by Grayson Schaffer | on October 22nd, 2009 | in Features, Training, Video Clips
Safety Note: If you're going to try this with a bottle, make sure your dog's hold is bulletproof.
Some folks were asking whether Danger can do that beer trick from last week’s short movie, “In the Face of Danger,” in one take. Yes, and chances are your dog can too. Training a dog to connect tricks or behaviors end to end is called chaining. Typically, these types of linked behaviors are taught back to front, or what’s called back-chaining. Basically, you start with the last part of the task, train that to proficiency, and then add the next-to-last part. Withhold your reward until the dog completes both of these well-polished tasks end to end. It won’t take him long to figure out that the criteria have been raised and he now has to do two tasks before the reward. Then add a third link in the chain and so on. When you see dogs performing complex, apparently human-like tasks on TV, this is generally how they’re taught.
OK, here’s the video.
How To Teach Your Dog To Fetch A Beer from Walker Parks on Vimeo.
by Grayson Schaffer | on September 18th, 2009 | in Features, Training
Dave, Dave, and Hank after a successful training run
And now for something completely different. In August I joined the Los Alamos–based Mountain Canine Corps to start training Danger for search & rescue work. The group is made up of volunteer trainers and a dozen or so dogs of various breeds and mixes. When someone goes missing in the woods, the New Mexico state police give them a ring and dogs and handlers get dispatched to the scene for a search.
The team trains dogs in three disciplines: tracking, air scent, and cadaver work. Tracking dogs need to follow an aged scent—sometimes days old and overlaid with other odors—for long distances. Air scent dogs pick up scent on the breeze and then home in on the source. And cadaver dogs, well, look for dead people and more importantly, bits and pieces of dead people.
Danger is just getting started on tracking, which he loves. The process of following a scent track is a simple chain of behaviors, just like retrieving. I say this now as a sort of personal reminder. Retriever training is often made unnecessarily complicated with complex drills and equipment. So to avoid that, we’ll take a positive-reinforcement approach to tracking, learn to understand Danger’s body language, and attempt to fix one problem at a time, while raising our criteria slowly so that he’s always successful. The same requirement of 80% proficiency before moving on to the next step that we’ve been following with Assistance Dogs of the West applies here, too.
Retrieving: Acquire a line, run a certain distance, stop and hunt for an object by scent, pick up the object and return, deliver the object to the handler. Tracking: Acquire a scent, follow a scent, find a subject, return to the handler, alert the handler, return to the subject with the handler.
To get some perspective on how to get started, I called Steve White. White, who’s based near Seattle, is well-known for training police and search dogs using positive reinforcement methods. As with hunting dogs, police dogs still tend to be trained using force-breaking methods. In that regard White, like our gun- and adventure-dog guru Mike Stewart, is ahead of the curve. He’s also spent enough time around other trainers to know what we’re all up against: “The only thing two dog trainers will ever agree on,” says White, “is what the third trainer is doing wrong.” So true.
White recommends starting search dogs on tracking, rather than air scent. “In my experience, we generally don’t have a dog do any air scenting until his tracking is good.” I’d been told this before, but White’s explanation really hit home for me because it deals with a dog’s natural hunting ability. “Dogs are hardwired to be efficient, effective hunters,” says White. “Wolves, foxes, and domestic dogs, tend to combine tracking, trailing, and air scenting, but the vast majority are of their hunts are successful with air scent and maybe a little tracking at the end.” I’ve seen this dozens of times, now, hunting in the upland with Danger: He’s quartering methodically, then his nose goes up and catches something on the wind; he quarters more aggressively, then his nose goes to the ground, and a few seconds later the birds flush.
A lost hiker, though, could be miles away, and the scent could be faint. A dog on the hunt would probably do better to ignore such a faint scent and keep looking for something fresher. I saw this first-hand yesterday, when Danger was tracking nicely until his trail intersected with one that another handler had just walked down. Danger switched off the aged track and went frantically off on the much fresher track. Clearly, we had him working above his ability level and needed to lower our criteria to ensure success. For starters, White recommends simplifying the terrain and removing every odor that’s not the one you want tracked. “Start on a hard surface like asphalt,” he says “then move to concrete, gravel, and finally grass.” Each of these surfaces holds progressively more scent. White recommended a beginner’s drill:
Have a subject lay treats a few feet apart on an uncontaminated stretch of asphalt. (The subject is also laying down scent while doing this.) The track should be straight and then end without a subject for the dog to find. The idea here is to make the faint (we’re on asphalt, remember) scent of this one human predictive of food. We want to reinforce the dog for the act of tracking. Then, gradually lengthen the distance between treats so that the dog has to follow the scent bridges from one treat to the next. Gradually, we’ll lengthen the track, add turns, and increase the space between treats. It’s a good idea to start your dog on lead, but if you’re having to give the cord anything more than the occasional nudge, you need to reduce the difficulty level for the dog.
OK, so that’s our plan for Sunday’s practice. But not so fast. What about that refind? The conventional wisdom on teaching a dog a behavior chain is to work back to front, called backchaining. We’ll go more into chaining in the weeks to come, but the basic theory here is that the dog learns the proper way to cross the finish line and then starts further and further from it. White says it’s critical to have your refind in place by the time you start running tracks with live subjects at the end. If you teach the dog to track and then offer a big reward without a refind, and then change thee the rules to require that refind, “The dog thinks this is a buzz kill,” says White. “And in actual neurochemical terms, that’s exactly what it is. The dog isn’t searching to find a person, its seraching to satisfy its neurochemical urge. He wants that good endorphine buzz, and you’ve just taken it away from him.” Bummer.
The good news, though, is that training a refind is much easier than reinforcing and refining a dog’s tracking ability. The refind is simple mechancics and repetition. To train the refind, we’ll use what’s called a runaway, a short track to an easy find where the dog has seen the subject leave. The steps are:
- Dog finds the subject. Subject marks the behavior and rewards it.
- Handler calls the dog back and requires a sit (the alert), marks the behavior and rewards it.
- Handler gives the cue “show me,” and the subject calls the dog back.
- When handler and dog are back at the subject, the handler delivers a jackpot-sized reward.
OK, so that’s the order of business: Get Danger’s refind up to 100 percent and reinforce him for slowing down and focusing on the track at hand.
Information about Steve White’s videos and courses can be found on his Web site, i2i K9