Why Reward-Based Training Works

Dogs rely primarily on non-verbal communciation from us, such as reading our body language. To effectively train your dog it is much easier, more effective and more humane to teach it what to do by rewarding it for appropriate behaviour than it is to teach it what not to do

Dogs rely primarily on non-verbal communciation from us, such as reading our body language. To effectively train your dog it is much easier, more effective and more humane to teach it what to do by rewarding it for appropriate behaviour than it is to teach it what not to do by punishing it.

So, how do dogs learn?

Dogs primarily learn by making associations between events that consistently occur in association with each other. For example, Pavlov's famous experiment was designed to measure dogs' salivation response to food. He also found, quite by accident, that they also began to salivate in response to a light that was turned on immediately before the food appeared. The light predicted that the food was coming so it became associated with the food and evoked salivation by itself.

This is known as 'classical conditioning', where the animal learned an association between two stumuli. You probably see examples of this type of learning everyday, such as your dog getting just as excited about you picking up the lead or putting on your shoes as the walk itself or your cat running into the kitchen at the sound of a can opener.

Dogs also learn by 'operant conditioning' whereby behaviours that result in positive outcomes will be more likely to occur in the future. Training using 'positive reinforcement' is an example of this.

What is positive reinforcement?

Presentation of a reward, such as food, to your dog immediately following a behaviour, such as sitting, will make that behaviour more likely to occur in the future.

For positive reinforcement to be effective the reward must immediately follow the response. Delays of a even a few seconds can dramatically decrease the reinforcing effect of the food - in this time another behaviour may have occured and the dog doesn't learn to associate the reward with the correct response.

A dog will also perform a behaviour to obtain secondary rewards consistently presented with the primary reward or reinforcer. For example a pat on the head or the words 'good dog' can be paired with the food presentation. Once the association between the two stimuli or reinforcers have been made, then the dog will perform the command for just the pat or words on their own from time to time but if it done too frequently, the association will be lost.

Another misconception is that reassuring a dog that is anxious or fearful e.g. when it is shaking or hiding during a thunderstorm is rewarding it. A dog that is in a fearful state is acting on reflex to protect itself from a real or perceived threat and is not using the thinking part of its brain. Reassuring the dog in a soothing voice can help to calm the dog down, but this is not going to encourage it to be afraid in similar situations in the future.

A fearful dog is not in a learning state, which is why dogs with anxiety disorders often need anti-anxiety medication. The medication helps calm the dog enough so that behavioural modification techniques can be implemented, such as teaching the dog a new behaviour (such as sit and look at the owner) in the presence of the fear-causing stimulus to replace the destructive or undesired behaviour.

Why shouldn't I punish my dog?

Punishment is not an effective means of learning. Consider the dog that has made a mess in the house while you've been away all day.

For punishment to be effective it must be:

  • applied within one or two seconds of the inappropriate behaviour;
  • applied every single time the behaviour is performed; and potent enough that the dog will seek to avoid it in the future but not be so aversive as to frighten the dog.

In reality this is not possible so punishment will only end up causing more harm than good in most cases. Dogs may react to punishment by becoming fearful and anxious in their owner's presence, unwilling to learn, or even aggressive. When a person punishes a dog it can seriously affect the human-animal bond.

The AVA position on reward-based training

The Australian Veterinary Association (AVA) believes use of positive reinforcement (reward-based training) is the most humane and effective training method as it avoids undesirable behavioural side effects, such as increased anxiety, escape and avoidance, defensive aggression, fear conditioning and learned helplessness. Positive reinforcement also makes training more enjoyable and helps to improve the bond between the trainer and the pet.

For example:

  1. Sassy jumps up to greet people: her owners have tried pushing her down and kneeing her to knock her off balance when she jumps. This has not worked, in fact she now jumps from further away to avoid the knee. Sassy should be ignored if she jumps and only receive attention (including eye contact) when she has four paws on the ground. Only when she is standing or sitting should she be rewarded with attention and treats.

  2. Fred likes to sit on the kids' artwork when they have it sprawled on the floor for colouring in, and often chews on the textas. They have tried pushing him away, saying 'no' and chasing him when he chews on their textas. Instead, Fred can be lured onto a special mat or cushion with food and rewarded when he gets on the mat. When he takes a texta they can ask him to come to them, sit and then swap the texta for a tasty treat or chew toy. Ultimately they can treat him to 'go to bed' and to 'give'. Fred can be trained to go to his mat and stay there using lure and reward methods that will gradually build the duration he can remain settled there.

  3. Pumpkin growls and bites her owner's hand when she has her harness put on. The owner has tried pulling her hand away, saying 'no' and smacking Pumpkin. The problem is getting worse and Pumpkin is biting sooner and harder. The owner should consider trying a collar instead of a harness, to remove the source of the problem and start a program of counterconditioning and desensitisation to the harness. This involves giving Pumpkin treats whenever she sees the harness (a distance away from her that does not cause her to growl). Gradually the harness can be brought closer, with treats given for calm, non-fearful behaviour from Pumpkin. With correct timing and much repetition, Pumpkin will associate the harness with treats and be happy to see the harness. In tiny increments, the harness is brought closer and closer and eventually placed on Pumpkin. Pumpkin can also be taught with treats to enjoy being handled.

(Information from 'Reward-based training: a guide for dog trainers' published by the Australian Veterinary Association. www.ava.com.au)

comments powered by Disqus

Recent Posts