Remember that long debate on microshaping?!!! Here's a formal definition from Kay Laurence....



WHAT IS MICROSHAPING, ANYWAY?

Dear Clicker Friends,

This month I'm welcoming a guest author to the Letter from Karen,
British clicker training innovator Kay Laurence. Please enjoy this
wonderful discussion of what Kay has dubbed "microshaping."˙˙Karen
Pryor

___________________________________________________________________________

Focus on the student and the process

Someone recently asked me how the term "microshaping" developed, and
what, exactly, it means. I think I probably originated that term
some years ago. At that time, it rather alarmed me that much of the
shaping I was witnessing was focused on achieving the end behavior,
rather than on the learner or on the teaching process˙˙in my
opinion, the more important elements.

Folks became fixated on getting the dog to put a paw on its
face/walk backward/retrieve an object/stop on a mat, etc. As humans,
we tend to fo! cus on our personal needs; trainers were not seeing
what was really happening, aside from getting the behavior they
wanted.

Dogs could keep up with this agenda only if they were very, very
skilled at puzzle solving, were in the top 10% of bright souls who
could cope with lots of unreinforced trials, and/or had oodles of
stamina and creativity with which to "find" the solutions.

My Gordon setters could check off some of those boxes˙˙they are
very creative, but poor at remembering their successes, and easily
distracted with jackpots. My collies lack creativity and would
prefer to work with clear guidance (targeting). But, both breeds
have demonstrated a very high sensitivity to the lack of
success˙˙which to them is "no click." And, I'm a good shaper!

Bad shaping leads to confused dogs, no matter how talented the dogs.
Even with resilient dogs and good shaping, dogs take failure
seriously.

It rather alarm! ed me that much of the shaping I was witnessing was
focused on achievi ng the end behavior.

Why we need microshaping

Microshaping evolved to serve two purposes:

1. The first purpose was to make the operator (clicker trainer)
observe the quantity of behaviors that were offered without
reinforcement. These behaviors can be as minor as changing balance,
a shift of weight, deep sighs, a flick of the tail, blowing of the
lips (a Gordon favorite), or lowering the head an inch. All of these
are offered behaviors with no click, no information.

Many videos and training sessions I witnessed that were considered
"good shaping" at that time demonstrated an average 70% failure rate
of offered behaviors to successful behaviors. In other words, only
30% of the offered behaviors were clicked.

Aacck! I would not want to be on the receiving end of that program!

2. Secondly, I wanted to invest time in a building, or acquisition,
process that resulted in very solid behaviors that needed almost no
practice sc! hedules, contained skills pertinent to the behaviors,
and most importantly, only contained the knowledge of the perfect
execution of that behavior.

I wanted to invest time in a building, or acquisition, process that
resulted in very solid behaviors.

An example of this process would be teaching a dog to walk backward,
away from the handler, in a straight line. This is a traditional
freestyle move, but is also useful for teaching mobility, self
awareness, and coordination. (Freestyle is an excellent ground for
developing great clicker trainers, and great clicker dogs.)

In workshops and seminars, I was seeing dogs struggling to achieve a
top quality, flowing, balanced movement as they moved backward.
Often the behavior was taught with pressure, causing dogs to hunch
and dodge backward in avoidance. It also exaggerated sideways
movement, since the dogs were seeking to move out of the way, which
they normally do sideways. I saw many do! gs that had been shaped to
move backward by stepping back with the fro nt feet˙˙which is the
easier movement for the trainer to see. Again, there were some very
weird actions, lots of over lifting of the front paws, kinked back
heads, hunched neck and shoulders, and more. All of these odd
actions were a result of the trainer's overzealous desire to achieve
backward movement.
Movement and patterns that work

I spent some time looking at dogs that achieved a pleasant looking
movement, filled with comfort, strength, and obvious enjoyment. The
movements of these dogs mostly originated from self taught
behaviors. Throw the ball, throw the ball, throw the ball, throw the
ball, throw the ball, throw the ball...demands that resulted in...A
Thrown Ball! Yay!!

So, with or without the owner's understanding, the dog's movements
eventually prompted the owner to throw. Those same movements,
sporadically reinforced by those throws, settled into repeated
patterns. However, since the dog was generating the moves
spontaneously,! the moves had a smoothness and energy that
human-shaped behavior often lacks.

Those dogs were some really great movers!

So I began to shape the next generation of dogs with the focus on
relaxed carriage, flowing movements, confidence, and attaining
distance.

So I began to shape the next generation of dogs with the focus on
relaxed carriage, flowing movements, confidence, and attaining
distance. Right at the outset I worked on muscle patterns, rather
than on the behavior. I developed a physique that would relieve all
stress when executing the movement, and I taught in "micro" slices
to develop a fixed pattern action that, with repetition, would
produce the desired final behavior.

I have one collie that can't stop backing once cued, but otherwise
has some great actions that require absolutely minimal regular
training. Microshaping built a dog that opens the movement with a
back leg action, not a fron! t leg action, and has no history of
anything else but the cor rect fixed pattern action. So, given the
cue "backing" there is only one outcome.

The end result

Microshaping is not a skill for the beginner clicker trainer, and
does require us to deepen our understanding of building behaviors.
We need to keep slicing extremely fine layers that serve the two
purposes:

1. Building only the desired actions/behaviors/skills

2. Working with a 95-100% rate of reinforcement on offered
behaviors. The goal is to build tremendously confident and skilled
dogs, never damaging even the most sensitive dog's desire to learn.

More resources˙˙start practicing!


If this answer falls short, more detailed explanation can be found
in my book, Clicker Intermediate Trainer: Level 3. There is also
ClickFlick video version available. If you have been to J! esús
Rosales-Ruiz' lectures, he uses this process regularly to
demonstrate shaping, with a collie paw tapping a green die. To
inspire the shaping-obsessed folk (like me!), I have a whole book on
shaping challenges, building FPA, and other cool stuff. (Now taking
orders in the US for Learning Games, available from KPCT in
February.)

Keep in mind that I have done the difficult part, breaking down the
behaviors into the steps and skills needed for the desired
outcomes˙˙all you have to do is work through the recipes!

Happy Clicking!

Karen Pryor

Sunshine Books, Inc.
49 River St., Suite 3
Waltham, MA 02453
1-800-47-CLICK(2-5425)

© 2006, Karen Pryor Clickertraining (KPCT)TM

--

Hi Catherine,
I think Kay's definition is what most of us here would consider to be just good shaping? That's why I was a bit disturbed when she introduced this term. However, I've spent quite a lot of time recently with one of her instructors (I'm hoping to spend time Kay herself later this year) and I think the term was introduced to help people rather than dogs. As kay says in her article, you can't expect novice CTers to analyse muscle movements to start off with so at this level its just called shaping. Once you get experienced, you move your shaping to another level - microshaping. Done properly this means that you need to acquire a high level of understanding of anatomy and locomotion. This is expected if you are involved with Kay!!

Alex Kurlands definition of mucroshaping is very different as we discussed on another thread. She alternates between the behaviour she's moving towards and and a previously well learned behaviour. This is what she calls microshaping (my view is she needs a new term) and she uses it to keep the rate of reinforcement up. Now a question - do you think its ever going to be possible to look at a horse the way you can look at a dog?? - what I mean by that is that you can pretty much look at the whole dog when you shape and see those tiny muscle movements. However, with a horse, there's a whole lot more to look at. My impression of Alex it that she does have some understanding of movement and anatomy and has spent time breaking things down so that she can teach dressage type movements in a clicker compatible way. However, I would have to say that the canine 'microshapers' I have spent time with this year blow Alex out of the water in terms of their ability to shape. However, maybe this is just because of the size difference???? What do you think??
All these things aside though, I think it would be a whole lot simpler if people just stuck to the term 'shaping'.
Cath

Hi Cath
Sorry, I keep meaning to come back to this one and kept forgetting!

I totally agree with you that microshaping and shaping are - or should be - one and the same. And AK redefining the term confuses things yet further. By definition, when you shape something you are starting off with the basic behaviour and then you incrementally refine it - that could be through numbers of repetitions and use of VSR or it could be through subtle muscle stuff or whatever.

I don't know about dogs (although would love to hear more about the microshapers you've been spending time with, especially if you've found any youtube footage of them!) but as far as the horses go I always feel uneasy about why we need to micromanage their movements so much. For example, on one of AK's videos and also on the clickryder forum there has been lots of talk of "The Pose" which is essentially rewarding the horse for standing and tucking his head into a false outline. It might look pretty (if you like that sort of thing!) and maybe you could use "microshaping" to further develop the false outline into a true outline but the reality (well, my opinion!!) is that we are using the clicker in place of draw reins. I know Alex has a classical dressage background and I'm sure this was never her intent but people being people do have a tendency to fixate on the head carriage and ignore the rest of the horse. Why give them a tool that enables them to do this with a clearer conscience in the name of positive reinforcement?!

I would also question whether the need for microshaping is because of our inadequacies as trainers. Yesterday it was pissing it down while I was feeding Jak and then as we moved from one field to the other he broke into a beautiful shoulder-in at trot, purely because he wanted to get from A to B quickly and without facing into the rain. He doesn't need me to shape (or microshape) his movement - he can already do it. I've spent hours (not recently!) trying to "teach" him shoulder-in and we never got anything as free-flowing as the one he did yesterday. Now that I can use CT would I be any better off? I doubt it. I would end up creating some approximation of what he can really do, get it on cue, and then spend ages "microshaping" to try to recapture the lovely subtleties.

So while I think CT and refined shaping are great tools for creating some great behaviours, I also wonder about whether they are self-fulfilling prophesies. We need them for the refinement because they are such coarse tools to start with?? I'm kind of playing devil's advocate a bit here because I have also always believed that you could do something like combine David Dodwell's "Horse Morse Code" with CT and get some really nice classical dressage out of it. But for me with Jak, I've always preferred to go out for a hack and experience his true passage when he tries to pass some pigs, experience his true collection when he's behind another horse having an inconveniently slow canter, his true lateral work when there's a plastic bag in the hedge, his true "airs above the ground" when he's in a strop and so on. Or am I just being "bah humbug" because I'm crap at dressage and have a horse who doesn't want to know?

Back to the difference between horses and dogs - not sure, it might be down to size and our ability to see what's happening with a dog. Conversely though, we can ride the horse and feel that level of subtlety instead. Or maybe that people have been doing more advanced CT stuff with dogs for much longer. But perhaps also down to the sorts of things we want to do with our animals. People want to control their horses for reasons of safety, peer pressure etc. Why do people want to control their dogs to this level? Is it more about competitive obedience rather than fear or would you say that the motivations are more similar?

I don't know the answers, just trying to muddy the waters with a few more thoughts! Interesting subject though, what do others think?

Catherine