[Original post on July 29, 2015]
It is a typical training session, and the handler-dog team are working on complex handling sequences in a full course. The dog is a well-trained agility dog, and has excellent understanding of obstacle performance criteria, and the team is progressing through the levels of competition. However, there are still mistakes to be made as the challenges in sequences increase.There is another team, very novice, who are just starting to put the pieces together and work on sequencing. The dog has a decent understanding of obstacle performance, but still makes mistakes here and there.
The above are very common and normal challenges we face in agility training, and even though the teams appear to be very different, the underlying issue is very much the same: learning the criteria of a behavior under new contexts. What handlers often have challenges with is how to effectively and efficiently address these mistakes while continuing to build the dog’s motivation, understanding and joy in the sport, trust and joy in the team. Over the years, I have learned some generalizations on how to deal with this, both through experience and from the instruction from some experts (Bob Bailey, Susan Garrett, and Dave Munnings, in particular).
With performance mistakes, it’s important to figure out why that mistake occurred (in real time). For example, an advanced agility dog comes down a speed line of jumps, straight to the weaves. The handler calls for weaves with great timing, and the dog dives into the weaves, but is unable to collect and bend in time to catch poles 2-3. In this case, the dog failed to collect enough to get the second turn in the weaves. The line approach to the weaves was a fast line and straight approach (hard to decel on those – but that is the dog’s responsibility provided the handler provided timely instructions). So returning to the line prior to recreate the speed would be appropriate. The dog nails the entry and continues in the weaves, but then pops out at the 3rd last pole. The second performance fail was due to commitment to the end, so sending to the first weave from where the handler is (not go all the way back to the start of that line) would have been better.
Repeatedly adding in extra obstacles or large approaches to get one right isn’t appropriate for dog training and only delays the chance for reward (thus increases frustration for the dog). Especially for novice dogs, recreating long distance approaches (for some that can be just 7’) without repeating previous obstacles can be problematic.
The best thing to do is to figure out the aspect of the mistake and recreate it. I have seen handlers repeat the previous obstacle or two regardless of nature of failure to recreate the full context of the mistake. This is not ideal dog training as it doesn’t aid the dog in understanding the nature of their mistake (source: Dave Munnings). As an instructor and trainer, I have only seen a couple of dogs who can handle that ‘tack-on’ without increasing their frustration, but for most dogs, it results in frustration and diminishes their motivation to play with their handler. [Read: decrease their speed, increase sniffing, increase barking at their handler, going for a zoomie, checking out and looking for something more rewarding like other dogs/rewards or switching into herding behaviors.] FYI, Kelsey is not one of those dogs who can handle it, which is why if reward time delay is growing I will break it up, and fluff up her neck or play with her frustration (it’s a game I have developed with her to help burn off her frustration invigoration). Ideally, I have a reward on me or around on the ground, and I can ask for a simple behavior (hand touch) and reward that.
So, generally speaking:
speed with failure to collect then you should recreate the speed of approach (which may require the additional obstacle, but not necessarily);
challenge of decision based on angle of approach, recreate the angle of approach (do not add the previous obstacle);
challenge of handler location to the obstacle, recreate that handler position (do not add the previous obstacle);
challenge of completing the obstacle to criteria, repeat obstacle only (do not functionally reward by continuing/ignoring the lack of average-or-better performance, and do not add the previous obstacle);
and for the truly novice dogs, the challenge of committing to the obstacle, simply hold the commitment point and wait for the dog’s correct choice (do not add distance, do not add previous obstacle).
For the dog training aspect – general rule for mistakes:
Dog makes a mistake – quickly make the above decision for how to repeat the obstacle to “Recreate the Opportunity to Make the Correct Choice/Performance with the Necessary Conditions that Led to the Failure as Proficiently as Possible.”
Verbally praise *only* for getting it right if it is a truly learned behavior like performance criteria (source: Susan Garrett); if it is a newer behavior like complex chain (angle, speed, distance of handler), reward that correct choice; then
Repeat in sequence and reward (source: Susan Garrett), and then continue (either by starting from that last obstacle, which could be like picking at a scab, or even better, transition from the reward to the following obstacle (flowing right from reward/relationship building moments into sequencing is REALLY important – source: Lynda Orton-Hill).
It is critical that you be aware of how much time has elapsed since your dog has received a reward. Bob Bailey comments that good training has a really high reinforcement rate until the behavior is truly learned. If your dog is making mistakes (along with you), then it means the behavior *in that context* isn’t fully understood. Tacking on more time before rewards are received is demotivating and doesn’t aid in learning. Sure, to compete, your dog has to run a full course without a reward, yes, but even then, the time gap for reinforcement is 27-60 seconds at most. In training, be aware that repeats and then continuing on without reward (of some nature – be it petting, toy, treat) tack on time.
I’m not saying that you should reward subpar (not: average or better) performance to increase reinforcement rate artificially, but do be aware of needlessly adding in stuff that adds time and diminishes reinforcement rate. At a camp in 2012, Greg Derrett talked about a “Handler is the dummy reward.” This is where the handler makes the mistake, so s/he asks for a nose touch immediately and rewards that. This helps to reset the reinforcement rate clock and helps mitigate the growth of frustration for the dog.
Finally, it is also critical to know if you are over-facing your dog (presenting them with a challenge that is too far beyond their understanding of the behavior). If after several repeated fails (speed approach at an odd angle, with handler caught behind), figure out how to deconstruct that behavior and test an element of it (odd angle without speed nor handler behind). If that is easy (reward), then try a different element of that behavior (speed and odd angle with handler beside) (reward). For the really advanced (Masters/International) courses, you will find fails associated with multiple challenges in a single obstacle. Take the hardest part of the challenge for your dog and do that part; for me, I try to keep my position in the original context (positional challenges are the hardest for Kelsey), and just repeat the one obstacle, all the while be aware of the reinforcement rate clock ticking.
I hope this helps clarify a common issue in running sequences. It is all simply dog training with multiple layers of criteria and understanding. Make it efficient, clear, fun and be aware of that clock!