Learning About Dogs - Teaching Dogs Article:

Teaching Dogs Vol 3 Issue 4

Free shaping sessions give us a window into the thought process of the learner. We can see how they make their choices towards solutions, we can equally see when they have run out of ideas. I can even watch a dog shaping and notice their ability to hold a thought. They may be shaping with an object and move away to collect the food. In doing so the object moves out of sight. Some dogs are very aware of the object and will turn back towards it as they eat, others get entirely immersed in the "mmmmm … foood" moment and completely forget the object. These dogs often emerge from the moment to look at the trainer with the precious "I know I was doing something but whatever it was it's completely slipped my mind now" face.

That uniquely canine skill of independently moving each eyebrow often gets triggered by the "where was I … ?" moment, easy to observe with the tan eyebrow of the Gordons.

The ability to "hold the success" in their minds during eating gets stronger as the reinforcement history increases. The more they get reinforced for touching the brick the quicker they return to it after eating, in fact some dogs return to it whilst eating - I use this as a measure of secure knowledge of that criteria and an indication of readiness to move the criteria up.

Where you place the food has very serious implications on the progression and pace of the lesson. It lets you develop the memory skills or push the learning beyond the capacity of the learner.

One observation that I bring up regularly on most shaping sessions, whether it is a dog or person in the shaping puzzle, is the process of "going through the click".

We may be playing our table game Genabacab, and the trainer is shaping the learner to touch a specific object. They have the correct timing of clicking the finger as it makes the lightest touch with the object, but the person continues to push or roll or pick up the object.

We may be shaping a dog to place a paw on a cone, or step on the first tread of a stair case but the dog goes through the click to push the cone, or run up the staircase. In nearly all cases neither the trainer or the learner are beginners in clicker training. They are demonstrating their understanding of "click marks the behaviour" by completing the behaviour. We can question the person and be sure they understand that "what they get clicked for is what is correct". So why are they continuing on to a further behaviour? In both dog and person the click is NOT ending the behaviour. It seems to be marking the intended behaviour.

If we leave either learner to progress the behaviour can be seen to get stronger, a larger push, and faster run up the stairs demonstrating that what the learner "hears" the click for is NOT what the trainer intended.

When learners offer "gross" behaviours we are limited as shapers to refine our teaching, if the gross behaviour is not the wanted solution we can come to frustratingly quick end of the session.

Very often we use one behaviour in a short form to capture another behaviour. Hundreds of behaviours are mini versions of larger behaviour where the click has been used as a scalpel to halt progression. A dog lifting a paw to touch a cone is the beginning of a wave, a dog sinking their shoulders to lay down is the beginning of a bow, a finger touch to an object is the opening to push, roll, flick, pick up, rotate. We need miniscule behaviours to allow us to change direction or develop at a different pace. Being able to freeze the small behaviours allows the learner to listen with greater precision.

To begin to teach the learner to "listen" to the click, the reward delivery must interrupt the progress through to the gross behaviour. For the dog intent on whacking the object across the room, an early click without any time delay to deliver the food behind the dog can halt progress. From the human (slow) eye it will seem the food is being thrown as the click happens. This would do no harm for the first handful of clicks. As the dog hears the click they will freeze if they see the delivery hand out of the corner of their eye moving fast to throw the food away. The first few times they may go through the click and ignore the hand movement, but very shortly you see the dog watching the delivery hand as they approach the object. The test will then be a second delay between click and delivery to see if the momentum has stopped on the click.

I prefer to deliver the food a few paces away from the behaviour to set the dog up at the start of the movement again. This allows me to monitor similar movement and compare the progress.

At the last Clicker Expo I held a Micro Shaping workshop and my wonderful learner was a 7 month old Rough Collie, Quince. The objective was to teach him to place his front feet on the first step of the four steps leading up to the speaker's platform. This an essential behaviour for a guiding dog, who will stop at this point waiting for a cue from the handler to progress on the staircase.

Quince recognised his moment of fame and wanted The Stage. Staircases were invisible let alone the first step, on approach to the stage he heard the click and flew the rest of the way towards my lap. (Seven months old and he hadn't worked out that he had Grown). If you have been teaching a dog to come towards you for delivery from your hand when they hear the click, this can invite a surge in that direction on the click. You can take the food to the dog in situ on the click, but this is not quick enough to prevent a "go through" the click, where the click reinforces intent. If you try to speed up the direct delivery to dog be very careful you don't unduly alarm the dog as the action will become associated with "click".

With Quince I had to make a very large movement of throwing food behind him, even before I clicked again, to develop his awareness of my movement to deliver food. This was quite a new concept to him. This just took a few repetitions, and I re-introduced the click as he approached the staircase. I was using a visual click to interrupt the behaviour, but also timing this with the audible click.

At seven months old he would not have been highly sensitive to what his body was doing, but rather going through life with zero finesse driven by the petrol known as "intent". Kent still reverts to that mode, his intent is to passionately say hello, his awareness of my preferred way of achieving this often evaporates in the moment leaving only "intent". Love'em!

We successfully shaped Quince to place his front feet on the staircase, in about 4 one-minute sessions. His body was still a-quiver with the intent of the flying Stage Entrance, but he was holding control for a second or two after the click. An excellent exercise for a young mind to not only take greater notice of the exact moment of the click, what they were doing at exactly that moment, but also to hold the moment between click and delivery. I firmly believe the greatest benefit of shaping is in developing the learner than simply a method of the trainer acquiring an un-lurable behaviours. Learning to be shaped is full of the same learning skills that life throws at you - especially for seven month old boys.

For people who become intent-clicked we have to find different strategies to freeze the moment. You can employ the same diversion with the delivery, but this only works if the learner is highly motivated to want the reward. I may be wicked and set up a system where the learner has to catch the reward before I drop it into a jar, or give it to another person first. Somewhat slightly punishing, but it focuses their mind on hearing the click, and if they were not very quick at responding to collect the reward it would be gone.

On the other hand ……

A little voice, that sounds ever so like Mabel saying "hang on a minute….", is saying to me; "but clicking an intent could be useful".

I know when I first started shaping with her, I would put out several interesting objects to stimulate her imagination (read 9 months old here and entirely self orientated). By clicking when she looked at a particular object she would collect her food and then travel straight over to that object. Asking her for a paw wave, and then clicking when she looked at the stool, would guarantee her going to the stool to tap it. We seem to have developed a process whereby I am as certain as I ever can be of Gordon thought process that what she is looking at is what she is thinking about. It is not exactly clicking her intent to do a particular process on arrival but we are able to jump many steps in the shaping.

I'm sure many of you have this understanding between yourself and your dog. I rather think the dogs are better at understanding our intentions for them than the other way around, but it is still an area I would like to explore.

Can we use a different marker to confirm intent?