I recently found Cyber Rally-O and their Dance Division. This is an organization that allows you to obtain titles in Rally and Freestyle via video submission. I was really excited by these, and thought that the preliminary dance patters (which are just heeling) would be the perfect kick in the ass to start to build duration into Perrin’s heel. I am bad at following through with training plans if I don’t have a goal to meet. I love teaching new things, so finishing old behaviours sometimes falls to the wayside.
The challange: the heeling patterns are much longer than what Perrin and I are capable of right now. Perfect, we need to build that anyway!
There are three heeling patterns that need to be completed to get the Preliminary title: 90 Degree Foxtrot, Figure Eight Waltz and Scenic Route Tango (all can be found here). I don’t yet have a space big enough to film these in (or, more honestly, an area big enough to film these in where Perrin is currently comfortable working), so until I get that peice into place I figured I would put some thought into how I am going to approach training these (I keep saying I need to do more of this, so here I am starting). I decided to start with the 90 Degree Foxtrot, as it seemed like it will be the easiest for Perrin with his current skills.
I have been thinking about how to approach this for a while. I still keep Perrin’s heeling sequences quite short and with an abundance of pivots or small circles. Cyber Rally O does allow you to use food rewards at very specific times, but honestly, I find the rules confusing and am not confident on when I can or cannot give a food reward without being NQ’d. It is clear that you can stop between exercises to give a food reward, but stopping in the middle of a pattern that is supposed to flow would both confuse me for what is coming up next and drive my A-type side insane. I figured that it isn’t a bad idea to try to build a whole pattern without. I can always add reward points in later if I need to.
That led me to my next dilemma. If I am not going to use treats in the middle of the sequence to break the pattern up, then I need to physically break down the sequence into bits. I have been thinking of the best way to approach this.
- Break several sections out, practice each seperately, then chain them together?
- Work sequentially through the pattern, adding one more station as the last becomes proficient?
- Same as number 2, but going backwards through the sequence?
I think that all three are valid and would work. Number 1 would be perfect if there were certain sequences of behaviours that I would like to group together frequently, however most of the 90 degree foxtrot is essentially the same sequence: heel forward, left (or right) pivot, so I didn’t see much value in this approach for this particular heeling pattern beyond practicing lots of left and right pivots (which we already do). Of the remaining two choices, I am going to try number 3. This is just good old back chaining.
The reasoning behind the advantages of back chaining are well understood in the dog world in general. The reader’s digest version is that by practicing the last (already fluent) behaviour in the chain first, each additional (fluent) behaviour becomes a cue for the next, all of which the dog knows are going to lead to the reward with the last behaviours having the greatest reinforcement history (for an infinitely better explanation try this).
So in the game plan will be to start with station 16, and have an awesome toy/treat/play event to reward. Once that is the way I want it, then I will add 15 and do the same toy/treat/play party at the end. Then add 14 when we are ready and so on, always having a big party with lots of treats at the end. This will be done when I am at the training building where Perrin is comfortable working with excellent attention. After we have built at least a few stations in I will take it on the road and try it other places, while moving back to station 16 and building it up again.
Problems I may run into:
- Individual behaviours breaking. Perrin knows how to do all of the behaviours in this pattern, but as I lower the rate of reinforcement by back chaining, their quality may begin to degrade.
- Plan: Continue practicing the individual components at other times to keep them strong. If a behaviour does break, stop with the pattern for a while, go back to practicing the individual behaviour until it is solid again, then return to the pattern but start at an easier point (e.g. if the behaviour broke at 10, go back to 14 and start working backwards again).
- Perrin starts to lose interest and disengages. Rate of reinforcement is too low. Too many behaviours were added too quickly.
- Plan: Step backwards to the beginning and add behaviours more slowly. Consider upping the value of the end reinforcer if appropriate.
Once I test how this approach goes, and see what kind of skills Perrin gains, and lessons I learn and then I will come back, adjust this plan/hypothesis with my learnings and then plan for the other two patterns!