There are some basic principles of a behaviour modification protocol which lead to it being successful. The principles of behaviour in the Training page should be read in conjunction with these.
The following are methods of changing undesirable behaviours.
Behaviours are presented because they were previously reinforced. Withholding the reinforcer leads to extinction because the behaviour is no longer rewarding. Behaviours which have been reinforced intermittently, even rarely, will be more difficult to extinguish. Training a new behaviour can be more effective than extinction alone in these cases.
2. Differential reinforcement
While waiting out a behaviour can lead to extinction, sometimes it is self-rewarding (e.g. excited running around instead of sitting) and won’t naturally become extinct, or has been intermittently reinforced. In these cases, the antecedent stimulus (cue) can be used to trigger a new behaviour, for example sitting instead of jumping, because it is impossible to jump while sitting.
3. Give the behaviour a cue (stimulus control)
It might not always be possible or desirable to completely extinguish a behaviour which is very rewarding to the dog. One common example with dogs which are undergoing training is a desire to jump up on the handler, either from the floor or the table. Close contact with the handler is very rewarding to the dog and while it’s not compatible with training, it might be possible to completely extinguish the behaviour in the time given for training. Instead, the behaviour can be put on cue (e.g. ‘up’). Once the behaviour is under stimulus control (i.e. only happens when the ‘up’ cue is given), it should only happen when the cue is given, leaving you free of jumping up behaviour during training. You may reward the dog with the opportunity to jump up once training is over.
4. Reinforce other desirable behaviours
When a behaviour is difficult to extinguish or replace, rewarding the dog for not doing the behaviour can be effective. For example, a dog that is excited and jumping up can be rewarded each time it doesn't display the jumping behaviour. Dogs will work to achieve reinforcement in the most efficient way, so if there is an easier way to be rewarded, they will soon stop the undesirable behaviour.
5. Change the motivation for the behaviour
Removing the stimulus for the behaviour (the antecedent stimulus) can prevent the behaviour from occurring. This can be one of the most straightforward behaviour change strategies because the stimulus is under your control. For example, if the sight of other dogs exercising causes dogs in home pens to become excited, moving the exercising dogs or blocking them from view can prevent excitement and barking.
6. Consult a knowledgeable colleague.
While too many trainers can complicate a training protocol, asking a knowledgeable colleague to troubleshoot a problem you’re having, or even to observe what you are doing can give you a fresh perspective and allow you to progress. Time for training is often limited so getting stuck is not in the interests of the handler or dog. If video recording your training sessions is an option, watching yourself back can let you quickly pick up on unconscious signals and inconsistent reinforcements that you have been giving the dog.
There are some common reasons why training using positive reinforcement might fail to work. These aren't issues with positive reinforcement as such, rather with how the technique is applied, and can be easily remedied.
1. The dog is physically unable to exhibit the behaviour
If you are having trouble getting the dog to exhibit a behaviour, for example ‘sit’ or ‘lie down’, the dog might have a physical problem which prevents it from assuming the position you require. Physical problems, stiffness or other health issues might make the position difficult. If you think this is a problem, consult a vet to have the dog’s health evaluated.
2. The dog or trainer is not experienced with training
A trainer who lacks experience may be caught out by the dog suddenly progressing in the shaping plan, meaning that an opportunity to reinforce the desired behaviour is missed. Alternately, timing or consistency might be lacking, meaning that the dog is not rewarded for the correct behaviour. However a dog which is inexperienced in responding to positive reinforcement training, which many laboratory-housed dogs may be, will have difficulty progressing rapidly. A dog which is accustomed to ‘freezing’ or otherwise inhibiting its behaviour may have difficulty exhibiting new behaviours and so training might progress more slowly. As dogs become more proficient at learning, they learn new skills more rapidly. “Learning how to learn” is a necessary part of the process for dogs which haven’t been trained with operant conditioning in the past.
3. The the steps between shaping stages are too big
Without having a schedule of training steps, or observing how rapidly the dog responds to training, rushing can lead to the dog failing to respond. Ensure that the dog is responding at around 80% reliability before moving to the next step.
4. The trainer is accidentally reinforcing the wrong behaviour
Timing is everything. Reinforcing too quickly or too slowly can lead to capturing of the wrong behaviour, for example when training 'sit', the dog is reinforced with the hindquarters only part of the way down. Having pre-agreed training steps and a clear description of the target behaviour can prevent this problem.
5. The reinforcer isn't effective
If a reinforcer is chosen which isn't rewarding to the dog, it will fail to reinforce the behaviour and therefore behaviour change won't occur.
6. Aversive training techniques have previously been used
Use of aversive techniques such as positive punishment and negative reinforcement can lead to response inhibition, where the dog fails to exhibit new behaviours, or learned helplessness in which the dog has shut down. In these situations, classical conditioning is necessary before training can begin in order to change the negative emotional state associated with training.
Training and altering behaviour through shaping can result in some very rapid, positive changes. Sometimes, it doesn’t go as planned. It takes a long time and a lot of experience to become a proficient trainer, and even then, dogs can behave in unexpected ways. While experience of a variety of training situations and dogs will help you plan for unexpected events, there are a number of troubleshooting tips which apply to most situations.
1. Dog is not interested in taking food rewards.
2. Dog does not seem to be interested in taking part in training.
3. Dog is over-excited or does not want to cooperate.
4. Dog which had been progressing well suddenly goes backwards in training stage.
5. Dog progressing through stages of learning new behaviour but continues to show signs of stress or negative welfare during training.
Examine the reasons for the problems. Is the dog consistently nervous, even when in the home pen? Such dogs will require additional input, such as positive interactions and hand-feeding sessions. Dogs which fail to respond to training will not adapt well to study life and it is much better to assign the dog to a different, shorter or terminal study if this is an option. Is the dog happy when in the home pen with other dogs? Beginning socialisation in the home pen or exercise with a more confident pen mate might help. Is it specific to one handler? Sometimes particular dogs and handlers might not be a good match, so switching with another handler is a good idea in their situation if it is possible.
The videos below demonstrate some specific troubleshooting techniques for common issues encountered in training.
Links to publications and book on the training of dogs and behaviour modification can be found in the Publications section.