Written by Ash B. · 7 min read

If you’ve spent any time around a trainer who works in force-free, non-aversive methods, you’ve heard some version of “reward what you want instead.” That’s the basic idea behind differential reinforcement, and it’s genuinely one of the most useful tools available for changing a dog’s behavior without adding pressure, corrections, or fear into the mix. But the mechanics underneath that simple advice are a lot more layered than “give treats for good stuff,” and if you skip the layers, you end up with a dog who has learned a shiny new trick while the original problem behavior is still fully intact, just waiting in the wings.

This post is for the owners who’ve already tried the surface-level version of this, the ones who taught a “sit for greeting” and still have a dog launching at guests half the time. Let’s get into why that happens and what actually fixes it.

What Differential Reinforcement Actually Means

At its core, differential reinforcement means you reinforce one specific category of behavior while putting the unwanted behavior on extinction, which just means you stop giving it whatever reinforcer has been keeping it alive. No punishment, no interruption tools. You’re not stopping the jumping directly. You’re making the jumping stop working, while making something else start working really, really well.

That distinction matters. A dog jumps on guests because it has, historically, gotten attention, eye contact, hands, sometimes even a shove that (to an aroused dog) reads as play. If jumping stops producing any of that and a hand target produces all of it instead, the dog’s behavior shifts because the payoff shifted, not because you scolded them out of it.

The Four Flavors of Differential Reinforcement

There isn’t just one way to apply this. Depending on the dog and the behavior, you’ll reach for a different schedule:

• DRA (Differential Reinforcement of Alternative behavior): you reinforce a specific replacement behavior while ignoring the target one. Rewarding a “touch” cue instead of jumping up for attention is a classic example.
• DRI (Differential Reinforcement of Incompatible behavior): you reinforce something the dog literally cannot do at the same time as the problem behavior. A dog lying on a mat can’t simultaneously be up on the counter or bracing paws on the dinner table. The behavior itself blocks the other one.
• DRO (Differential Reinforcement of Other behavior): you reinforce the absence of the target behavior for a set stretch of time, regardless of what the dog is doing instead. A frustrated barker gets rewarded just for going 10 seconds without a bark, whether they’re sniffing, sitting, or staring at a squirrel.
• DRL (Differential Reinforcement of Low rates): you reinforce the behavior when it happens less often. A dog who nudges you for attention gets rewarded for an occasional gentle nudge, but rapid-fire nudging gets nothing.

These aren’t interchangeable. A dog with a habit of resource guarding a bed isn’t a great candidate for DRI training a “down” on that same bed. A dog who barks at every passing car is a much better DRO candidate than a DRA one, because you’re not necessarily trying to install a brand new behavior, you’re rewarding the gaps of quiet that are already happening naturally.

You Can’t Skip the Functional Assessment

Here’s where a lot of well-meaning owners go sideways. Differential reinforcement only works if you’ve correctly identified what’s currently maintaining the unwanted behavior in the first place. If you reinforce an alternative but the original behavior is still getting paid somewhere else, you haven’t solved anything. You’ve just added a second trick to a dog who is still getting rewarded for the first one.

Research looking at jumping up in companion dogs found that owner-delivered consequences, things like attention or access to something the dog wants, are almost always what’s fueling the behavior. Once trainers identified the exact reinforcer and cut it off while building up an alternative behavior in its place, jumping dropped significantly. The extinction piece and the reinforcement piece had to happen together. Neither one alone did the job.

This is the part people skip. They start rewarding “four on the floor” without ever cutting off the guest’s habit of petting the dog mid-jump, or without addressing the fact that the dog jumps way more when the door opens and way less when it’s just family in the kitchen. If the functional assessment is wrong, or incomplete, the whole protocol quietly fails while looking, on paper, like it should have worked.

The Weird Side Effect Nobody Warns You About

Here’s the part that surprises even experienced trainers. Differential reinforcement of alternative behavior, done well, can actually make the original problem behavior harder to extinguish if something disrupts the new routine.

The mechanism behind this is called behavioral momentum. When you pour a lot of reinforcement into a specific environment (say, the front entryway, where you’re now heavily rewarding a hand target), the overall reinforcement density in that space goes up. That richer environment can paradoxically make the old jumping behavior more resistant to extinction if the new reinforcement schedule ever gets interrupted, like if you run out of treats, get distracted, or a guest doesn’t play along with the new routine. The dog gets frustrated, momentum kicks in, and the original behavior can come roaring back with more intensity than before you started training at all.

This isn’t a reason to abandon DRA. It’s a reason to understand why “it was working and then suddenly it wasn’t” is a real, documented phenomenon and not a sign that your dog is being stubborn or that positive reinforcement doesn’t work on them.

The Fix Is Where You Train, Not Just What You Train

The research-backed solution here is something trainers already do instinctively but now has a name: separate-context training. Instead of teaching the new alternative behavior for the first time in the exact scenario where the problem happens, you teach it somewhere completely different and low-pressure first.

Practically, that means training a rock-solid “place” cue in a quiet living room with zero distractions, long before you ever try to use it while the doorbell is going off and guests are walking in. Once the behavior is fluent in a neutral context, transferring it into the high-arousal scenario is far less likely to leave the old behavior lurking underneath, ready to resurface the second things get frustrating.

This is also the scientific backbone behind why “splitting” criteria (breaking a big ask into tiny, boring, repetitive steps in a controlled setting before ever raising the difficulty) isn’t just a nicety for impatient trainers. It changes the actual resistance of the old behavior to extinction. Skipping it doesn’t just slow progress, it can undermine the whole protocol.

What This Looks Like Day to Day

If you’re working on a jumping, counter-surfing, barking, or attention-seeking behavior at home, here’s the sequence that actually holds up:

• Figure out the real reinforcer. Attention, food access, and social contact are the usual suspects, but be honest about which one it is for your dog specifically.
• Choose the right schedule for the behavior. Incompatible behaviors work great for physical conflicts like counter surfing. DRO works well for behaviors that just need gaps of quiet, like barking.
• Teach the alternative somewhere boring first. Living room, no distractions, no guests, no doorbell. Get it fluent before it has a job to do.
• Cut off the original reinforcer consistently, everywhere. If jumping still gets attention from one family member even occasionally, the whole extinction process resets.
• Expect a possible extinction burst or a return of the old behavior under stress, and don’t panic when it happens. It’s momentum, not failure.

None of this requires an aversive tool or a raised voice. It requires patience, consistency, and a willingness to train the boring version of the skill before you need the exciting version of it.