Cat & Dog - Use of Positive Reinforcement (Rewards) in Training

When we speak of training our pets, what we usually mean is that we want them to learn a specific task or behaviour. The best way to train pets is through the proper application of learning theory but it is important to understand which methods to use and when, how to use them and what will happen if they are used incorrectly.

How does learning take place?

Learning occurs by establishing a relationship between a behaviour and another event. If the pet experiences the other event as pleasant, the associated behaviour is likely to be strengthened – for example if when the dog puts his paws on the kitchen counter he finds food, he is likely to keep reaching up onto the counter. If the dog experiences the event as unpleasant, then the associated behaviour is likely to reduce. For example, if when he jumped up at the counter for the first time he knocked a pile of crockery to the floor and was scared by the crash, he is less likely to jump up at the counter again.

When something increases the likelihood of a behaviour occurring, it is referred to as a reinforcer. When something reduces the likelihood of a behaviour occurring, it is referred to as a punisher.

Reinforcement and punishment can both occur as a result of an event occurring, or an event stopping, the following examples will make this clearer:

Positive reinforcement is the term applied when a behaviour increases due to something that the animal likes being added to a situation:

• Something the pet likes happens → behaviour is strengthened → positive reinforcement
• (Dog goes to lie in his bed and owner takes him a chew toy – going to bed is reinforced.)

Positive reinforcement is the mainstay of effective, welfare friendly animal training

Negative reinforcement is the term applied when a behaviour increases due to something that the animal dislikes being removed from a situation:

• Something pet doesn’t like stops happening → behaviour is strengthened → negative reinforcement
• (Dog barks at postman (intruder onto territory) and postman goes away – barking is reinforced.)

Positive punishment is the term applied when a behaviour decreases due to something that the animal dislikes being added to a situation:

• Something pet doesn’t like happens → behaviour is reduced → positive punishment
• (Cat jumps up onto kitchen counter and feels sticky double sided tape under his paws – jumping onto the counter is punished.)

Positive punishment runs the risk of inducing fear and should be avoided as much as possible in the process of animal training

Negative punishment is the term applied when a behaviour decreases due to something that the animal likes being removed from a situation:

• Something pet likes is taken away → behaviour is reduced → negative punishment
• (Dog receiving attention becomes excited and jumps up at person so person withdraws attention – jumping up is punished.)

What kinds of things will my pet consider as positive reinforcement?

They will differ from dog to dog and from cat to cat. For some it may be social contact, a play session, a fun toy, a walk, or a food treat. The key is to select the reward that motivates your pet.

If you choose to use food as a reward, it can be useful to feed your pet, depending on its age, at one or two scheduled meal times. Training sessions can then be performed just prior to a mealtime when your pet is a little hungry as most pets can then be motivated with pieces of their daily diet food. However, it is important not to wait until your dog is really hungry before training since a significant lack of food can make it difficult for an animal to concentrate. In the same way, toys, play sessions and affection can be partially restricted outside training time so that your pet is 'hungry' for these rewards. Obviously it is not appropriate to significantly deprive pets of things which they value outside of training times, but partial restriction can sometimes be helpful in situations where you are finding it difficult to motivate your pet.

Some of the animals that are hardest to train are those that are difficult to motivate and these dogs may do better with a few special treats (or a very high value toy) that are saved specifically for training sessions. Some treats that many dogs like include hot dog slices (frankfurters), small morsels of cheese, liver that has been baked or fried, or pieces of dog food sprinkled with flavouring such as powdered cheese or garlic. Cats may be tempted by fish or liver paste or pate (put some on a spoon for them to lick off), tinned tuna mashed fine so that they can take a small bite, or possibly even a little plain yoghurt.

It is helpful to consider these titbits and biscuits not as treats but as 'training rewards'. If there is no good reason to give your dog a treat, then it is better not to give it anything – excessive delivery of treats, rather than rewards, will fill your pet up, can lead to weight gain and accomplishes nothing. If these rewards are saved for training they are more useful, and the pet will learn to associate the rewards with the desired behaviour. Whenever you are giving your pet something of value, from food to exercise, first give a command so that each reward can be earned.

This approach also teaches good manners to dogs – a dog that is sitting to have his lead attached cannot be leaping around like a lunatic and running the risk of injuring someone.

How do I properly use positive reinforcement?

The proper use of positive reinforcement is more than just giving a treat or a pat on the head, the timing of the reinforcement is very important. Remember that your pet is behaving all the time. So, you need to be sure to reinforce the behaviour that you want and not some other unintended behaviour pattern. Therefore, closely associate the reinforcement with the behaviour you wish to increase. Reinforcement should occur during or immediately following the behaviour. Any significant delay between the behaviour and the reward and you run the risk of the pet engaging in another behaviour while you are administering the reinforcement. A good example of this is when you teach your dog to sit. You tell your dog to sit, and lure him into the position. If you take too long fumbling in your pocket to get the food reward you may find that your dog stands up before the treat is delivered. As a result you have rewarded sit and stand up as a sequence of behaviour and many dogs learn that as soon as they sit they should stand up again in order to receive their reward. 'Clicker training' can help to provide good positive reinforcement timing if you struggle to get a treat to your pet quickly.

Should I reward my pet every time?

The frequency of reinforcement is important. The rate at which behaviour is reinforced is called the 'schedule'. There are several different schedules of reinforcement.

1. Continuous reinforcement. Every time your pet engages in a behaviour it is reinforced with a reward. While this may sound like a good idea, it is actually less than ideal other than in situations where you are working to establish a completely new behaviour. If you reward a behaviour continuously over a long period of time then you can expect some problems when the rewards stop. When you cease rewarding the behaviour, it will stop through a process of extinction but the immediate reaction to the cessation of reward will be to intensify the behaviour as a result of frustration. To date the behaviour has always been rewarded and therefore there is a very high expectation of delivery of the reward. When it does not arrive the animal’s expectation does not match with reality and frustration follows. The increase in behaviour is sometimes referred to as a frustration spike or an extinction burst.
2. Intermittent reinforcement. The reinforcement does not come after each performance of the behaviour but intermittently. This may mean that instead of a reward being delivered every time a behaviour is performed, the pet gets a reward every third time, then perhaps two in a row, then maybe not until the pet has performed the behaviour five more times. What happens if you reward this way? Behaviour tends to be stronger and last longer since the animal cannot accurately predict which performance of the behaviour will result in the delivery of the reward.
3. Shifting schedule. Start training new commands or tasks with continuous reinforcement but switch to intermittent, variable rates as soon as your pet is responding consistently.

What if my rewards are not working?

There are many reasons why learning fails to happen, some of them are as follows:

• The reward you have chosen may not be considered reinforcing by your pet.
• You may not be reinforcing the correct task. Remember the example of sit and stand up. Be sure that the timing of your reinforcement is correct and occurs during or immediately after the behaviour you wish to increase.
• You may be phasing out your reinforcement before your pet has adequately learned the new behaviour. Go back to basics and reinforce on a continuous schedule for longer so that you can be sure your pet understands what to do.
• You may be repeating commands several times or in different ways and thus confusing your pet. eg some people say 'sit down' - should the dog sit or lie down?
• You may be asking too much of your pet for his capacity at that particular time – for example the environment you have chosen to work in may be too distracting for him, or he may not be physically capable of performing a specific action, for example if he has sore hips and you ask him to sit. Greyhounds and similar build dogs often struggle to sit due to the size of their thigh muscles

If you are sure that none of the above hold true, you may need help from a professional in the field, please speak to your veterinary surgeon who can arrange a referral to someone appropriately qualified to help you.

What type of rewards should I use?

Rewards do not always have to be food. For many pets, owner attention can be a reward and for a dog a walk in the park or a game of fetch can also be rewarding. What is important is that it is appropriate and motivating for your pet. Remember, you need not give a 'special' reward such as food each time your pet performs a task but always acknowledge good behaviour if only with praise or affection.

It is a very good idea to make use of the rewards offered by daily life to help your pet understand that responding appropriately to you pays off. For example, sitting to be allowed off the lead for a run, lying quietly in the crate before being allowed out of the car or standing still for grooming – provided that your pet finds grooming pleasant.

Is there a wrong way to reward my pet?

Yes. We may reinforce behaviours that we do not want. Remember, positive reinforcement makes behaviour increase. So, there may be times when you actually think you are punishing when you are instead reinforcing behaviours. Examples include scolding in a friendly tone of voice, or petting a dog after it jumps up on you even though you do not like it jumping up. Giving any form of attention to a barking dog, a dog that is jumping up, a dog scratching at the back door, or even a fearful dog, may only serve to reward these behaviours. Sometimes people even give a bit of food, pat the dog, or play with it in an attempt to calm it down. What they are really doing however is reinforcing the problem behaviour. What is worse is that when these behaviours are rewarded occasionally or intermittently, the behaviour becomes stronger and lasts longer (see above).

Are rewards only used for training?

There are other situations where rewards can be helpful. For example, it may help a young or even an adult animal to learn to accept new people if that greeting is always coupled with the passive delivery of a food treat. This will help the pet to learn that new people bring something good and the world becomes a better place when they arrive. It is important that the new person does not give the treat to the dog directly from their hand as there is a risk that the concern about the new person could overshadow the positive qualities of the reward and lead to emotional confusion. In many cases, rewards can be used to encourage desirable behaviour outside of specific training sessions. For example, food packed toys may encourage a dog to chew on them and thereby reward him for being relaxed and still but also reward him for leaving the household possessions alone.

Is there any difference in the reward I should use for my cat compared to my dog?

Cats respond to training like dogs, however, they can be more difficult to motivate and they seem to need reinforcement at a higher rate than dogs to maintain performance. Food is often the best reinforcement for cats but many will enjoy play sessions with favourite toys as well. When food is used the process of finding small titbits of human food, or special cat treats with high appeal, may be more reinforcing then regular food. Try to find training treats that are quick to chew and swallow – usually softer foods, otherwise the training session is slowed down whilst each treat is carefully chewed. Remember to think of toys and snacks as rewards rather than treats.

As a result of their natural social behaviour cats are usually more independent than dogs and they have a looser social relationships with their owner. For this reason social rewards like attention and affection are often much less effective for cats.