One year ago, Elsie was picked up as a stray in Etobicoke by Toronto Animal Services. I retrieved Elsie from TAS as a foster for Speaking of Dogs – I was going on vacation in three weeks and thought “No way I’ll want to keep an 8yo shepherd! No problem taking on a foster now…” – but as the story so often ends she became a “failed foster” and took up permanent residence.
I don’t know what her life was before that, but I have no doubt that she had a damn good time chasing wildlife once she hit the streets! All last summer the bane of my existence, and the glory of Elsie’s, were the squirrels in the parks, the cats on the fence in the yard and both critters during neighbourhood walks. If Elsie thought she may perhaps just might have seen a bit of movement, or heard a little noise, that may have been a cat or squirrel her brain would disappear into an alternate universe. She would freeze, stiff as rigor mortis, and her eyes would go completely blank as she stared (and sometimes drooled). The other half of her reaction to critters was full-out hurling herself toward the suspected Squirrel Zone, sometimes screaming like I was killing her, and also running backwards like a crayfish, just about popping out of her collar (this is why she wears a sighthound martingale collar). While there were some squirrel-safe parks, other parks were off-bounds because she would disappear into the forest in hot pursuit of her fluffy-tailed friends. She would not only tree the squirrels, but also chase them from the ground as they ran through the leafy canopy 50ft above.
It’s easy enough to say “work at a distance, wait for her to look back at you and click and feed attention.” Sure, the basics are there. Decrease distraction with distance, mark and reinforce a desirable behaviour. But I had to ask myself: what on earth could I have possibly offered Elsie that was more potent than a squirrel, even at a distance?
The Answer? A squirrel, of course.
Premack’s Principle states that a higher-probability behaviour can act to reinforce a lower-probability behaviour. This basically means that anything a dog is apt to do on their own free will can act as a reinforcer for any behaviour that they may not do otherwise (i.e. just about anything we want them to do). Look at me = chase a squirrel. If you think about it, all the world’s a Premack: does a liver snap reinforce a sit, or does the behaviour of eating that liver snap reinforce a sit? I discussed Premack’s in a past post about Arlo’s recall training.
Training is not only about the oft-toted “consistency, timing and patience,” but also knowing what motivates the learner and how to use that thing/activity to your best advantage. I cringe when I hear complaints of a dog who “is overly food motivated” or “goes too crazy for the ball,” or “tries to tug on everything.” Having too many motivating/reinforcing things and activities is the least of a trainer’s worries! Certainly having a dog who can’t think when there’s potential to engage in chasing a ball – that’s Elsie – can be a pain at times and requires some work and criteria splitting of it’s own… but I can tell you that I’d far prefer to work through impulse control exercises and finding less potent reinforcers than scrounging for something – anything – that a dog likes and that I can control for use in training! (See posts on Arlo!) Accurately marking behaviours and controlling reinforcers are the cornerstones of training any animal. Without motivation on the part of the learner, there’s no point trying to teach anything.
This winter I worked a lot with Elsie on basic behaviours and becoming ‘operant’ – Skinner would roll in his grave at the use of that word in this context – thus forming a better foundation for our squirrel work this spring. Simply building a better working relationship with Elsie (i.e. Elsie knows that I am a source of reinforcing things and activities, and she can control the output by offering behaviours) has put us miles ahead of where we were last summer. The second part of the training utilizes Premack’s Principle, as I mentioned above. How do I reinforce looking away from a squirrel/cat? By letting her chase the squirrel/cat! Chasing vermin trumps any food or ball chasing that I could offer.
The basics are the same as in any training. Mark and reinforce. For the purposes of super-charging my clicker I have chosen mainly to use a click to mark attention, though I will occasionally mark with a ‘yes’ if I don’t have the clicker in my hand. The reinforcer comes from me in the form of a release to go chase the critter – “After it!” Thus, attention to me/away from the squirrel has been reinforced via access to the Most Sacred of Activities, squirrel/cat chasing. (I had started training with the release “go kill!” but I realized after a time or two of using it that it could be poor public relations to tell my dog to ‘go kill’ and have her run of after something, especially a domestic kitty…)
I’ve approached this training in three different scenarios:
First, in the parks. To start, I slide the handle end of the leash through the loop of her martingale collar. This way if she shoots backward she isn’t going to slip her collar and I have the ability to limit her access to critter chasing before she “earns” it. As I described above, I will mark attention on me with a click and release her to chase. I’ll drop the handle of the leash and hold the clip so the leash slides out of her collar as she runs away. Since she is most often squirrelly in forested areas and she is running full tilt, I don’t want her leash to get caught up on a branch if I just drop the leash as she runs off.
Now, there is an obvious problem in this scenario, and that is what happens after the leash comes off. I admit, it is very difficult to get a lot of trials in each trip to the park since each trial takes a good few minutes. The bigger problem is that of danger to Elsie if she goes too far or gets to a road. Generally, she does stay in the vicinity, however I am careful which parks she visits off-leash. I must know the area well and know that there are no major roadways that she will reach. I do recognize a degree of risk – as there is any time a dog is off-leash – however I would prefer to have her running around like a lunatic after I release her with the ultimate goal of being able to call off a potent distraction rather than having life-long management in forested areas where she is apt to give chase. Everything is a balancing act of risk-benefit and this is the route I’ve chosen. That said, the risk of her being hit by a car is far less than the risks posed by a common tool used in dealing with predatory sequences – that being the shock collar. (See also the series of posts about my introduction to dog training)
The second scenario is training on-leash in the neighbourhood. There are about 10 000 000 cats in my neighbourhood and Elsie wants to eat them all. We will encounter at least a few on each leashed walk through the neighbourhood and now, utilizing Premack’s Principle, I can put them to good use. Similar to the park situation, I will mark attentive behaviour and release Elsie with “After It!” Needless to say, I have to go with her on the chase! The trouble with this scenario has everything to do with a cat’s ability to give some gnarly stink-eye and disregard their own safety. With some work, Elsie will now take food around cats so I am able to reinforce attention with food before releasing her to chase (also building value into the food) – at which point the cat is gone or is far enough away that it is not at risk. Although, this is also in large part due to that I’m not a sprinter. In both cases, Elsie is sure the cat is still around and will chase with a lot of enthusiasm (=reinforcing!). I imagine that she is also feeding off of opposition reflex triggered by the tight leash, though this is a natural by-product of me being a slow human being consensually dragged around by a fast dog.
The third scenario takes place in the yard. There is a cat and squirrel superhighway on the back fence, so I can train here the same way as in the park and simply release her into the yard to chase. This is very convenient because training situations find me rather than the other way around. Below is a short video showing how I execute this:
I’m finding that Elsie is reorienting to me more frequently, is less intense, and ‘comes down’ faster than she has previously. There are times that I release her to go “after it,” and she doesn’t give chase. This poses an issue all of its own, as the release will lose its charge if I use those sacred words and don’t provide a squirrel or cat to chase. I never thought that NOT chasing a small fuzzy would be a problem! This training is in its infancy at the moment. I would like to get to the point that she will recall from a squirrel in the park after she’s ‘locked and loaded,’ as well as use these predatory behaviours to reinforce more complex obedience behaviours. To do this, I will have to split criteria to ensure that Elsie has success every step of the way. In the same way that using the ball as a reinforcer has boosted Elsie’s heel work, using predatory behaviour to reinforce obedience work will have a similar, or dare I say more intense, effect. It is certainly more difficult to control reinforcement since cats and squirrels don’t fit in my treat pouch, but I also know that chasing prey is an intensely rewarding experience for Elsie and she’s going to do it regardless of whether I appropriate it for my own purposes. So why not use it to my own ends and let Elsie get her kicks at the same time? In the end, it has diminished what can become a very adversarial topic of conversation between dog and owner.