Monday, 9 March 2020

How to use of rewards in dog training

Almost everyone uses rewards in dog training.  Or they think they do.   Unfortunately, some people are deluding themselves.

free pic by

A reward is something of value to the recipient

I could offer you a reward for correctly completing a puzzle,  but if that reward was a square of old cardboard,  you might be disappointed.  If the reward was to gently stroke your ear,  you might be pretty annoyed.   Yet we do this to dogs all the time.   We give them rewards that have little or no value to a dog.
Some traditional trainers tend to train using verbal praise and gentle stroking as a reward.  Maybe you have a dog that craves human contact and is ecstatic about having his ears fondled.  If so,  you are lucky.  The fact is many dogs do not value ear stroking that highly.   They might quite like it indoors whilst you are watching TV in the evening,  but outside,  where there are squirrels to chase?
I don’t think so.
If we accept that we need to reward our dogs when they obey us,  it makes sense to spend a little time thinking about what the dog actually finds rewarding.  Is he crazy about tennis balls?  Does he love his tug toy?  Is he wild about retrieving?  Is he an ace at Frisbee?  Is he obsessed with food?

Failing to use the ‘right’ rewards leads  to punishment

Unfortunately, many people struggle on for years trying to train dogs using a kind word of praise and a gentle pat as a reinforcement for good behavior.   This is not always sufficient on its own to motivate a dog and thus trainers who are successful with these methods also tend to use quite a bit of punishment.  Sometimes a lot more punishment than they realize.
If you like the idea of training with the minimum amount of punishment,  think hard about how you are rewarding your dog.    The reason that people become disillusioned with reward-based training and turn to punitive methods, is normally because they are using the wrong rewards.

Using food

Many people are afraid to use food as a reward because they worry that they will be stuck with it forever.   Actually,  rewards are only used profusely in the early stages of training and should be diminished over time to just the occasional random treat.
With many dogs,  once the basic foundations of obedience are established,  food can, and should be, exchanged for more interesting rewards based on activities that the dog enjoys such as fetching toys or balls.
For many dogs,  food in the early stages of training is the most attractive and most convenient reward you will have at your disposal.   Don’t be afraid to use it.

How to use reward markers in dog training

If your dog hasn’t got a clue what you want him to do,  it may be time for a ‘reward marker’!
Dogs modify their behaviour in response to the consequences that follow it.  This wonderfully simple scientific ‘law’  enables us to change a dog’s behaviour by changing the consequences of their behaviour.  But whilst we know exactly what we are rewarding our dog for doing,  the dog is often in the dark.  Reward markers show the dog what his trainer wants him to do and speed up the learning process beautifully.
The problem with our helpful ‘law of consequences’  is that it comes with a hefty proviso.
The ‘consequence’ must take place within seconds of the behaviour we want to modify, or it will have no effect on it at all. This means that it is often difficult to change behaviour effectively because dogs don’t behave in isolation.  One behaviour follows another and another.
A dog sits, then gets up, then wags his tail,  then barks, then play bows, then sits, then turns in a circle.   All in rapid succession in the space of a few seconds.   How on earth do you reward the ‘sit’  effectively when there is so much else going on?   It is hardly surprising that dogs get confused.
The secret to accurate rewarding is the use of a ‘reward marker’.

What is a reward  marker

The marker is a sound or signal that is so closely associated with a reward,  that to all intents and purposes it takes on the role of the reward in reinforcing the behaviour that we want the dog to engage in.

How do we use a reward marker

We use the signal we choose as our marker to define the precise moment that the dog does what we want.  For example,  when we teach a sit,  to begin with, the dog may sit and then get up again very fast.
To make sure he knows we are rewarding the sit, and not rewarding him for getting up,  we ‘mark’  the moment his bottom touches the ground with some kind of precise signal.   It then does not matter if the dog has got up again when he is rewarded because the marker has reinforced the required behaviour.

Different types of marker

A marker can be anything that gives an instant, rapid, clear and consistent signal.  Suitable markers depend on the animal being trained.  Marine mammals such as dolphins are often trained using ‘whistles’ as reward markers.
This is not ideal for dogs because we use whistles as cues or commands.  Using them as markers too could be confusing.   Clickers have become the most popular reward marker for dogs and with good reason.  Clickers fulfil all the criteria (rapid, clear consistent) for a good marker and they are cheap and portable.
Some people prefer to use a spoken word or sound as a marker.  And this can work.   If you decide to use a word,  make sure you say it in a consistent pitch and tone,  and deliver the word clearly and precisely.  A short snappy word that is not in constant use in your daily conversations with your dog is essential.

Building the association

Before the reward marker can become a useful source of information for the dog,  the trainer needs to make sure that the marker and the reward are inextricably linked in the dog’s mind.  The trainer builds an association in the dog’s mind between the marker and the reward,  by repeatedly delivering them one after another,  first the marker, then the reward.
With clicker training,  we call this process charging the clicker  And it is simply a case of repeatedly making the click and immediately following each click with a tiny treat (you will normally see this referred to as C&T)

Using reward markers

Once the marker has become associated with the reward we can use it to define for the dog,  the exact behaviour that we want him to repeat.   The dolphin trainer blows her whistle at the highest point of the dolphin’s leap into the air,  he then swims over for his fishy reward.  The dog trainer presses his clicker just as the dog is pressing his rear end to the ground,  and then throws the dog his treat,  or throws him a ball.  In both cases the required behaviour,  be it leap or sit, is reinforced,  ie is more likely to occur again in the future.
You can see that the marker also buys us a little time,  it enables the trainer to reward the dog up to several seconds after the event,  provided that the event itself was clearly marked.

Can you train without reward markers

Yes, you can, and many traditional trainers do so.   However,  bear in mind that without an effective marker it can be difficult to time rewards accurately.  And if a reward is not timed accurately it is likely to be ineffective at reinforcing the behaviour that preceded it.
This in its turn is likely to lead to the trainer becoming dissatisfied with rewards (and to assume that the use of rewards is ineffective)  and more reliant on aversives.
Accurate reward markers enable us to train more kindly and with fewer corrections.