Sunday, July 03, 2011

Hope's sheepdog-type plan rough draft thingiemawhotchie

So... best intentions aside I haven't blogged for weeks. 

However lots has been going on.  With Hopie we'd trying to figure out how to train herding using R+ based training wherever possible and to teach her to become more comfortable with human applied pressure for pressure and release.  I felt like her instincts were keen and she loves the actual work but the only real sticking point was in her not faring well with even very, very, very minor aversively taught approaches.  I had some basic ideas but fleshed it out with a discussion from some lovely folks and (hopefully!) between it we've got a good plan going.

So here is the attempt to write out the plan (something I've been remiss on...) for her and some of what we've been working on.  Worst case, it doesn't improve things but we're no worse off and best case it may work super for us. (Certainly hope so!)  Either way for a dog for whom the tools 'inside the box' aren't bringing out her best game, 'outside the box' however oddly nontraditional it appears has little to loose.

So the areas of trouble for her mainly include:

Behaviours not fluent before being asked for at working level
Dry training had been in the plans for awhile and after talking to some people who have taught working/trial dogs with R+ I came away even more certain. I decided to start dry until many behaviours are fully/very very highly fluent so I can focus just on the dog, without also having to contend with a number of other independent minds who are not stationary and more difficult to reset for multiple finely sliced reps of the same behaviour. The dry-work has the benefit of being able to practice single behaviours in well split easily achievable criteria in a controlled setting which allows the behaviour to become solid, low latency and have significant duration, distance and distraction proofing added to it rather than try to attain those on the fly with the sheep and concurrent with everything else.  Likewise you don't have to rely on other half-trained behaviours at the same time. ;)

Some of this is not new or exclusive as certainly there are a fair few well known working dog trainers who teach a down/stop/come/flanking dry. I'm also wanting to teach to higher standards:
* come in (pretty self explanatory)
* back (move directly outward from the sheep 'target'- not sure if I want this to simply be stepping back or actually turning around or if it even matters... pondering atm!)
* directional modifier cues (not flanking commands - turning out the whole head/shoulder orientation to widen the future trajectory)
* walk  - meaning along whatever path your head/shoulders are currently facing until told to do something else. I'm thinking that this can get paired with directionals - walk on, walk up, walk by, walk away) for when I specifically want her to do the movement at a slower than normal pace.
* outruns
* angles to come around sheep on - has previously been something of an issue
* and while not intended to be taught on cue per se but shaping angles for casting out on stationary objects at increasing widths and distances, proper shaped flanking

And maybe other stuff as I think of it or when I trip across the need for them in levels beyond started.  She does know many of the cues in the obedience/general context very well but the plan is to practice even those in a herding context and at distances to help them become more as high quality/lickitysplit/'no matter what' as they are in other areas.

One of the things we were seeing consistently was that she while she was okay giving the sheep some distance she wanted to work relatively close to me and became uncertain if the sheep's distance meant she was asked to work further distances relative to me.  Doubly if she was at all uncertain.  As well as making it difficult to teach/refine behaviours, it also meant she tended to quarter in and bring the sheep up too close to me for comfortable movement to stay within that relative distance and may also affect lining up for obstacles. 

While I've varied my rewards quite a bit - food, toys, games, planted environmental rewards, opportunities to access things - it seems I've unintentionally trained her that there is a 20-ish meter zone that she is to work in and expect reward delivery within relative to me.  (Since so far nothing we've trained or done has really required a functional/reliable working distance beyond that beyond a recall.)  Because of that and how sensitive she is, working outside that distance especially where the behaviours are not trained to fluency is worrisome.  While for most dogs this may not be significant, for her it's a problem.  While not directly applicable to herding general distance work may extended her safe/familiar/rewarding/comfort zone so that if there IS a stressor added it's not on top of uncertainty due to distance.   Likewise generic (non herding specific) distance (remote?) reward delivery to ensure she is aware that rewards can be delivered at that distance regardless of relation to me.

Hope is incredibly sensitive to social approval and her perceiving it to be withdrawn makes her anxious.  She is surprisingly confident with external pressure (ie that applied by the sheep and cattle in general or individuals in specific) which speaks to it not being pressure in general she is over sensitive to.  Because of this, I do not believe (natural) pressure itself is the only issue, the timing of the pressure or the degree of it but rather the combination of the behaviours not being fully trained ones and the pressure (coming from human/stock stick) being taught as an aversive to 'correct' her into guessing right to get the release.

The direct seeking-in I think is more of an appeasement behaviour over not knowing what is wanted, anxiety re: distance and anxiety re: aversive pressure and an attempt to use a fall back with a high history of being 'right' everywhere else in life.  Actions like shooing her out, withdraw eye contact or continuing to request a yield (continued pressure) becomes perceived as failing to be able to appease. The more appeasement is withheld the harder the dog tries to appease and the higher the stress raises, until shut down will be the inevitably hit when NOTHING is going to be able to make things right.

Yielding - Central to the pressure issues imo is that at the moment the yield with a stock stick is taught with P&R being taught as a mild aversive.  For most dogs combined with the heavy draw of the sheep it is mild enough that especially with precise timing in the application, the yield gaining removal and the amount of pressure being applied on an 'as mild as possible' basis for the individual and ideally passively (ie the pressure is held at the lowest possible useful level, so the dog can't gain access but is simply waited out vs successively applied increases of pressure) the dog works through any stress.
    For some dogs like Hope, even mildest aversives are enough to be significantly off putting but there is no reason the action (yielding) or reacting to pressure must be taught with an aversive.  That was the 'reframing' context I was dithering on!!!  The same action (move off/yield to either SS pressure or body signal pressure) can be taught in a slightly different manner utilizing R+ to shape the yield and then delivering the reward at or just beyond the peripheral zone.  (ie delivery in place or tossing behind/on the outward trajectory the dog may be asked to move next).  The result should be a dog who is comfortable with the pressure being applied as positive information - which was pointed out to me (I'd not realized, not being that far along) may be very useful in more advanced classes where you are sorting/shedding or instances where you want to have the dog come into the pressure you're putting on the stock.  I likes!

So that's where we are now, working on the dry stuff and then going to start scouting around for a few new sheep who are veerrryyy docile and like to stay together quietly and some quackers now that we can. :)

No comments: