Reward-Based Doesn’t Mean Rule-Free

“My dog needs to know when he’s wrong.”

“Dogs need consequences.”

“You can’t just let them do whatever they want.”

I agree with two of those statements as dogs do need boundaries and their behaviour does have consequences. What I don’t agree with is the assumption that those consequences need to hurt, frighten or intimidate the dog in order to matter.

Reward-based training is sometimes presented as though the alternative to punishment is standing politely at the edge of the room while your dog eats the sofa. It isn’t. I interrupt my dogs, I prevent them doing things, I remove things they shouldn’t have, I use leads, gates, closed doors and management, I end interactions when arousal has gone too high. There are rules in my home and there are behaviours I expect my dogs to learn.

What I don’t use is pain, fear or intimidation to teach them. It is simply being very clear about what I want my dog to learn from the consequence.

The Common Belief or Problem

Dog training discussions often get reduced to two imaginary choices, at one end is the “firm” trainer. There are corrections and there are consequences. The dog learns that unwanted behaviour will have an unpleasant outcome. At the other end is supposedly the reward-based trainer, there are treats and there are no rules. The dog does whatever they like because apparently saying no will cause their entire emotional world to collapse. Neither picture is particularly useful.

Reward-based training does not mean removing structure from a dog’s life and current veterinary behaviour guidance recognises that dogs benefit from routine, predictability, clear communication and consistent expectations. Reward-based training guidance makes the same point: boundaries can be taught without fear, intimidation or pain. The real disagreement is not:

Should dogs have boundaries? It is: How should we teach them?

Dogs Can Learn Rules Without Understanding Them as Morality

Dogs are extremely good learners. They learn which behaviours make useful things happen. They learn which events predict other events. They learn where opportunities appear, when they disappear and which humans behave differently in different situations.

What we should be careful about assuming is that dogs understand our household expectations through a human moral framework of:

Good, bad, right, wrong or naughty

Consider the dog who sleeps on the sofa every afternoon while their owner is at work but never attempts to climb onto it when that owner is home. It is tempting to say:

There is a simpler learning explanation. Perhaps getting onto the sofa when the human is present has repeatedly led to being removed, told off or otherwise interrupted so getting onto it when the human is absent has not.

The dog can learn:

Human present + sofa attempt = unpleasant or unsuccessful.

Human absent + sofa = comfortable nap.

They do not understand that furniture access is morally wrong, it is just unsafe when the human is home. Veterinary behaviour guidance specifically warns that behaviour suppressed by consequences delivered only in the owner’s presence may simply continue when the owner is absent. The dog has learnt the contingency and that may not be the same lesson the human thinks they taught.

The Learner Is Never Wrong

This is one of the principles I find most useful in training.

The learner is never wrong.

That does not mean every behaviour is acceptable. It means behaviour tells us what the dog has actually learnt, not what we intended to teach. For example, if my dog pulls on the lead, I can call them stubborn or I can ask whether I have genuinely taught loose-lead walking strongly enough in this environment.

If my dog comes beautifully when called in the garden but ignores me when a deer appears, I can punish the failure or I can recognise that I have taken a behaviour trained under easy conditions and asked it to compete with something much more reinforcing before it was ready.

If my dog jumps on visitors and every visitor immediately talks to them, touches them or pushes them away, the dog may be getting social interaction every time they jump. Again, the learner isn’t wrong the training picture simply isn’t producing the behaviour we wanted.

Many behaviours owners describe as “problems” are also perfectly normal canine behaviours occurring in inconvenient places: chewing, digging, pulling, scavenging, jumping, chasing and barking among them. We say we love dogs but in reality we really don’t like much about what a dog is or what a dog does. Other behaviour problems can involve fear, anxiety, pain, genetics or medical factors and should not be blamed on poor training.

The useful question is not:

It is:

A Consequence Is Not Automatically an Aversive

The word punishment causes enormous confusion in dog training because it has a technical meaning in learning theory. Punishment is simply a consequence that makes a behaviour less likely to happen again.

Positive punishment means adding something unpleasant after behaviour.

Negative punishment means removing something the animal wants after behaviour.

So if a dog repeatedly jumps up for attention and attention immediately disappears when they jump, jumping may decrease.

Technically, that can be negative punishment too. That is obviously very different from jerking a lead, applying electric stimulation, causing pain with equipment or frightening a dog into stopping.

This is why I generally prefer to say exactly what I mean:

If play becomes too rough, play may stop and if a dog is wildly over-aroused and unable to cope with an interaction, I may calmly remove them from that situation and give them somewhere quiet to decompress.

And if a dog grabs something unsafe, I am going to intervene. If they are about to run into traffic, I am not going to stand there contemplating learner autonomy. Safety, management and boundaries all matters but what matters just as much is what the dog learns from our intervention.

Yes, Punishment Can Stop Behaviour

This is important because pretending otherwise makes the reward-based argument weaker, not stronger. Punishment can suppress behaviour, that is part of its definition.

If a dog barks and an unpleasant consequence happens the barking stops now, from the human perspective, that can look extremely effective. The irritating behaviour disappeared quickly but:

That is especially important when behaviour is being driven by emotion. Imagine a dog who sees another dog and feels frightened, threatened or frustrated so they bark and lunge and the handler delivers a painful or frightening correction.

The dog stops barking but what has changed?

We know the barking stopped but we do not automatically know that the dog now feels safer around other dogs. Stopping the visible behaviour is not the same as improving the emotional state that produced it. Modern veterinary behaviour treatment for fear, anxiety and aggression therefore often focuses not only on changing the behaviour but on changing the dog’s emotional response and teaching a safer alternative response.

A quiet dog is not necessarily a comfortable dog.

What Else Might the Dog Learn?

Learning is happening all the time, including learning we did not intend so suppose another dog appears. Your dog reacts. A painful correction follows. You intend to teach: “Don’t bark.” but another association is also available: Other dog appears → unpleasant thing happens.

Pairing another dog, stranger or other stimulus with restraint or pain can establish or strengthen a negative emotional association with that stimulus. That does not require the handler’s timing to be terrible either. Both events are occurring together so the dog is free to learn about both. The dog may also learn something about the handler, hands approaching may predict discomfort, the lead tightening may predict pain and certain equipment may predict unpleasant events.

The dog may become less willing to display early warning behaviour, again, the visible behaviour may decrease while the emotional problem remains or becomes harder to read which means the human problem can look solved but the dog’s problem may not be.

The Guilty Look Isn’t a Confession

This is another place where human morality gets projected onto dog behaviour. You walk into the room and the bin has been emptied. Your dog lowers their head, turns away, pins their ears back and makes themselves small. They look guilty.

Surely they know they have done wrong right? Not necessarily. In Alexandra Horowitz’s well-known experiment, dogs showed more of the behaviours owners interpret as a “guilty look” when their owners scolded them, regardless of whether the dog had actually done wrong. The effect was particularly strong in dogs that had not committed the supposed offence but were scolded anyway. That does not prove dogs are incapable of experiencing anything resembling guilt but it does show that the guilty look itself is poor evidence that the dog understands they have committed a moral offence. A much safer interpretation is:

Reward-Based Does Not Mean Rule-Free

My dogs have rules. Their rules are:

You cannot chase wildlife.

You need to come when called.

You cannot drag me down the road.

You cannot jump all over somebody who does not want to be jumped on.

There are things you cannot eat.

There are places you sometimes cannot go.

There are times when play needs to stop.

There are situations where you need to settle.

Living successfully with humans requires dogs to learn a remarkable number of strange human expectations and these rules are not the problem.

How we establish them is the important bit.

A dog can learn:

Those are boundaries but they are also learning opportunities. I do not need my dog to be afraid of the consequence of making the wrong choice in order for the right choice to become worthwhile.

Interrupting Behaviour Is Not a Moral Failure

Sometimes people hear “reward-based” and become frightened of interrupting their dog at all. I don’t think that is useful either. If my dog has a cushion they are about to destroy, I may remove the cushion and If arousal has gone through the roof and the dog cannot make sensible choices anymore, I may end the activity and give them a quiet space to reflect their life choices. If a dog is repeatedly rehearsing something unsafe or inappropriate, I may interrupt and redirect them. And that is not the same thing as deliberately frightening or hurting the dog.

The purpose is not:

It is:

There is a huge difference between preventing behaviour and punishing an animal emotionally for behaving.

Management Is Part of Training

Management gets dismissed surprisingly often as though it means the training has failed. It hasn’t. Management prevents the dog rehearsing behaviour while learning is still underway. If my dog raids the bin, I can secure the bin while I work on the underlying behaviour and if recall is unreliable around wildlife, I can use a long line while training. If my puppy jumps on visitors, I can manage the greeting while teaching another way to say hello and if my dog reacts to other dogs, I can create enough distance for learning to happen instead of repeatedly putting them into situations where they explode and then punish them.

Management should sit right at the beginning of many treatment plans because preventing repeated rehearsal protects welfare and safety and stops the unwanted behaviour being strengthened further. Management is arranging the environment so the dog can actually learn.

Teach What You Want Instead

“Stop doing that” is incomplete information. If I do not want the dog jumping, what do I want?

Four feet on the floor? Sit? Go to a mat? Bring a toy?

If I do not want pulling, what does successful lead walking look like?

If I do not want barking at the window, what should the dog do when somebody walks past?

If I do not want the dog stealing cushions, where are their legitimate outlets for carrying, chewing and play?

Modern behaviour modification commonly uses response substitution: teach a desirable response that can replace the unwanted one and that is much more useful than waiting for failure and then punishing it.

So my training question is usually:

Then I teach that and while I am teaching it, I manage the environment so the unwanted behaviour is not being practised twenty times a day.

What About the Quick Fix?

This is where aversive training can be very seductive. Visible behaviour can sometimes be suppressed quickly and that feels satisfying. The barking stopped. The dog stopped pulling. The dog no longer lunges. The human problem has disappeared. Changing emotional responses and building new behaviour can take longer because you are trying to do more than remove the symptom.

You are trying to help the dog feel differently so you teach another response, practise it enough that it becomes reliable and make it work when life becomes difficult.

A controlled 2020 study of 63 dogs with off-lead behaviour problems compared electronic-collar training with training focused primarily on positive reinforcement. The reward-focused group performed as well as or better than the electronic-collar groups on several measures of training performance, and the researchers concluded there was no evidence that electronic collars were necessary for those problems. Other research has found greater stress-related behaviour, greater post-training cortisol changes and more pessimistic responses on a cognitive-bias test among dogs trained with higher proportions of aversive methods.

So the question should not simply be: “Can this make the behaviour stop?” It should be: “What else is this teaching, what is the welfare cost, and is there a safer way to achieve the same goal?”

We Hold the Power

Dogs get very little choice in their world because we decide:

• where they live;

• when and what they eat;

• when doors open;

• where and when they walk;
  

• who they interact with;

• when they receive veterinary care;

• whether they reproduce;

• what behaviours are permitted;

• and what happens when they fail to meet our expectations.

That is an enormous amount of power we have. I don’t think that gives us greater entitlement to control dogs through fear, I think it gives us greater responsibility to teach fairly.

Dogs arrive in our homes behaving like dogs, they pull towards interesting smells, they chase movement, they bark, they chew, they dig, they jump, they investigate and they eat things they find on the floor. They sometimes also make spectacularly inconvenient decisions.

Our job is to help them learn how to live successfully in a human world without punishing them simply for being dogs. Clear. Calm. Consistent. Kind.

Those qualities are not incompatible with boundaries and they are how I want those boundaries taught.

What the Evidence Does Not Prove

Reward-based training does not mean dogs never experience frustration. It does not mean every request is optional and it does not mean humans should never interrupt behaviour.

It does not mean leads, gates, barriers or physical management are inherently unethical and it does not mean every behaviour problem is caused by the owner. The latest scientific evidence gives us a more useful position.

Aversive training carries recognised welfare risks, including increased fear, anxiety, avoidance and aggression, while reward-based methods are effective and avoid intentionally using those risks as part of the teaching process. AVSAB’s current position recommends reward-based methods for all dog training and behaviour modification and advises against the use of aversive methods. That does not require pretending dogs can live without rules. It requires separating rules from fear of breaking them.

What This Means for Owners and Their Dogs

If your dog is doing something you dislike, start by describing the behaviour rather than judging it.

Instead of:

Try:

Instead of:

Try:

Instead of:

Try:

Then ask:

• What does my dog get from this behaviour?

• What predicts it?

• What has reinforced it previously?

• Is emotion involved?

• Could fear, frustration, pain or excitement be contributing?

• What would I like the dog to do instead?

• How can I prevent rehearsal while I teach that?

• And what will my intervention teach the dog about me, the environment and the thing they are reacting to?

That is the part that matters.

Final Thought

My dogs have boundaries and they have rules. Their actions have consequences and sometimes I interrupt them and sometimes access to something ends. Sometimes I manage the environment because they are not ready to make the choice I need them to make. None of that requires me to deliberately hurt or frighten them to make them comply.

The question I care about is not simply:

It is:

Because the quickest way to silence a behaviour is not automatically the best way to solve the problem behind it.

I want dogs who understand what works and dogs who feel safe enough to communicate. Dogs who have learnt the skills they need to live comfortably alongside us and I want their behaviour around me to be built through learning and trust rather than maintained because they are worried about what will happen if they get it wrong.

You have consequences.

You have boundaries.

You just care about what the dog learns from them.

Sources and Further Reading

Educational Disclaimer

This article is for general educational purposes only. It does not diagnose an individual dog or replace veterinary or behaviour assessment.

Behaviour can be influenced by learning, emotion, environment, physical health and individual history. Sudden behaviour change, significant fear, aggression, pain-related behaviour or other welfare concerns may require veterinary assessment and support from a suitably qualified behaviour professional.