When we begin teaching a new skill, we reward the dog each and every time he gets it right.
But, over time, to ensure that the new skill is maintained many trainers make rewards more intermittent or occasional.
We often refer to this as ‘fading’ rewards. People usually direct their enthusiasm for fading rewards at food.
They don’t seem so worried about getting rid of other rewards like toys, games, or opportunities to run about and play.
Why do people fade rewards? – 1: bad reasons
There are good and bad reasons for fading rewards. Here are some bad reasons to fade rewards in dog training
- Because your friends don’t use food
- Because your trainer says your dog is too dependent on food
- Because you can’t be bothered to carry food around with you
- Because you are worried that your dog is not working for you out of love or respect
Many people start off by training with food, with the clear goal in mind, of getting rid of food as quickly as possible. This inevitably leads people to fade rewards long before the dog is ready, or far too quickly, or even to discontinue rewards altogether.
But my friends don’t use food
There are still many people out there, that don’t reward their dogs with food, if at all.
There are also many people out there that still use a lot of punishment in dog training, or that fail to train their dogs successfully. If you don’t want to fall into either of these camps, you will need to use effective rewards for your dog, and food is one of the most convenient effective rewards available to you.
So try not to worry about what other people are doing
My trainer says food is cheating
Many traditional dog trainers view food as cheating. This is of course entirely up to them.
To avoid being harassed by a trainer over your use of food, you need to pick a modern dog trainer that teaches according to the principles of behavioural science.
Carrying food is a lot of trouble
I used to think that food was an annoying extra to take out with me, but you know, it really isn’t a good idea to think this way.
And that way lies trouble.
Try to get used to having food around when you are dog training.
Buy yourself a nice treat bag. One that is easy to use, easy to wash and easy to clip and unclip from your belt or pocket.
Have a tupperware box in your fridge with food in it, so that you don’t have to scrape about looking for treats at the last minute. Be prepared, it will make life much easier.
You want your dog to obey you because he loves and respects you
So, you are happy to use food to begin with? But ultimately your aim is to have your dog working for love. Or respect. Is that it?
Many people feel this way, but it is a sign that you are not really on board with how dogs learn and what makes dogs change their behaviour in any kind of permanent way.
Sadly, there is no way to make a dog ‘respect’ you in the way that people may respect one another. He doesn’t have the intellectual capacity for this. What people really mean when they say their dog obeys out of respect, is that the dog fears the consequences if he does not obey. That isn’t respect, it’s intimidation.
As for love, unfortunately no one has ever yet managed to motivate behavioural change in a dog through love alone.
We have proven beyond doubt that the motivators for behavioural change in dogs are the expectation of a great outcome (which increases behaviour) or the expectation of an unpleasant outcome (which diminishes behaviour). We call these reinforcers and punishers. And they are the tools you have to choose between when it comes to training your dog.
Why do people fade rewards? – 2: Good reasons
Before we look at how and when to fade rewards in dog training, let’s look at some of the good reasons for doing so. We can fade rewards
- In order to get the dog to try harder
- In order to make behaviours more reliable
- In order to get longer behaviour chains
Sometimes we want to encourage dogs to try harder, to jump higher, to run faster, to sit more quickly etc.
If we begin to deliver rewards intermittently, or stop delivering them for a brief time, this often triggers a burst of effort from the dog as he tries to get the rewards coming fast and thick again.
This means that we can be more selective, move the goalposts a bit, or ask for something a bit different by rewarding only the best effort, or a particular aspect of the behaviour that we want to develop
Strictly speaking this is witholding rather than fading a reward, but it is all part of manipulating the way we deliver rewards and it is essential to the process of shaping.
When we fade rewards for a well learned skill, we also create an opportunity to offer much better rewards (by comparison) for new or more challenging behaviours.
It has been shown in experiments that animals work harder for intermittent and unpredictable rewards. The evolutionary advantage of this characteristic is that it enables animals to be persistent in the face of environmental challenges.
You can read more about this phenomenon in this article: The gambling effect
It should be said that not all dog trainers believe this aspect of learning theory is particularly relevant to dog training, and some trainers nowadays do not fade rewards much, if at all.
I still do fade rewards, but I am open to the suggestion that continuous reinforcement may be a better option for maintaining reliability in some aspects of training.
Getting longer behaviour chains
When we first start teaching long duration behaviours, getting a dog to heel for long distances for example, or to lie down for long periods of time, we need to go to the dog and rewards him frequently.
To lengthen these long duration behaviours, you need to ‘stretch out’ the gaps between rewards. So that the dog is working longer and harder for less.
Bear in mind though, that this only works up to a point. Take this too far, and your behaviour will break down.
How to fade rewards
There are two main ways to fade rewards. We can reward less often, but we can also reduce the value of the rewards that we do offer.
If you are using high value rewards, this second approach is a good way to begin.
If I teach my dog to ‘sit’ in my kitchen, to begin with I might reward him with a piece of chicken for each and every sit. Once the ‘sit’ is well established, I can begin switching some of the chicken rewards to a little piece of kibble.
Once most of the rewards were kibble, I would then introduce the first approach and begin to omit some of the rewards entirely. So that the dog is being rewarded for some sits, and not for others.
This method of ‘fading’ from high value and frequent rewards to low value infrequent rewards is used in exactly the same way for any established skill at a given level.
When to increase rewards
Knowing when to increase rewards is as important as knowing when to fade them.
When we want to make things more difficult for the dog in training, teaching him a longer sit (stay) for example, or teaching him to sit outdoors, then we reintroduce continuous rewards.
If the task is a difficult one (working with distractions for example) then we also increase reward value.
If we were teaching the dog to stay whilst another dog walked past him, we would ditch the ‘kibble’ and return to the juicy chicken rewards.
Don’t forget, if you are using an event marker in your dog training, you need to omit this whenever you omit the reward.
So, if you are clicker training, don’t ‘click’ if you are not intending to ‘treat’.
The problem with fading rewards
The most common problem with fading rewards is that people try to go too far or too fast.
Remember that ‘fading’ rewards does NOT mean getting rid of them completely.
If you ditch rewards completely, your trained behaviour will gradually die.
Maybe not today, maybe not tomorrow, but it will die. Never stop rewarding completely, any behaviour that you want your dog to carry on doing for the long term.
Remember also that rewards do not need to be food based, though food is very convenient. You need to be objective about whether or not you are reinforcing your dog’s behaviour effectively. If the behaviour is breaking down, you probably are not.
You’ll find lots of examples of how to fade rewards in Pippa’s book Total Recall a complete recall training programme for puppies and adult dogs
This article was originally published in 2013 and has been extensively revised and updated for 2015