Since the methods we’re using to train Spencer — mostly based on clicker training — are not that well known among the general public, I thought I’d give a few more details on how the car has been helping to shape his overall behavior.
- Whenever Spencer looked at someone outside of the car, we called his name and clicked when he looked at us instead of the person.
- The click tells him that the behavior he just did is something we want and is followed by a treat.
- We called his name to interrupt a potential reaction and teach him an alternate behavior. You want to reduce the number of times your dog rehearses an unwanted behavior. (Alternately, you can use this sort of technique to prevent such behaviors from occurring in the first place.)
- Whenever Spencer looked at someone, we clicked and gave him a treat, just for looking (again, before he had a chance to react).
- Whenever Spencer would look at someone, we would wait for him to look away (often at the treat bag that sits near the gearshift for easy access while driving). We would then click and treat.
- Over time, our criteria for rewarding him have gone up: whereas once we would have treated for someone across the street, they now have to be nearer. We have also gradually reduced the number of clicks even for nearby cases because he now knows what’s the right behavior. (But we still often treat to prevent slippage if he is startled by someone’s unexpected movement.)
- We’re currently working on a new stage in the parked car where he isn’t rewarded just for looking away: now he has to look away and put his head down on the arm rest or if his head is already down, not lift it. If he does either of these, he gets clicked and treated.
- The result is that he is starting to leaving his head down and just follow people with his eyes or even ignore them. He’s learning that even if he doesn’t react, the people will go away.
Another variation on this has been learning not to react to people loitering near our car. People walking by are less threatening, because they have a trajectory that he can extrapolate and realistically anticipate their departure. He doesn’t know what people loitering are going to do, and this is scary. A general principle of clicker training is that when one dimension gets harder, you make the other dimensions easier. So Spencer is on an earlier stage of clicking and treating for this. We’re somewhere around Stage 2. He did really well the other day by not reacting when a guy was practically sitting on the hood of our car smoking.
Yet another variation we’re dealing with is sensitivity to proximity to “his” door. Apparently he’s figured out that if we can get to him through the door he uses to get in and out of the car, so can Scary People. (Since parking is on the right-hand sidewalk of our street, he almost always gets in and out of the same door.) This only seems to come into to play when we are backing up or parking and there is someone near the rear right wheel. If we are parked and doing the passerby exercise, he is used to people arriving from behind and walking by. The challenge of this variation is that it is extremely difficult to parallel park and simultaneously click-and-treat if someone appears in this zone.