Reward-Based or Aversive Training: What the Studies Found

Researchers in Portugal filmed 92 dogs during ordinary training classes and tested their saliva for cortisol, a stress hormone. Dogs from schools that relied heavily on aversive methods “displayed more stress-related behaviors, were more frequently in tense and low behavioral states and panted more during training, and exhibited higher post-training increases in cortisol levels than dogs from Group Reward” (Vieira de Castro et al., PLoS One 2020).

While that is one study, it is the most careful one we found on companion dogs in real training classes. This page explains what it and the other evidence show, and where the evidence is weaker than either side admits.

---

1. What the two methods are

Reward-based training marks the behaviour you want and rewards it. Aversive training uses something the dog finds unpleasant to stop unwanted behaviour. AVSAB, a society of veterinary behaviour specialists, describes aversive methods as those that “rely on application of force, pain, or emotional or physical discomfort” (AVSAB, Position Statement on Humane Dog Training, 2021).

It lists examples, including “leash jerking, physical force”, shouting, and forceful manipulation such as “alpha rolls”.

Many trainers mix the two and call it “balanced” training.

---

2. What the Portuguese study found

The study recruited dogs from seven training schools around Porto: three reward-based, two using some aversive methods (the “mixed” group) and two using many (the “aversive” group).

• During training, dogs in both the aversive and the mixed groups showed more stress behaviours, such as lip licking and yawning, than dogs in the reward group.
• After training, cortisol rose more in the aversive group than in the reward group.
• Outside training, dogs from the aversive group were more “pessimistic” in a cognitive bias test, a task that measures whether a dog expects good or bad outcomes.

The authors concluded that aversive methods, “especially if used in high proportions, compromise the welfare of companion dogs both within and outside the training context.”

---

3. What it can’t prove

The authors are clear about the limits:

• It wasn’t randomised. Owners chose their schools. The authors write that “we cannot infer a true causal relationship between training methods and dog welfare. To do so would require a randomized control trial.” They add that such a trial would raise ethical concerns.
• It measured welfare, not results. The study “did not compare the efficacy of training methods.”
• Volunteers may differ. The authors note that “a volunteer bias cannot be excluded”.

Our note: The differences may reflect the dogs rather than the method, for example if owners of harder-to-handle dogs chose aversive schools. The authors collected data on possible confounders.

---

4. Does reward-based training work as well?

This is the question owners ask most, and the evidence is thinner. The Portuguese authors summarise the research as “scarce and inconsistent”: some studies favour rewards, one favours aversive methods, and three find no difference.

The best direct comparison we found is a trial of 63 dogs with problems such as poor recall. Trainers who focused on rewards got better responses to “Come” and “Sit” than trainers using e-collars (China et al., Front Vet Sci 2020). The trial had a weakness of its own: the reward-trained dogs were trained in a different season and weren’t randomly allocated. → E-Collars, Prong and Choke Collars: The Evidence and the Law

AVSAB’s reading of the whole literature is that “Reward-based training methods have been shown to be more effective than aversive methods.” It also cites survey studies, while noting that “Survey studies cannot differentiate between causation and correlation”.

Our note: AVSAB is a professional society, and its statement isn’t peer-reviewed.

---

5. What this means for you

• Start with rewards. Both the welfare evidence and the main veterinary behaviour society favour them, and there’s no good evidence that aversive methods are needed.
• Ask a trainer what they do when a dog gets it wrong. The answer tells you more than a label like “balanced” or “positive”.
• For fear or aggression, see your vet first. AVSAB says these need “a treatment plan that includes environmental management, behavior modification, and, in some cases, medication.”

Our note: “Ask a trainer what they do when a dog gets it wrong” is our suggestion, not from a source.

---

6. What we could not verify

Commonly said	What we found
“Aversive training is proven to cause behaviour problems”	Not proven. The evidence links them, but the studies can’t show cause.
“Reward-based training doesn’t work for serious problems”	Not supported. AVSAB says there is no evidence aversive training is necessary.
“Balanced training is the best of both”	Not shown. The mixed group in the Portuguese study showed more stress than the reward group.

---

Sources

• Vieira de Castro AC, Fuchs D, Morello GM, Pastur S, de Sousa L, Olsson IAS. Does training method matter? Evidence for the negative impact of aversive-based methods on companion dog welfare (PLoS One 2020; full text read)
• China L, Mills DS, Cooper JJ. Efficacy of Dog Training With and Without Remote Electronic Collars vs. a Focus on Positive Reinforcement (Front Vet Sci 2020; full text read)
• American Veterinary Society of Animal Behavior: Position Statement on Humane Dog Training (2021; professional society position statement, not peer-reviewed; read in full)

Always confirm with your vet before acting on anything on this page.