KEY NOTES

Introduction
All species of mammal use the same basic rules in order to learn about the world they live in. That includes human beings too!

The rules have been investigated and understood for a long time and are described by the term 'associative learning'. All this means is that we learn by learning about the association between individual events and their consequences.

Here are some basics:

• If two things happen at the same time then we assume there is a connection between them.
• If I do something and there is an immediate effect then I will learn that there is a connection between what I did and what happened.
• The more often events happen at the same time the more strongly we suspect a connection.
• The more important the events are, then the more likely we are to make the connection quickly - because it matters to us.

We have all experienced the situation where we press a switch and suddenly there is a loud bang in another room. Your first thought is: 'What did I do?', even though we know there is no connection. If it happened every time we pushed the switch then we would soon assume a connection. If instead of a loud bang, a brick fell on your head you would learn the connection a lot quicker because it is important to avoid the brick!

The important difference between animals and us is animals learn about what is going on right now, not what happened a week ago. They need to react quickly because, unlike us, they are constantly under threat or are looking for opportunities that may be gone in an instant. For this reason animals are most able to learn about what is happening now or has happened in the last two seconds. If two events often occur together then an animal will learn a connection. If an animal does something then it will learn about what the consequences were if they happened within two seconds.

What is the importance of this?
Firstly it means that rewards need to be given quickly; any longer than a couple of seconds will be too slow and you may not in fact be rewarding what you think. Secondly it means that if you give a reward you are actually rewarding what is going on right now. For example, if you have asked your dog to sit, but as you go to give the reward the dog stands up what are you rewarding? The 'sitting' happened five seconds ago and is ancient history to the dog. It was standing up as the food went into its mouth, so it may think that it is being rewarded for standing up.

The importance of timely punishment.
Punish as the dog starts to do something wrong and you will have a greater effect and need to be less nasty. If you wait until after the dog has done wrong there will be a problem; the dog may have enjoyed breaking the rules and may balance the punishment against the reward. The dog may also not associate the punishment with what you want to punish. The typical case is the dog that won't come back. He has run up and down the field for three hours with you running behind shouting. Finally he comes back. What do you do? If you shout and get nasty then what are you punishing? You are punishing the thing that you do want - coming back!

Reward and punishment
These are two of the most misused words in the English language! Here are the technical definitions:

• Reward: A reward is anything that causes an increase in the likelihood of a behaviour happening again in the future.
• Punishment: A punishment is anything that causes a decrease in the likelihood of a behaviour happening in the future.

If you tell your dog off for doing the same thing every day for five months and at the end of that time the dog is still doing it the same as before, are you punishing the behaviour? The answer is NO! What you are doing is having no effect at all, so you are neither punishing nor rewarding.

Worse still, if the punishment is delayed the dog may just come to regard you as a bit of a maniac. If when you came home your dog had chewed up the remote control for the TV you would get pretty upset. However, shouting and smacking the dog will not help because the dog cannot make a clear enough association between what it has done and the punishment that is happening. It chewed the remote control just after you went out five hours ago and now you have turned nasty. This is not to mention the fact that a remote control might be important to you, but to a dog it is valueless. You will undoubtedly be angrier because it is a £40 remote control that has been chewed rather than a 30p newspaper but the dog cannot understand this. To the dog it just looks like when you come home sometimes you are nice, whilst at other times you become threatening and potentially violent. The dog cannot understand why you are so unpredictable but falls back on natural behaviour to comfort itself. Unfortunately chewing is one of these natural comfort behaviours so the dog may become even more destructive!

When you realise that dogs don't understand what is going on, but that their natural reaction to a threat is to submit or slink away and hide then it becomes obvious that a dog who looks 'guilty' may simply be trying to appease you and avoid punishment; it has nothing to do with knowing it has done wrong!

Another example ... "My dog has started to jump up at me when I come in through the door. I look at him and say: 'No. Bad dog' and push him away, but he is getting worse."

What the owner is doing has actually caused the dog to jump up more...it is therefore not punishment, it is reward!

How can you use punishment and reward effectively?
Punishment: Imagine the situation where you go out of your front door and a brick falls on your head. You look up and there is nothing going on that might explain why. The next time you go out of the front door the same thing happens, and again you have no idea why. You get a builder to check the house, and still you have no idea what is going on. After a week you decide to go out through the back door from now on because you are fed up having a sore head. After a month of leaving via the back door you decide to try the front one again...and a brick lands on your head. Because having a sore head is unpleasant and it is just as easy to use the back door you choose never to go through the front again. Picture the same situation again, but this time when you look up there is a crane looming over the house. You go back in, and decide to avoid going through the front door again until the crane has gone.

What does this story tell us? It shows that the best punishment is one where the exact mechanism for how it was delivered is unclear. If every time your dog chews a book you bop him on the head with a newspaper, the dog will learn not to chew books while you are around. If every time your dog chews a book there is a cap underneath it that goes bang when the book is picked up, then the dog will learn not to chew books at all. We call this 'remote punishment'; this is punishment that is very closely related to the behaviour that we don't want, but is completely impersonal (nobody was seen to be involved in it). The best forms of punishment are therefore quick, effectively stop the behaviour and are not connected with a person.

Reward: Imagine another situation: at the end of the month you get a bonus. The bonus is related to success, and your boss is always keen to tell you what you need to do to get the biggest bonus. He is also honest; so when he tells you there is a good chance you will get a big bonus you believe him. One month he says: 'You have been working really hard this month, and the bonus is going to be really good. I don't know how much it will be but it could be 30, 40 maybe even 50% bigger." You think: 'Wow I am doing really well; this weekend I was going to go away but if I do a bit of extra work maybe I can get an even bigger bonus!' Your heart is set on a 50 or 60% bonus. At the end of the month the bonus comes, and it is a good one, perhaps not as good as the boss made out but still worth having. Instead of getting £200 you get £260 (30% increase).

Now think of an alternative situation. You are doing well, and have made a loose arrangement for a weekend away. The boss comes in and says: 'We have a lot of work that needs doing, and I want some volunteers to stay in over the weekend to clear the backlog. There's money in it for you; I will give anyone who volunteers an extra £60 bonus.' You decide you don't want the money and keep to your arrangement for the weekend.

This is the difference between a reward and a bribe. Bribes don't work unless they are really good because you have the chance to weigh up what you are willing to do. In the second situation above, somebody will have stayed behind to work, but they certainly negotiated a better bonus than the £60 the boss offered. In the first example you were already familiar with the idea of the bonus and had been 'trained' about how to work to get it.

It's exactly the same with training dogs. If you offer a dog food for doing something, beware that you may be presenting a bribe, and that the dog may attempt to negotiate. As you offer the food you may see a familiar expression on your dog's face: 'Make me a better offer!' If however, you train the dog for food rewards in very simple easy situations first and then gradually build up to more difficult ones - never using the food as a bribe - then the dog has not had a chance to negotiate and will assume that the food reward is going to be good, without having to see it. There is more information about this in the next module: 'Training using food rewards'.

© Jon Bowen 2000