The Revision Guide for Student Nurses (Part I)

Training With Food Rewards


The most effective way to train an animal is by using some form of reward for good behaviour. If, at the same time, you can find some appropriate way to punish the bad behaviour, then that is good too.

Food is an effective form of reward but the problem is that the dog can become dependent on food treats and will bargain with you. It becomes a kind of negotiation where the dog sizes up the reward and decides whether to obey. No reward equals no obedience. When you try to stop using rewards the dog stops obeying. This is entirely avoidable, and the aim is to have a dog that needs only occasional food reward.

The first thing to do is make sure you are using the right rewards. Some dogs will not work for treats; some prefer games or contact. Some dogs will not work hard enough for a food treat but will work very hard if the only time they get fed is during training. Find out what it is that your dog likes and then withhold it in all but the training situation. The following illustration is about food rewards but it is equally appropriate to games or other rewards. This is not an instruction on how to train; it is about how to use reward. The actual process of training is dependent upon what the dog is being taught.


The rules for using reward

  • Stage 1. Decide on the exact behaviour you are going to reward. Look at the behaviour the dog is doing and decide whether it is what you wanted or asked for. If it is, then reward within two seconds. Repeat this until the dog is responding reliably. In the situation of obeying a command, ask yourself: 'Is the dog doing what I asked or not?' If the dog flopped down when you asked him or her to sit, would you give a reward? No!

During the training process you should repeat the sequence of 'command (what you tell the dog to do), response (what the dog does), reward' over and over again until the dog has a pretty good idea of what to do. Then move onto stage 2.

  • Stage 2. Performance-related pay. For the first few performances all you want is something close to what you want, then you can start to perfect it. During Stage 1. a sloppy or slow sit is enough, but during Stage 2. we want to give a better reward for perfection. Decide what the ideal response is (a quicker sit, more concentration in a distracting environment). From now on responses that are closer to the ideal get a better reward (more food, nicer food, a cuddle or game as well as food etc). Repeat until the response is perfect.
  • Stage 3. Create a gambling addict. Once the behaviour is perfect then you need to reduce the dependence on reward and make the learning more permanent. Put the dog onto a random reward schedule. This sounds complex but is in fact very simple. Ask the dog to perform several acts of obedience but only reward a few of them. Sometimes reward the first time, sometimes the fifth and so on. The dog has no idea which time the reward will come, but it knows that there is a good chance that a reward will come in the end. Always say: 'Good dog', but deliver the actual reward on between 1 to 5, then 1 to 10 and then 1 to 20 performances. The amount of reward is different too; if you deliver a reward after two performances then double the reward, after five trials the reward is five times as much. This is just like a slot machine; the pay out is random but when it comes, it is really good. There has to be an upper limit to this reward; if you get down to rewarding every 50th performance, you cannot give fifty bits of food. But this is exactly the same as the slot machine: the top prize pay out is never as much as the money put into the machine. After a while the dog is not working for an actual reward. It is working for the anticipation that one might come.

Once the dog is on this kind of random reward patter, you can also alternate the reward; sometimes give a pat on the head, sometimes food and sometimes a game. This also increases the addiction because the dog can never be sure what the reward will be.

Concentrate on training one thing at each session, and don't start giving random rewards until the dog is doing all of the things you want exactly as you would like. Once you have a whole set of perfected behaviours you can go onto Stage 3. for all of them.

If you are using 'clicker training' it is important to remember that the clicker must always indicate the delivery of a reward. To wean dogs off reward with a clicker you follow the same routine as above. Start by clicking every behaviour you want, then click better behaviours and give a better reward after the click. Finally click and reward randomly and give extra good rewards just as mentioned above.

Just a couple of final tips:

  • Always precede every command with the dog's name said in a high-pitched musical voice ( e.g. 'Freddy, come').
  • When training at home try to use a very quiet voice so that your dog gets used to concentrating so he can hear what you are saying.
  • Don't shout or get cross, because the dog will then be more interested in the tone of your voice and will ignore what you are saying.

It is easy to understand why. If someone shouted angrily at you in a foreign language, what would you do? I would think about running away! That is exactly what a dog will often think about too. By talking quietly and always using the dog's name you are teaching him or her to listen to what you are saying, which enables them to concentrate better.

© Jon Bowen 2000