20/05/2026
Positive reinforcement has become the dominant language of modern dog training. In many ways, that is a good thing. Used correctly, positive reinforcement is rooted in sound behavioural science and gives trainers a clear framework for teaching dogs what behaviours are worth repeating. The problem is not positive reinforcement itself. The problem is the way it is often misunderstood, oversimplified and treated as a complete philosophy rather than one tool within learning theory.
At its core, positive reinforcement is straightforward. A behaviour occurs, something the dog values is added, and the likelihood of that behaviour happening again increases. That is all it means. The word “positive” does not mean kind or gentle. In behavioural science it simply means something is added. “Reinforcement” means the behaviour increases in frequency. 
This principle comes from operant conditioning, largely associated with B.F. Skinner’s work on behavioural consequences. Dogs learn through outcomes. Behaviours that improve the dog’s situation tend to repeat. Behaviours that do not tend to fade away. 
That is why food, toys, praise and play work so well in training. If a dog sits and receives something valuable, sitting becomes more likely in the future. Repeat the process enough times with good timing and consistency, and the behaviour becomes increasingly reliable. Eventually it becomes habit. In many cases, it becomes automatic and can later be placed under cue control.
This is where good trainers separate themselves from people who merely hand out treats.
The objective of reinforcement is not endless payment. The objective is learning.
A properly reinforced behaviour should move through stages. Initially, the dog is discovering what behaviour produces value. Then the behaviour becomes more fluent and reliable. Over time, repetition and consistency build a habit pattern. Eventually, the dog responds automatically because the behaviour has become deeply conditioned and historically successful.
That is how real learning works.
You should not need to spend the rest of a dog’s life bribing it into basic obedience. If a dog fully understands sit, down, stay or place, those behaviours should occur smoothly and automatically when cued. The dog is no longer performing because it has spotted chicken in your hand. It is responding because reinforcement history has built a strong behavioural pattern.
This is something many modern trainers seem reluctant to admit. Large parts of the force-free world have unintentionally created dogs that are permanently dependent on active reinforcement. Owners walk around carrying treat pouches, tug toys and food rewards for years, constantly negotiating with the dog for compliance. The training never really ends because the behaviour never truly stabilises.
That is not sophisticated training. It is prolonged management.
Of course, rewards still exist in the dog’s life. Good trainers play with their dogs, interact with them and provide enjoyable experiences. But that is very different from continuously paying for already trained behaviour. Once behaviour becomes automatic, reinforcement becomes far less frequent because the learning has already taken place.
When teaching something new, reinforcement becomes important again. That is exactly how learning theory predicts behaviour should work. You reinforce new responses so the dog learns that the new behaviour improves its situation.
Where the discussion becomes uncomfortable for many reward-only trainers is when we look at competing reinforcers.
Learning theory does not say reinforcement magically overrides every other motivation a dog possesses. In fact, behavioural science suggests the opposite. Dogs will always gravitate towards the behaviour they perceive as most valuable in that moment.
This matters enormously when dealing with instinctive, self-rewarding behaviours.
Predatory behaviour is the obvious example. Chasing livestock, hunting wildlife or pursuing moving animals can be intrinsically reinforcing to certain dogs. The behaviour itself delivers enormous reward through genetics, adrenaline, dopamine and fulfilment of instinctive motor patterns. The dog does not need a handler to reinforce the behaviour because the behaviour reinforces itself.
That changes the equation entirely.
You can absolutely use positive reinforcement around livestock. You can reward disengagement, reinforce recall, build alternative behaviours and work carefully around thresholds and arousal levels. Those things can all improve outcomes and increase control. But there is an important reality many trainers refuse to acknowledge: if the dog ultimately values the chase more than anything you are offering, the chase will win.
Every time.
This is not anti-science. It is learning theory in its purest form.
Reinforcement only increases the likelihood of behaviour relative to competing consequences. It does not erase competing reinforcement histories, nor does it extinguish instinctive rewards simply because food is present. Even reward-based literature acknowledges that behaviours such as chasing can be intrinsically rewarding and difficult to extinguish through reinforcement alone. 
That is where ideology starts to clash with reality.
Many trainers speak as though reinforcement alone can solve every behavioural issue if timing, threshold management and reward selection are good enough. Sometimes that is true. Sometimes it is not. A dog with extremely high predatory drive may still choose the self-rewarding behaviour despite excellent reinforcement history elsewhere.
Dogs are not robots. They are living animals driven by competing motivations.
The irony is that learning theory itself already explains this perfectly well. Behaviour is governed by consequence. If chasing sheep produces the highest perceived value to the dog, then that behaviour remains highly reinforcing. Pretending otherwise does not make it less true.
This is why balanced trainers often argue that reinforcement alone is incomplete. Reinforcing desired behaviour is essential, but there must also be clarity around behaviours that are unacceptable. Animals learn from both reward and consequence. Operant conditioning has always included reinforcement and punishment within the same framework. 
That does not mean harshness, intimidation or mindless correction. Good training should always be fair, proportionate and understandable to the dog. But the idea that all unwanted behaviour can simply be out-rewarded is not supported by real-world behavioural complexity.
A dog ultimately performs the behaviours that have proven most worthwhile throughout its life.
That is the actual goal of training. To create clear habits and reliable behavioural patterns that the dog understands are beneficial. Not endless bribery. Not constant negotiation. Not carrying food forever in the hope the dog continues cooperating.
When training is done properly, behaviour becomes fluent, reliable and automatic. Reinforcement teaches the lesson. Consistency solidifies it. Habit maintains it.
And that is the point many people seem to have forgotten.
