Shock Collars vs Positive Reinforcement: What the Science Says

Search for dog training advice and you will quickly hit a fork in the road: aversive tools like shock collars on one side, positive reinforcement on the other. Marketing for e-collars promises fast, reliable results. So what does the actual research say? At Bark Science we ground everything in peer-reviewed behavior research and use reward-based methods only — and we will be honest about where the science is strong and where it is still thin.

The headline finding

The scientific evidence has reached a meaningful consensus: reward-based training is at least as effective as aversive methods, and carries substantially lower welfare risk. Across multiple studies, aversively trained dogs show more stress and fear without any corresponding gain in obedience. That combination — same results, more harm — is the core of the case.

What the welfare studies found

In a well-known study of 92 companion dogs (Vieira de Castro and colleagues, 2020), dogs trained with high proportions of aversive methods showed significantly higher post-training increases in salivary cortisol — a stress hormone — than reward-trained dogs. They also displayed more stress-related behaviors (lip licking, yawning), were more often in tense, low behavioral states, and panted more during training.

The effect reached beyond the training session itself. In a cognitive bias task, the aversively trained dogs were more "pessimistic," indicating a more negative long-term emotional state outside of training. An owner-survey study by Hiby and colleagues found that punishment was associated with an increased incidence of problem behaviors — a welfare concern with no obedience benefit; in fact owner-rated obedience correlated with the number of tasks trained using rewards, not punishment.

What about effectiveness?

This is where honesty matters. The dominant claim for e-collars is that they work better, and the broad evidence does not support that — reward-based methods match or beat aversive ones on obedience while avoiding the stress costs.

But the picture is not perfectly clean, and we will not pretend it is. One controlled chasing-behavior study found that an e-collar group stopped chasing a lure after just one or two sessions, while none of the dogs in the non-aversive groups stopped — a genuine exception worth acknowledging. And a small controlled study comparing leash-walking equipment found no statistically significant welfare difference between a prong-style collar and other tools in that setting. These contradictions exist; the responsible thing is to surface them, not bury them. They do not overturn the overall weight of evidence, which points toward reward-based methods.

The veterinary consensus

The American Veterinary Society of Animal Behavior (AVSAB) recommends that only reward-based methods be used for all dog training, including for behavior problems, and its 2025 board statement says aversive methods — e-collars, prong collars, choke chains, and physical or psychological punishment — should not be used under any circumstances, regardless of the tool or the trainer's experience. The AVSAB position even states there is no role for aversive training in behavior modification. Organizations like the RSPCA and BSAVA take similar positions, and several governments have moved to restrict these tools on welfare grounds.

The dominance justification doesn't hold up

A large part of the cultural case for shock and prong collars rests on dominance theory — the idea that you must "be the alpha" and use force to assert control. That foundation has collapsed. The dominance model came from studies of captive wolves, and the biologist who did the most to popularize it, David Mech, later repudiated it: he studied a wild wolf pack for 13 years and never observed a single contest of dominance, finding instead that wild wolves live in cooperative family units. Dogs, moreover, are domesticated animals genetically distinct from wolves, so wolf-pack models are a poor framework to begin with. When the rationale for a tool turns out to be a debunked theory, that is worth weighing heavily.

Why the welfare cost also hurts results

There is a practical thread connecting welfare and effectiveness: stress and fear inhibit learning in all animals. A dog that is anxious from a correction is a dog that learns more slowly and may grow more fearful or reactive toward whatever it associates with the pain. There is also a specific risk with punishment and aggression: intimidation may suppress a behavior in the moment, but because it never addresses the underlying emotion, the dog can become more likely to react defensively later. So the welfare-friendly choice is usually the effective choice, too.

What reward-based training looks like in practice

Choosing reward-based training does not mean having no structure or letting your dog do whatever it wants. It means using the science of how dogs actually learn: marking and rewarding the behaviors you want so they happen more often, managing the environment so your dog rarely gets to practice the behaviors you don't want, and — for fear-based problems like reactivity — changing how your dog feels about a trigger through counterconditioning rather than punishing the reaction. It is structured, deliberate, and backed by the weight of the evidence. The result is a dog that cooperates because the choices you want pay off, not because it is afraid of the alternative.

Choose the evidence-based path

If you want results without the welfare cost, reward-based training is the approach the science supports.

Take our free Reactive Dog quiz to get a confidential read on your dog and a reward-based plan built from the research.