When you are ready to move on from a basic training plan, one with simple goals and behaviors defined step-by-step, it's time to learn about expanding behavior, record-keeping, and those inevitable mistakes.
Building duration or distance—randomly
A sticky point for many trainers seems to be how to increase difficulty steadily enough to maintain progress, but not so quickly that it discourages the animal. While the 300-Peck method is great for some situations, the rigid mathematics of it disturb me; call it a personal failing, but I can't maintain a 300-Peck program. Instead, I rely upon random variables to keep both my interest and the animal's while making steady progress.
Raid a dusty board game for dice. Dice are a wonderful way of generating random training sessions! To introduce duration, for example, you can roll a single die to vary your criteria between one and six seconds. As your dog progresses, add more dice—or dice with more sides. This way your dog never knows how long he'll be holding the position, and you won't find yourself falling into predictable patterns that will quietly undermine your training.
(I did know someone who wrote a computer program to generate a sequence of random numbers between fixed upper and lower parameters. That may be too random, and is clearly beyond the call of duty!)
I resisted writing plans and tracking data for a long time, because I feared it would be time-consuming and tedious. (I also have a fundamental aversion to math.) Here are some tips I found to make my record-keeping simpler and more useful.
- Use a very simple plan/data sheet. I designed mine to have a single line per session; no lengthy journals for me! I define criteria in a few words and then track attempts and success.
- Work in simple units. I am far too lazy to do more math than necessary, so most of my formal training is done in bursts of 5 or 10 repetitions. This makes it incredibly simple to convert logged data into success ratios. We have 7 correct out of 10? We're not at 80% yet. We have 4 out of 5? We're ready to move on!
- Use numbers to reinforce your own good record-keeping. Look at all those high success ratios! Aren't you proud of yourself and your dog? You can only see this hard proof of success if you document it.
Record your training
Training plans and record-keeping are nearly inseparable. I already know every excuse for not keeping training data—I invented most of them! The fact is, I found that my training progressed much faster when I was keeping records.
Most of us simply don't remember things as accurately as we'd like to think. We may know that it's time to raise criteria or move to the next step when the subject is successful 80% of the time. But without real data, we think "he does it most of the time" when he actually has only 60% success, or 95% success.
With loose data-keeping, the imagined 80% success rate—really a 60% success rate—means that the dog is not yet "good enough" to ask for more, and you would not likely be successful at higher criteria. On the other end of the spectrum, the imagined 80% success rate—really a 95% success rate—means that the dog is pretty certain exactly what is expected and correct. He'll be frustrated if you "change the rules" and require more behavior for the same reinforcement. It will be harder to require a greater effort for similar reinforcement in the future.
Tracking success rates means that you are able to increase criteria just when the sweet spot of 80% is reached. Keeping records also means that you can identify trouble spots instantly and fix the root of the problem, rather than muddling through an effort to broadside the solution.
Dog has an unreliable stay when the car door opens? Let me check his records. Ah, yes, it says here that his success rate dropped to 50% while we were working on raised surfaces a few weeks ago. I don't need to work on varying distractions or different locations—I can jump straight to the real problem, which is that the dog wants to reach the ground, and thereby save a lot of time and effort!
Be specific when tracking your training; this is where your training plan blends seamlessly. "The dog can reliably stay 5 seconds at a distance of 10 feet." is a lot more useful than "He can hold it for a short while a little ways away."
I'm not a terribly organized person, so I've had to develop a lot of tools to train myself to keep records. The biggest reinforcement, however, is seeing my dog learn a behavior fluently and quickly!
Adjust as needed
Of course, most dogs don't read the training plan, and the best plan can need adapting. Sometimes you may reach what seems to be an impasse in training. Using the example from Part 1, what if the dog, once sitting on the mat, offers no further variation? Or what if the dog doesn't even look away from the handler to start?
I learned from Bob Bailey that all training problems are rooted in imprecise timing, inappropriate criteria, or insufficient rate of reinforcement. Timing, criteria, rate of reinforcement is the chant that runs through my head whenever I face a sticking point. A training plan is about simplifying and formalizing the "criteria" part of the process. If the subject isn't progressing, stop training and think about the plan for a moment. There are several ways to get an errant training plan back on track.
Break it down—further!
It's easy to "lump" criteria into big chunks. After all, it's obvious to us—we already know the right answer! Try to split each step of your training plan into two or three smaller steps. For trouble spots, try to carve it up into five smaller steps. (I once listened to a student describe her failure to keep a dog on a mat while visitors entered her home. After asking her to describe her progression, I took one of her intermediate steps and broke it into an additional 18 training pieces! Those mini-steps were much easier for her dog to achieve.)
I see many trainers waiting for a "deliberate" movement to click. But if the dog accidentally falls into my criteria, I'll certainly take it! A very common example of this is if a dog is not lowering himself to a mat. I can click stepping onto the mat, and then drop the treat onto the mat itself. Most dogs keep their heads low as they chew the treat, and, yes, I can click the head low to the mat!
Change the picture
If the animal can't come up with a different behavior and you can't come up with a different incremental behavior, change the scenery. This can be as simple as approaching from the other side of a settle mat, or as dramatic as going to a different building. Sometimes a change like this can make a behavior easier to perform if there is an environmental influence. And, sometimes it just jars you out of a rut!
Observe your animal, and avoid fixed "recipes"
It's important to note that there are many possible training plans for any given behavior. Some may be more efficient than others, some may work better for some temperaments than others, but there are many possibilities. When asked how to train a behavior, Julie Shaw, Senior Animal Behavior Technician at a leading midwest veterinary university, often answers, "I don't know. Let's find out!"
With a dog I know well or with an animal with foundation skills, I can take shortcuts that aren't possible with an unknown or naive subject. A dog with a natural preference for mouthing an object might need different steps than one with a natural preference for pawing.
You should be able to recognize quickly whether your plan is appropriate or not. Some steps may need many sessions to build up to 80% success; other times you may ask only two repetitions and then move on to the next step. Let your animal tell you about your plan!Let your animal tell you about your plan!
Some of the most annoying hours in my school career were spent listening to an instructor drone interminably through explanations of a concept I already understood. Remaining at a low level of criteria or performance can be intensely frustrating to a learner. In Don't Shoot the Dog, Karen Pryor points out that failure to move ahead when the student is ready can be just as aversive as any punishment.
Likewise, when your student seems to leap ahead of your training plan, don't feel that you must go back and take each step in turn. If the dog is ready to jump the high bar and clearly enjoys doing it, you may be punishing him if you keep insisting he walk over the low bar.
Having a training plan is very handy in this situation as well. When the dog makes a great leap forward, you're ready for him. You'll avoid those awkward moments of, "Wow! He's doing it! And now what?" followed by a de-motivating pause as you decide what to click next.
Fluency in training plans
As with all skills, a trainer becomes more fluent in training plans the more often the plans are used. I do not write out a formal plan for each behavior I train, but I do have a series of steps in mind as I'm working. If someone asks in the middle of a session, I can list my last several criteria steps and those next several I'm working toward. I can spell out my exact criteria and explain why I've chosen it.
For complex, novel, or important behaviors, I create a formal plan (refer to the sidebar for one example). There is no substitute for a good training plan. For common behaviors, I have a training recipe that, like most of my kitchen recipes, will vary according to what I have available. Settling on a mat, for example, looks very similar each time I do it, though each dog brings his own individual proclivities. I've trained that behavior so often that I hardly have to think about my plan anymore, even while adjusting on the fly. Fluency works at both ends of the leash!
Success, with the support of a plan
Remember the training partner who felt intimidated at the thought of writing a training plan? She recognized her problem and reported the following to me, "Muddling allows me to wobble on my criteria without consistency. A training plan holds me accountable and gives me clear guidance."
With that guidance, the benefits of training with a plan are clear: simpler steps, faster progress, and more reinforcing results! Good luck with all of your plans this year.
How to Write a Training Plan—Part Two
Very good article. It got me to write a training plan for my dog. I am trying to teach a very social lab to walk past another dog. Not an easy task. But this article gave me an idea of how to do it.
Excellent! Happy training!
Excellent! Happy training!
Post new comment