Saturday, August 13, 2011

Teaching Vanya not to chase livestock or deer

Vanya and livestock/wildlife chasing
On our small farm, we have chickens. On 3 sides, we’re surrounded by DNR wildlife areas where deer and hunters are abundant. On the 4th side, our neighbor owns a pasture that usually has 40 or 50 heifers (young female cows) grazing.

 Our older two dogs, a pitbull named Tiva and a husky mix named Juneau, never chased critters once they were adults. Soon after I adopted her,  Juneau lost her desire to chase deer when she caught up to one and it kicked her in the forehead hard enough to nearly take her eye out. Tiva chases small rodents, but with deer, chickens, and cows, she was always content just to wallow in their poop.

But when we adopted impulsive, hyper-aroused young Vanya, it was soon clear that he liked to chase livestock. So we kept the chickens in their yard and eventually put up fencing around all 20 acres. But a determined dog can always get through, especially when the surveyor leaves a gate loose.



I did a series of foundation exercises with Vanya, hoping to reduce his arousal around livestock. I started with our chickens and then moved to the neighbors’ heifers (from our side of the fence):

Foundation work
  1.  I started by doing the relaxation protocol far enough from the chickens so that Vanya could still stay relaxed.
  2. Then I moved on to LAT (Look at That) with the chickens staying still (eating), starting far enough away so that Vanya could still stay relaxed around them; slowly decreasing the distance.
  3. Then LAT with the chickens running around, starting far enough so that Vanya could stay relaxed near them. I did all this as well with the cows on the other side of the fence. 
  4. For our chickens, I then took Vanya inside the yard with them. I walked  back and forth perpendicular to the chickens, me armed with a clicker and cheese whiz (his favorite treat), the chickens staying reasonably still eating their food. Vanya was on a line attached to my waist. I made sure to stay far enough from the chickens so that Vanya could glance at the chickens then back at me for his cheese. We weren't getting closer to the chickens in this exercise.
  5.  Then I walked  closer to the chickens, with lots of clicks and cheese for Vanya staying calm and loose leash. If he ignored the chickens, good things happened. If he focused on them (alert, excited) we went backwards--a sort of penalty yards game.
He got a lot calmer with both the chickens and cows, and I thought: success! Wrong. A couple chickens made their way outside the yard, and he chased and caught them. He didn’t hurt them, only damaged their dignity by licking their bellies. He dropped them when my husband noticed and yelled. 

Much more dangerously, he squeezed through the cattle fence and got in with the cows. Initially, they chased him and he responded in a sensible way: he ran away. But then he started barking,  and they turned and scattered,  and that was when he started chasing them (I wasn’t around to witness this). My husband ran after him and caught him, but only after he grabbed a heifer’s nose and swung from it. At that point, we put up another line of fencing and tried a lot more “look at that” games from a distance to get him to calm down around the heifers. He calmed down just fine, as long as they were standing still and he was on the other side of the fence. But eventually he somehow got through the double fencing, chased the heifers again, and got kicked hard enough to get a lung contusion and nearly die.

At this point, I decided I had two choices: keep him on a line for the rest of his life since LAT and the relaxation protocol weren’t enough to discourage him from cattle chasing. Or else I could try an e-collar. When a trainer I respect suggested the e-collar to me, I was initially horrified, imagining that the shock was like the kick of an electric fence. But I decided that leashed confinement, not low-level stims, would be the greater aversive for this dog, and since I try to follow a “least invasive, minimally aversive” training philosophy, I figured it was time to learn more about an e-collar.

Theory and Practice:
I started by reading Steven Lindsay's chapter on  e-collars in Vol 3 of his Handbook of Applied Dog Behavior and Training (Chapter 9). Lindsay is a well respected primarily-positive trainer--I think he was an early promoter of the "Least Invasive, Minimally Aversive" philosophy.
 From Lindsay, I learned that modern e-collar training is very different from old style shock collars, and I learned that he believes it can be significantly less painful than a gentle leader.

Early shock collar training worked on a positive punishment principle. The trainer would wait for the dog to do something he wasn’t supposed to do, and then he’d shock the dog (+) to stop the behavior and reduce its chance of happening again (P). The problem with this approach is that you might need to use a very high level of shock to be punishing enough for a very rewarding behavior like cattle chasing. Other problems: the dog might not have a clue what the shock means, so he goes running off into the next county, trying to get away from pain. Or the dog associates the source of the shock with whatever happens to be nearby. The largest problem for me is that this approach typically involves working above threshold, well out of the dog’s ability to respond operantly (in other words, to learn that his or her own behavior can control the world around him, or at least that shock).

Modern e-collar training works in a very different way. Instead of tossing the dog in a situation where he zooms above threshold and then zapping him into submission, you  introduce the e-collar into a very low-distraction environment. You put the dog on a long-line and then you find the dog’s “working level” by tapping on the collar at the very lowest levels of electric stimulation, until the dog responds to the tap, usually with a twitch of her ear or a slight head tilt. (Of course, you first test these levels on your own neck or wrist, so you can feel what they feel like. It’s much, much lower than the shock you get when you walk across a dry carpet. It feels to me about like a light pin-prick. It’s noticeable and a bit annoying, but not painful.)

The dog then learns that she can turn off the taps (the stim) by coming toward you. You guide her with a slight pressure on the line, and the instant she takes a step toward you, you stop the light taps and mark the desired behavior (I use a mouth click) and then reward that behavior with a treat when she arrives at your side. As far as quadrants go, you can interpret this in several ways:

-R: you’re removing the stim (-) to increase the frequency (R) of coming toward you; or
-P: you’re removing the stim (-) to decrease the frequency (P) of blowing off your recall cue.

You repeat this dozens and dozens of times in a very undistracting environment. To see this in action, here’s a good youtube video from the trainer Mike Loesche:

The dog in Mike’s video appears to be showing some very slight stress with the stims, but those low levels of stress are acceptable to me if they allow my dog off-leash options. Mike starts with a level 8 (out of 127 levels on his dogtra collar) in the video, and I personally have a hard time feeling that on my wrist or neck. When he increases distractions in the video, he increases  his stim to level 14, which I can feel on my wrist as a pinprick.  (For comparison, my hyper-impulsive pittie Vanya's working level varies from 10 to 16, depending on the distractions.


 These levels don’t elicit signs of pain, unlike the gentle leader or even a regular flat collar.

This should sound a bit like training a dog on the gentle leader, except for an important difference: the e-collar is much less aversive for my particular dog than a gentle leader tightening around his muzzle, and much easier for me to control the level of aversiveness. You have 127 different levels of stim on a good e-collar, and you can start and end the stim precisely, unlike a gentle leader. (I dislike the GL, not because it uses P- or R-, but because with Vanya it has a bad tendency to send him spiraling over his threshold, unlike the collar). .

Equipment: I use a dogtra 280 ncp with a half-mile range, vibration, and 127 levels of stim. It's about $200. Paying the money for a higher quality collar with at least 100 levels is essential if one is going to try e-collars. Dogtra is one of the  reliable brands. I started with a cheaper brand (Sportdog) and it was a waste of money. It only had 8 levels, so I couldn’t find the lowest possible stim level that the dog can perceive without pain.

 Plus it only had ¼ mile range—not enough—and it wasn’t reliable enough.

Introducing the e-collar around livestock: teaching the recall again
  1. First I found Vanya’s working level with no distractions
  2. Then I trained a new recall cue (here), using the methods in Mike’s video above.
  3. Then I introduced very low levels of distractions for the recall, with Vanya on a long line and great treats as rewards for responding to the stim by coming toward me.
  4. Then I did this well outside the chicken yard with the chickens calmly eating. Then we moved slightly closer to the chickens, and then a little closer.
  5. Then I had the chickens running around, us well away.
  6. Then we moved closer to running chickens
  7. Then we moved inside the chicken yard
  8. Then we moved to bike riding through the running chickens
My goal, through all of this, was to stay far enough under threshold so Vanya got thousands of reps, all at low stim. I built a foundation of conditioned responses, in other words.
Then we moved on to practicing the recall this outside the heifer pasture with the heifers far away, standing still. Then closer, with the heifers running around.

Teaching Vanya to ignore livestock and deer:
So now I had a very good recall, even with cows and chickens (and deer) nearby. But I also wanted Vanya to leave critters alone even when I wasn’t there to recall him. All of the above work had taught him a great deal of calmness in the presence of these critters, but I also decided to work with what’s called “the crittering protocol” (Lou Castle came up with this). This uses exposure to increasing levels of distraction to teach a dog to leave a particular critter alone, even if there’s no person nearby to give a recall or leave it cue. Here’s Castle’s full protocol: http://www.loucastle.com/critter.htm

I didn’t follow his exact protocol, but I did follow the general outline:
  1. Always stay under threshold, where the dog can see the critter but not go above threshold
  2. Start, for example, with a cow 100 yards away, standing still. Have your dog on a leash. Imagine you’re on a football field. The cow is standing at one 0 yard line. You’re standing on the other 0 yard line. You walk back and forth with your dog (leashed) at your zero yard line. If your dog calmly observes the cow, lovely. After a couple of calm strolls back and forth on your 0 yard line, you walk diagonally to your 10 yard line (so you are 10 yards closer to that cow). (Lou has the cow walking back and forth on her 10 yard line, but I didn’t introduce that until much later.)
  3. The moment your dog gets close enough to begin to react slightly to that cow (ie, for Vanya, his tail goes up a little, and he begins to stare, which always precedes his chase), you begin tapping on the e-collar at the dog’s very low working level while walking backwards  directly away from the critter. As soon as the dog stops staring at the critter and looks back toward you, you stop tapping and stop walking backward. You use that distance as the new line, and you begin your strolling on that line (so, say, you got to your 30 yard line before the dog stared at the critter. You begin to tap and also walk backwards, and you got 5 yards before the dog relaxed and looked away from the cow. You’re now at the 25 yard line, and you walk back and forth on that a few times before going up to the 35 yard line.)
  4. Continue until the dog figures out that getting too interested in the cow means he goes backwards away from the cow (and gets an annoying tap on his collar), while staying calm means he gets to go closer. Continue until you and the dog are calmly standing next to the calm cow.

Then repeat from the beginning, 100 yards away, this time with the cow trotting back and forth on her own 0 yard line (a critter in motion is much more distracting than a calm critter).  This protocol works by keeping the distraction levels very low, so there’s no need for high level shocks (no P+, in other words). When the dog stares, you begin stimming, and you remove that stim (-) in order to decrease the behavior (P) of staring.

With deer, I couldn’t find any willing deer to stand calmly on a line while I went through this protocol. So a few times Vanya did break and chase when a deer came bounding by right in front of us, and I did give him a higher level of stim to get him to recall. (When I do that, I always check those levels of stim on my own wrist. They aren’t pleasant, but they aren’t awful.)

This is really quite similar to the penalty yards approach for teaching a dog to walk on a loose leash and to greet politely. 

Some questions people have asked me:
Q1: Does Vanya think the stim comes from me or some place else?
A1: I have no idea, and because the stims are so low, I don't think it matters. If the stims were big,  nasty shocks, then I wouldn't want him to think they came from me, but rather from the object he's chasing. But since I use a low-stim protocol, it doesn't matter.

Q2: Is Vanya stressed by the e-collar?
A2: Not by the collar itself. He's happy to see it come out. Actually, he wears it each and every time he goes outside. 

Q3: Well, so is he stressed by the stims?
A3: Sure, probably. But he's much less stressed by the stims than he is by being on a leash, being confined, or being in his gentle leader. When he gets a stim, he notices it, of course. But it doesn't make him look subdued the way a long line pull does, and it certainly doesn't make him yelp and shriek the way a gentle leader tightening does. Vanya is a very vocal, expressive dog. He sings, warbles, shrieks, screams, yelps, snorts, yodels. When he gets a stim, he runs to me for his treat. He might cock his head slightly or perhaps twitch an ear. But mostly he just comes for his treat. 


More importantly, I firmly believe (and abundant research supports this idea) that stress is a good thing for animals (including people) when the animals learn a behavior that can end the stressor.  Learned helplessness and other fallouts from stressors result not from the momentary aversive, but from the inability to make that aversive stop. When a dog, or a person, learns that a given behavior will end the stressor--and the world doesn't end!--resilience increases. (An interesting study along these lines is Seery et al's recent paper showing that happiness is highest in those people who have faced intermediate levels of adversity. Too little adversity in your youth, and the first little bump in your road throws you into depression. Too much adversity in your youth, and you learn to give up, not to be resilient. See Seery, M. D., Holman, E. A., & Silver, R. C. (2010). Whatever does not kill us: Cumulative lifetime adversity, vulnerability, and resilience. Journal of Personality and Social Psychology, 99, 1025-1041.)

Q4: Did I stop using the ecollar once I trained him?
A4: No. Where I live and work, there are tons of things I can't control: deer, bear, hunters, loose dogs. I want off leash control, so I put the collar on him when I let him outside into our fenced fields, and when I walk with him off our land (off leash, yes, but only in places where I'm fairly certain no off-leash dogs will show up).

Q5: Did I use these protocols to deal with Vanya's dog-reactivity?
A5: No.  Vanya's dog reactivity has complex roots, and because he can quickly  zoom over threshold near another dog, I don't use the e-collar around other dogs. With deer, if I have to give him P+ because he's chasing a deer, I don't much mind if he comes away with a fear of deer or a superstitious hatred of deer. But I don't want to ever get him in an over-threshold situation with another dog and end up superstitiously associating that other dog with something aversive. When he's playing with another dog, I do sometimes use LOW levels of stim to make sure he recalls to me for his time-outs. But this is under very controlled circumstances. 

Q6: What else do I use the e-collar for?
A6: I don't teach obedience with it. I just want Vanya to recall whenever he feels it, so I keep it simple. I do use it for TENS work on my back to relax stiff muscles (seriously--at Vanya's working levels, the e-collar is great for my tight back muscles). 

Q7: Would I recommend it to another person?
A7:  Each person needs to figure out for her own dog what is least invasive and minimally aversive. I do recommend reading Steven Lindsay's chapter to people who are having trouble with their dogs and  livestock chasing. 


Q8: Can things go wrong with the ecollar?
A8: Yes. Since it's a powerful tool, the temptation is always to increase distractions too quickly and go over threshold. Bad, bad idea. I don't let other people use the ecollar on Vanya, because I'm afraid they'll zap him. High levels of shock can have bad fallout, just as high levels of any aversive can. But that doesn't mean low levels have fallout. The main danger is being impatient and setting your dog up to fail. 


Q9: Do you consider using the e-collar to be an admission of failure?
A9: Nope, not at all. But the couple times that I have  put Vanya in a position where he went over threshold (ie, a running deer) and I needed to zap him: yes, those were failures. Just as putting your dog in a position where he ever goes over threshold is undesirable. But life happens. Most dogs are resilient, and we can teach dogs to be more resilient by exposing them to small stresses and showing them that they can respond successfully. (When I say "over threshold", I don't mean a little stress, a little shrieking and yodeling and carrying on. Vanya can learn perfectly well when he's tossing a hissy fit. I mean that state when a dog is so over-aroused and overwhelmed that learning can't take place. Imagine your limbic leap when a rattlesnake rattles next to your foot. Your conscious brain isn't processing; you are simply reacting.)


Q10: Hasn't Vanya been made miserable by this horrible tool?
A10: No. He loves it, because to him, it means he gets to run. And running off leash is his joy. So I don't feel any guilt whatsoever for choosing to use it. I think it works on very similar principles to other training protocols common in the positive training community--work under threshold and develop the foundations that create a conditioned response to a stimulus. So then, when high level distractions show up, your dog has a conditioned response. You don't need a t-bone steak to distract your dog if you've done your clicker foundation training correctly, and you don't need a painful shock to train your dog if you've done your e-collar foundation training correctly.