I see these debates pop up again and again on lists... VSR vs. continuous, click/reinforce vs. clicking w/o additional reinforcement, laboratory vs. real world, an so on. Ultimately nothing is ever decided because each side has points in its favor. Every method has a valid use.
What are your goals? The method you use to get a reliable recall in a house dog may or may not be the same method you use to get a recall in the ring, may or may not be the same method you use to get a perfect 200 in the ring, may or may not be the same method you would use with your service dog, and may or may not be the same method you would use to get a behavior in a bomb-detection dog where your very life is on the line.
So when you start evaluating the information out there, first define for yourself exactly what your goals are. Strong behavior? Reliable behavior? Precise behavior? How strong, how reliable, how precise? There's a difference between good-mannered house dogs, dogs who score 170 in the obedience, dogs who score 200 in the ring, and dogs who day after day perform life-or-death tasks. What are *your* goals?
Now look at the goals of the people who are providing the information. Differing goals doesn't automatically mean that you can't learn from someone -- or even that their solution isn't the best for you. But weigh the different options. Find people who are doing what you want and see what they're doing. Then evaluate what they're doing and see if it appeals to you. If there's something that doesn't sit right with you, don't do it! If you think something is too complex for you to do correctly, don't do it.
Be as objective as you can when weighing other people's solutions. If you're looking for a well-mannered pet or just want to have fun and don't care about scores, you may want strictly "easy and fun." But if you want super high scores or need to produce uber-working dogs, then you have to be more discriminating. Here's a huge question to ask:
HOW DO YOU KNOW THIS METHOD IS BETTER?
I like what I'm doing now. I'm positive it's better than what I did with Rain. How do I know? Well, Pax learned faster. How much faster? Well, let me think. How many sessions did it take Rain to learn? How many reps? What kind of reinforcement schedule -- and what quality of reinforcement? What were the criteria I used for each? Did I do any lumping with one or the other? Did I have to go back and redo something? Did I shape or lure or capture or model?
You know what? If I don't have data -- cold, hard data -- then, frankly, you can take my opinion as just that. Opinion. A nice anecdote. Of course I like what I'm doing now. If I didn't I wouldn't be doing it. Without data, I may believe it's faster or better, but I don't know how or why or even if it's a true assumption. Also, without data, I cannot draw the conclusion that what I did then or now is efficient training worth emulating. So it works. Big freaking deal. Lots of things work. What works best?
I personally choose to follow the teachings of the Baileys. I chose them for reasons that may or may not be related to your training goals.
* The skills they teach have proven to work in all the species and with all of the individual animals they attempted to train -- over 140 different species, 15000 animals. I run a mailing list that is not restricted to dogs. I want to recommend techniques that will work with any animal a person may inquire about training.
* They were first scientists and then business people. They kept data and endeavored to objectively find the most efficient method for training and maintaining exactly the behavior(s) their clients required. They stuck primarily to R+ and extinction for a reason -- because although aversives could give them a faster result, they spent more time fixing the fallout. I happen to choose R+ and extinction primarily for ethical reasons. However their experience gives me the confidence to stick to my beliefs and to recommend positive solutions to people who just want to solve problem behavior.
* Data. They always kept data. My goal is efficiency. Data enables me to draw conclusions and effectively problem-solve. I get lazy about it -- oh, yeah, I definitely get lazy about it. And when I do, and a problem crops up (and I get tired of screwing around), I go back to it, and I resolve the problem quickly and efficiently. Every. Single. Time. (You'd think I'd learn to maintain the recordkeeping, wouldn't you?)
* They trained for the highest level of precision and reliability. As an analogy, in their world, if one of their dogs didn't get a perfect 200, the dog and handler DIED. <grin> I'm *soooo* not planning to train my dogs to that level. But I figure if their method will get me to that level, then it'll definitely get me anywhere else I want to go.
* According to Bob -- and no, I don't know if this is from data they crunched (Ask him!) -- 97% of training problems can be traced back to one (or more) of these three problems: timing, rate of reinforcement, and ability to set/evaluate criteria. Based on that statement, before I recommend complex solutions or start playing with schedule of reinforcement or other advanced technique, I go back and examine those things. I also use that as a guideline when helping students or members of my list problem-solve. It hasn't failed me yet.
* After all their experience with all different situations and all different species and all different individuals, they came to an interesting conclusion: Simple Works. There were times that they needed NRMs and VSRs and other complex training techniques. And when they did, they used them. But that's the ONLY time they used them. This is important to me because the members of my mailing list vary in skill, dedication, experience, and ability. Even if they have lofty goals, I cannot guarantee that they will be able to apply complex techniques correctly. So I default to simple solutions, because simple solutions work.
Those are some of MY reasons. Your reasons, your goals, your experience may be COMPLETELY different. And that's okay. The choices I make don't invalidate your choices, and the choices you make don't invalidate mine.
If you're trying to evaluate a method, I suggest this...
Long- and short-term data. Watch your dog. Always, always watch your dog. If he's not having fun, then who cares what the data says? If you want precise or strong or reliable, define those objectively, and then keep data until you get there. Evaluate as you go. Don't waste time with things that don't work for YOU and YOUR DOG.
Just make it objective, test your hypothesis, and TRUST THE DATA, not the opinions and anecdotes. Even mine. ;-)
Copyright of all posts is the property of the original author. Please obtain permission from the original author before copying, quoting, or forwarding.
List and Site Owner: Melissa Alexander, mca @ clickersolutions.com