Positive reinforcement training

The Method

Positive reinforcement (R+) training, focuses on reinforcing (rewarding) behaviour you want repeated and ignoring unwanted behaviour. By definition, positive reinforcement occurs when behaviour is strengthened (reinforced) because a desired consequence (e.g. food, game) is added (+).

Mark the instant of correct behaviour and immediately follow with a reward.

Focus on the dog’s behaviour not the dog’s intent. You should notice and try to read body language and emotion accurately as you want the dog’s emotions (and your own) to be working for you, not against you, however, that’s all you need to analyse as you train the behaviour that you want. Stop trying to read your dog’s mind (you can never know what she’s actually thinking). Let go of frustration with your perception of the dog’s ulterior motives and look for ways to prevent it doing the wrong thing as you train the right thing.

The Skills

There are three fundamental skills to understand and develop.

1. Timing

When you instantly mark correct behaviour, and quickly deliver a reinforcer (reward),  the dog better understands what you want and is likely to repeat it.

2. Criteria

To mark on time you must know EXACTLY what behaviour you want. Criteria is the most important part of the method. Not only must you know what you want, but the dog must be able to achieve it quite quickly and easily. Failure after failure demotivates,  confuses and causes anxiety and frustration, for both dog and handler.

3.  Rate of Reinforcement

If the criteria is readily achievable, the behaviour can be practised a lot, reinforced a lot and thoroughly proofed. For efficient learning reinforcement should be happening quickly and often (aim for at least 8 successes out of 10 repetitions). The dog quickly learns which behaviour pays off and is less likely to offer a behaviour that doesn’t e.g. when a sit is being practised and reinforced, standing up isn’t being practised or reinforced. 

Proof the behaviour but don’t bore the dog

Basic training may seem quite quick but establishing new habits and proofing behaviours thoroughly takes many, many repetitions (hundreds if not thousands). Don’t be in a hurry to consider the dog is ‘trained’ and stop reinforcing. Training to fluency (quick, reliable, accurate responses in many contexts) takes time and numerous reinforcements.

For anything other than really simple behaviour, setting the dog up to succeed may need careful thought and planning.  Behaviours and behaviour chains are built up with systematic training in small logical steps gradually adding the four Ds of proofing … things which are different (locations, positions relative to you etc), distance from handler, duration of behaviour and strength of distractions.

The number of repetitions a dog can do well and enjoy in one session will vary. Sticking too long at one criteria and repeating any behaviour so often that repetitions become slower and slower and the dog becomes bored or stressed, is not efficient or effective training. Always quit while you’re ahead, with the dog keen to work and wanting more.

Proofing can make the training more interesting because you are raising criteria by making little changes which challenge the behaviour. However, remember this also makes the job harder for the dog.  A successful training session teaches certain skills but maintains the dogs’ working attitude and relationship with you.  

Keeping it simple

Get the behaviour.

Lures or targets are fine, however fade anything you don’t want to stay as part of the behaviour. Dependence on visible food or targets may not be what you want.

Train what the dog needs.

In my early clicker training days the focus was on getting the dog ‘thinking’ and ‘doing’. This often led to busy, busy dogs experimenting wildly and getting stressed and confused. Dogs who are already busy may benefit from learning passive behaviours like standing calmly, for a start, while less active ones may benefit from learning to experiment, with games like “101 things to do with a box”.

Start a training session with an end point in mind and stick to your target.

Mixing and matching behaviours within the session can confuse the dog about what is actually being trained. If the session isn’t working, stop, toss the dog a toy or treats and rethink your plan.

Get the behaviour on cue.

This means the dog does the correct behaviour in response to the correct cue, all the time.  Having a lot of behaviour, none of which is really on cue, is a recipe for confusion and frustration for dog and handler.

Train ONE criteria at a time

E.g. when getting ‘sit’ on cue, all you are doing is clicking when the dog sits after the cue. You aren’t asking the dog to sit for long or cope with distractions. You just click, reinforce and release for another trial.

When you raise one criteria you need to lower others

E.g. If you want the dog to hold position for longer, just stand beside him and reward from there. Don’t go a long way away because that would be asking for increased handler distance AND a longer sit. If the dog breaks, you have an unnecessary failure and if he becomes confused he will likely lose confidence and enthusiasm or become anxious, noisy and overactive. You are aiming for 80% success before you make the job harder. Once the dog can hold the position for some time you can begin to move away a little (add distance).

What R+ training isn’t….

It isn’t: just using food.  Anything the dog wants is a suitable reinforcer. Choose wisely. If the dog has no interest in biscuits try steak. If the dog can’t think straight for steak or a toy, try something lower key. Sitting while the leash is put on is reinforced by going for a walk. Every time you allow a dog to go for a walk after you have battled to put the leash on, you risk reinforcing overactivity rather than holding the sit.

The way the reinforcer is delivered can also strengthen behaviour.Food in position can reinforce stationary positions. Don’t let the dog bounce up out of a sit to get the treat, only feed in the sit, then release. Thrown toys can reinforce movement and speed.

R+ isn’t: all positive. No training can be. When you restrict a dog who just wants to leap on somebody and you wait for them to settle enough to sit, the dog may feel stressed. It may squeal and struggle in a way that’s probably not a lot of fun for the dog or handler but it may be a necessary stage to get through for learning to take place.

Unfortunately if some well meaning person approaches and pats your dog while this is happening, the battle may be reinforced; not what you want. If the dog simply can’t settle, the lesson may be too hard. He may need to learn to sit calmly further away from exciting people before he can be brought closer.

R+ isn’t: automatically fun and motivational for dog or handler. The skills are learned and many well intentioned handlers learning the ropes find themselves feeling frustrated with failure and/ or trying to understand why their dog is so worried. Don’t take it personally but do take it on board. Try something different. If you’re all out of ideas, there are numerous trainers you could ask, books to read and email lists you could join.

It isn’t: letting the dog do whatever she likes and hoping she will chance upon the right thing. Reinforcement doesn’t just come from you. If chasing bikes is ‘fun’, the dog will keep doing it. You may need to control the dog so she can’t practise unwanted behaviour while you teach alternatives. Remember LIFE is a TRAINING SESSION. Dogs running up and down the fence chasing bikes going past are learning, even if you aren’t there doing the training.

R+  training is…

It is: a method requiring a shift in mindset. Instead of punishing mistakes the trainer aims to avoid them, help the learner to be ‘right’ and reinforce that.  The learner must feel safe enough to try and to make mistakes, so the trainer can see what they are actually learning and change their teaching accordingly. Mistakes provide useful information.

It is: confidence building for the trainer when they take responsibility for problem solving and make progress. Frustration and resentment slide quietly away.

It is: confidence building for the dog when they feel safe to experiment and show what they understand, even when it’s not what the trainer intended them to learn!

It is: a method requiring time, patience, study and practice to learn effectively. Many mistakes will be made and there will be frustrations, however the commitment to control your temper and avoid blaming your dog, will pay off for both of you.

It is: a method that can be used to train behaviour as well as performance skills. Dogs who sit nicely have the sit reinforced by being allowed to greet people or play with other dogs. Dogs who normally react to other dogs get reinforced for walking past without reacting. Dogs who demand attention are ignored until they relax, and are then given calm attention to reinforce relaxed behaviour.

I think…

Dogs are unusual as they are one of the few animals that we train which also share our lives and our homes. Horses, goats, chickens and exotic animals can be trained in short focused sessions, then left to their own devices for the rest of the day, safely enclosed in their paddocks, cages or pools.

Our dogs are around us all the time when we are at home, perhaps closely supervised but often not, or supervised by the wrong people such as children playing roughly. The pet dog in the home has many more opportunities to practise and learn the wrong thing than most other animals and are often punished for their motives (e.g. ‘trying to get back’ at the owner) rather than for what they actually did, at the moment they did it. Training the behaviour that is actually wanted, is also often overlooked in the rush of busy human lives.

R+ training is a method, but its full benefit comes from using it as a philosophy. In this way the focus isn’t simply on the end product; what the dog can ‘do’.  Although important, these can be taught in other ways. As a philosophy, it teaches trainers to look for positive events and ways to solve problems. They realise that the way they live with their dogs is as much a part of the learning as a formal training session, because the dog is always learning about them and about the world. They focus on the learning process, its associated attitudes, and the relationship forged between learner and teacher. As well as tricks and manners, they look for confidence, enthusiasm, self control and trust between dog and trainer.