Thread: Variable Reward Schedules in Detection Work

Hi Melissa

I found that really all you can do is maintain a good training/work log. Which I know you do. Also I am sure what I am saying here, is not new to you, and are most probably implimenting at work.

As it is impossible to know if your dog is bypassing some odours whilst working, we can only really monitor our dog whilst in training, or the setting up of target odours in the dogs working environment.

I am not sure if this continues in Quarantine, but I found (many years ago) that whilst working, some handlers were rewarding almost every find a dog made. As most finds are usually fruit, the dogs were conditioned/imprinted to focus on apples and oranges, whilst the conditioned response for other odours slowly deminished. We need to set up target odours that we know the dog is bypassing in training. When the dog hits on a fruit, we verbally/physically reward the dog (no food reward), then the dog is presented with a new target odour that the dog is bypassing and rewarded with food.

Every dog is different, and so is every handler. Some may not even realise that their dog on a particular shift was rewarded 20 times for finding apples, even though it did indicate to say 30. But was not reward at all or maybe only twice for say plant material.

We really do need to treat each work run as a training run and still plant target odours that are not as "popular" as say apples and oranges.

For dogs that a continually missing certain target odours, we need to have an observer. Once our dog has say found 3 pieces of fruit, only rewarded once, then the observer presents a new not so popular target odour. The handler, if the dog passes it, then should immediately enforce the sit command and rewarded for taking interest.

If the dog did miss this target odour, and the handler enforced the sit. This target odour should then be presented again after another 2 positive finds with apples/oranges (but no food reward). And continued in this manner until the dog responds appropriately to the training target odour.

Once the dog has responded appropriately to the training target odour it is then again immediately rewarded if it then hits on another piece of fruit.

I personally don't believe there is a set ratio for witholding or giving a reward, as during work the finds are too random in the type of odour and its frequency. Sure its fine in a more controlled training environment.

Maintaining such a high positive find ratio with quaratine dogs is so much more difficult than say with customs dogs. Customs don't get the number of finds quarantine dogs do during a shift, and their target odours are only around 4 to 5. A customs dog may only get 1 or 2 positive hits in a run, but a quarantine dog may receive as many as 20 or more. But also in these 20 finds, a higher percentage is really one or 2 target odours (usually fruit). I feel many handlers may not be vigilant enough at keeping track of the type of target odours the dog hits on during each shift, and the numbers of each target odour indicated too and how often each target was rewarded for. We reward more for apples and less for plant material..we must therefore expect the response for apples to grow stronger and the response for plant material to become weaker.

Difficult dilema

Cheers

If the dogs were rewarded every single time it hit on target odour why would it even think about bypassing an odour for a so called easier target odour. Is it because the dog is confused about when or if it will even be rewarded for finding the target odour? If it detects an odour it is trained to find then our expectation is the dog will indicate at the source and it is the dogs expectation that it will be rewarded for finding it. Variable reward schedules in detection are crazy, it causes too much confusion in the dogs mind. Keep it simple and reward the dog for finding what you asked it to find.