Reward vs. Aversive:

What Science Says About Dog Training

Dog training is not just about getting your dog to sit or stop barking. It is about how your dog learns, how they feel during the process, and what kind of relationship you build together. The method you choose shapes all three.

The Four Ways Dogs Learn

In behavior science, a dog’s learning falls into four categories. These terms have precise meanings that differ from everyday speech:

• Positive reinforcement — we add something pleasant to increase a behavior. (Your dog walks calmly at your side; you give a treat and praise.)
• Negative reinforcement — we remove something unpleasant to increase a behavior. (Pressure from a prong collar is released only when the dog moves into heel position.)
• Positive punishment — we add something unpleasant to decrease a behavior. (A shock is delivered to stop barking.)
• Negative punishment — we remove something pleasant to decrease a behavior. (You stop walking forward when your dog pulls on the leash; movement resumes only when the leash is slack.)

Positive reinforcement and negative punishment are considered humane, non-aversive methods. Negative reinforcement and positive punishment are classified as aversive methods — they rely on discomfort, pain, or fear to change behaviour (Here is why the dominance theory falls apart).

What the Research Actually Shows

Aversive methods are not more effective — and often less so.

A 2014 randomized controlled trial compared three groups of dogs with recall problems: one trained with e-collars by industry-approved trainers, one trained by the same trainers without e-collars, and one trained by reward-based professionals. The result? The e-collar group showed no better obedience outcomes. They spent significantly more time tense, yawned more frequently, and engaged less with their environment — all signs of stress. A 2020 reanalysis of the same data found that reward-based training achieved equivalent results more efficiently, requiring fewer signals and producing faster responses.

A 2017 systematic review of seventeen peer-reviewed studies concluded that there was no evidence positive punishment is more effective than positive reinforcement, and that aversive methods jeopardize dogs’ physical and mental health.

Aversive methods carry serious welfare risks

Dogs trained with shock collars show elevated cortisol and heart rate during and after training. Meaning that, the stress does not stay in the training session. In one study, dogs trained with shock collars displayed stress behaviours — lowered body posture, avoidance, heightened vigilance — not only during training but also later, simply when encountering the trainer or the training environment.

Clinical surveys paint an equally stark picture. In a study of 140 dogs referred to a veterinary behavior clinic, confrontational techniques triggered aggression in a significant proportion of cases: alpha rolls elicited aggression in 31% of dogs, dominance downs in 29%, hitting or kicking in 43%, and shock collars in 10%.

Suppression is not the same as learning

Aversive methods often suppress behaviour — the dog stops doing it to avoid pain — without resolving the underlying emotion. This creates a dangerous illusion of a “fixed” dog. When a dog can no longer suppress their fear or frustration, the result can be an apparently sudden, unprovoked bite — what owners might call “out of the blue” aggression, but what is actually the endpoint of long-term emotional suppression. In extreme cases, repeated punishment can lead to learned helplessness: the dog stops offering any behaviour at all, not because they are calm, but because they have learned their actions do not matter.

Why You Should Avoid Aversive Methods

They demand perfect timing — and perfect timing is nearly impossible.

For punishment to work, it must be delivered within seconds of the unwanted behaviour. Delay it slightly, and you risk punishing the wrong behaviour — or worse, punishing your dog for something they did not do. Reward-based training is more forgiving of minor timing lapses, though optimal learning still requires quick delivery. The key difference is not that reinforcement tolerates delay, but that the consequences of poor timing with punishment are far more severe: fear, broken trust, and learned associations you never intended.

They do not teach your dog what to do instead

When your dog barks at another dog on a walk, they are not plotting mischief. They are reacting to an emotion — often fear, but sometimes frustration, overarousal, or excitement. Yanking the leash, yelling, or delivering a shock tells your dog what not to do, but it does nothing to address the emotion driving the behaviour. In fact, if your dog is afraid, adding pain or intimidation confirms their worst fear: This situation is dangerous. Their reactivity may actually escalate, not diminish, because now they have even more reason to want the trigger to go away.

They erode trust and damage your bond

Dogs trained with positive reinforcement show fewer stress signals and are more obedient on novel tasks than dogs trained with punishment. When training is predictable and enjoyable, your dog learns to trust you as a guide through difficult situations. When training is unpredictable and painful, that trust frays. The dog may obey, but out of anxiety rather than cooperation.

The Professional Consensus Is Clear

The evidence has moved beyond individual studies into international policy. In June 2024, the Federation of Veterinarians of Europe, the Federation of European Companion Animal Veterinary Associations, the Federation of European Equine Veterinary Associations, and the World Small Animal Veterinary Association unanimously adopted a joint position paper calling for a complete ban on the sale and use of electric pulse training devices, including shock collars.

Major veterinary and professional bodies — including the American Veterinary Society of Animal Behavior, the British Small Animal Veterinary Association, the British Veterinary Association, and the Australian Veterinary Association — have issued position statements recommending reward-based methods and opposing aversive equipment.

Governments have acted. Wales, Switzerland, Germany, Norway, Sweden, Finland, France, Belgium, Spain, Colombia, Quebec, and multiple Australian states and territories have enacted bans or restrictions on electronic collars, prong collars, or both.