Reward-Based vs. Aversive Dog Training: What the Research Says

Research on dog training methods points in one direction: dogs trained with aversive methods such as leash jerks, shouting or shock collars show more signs of stress and fear than dogs trained with rewards, and no study has shown that aversive methods work better. The evidence has limits – many studies are small or rely on owner reports – but it is consistent enough that veterinary behavior organizations recommend reward-based training for all dogs.

What the two terms mean

Reward-based training increases wanted behavior by following it with something the dog values: food, play, praise, access to something interesting. Unwanted behavior is handled by management and by teaching an alternative. How this works in practice is described in the guide to positive reinforcement training.

Aversive training relies on consequences the dog wants to avoid. In the research literature this usually covers positive punishment (adding something unpleasant after a behavior, such as a leash jerk or a shock) and negative reinforcement (keeping up pressure or discomfort until the dog complies). Many trainers use a mixture, which some call “balanced” training; studies often treat this as a separate, intermediate group.

Why the debate exists

For much of the twentieth century, dog training was shaped by military and police traditions and by the idea that dogs try to dominate their owners. That idea drew on early observations of unrelated captive wolves kept together in enclosures. Later fieldwork on wild wolves showed that packs are family groups led by the parents, and the “alpha” model is now regarded as outdated, both for wolves and as an explanation of dog behavior.

In parallel, work on operant conditioning, associated above all with B. F. Skinner, showed that behavior can be built reliably through reinforcement, and marker-based methods such as clicker training later carried those principles into dog training. Since the early 2000s, researchers have compared the approaches directly.

How researchers measure welfare

Dogs cannot report how they feel, so studies use indirect indicators:

• Behavior during training. Lip licking, yawning, a lowered body posture, a tucked tail, crouching, yelping and avoidance are counted from video. Reading these signals is covered in the guide to dog body language.
• Physiology. Cortisol, a hormone involved in the stress response, is measured in saliva before and after training.
• Cognitive bias tests. Dogs learn that a bowl in one location contains food and in another location is empty. A bowl is then placed in between. Dogs that approach the ambiguous bowl slowly are interpreted as expecting a worse outcome, a more “pessimistic” state.
• Owner questionnaires. Owners report which methods they use and which behavior problems they see.

Each measure has weaknesses. Panting or yawning can have other causes, cortisol also rises with excitement, and owner reports depend on memory and honesty. The findings carry weight because different methods lead to similar conclusions.

Key studies and what they found

Study	Design	Main finding
Hiby, Rooney & Bradshaw 2004 (Animal Welfare)	Questionnaire among dog owners in the UK	Reward-based training was associated with higher reported obedience; punishment was associated with more reported behavior problems.
Schilder & van der Borg 2004 (Applied Animal Behaviour Science)	Observation of German Shepherds in protection training with shock collars	Dogs that had received shocks showed more stress-related behavior, also later on the training ground when no shocks were given.
Herron, Shofer & Reisner 2009 (Applied Animal Behaviour Science)	Survey of owners attending a veterinary behavior clinic	Confrontational techniques such as hitting or forcing the dog onto its back frequently provoked an aggressive response.
Cooper et al. 2014 (PLOS ONE)	Controlled comparison of electronic collar training and reward-based training for recall problems	Electronic collars brought no advantage in effectiveness; the authors noted welfare concerns.
Deldalle & Gaunet 2014 (Journal of Veterinary Behavior)	Observation of dogs at two training schools	Dogs at the school using negative reinforcement showed more stress-related behavior and lower postures; dogs at the reward-based school looked at their owners more.
Ziv 2017 (Journal of Veterinary Behavior)	Review of the published studies	Aversive methods put dogs’ physical and mental health at risk; no evidence that they are more effective.
Vieira de Castro et al. 2020 (PLOS ONE)	Dogs from reward-based, mixed and aversive training schools in Portugal; video, saliva cortisol, cognitive bias test	Dogs from aversive schools showed more stress behavior and a larger cortisol rise after training, and responded more “pessimistically” in the bias test.

Stress during training

The most direct evidence comes from observation. In the study by Vieira de Castro and colleagues, dogs at schools using mainly aversive methods displayed more stress-related behaviors, spent more time in tense body states and panted more during sessions than dogs at reward-based schools. Their cortisol levels rose more after training compared with samples taken at home. Dogs from mixed-method schools fell in between on several measures.

Effects beyond the session

The same study is notable because it looked outside the training context. In the cognitive bias task, carried out away from the school, dogs from the aversive group were slower to approach the ambiguous bowl. The authors interpret this as a sign that the effects on emotional state are not limited to the lesson itself. Schilder and van der Borg had earlier observed that dogs trained with shocks showed stress signals on the training ground even when no shocks were delivered, suggesting the dogs had come to associate the place or the handler with the unpleasant experience.

Behavior problems and aggression

Survey studies link punishment-based methods to more reported problem behavior. Herron and colleagues found that confrontational techniques often triggered aggressive reactions in dogs that were already being seen for behavior problems. This matters for safety: a dog that is punished for growling may stop growling without becoming any less afraid.

Surveys cannot show which came first. Owners of difficult dogs may be more likely to resort to punishment. The controlled and observational studies help here, because they point the same way.

Effectiveness

If aversive methods clearly worked better, there would be a trade-off to discuss. So far, the research does not show one. Neither the controlled recall study nor the owner surveys nor Ziv’s review found an advantage for aversive methods.

Limits of the evidence

A fair summary includes what is not known:

• Few randomized experiments. Owners choose their training school, so the groups may differ in other ways, such as the owner’s personality or the dog’s history.
• Small samples. Many studies involve a few dozen dogs.
• Indirect measures. Stress behaviors and cortisol are indicators, not a direct readout of welfare.
• Method categories are coarse. “Aversive” lumps a sharp word together with an electric shock; intensity, timing and the trainer’s skill are rarely separated out.
• Little long-term data. Few studies follow dogs over years.

These limits call for cautious wording, not for dismissing the findings. The direction of the results is consistent across countries, research groups and study designs, and no body of evidence points the other way.

What professional bodies recommend

The American Veterinary Society of Animal Behavior (AVSAB) stated in its 2021 position statement on humane dog training that reward-based methods should be used for all canine training, including the treatment of behavior problems, and that aversive methods should not be used. Several countries restrict or ban specific tools such as electronic or prong collars; the rules differ by country and region, so check what applies where you live.

What this means for your training

1. Choose rewards as the default. Teach skills with food, play and praise; the dog training overview lists step-by-step guides.
2. Watch the dog, not only the result. A dog that complies while licking its lips, crouching or avoiding eye contact is telling you something about the method.
3. Ask trainers precise questions. What happens when the dog gets it right? What happens when it gets it wrong? Which equipment is used?
4. Change gradually if you have used corrections. Remove the aversive tool, lower the difficulty, and pay generously for behavior you want. A period of hesitancy is normal.

Common misreadings of the research

• “Reward-based means no boundaries.” The studies compare how rules are taught, not whether there are rules.
• “It only applies to easy dogs.” The recall study involved dogs with real problems, including chasing, and rewards performed no worse. AVSAB explicitly includes behavior problems.
• “If the dog obeys, it must be fine.” Suppressed behavior and good welfare are different outcomes.

When professional help makes sense

Fear, aggression and reactivity are the situations in which owners are most often advised to “get tougher” and in which aversive methods carry the greatest risk. If your dog growls, snaps, bites or lunges, see a veterinarian to rule out pain or illness and work with a qualified behavior professional who uses reward-based methods. A starting point for one common problem is the guide to leash reactivity.