained by the natural consequence of being called on and participating in class, not by earning points on a chart indefinitely. Without a plan for this transition, the behavior often doesn’t hold up once the formal system is removed.