So…what does it look like? Rewarding errors?
I taped Brito’s scent articles today. We have been able to progress from about four metal articles to seven in less than five sessions – simply by allowing him to make errors with support. Support is a euphemism for a free cookie:).
For Brito, the trick appears to be not rewarding an error more than twice in a row because while it is true that he’s happy, it’s also true that he is not getting trained! If he fails twice in a row then I need to do something to help him succeed – with articles I just pull most of them out of the pile and work up again. It’s working for him.
I “mark” correct choices with a combination of personality and throwing the food. I am much quieter when he makes an error but the actual cookie is the same.
I try to start so that he’ll be correct – whatever that might mean for your dog. Most important – if errors begin to crop up then change something.
Here’s a video of a session with metal articles – errors and all:
(Note that on the last attempt, I accidentally clicked before he had actually picked up the article so he didn’t complete the retrieve – that’s not a problem.)
I started rewarding errors with leather articles too, but I quickly realized that Brito does not understand leather articles. I went back to two articles, placing them extremely close together (so that he can compare), removing incorrect choices (spit on) and….I do not reward errors.
Separately we are working with two metal articles far apart (2 feet) so that he can learn to find the correct option without comparing two choices side by side. I do not reward errors there either, but I’ll simplify the game if needed.
So much interesting stuff!