We all know what a reward is, don’t we? Well, we know what we personally find rewarding to ourselves and we may think we know what is rewarding to our dog.... But do we??
What is rewarding can vary from dog to dog. For example, a reward for a beagle or bloodhound could be following a scent. For a Border Collie it might be herding a group of people, ducks, sheep or even other dogs. For a Labrador it might be retrieving or holding something in it’s mouth. Our job as trainers is to pick the reward that's easiest to use and is the most effective at getting our dogs to respond to us when they give us the behaviour we seek. In our day-to-day training, that's usually a tasty piece of food. However, be aware that the value of the food can vary from dog to dog and situation to situation, so test your dog out, see what he finds most rewarding in a given situation. When you are training for field work with your dog, there are great opportunities to use the dogs natural desires as rewards to reinforce the behaviours you want.
These are called “environmental rewards” and if used correctly can provide a massive boost to the training. For example, you might have trained your gundog to go out, pick up and return to you with the dummy. But the dog is unsteady in the wait before being sent out.. By getting a wait first, the send out becomes a big reward for waiting. With a pet dog it might be the opportunity to go for a good run so asking for a wait in the boot space or a sit on lead before being let free makes the release and run a great reward for sitting first. For our everyday reward training at home, when using tasty titbits, we have to use them correctly or we will end up with a dog who will only work for us when we have a reward in our hand.
There are three phases to using rewards properly. But the good news is that there are really only 2 phases you need to learn, the 3rd is not necessary except in specialised training. I call them Luring, Bribery, and Random phases. I have defined each in the topic header.
Phase 1 – LURING - when the treats are right in front of the dog’s nose/face and I am getting the dog to move into the desired position or motion I want. I use this to start most basic behaviour like sit, down, LLW, stand, weave, move to heal position, roll over, etc. In this first phase, we hold the treat in our hand in front of the dog to get him into the position we want. The dog has no idea what we want him to do so we need to use our hand motion with the treat to lure the dog into the sit (or whatever). Dogs learn hand signals (cues) much faster and better than verbal signals. Therefore, it’s important to build these hand signals into the luring at this beginning stage. After luring about 8 or 10 times put the treat in the other hand and try the hand signal without a treat. If the dog gives the desired behaviour then produce the reward from the other hand immediately. When your dog can give the behaviour repeatedly for the hand signal (with no treat in that hand) then you are ready to move into phase 2. If the dog does a few then stops, just go back to luring another 5 times and try again with the empty hand signal. Eventually, the dog will get the concept and give the behaviour for the hand signal without the treat in that hand.
Phase 2 – Bribery – when the treats are nearby but not in your hand or in front of the dog. I have the treats in a treat pouch or on a counter top that I can reach without moving. In this stage, we use the same hand signals as when Luring but we DO NOT have the reward in the signal hand. Keep the rewards on the counter, in a treat pouch or behind your back. Give the hand signal and wait, give the dog 15 or 20 seconds to respond, reward when the behaviour is given. The pause (15 to 20seconds) does not always occur but it often does. I call the it “thinking time” but it really is that the dog is not sure what it should do¬!! Since it is unsure you should say nothing, let the dog work it out, let it give a behaviour, it will most likely be the right one and the reward will help confirm in his mind exactly what is needed to get the reward. Being overly helpful with more signals is usually detrimental to learning for the dog. Continue to train in the bribery phase and as the dog learns more precisely what is expected of him, the response will become more exact and quicker. A sure sign of learning. This is as far as I ever go with my training. Moving to Random reinforcement is simply not necessary except for a very few specific areas. Bob Bailey says as much in one of his post many years ago and if it is good enough for Bob Bailey it certainly is good enough for me. However, read on so you know more about Random reinforcement schedules.
Phase 3 – Random Reinforcement – the does not get rewarded for every performance of the requested behaviour. The rewards are delivered at random times or at random repetitions of the behaviour. Up to now, we have been rewarding ever performance of the cued behaviour. This is Continuous Reinforcement. In Random Reinforcement, we reward now and then, maybe on the 2nd or 5th or 7th or 3rd performance. In other words then dog never knows which performance will earn the reward. Since humans are not very random beings it is best to use a random number generator app or similar. Otherwise, you will likely build a pattern into your training that you are not aware of. Both Continuous and Random schedules of reinforcement are very strong and powerful. If all rewards are stopped permanently then the trained behaviour will die out! But it will die a bit quicker in a dog who has been under a Continuous schedule than one who has been under a Random Schedule. This is because under the Random schedule, the dog never knows which repetition will earn the reward so he keeps trying longer.
Note: Random schedule of reinforcement is what gets people addicted to gambling. Many trainers say we must introduce random reinforcement as early as possible. This is simply not true. For almost all training, a schedule or Continuous reinforcement is all that is needed. This means we reward every performance of the requested behaviour as opposed to randomly rewarding a performance. In formal training sessions, we should certainly use a Continuous schedule or reinforcement because in all other situations in the home or on walks we, without realising, will be randomly reinforcing our dogs. That’s life and the way we are. When we are about the house we may ask the dog to sit or go in his bed, we may reward that with a treat, or a word or a stroke, or we may ignore it completely depending on how busy we are at that moment. Randomness slips into our daily life and routine even if we don’t think about it. So don’t worry about Random reinforcement, unless you are doing search and rescue, tracking or similar long term activities. Only then is it worth considering.
Alton Matherne
Accredited Animal Behaviourist
PRAMA Behavioural Trainer
Jan 2022
Around the Circle
I am part of a small blog circle with other dog trainer who all join me in writing about a topic. Next on the blog circle this week is Dan Salliss, Head trainer of Croydon Dog Club. Dan talks all about rewards and how we use these in dog training.
Comentarios