Copyright 2000-2005 by Carol Whitney and Robert E. Bailey.
Thanks to Bob Bailey for permission to post this article.
My original post is in a normal font. Bob Bailey's additions are in italics. I have removed quote marks from the original post, to make for easier reading.
I am including a few headings in addition to those in the original message, to make this article easier to read.
From: "Bob & Marian Bailey" <email@example.com>
Date: Wed, 16 May 2001 16:40:28 -0500
Subject: Re: [CS] Bailey Basics
Carol does pretty well with this. I would have let things go myself, but I have had enough private messages about "fine" points of what Carol calls The Bailey Basics, I guess I have to make a few comments.
Please, I am not fussing at Carol. She says things very well, and she has been kind in her words directed towards Marian and myself.
I will add to Carol's statements.
----- Original Message -----
Subject: [CS] Bailey Basics
The Bailey Basics, and Speed Trials, or Short-Burst Training Periods
Carol Whitney, Ph.D.
last revised Mon 11-06-2000 18:33
This article presents my personal understanding and interpretation of suggestions made by animal trainers Bob and Marian Bailey, Hot Springs, Arkansas (USA), in posts to email lists. Any errors in interpretation or paraphrasing are mine.
I'm including here what I call "The Bailey Basics" (this term has not been approved by the Baileys, but they haven't reprimanded me [yet] for using it).
A trial is one individual response (from the animal), to a cue or prompt from the trainer or from the environment.
A session consists of a single series of trials (repetitions, or repetitions of efforts), without a break.
A training period consists of the time you spend training the animal in a single series of trials and sessions.
Commonly the trainer takes a short break between sessions, during which s/he may consider any possible improvements in timing, or in setting criteria, and may make notes on the session just completed.
In our class we teach that we end a session or period by planning the next session or period. So, we suggest recording what just happened and plan what we are going to do, and then we are finished with the session or period.
It must be worthwhile for the animal to play your (the trainer's) silly little game.
I know, I know, to us (the trainer) the "game" may be important business, and we usually think of it that way. That might suggest to us that the animal should be concerned about it, or impressed with its importance, too (we all tend to be a little too impressed with our importance, don't you think <smile>?)
Well, with few exceptions, most of what we do with dogs has most probably little to do with how "dogs of old" (that is in the evolutionary history of dogs) earned their livings - that is, chasing things and killing and eating. So, to put things into some sort of perspective for our students, I often refer to humans and their "silly little games," and it is up to the human to make it (the game) so worthwhile that the dog will play, and play enthusiastically.
By the way, this is true of chickens too. My guess is that a million or so years ago, there were very few chickens running around pecking at circles, triangles, and squares, as they do in our class. However, our chickens do so enthusiastically (as our students will attest), at least most of the time (as long as the rate of reinforcement has been high enough).
During training, the Baileys recommend using continual reinforcement; that is, the trainer clicks and treats the trials that meet the trainer's current criteria.
Bob and Marian Bailey call trying to teach too much all at once, "lumping." They recommend we "split" behaviors into their smallest useful increments. "Splitting" lets us work with one bit of behavior at a time, and also helps speed up the responses, so we can get a high rate of reinforcement going.
We teach splitting rather than lumping, and we practice splitting, but with opportunistic lumping. If an animal gives us the entire behavior, we won't send it back - we'll take it. We don't necessarily change our criteria until the animal shows us that we can get more, but we will change and expect more if the animal proves that it can give us more.
Rate of reinforcement needs to be kept high, such as clicking and treating once every two to three seconds, so neither trainer nor animal loses concentration. If this rate is impossible to maintain, perhaps the criteria for the behavior in question need to be re-examined, and the behavior broken into smaller bits.
Short-burst, or speed, trials, are a very helpful design for training an animal to give small, specific and precise responses.
Both trainer and animal can then focus on one particular response. With speed trials, keeping records becomes manageable, the bits of behavior we work with become more easily observable, since what we click is what we get, and we can track performance, both ours and the animal's, to help us set criteria for success.
A typical schedule for a training period in which I use speed trials is:
You can count the number of trials, or you can set a timer and end the session, say, after 30 seconds or one minute.
If you hold ten treats, and do ten trials, you can tell how many trials were successful by counting the number of treats you have left of the ten you started with.
Remember that if you click by mistake, you feed the treat anyway; otherwise, the power of the click-sound to the dog begins to be devalued.
We require students to treat for every click, and coaches must remind the errant student when they renege on the "contract:" one click; one treat.
However! In the real world (while training), we don't suggest that the trainer do this. The first thing that a trainer should do when he or she makes a glaring and serious error of judgment in a real training situation (not in our class) and clicks for an unwanted behavior, is to go to the nearest hard (brick or the equivalent) wall, bang his or her head three times, and then continue training. If there are more than a few such clicks in a day, then that trainer should review such factors as concentration, planning, technique, or even commitment to being a good trainer.
In our classroom situation, we demand that each click be treated because we want the student to feel the full impact, the full anguish, and reap the full "reward" of his or her error - the chicken gets to eat for its mistake, while the trainer must look on while the chicken eats for its mistake (and for the trainer's too <smile>). We don't put E-collars on our student-trainers, but the howls new students give when forced to feed for a poor or unintended click would make you think we had just hit them with at least a thousand volts. This has the additional benefit of making the student appreciate more fully the value of the food as reinforcer, because sometimes the trainer sees the delivery of the food definitely increase the rate of a given (unwanted, but reinforced) behavior.
After one session (set of trials), take a short break. You can make notes on how many of the trials were successful, and think about criteria for the next session. If the animal succeeded in eight out of ten trials (about 80%), you could try one more session without raising criteria, or you might change one criterion for the next session.
The most effective training requires changing only one criterion at a time. Notes are important, to keep track of progress, and to help in planning what to do next.
One training period consisting of speed trials should probably consist of about three sessions. Two or four are okay, or you can use just one as a kind of refresher, or as a way of keying up, or tuning up, animal and trainer together.
When doing speed trials, learning is most effective when the training period is kept short. I aim for no longer than three minutes for the entire training period.
I don't believe in just saying these things; I have to try them, and test them for myself. In the few years I've been doing this, I've found these training methods to work as the Baileys suggest. As Karen Pryor points out, it's not only the animal who learns; the trainer learns just as much.
Setting criteria so the animal can succeed easily (eight out of ten times) is essential, to make these sessions fun for both trainer and animal. If sessions begin to falter, check criteria, make sure you're clicking on time, and make sure you're keeping the rate of reinforcement high.
In my personal experience, speed trials, or short-burst trials, help both trainer and animal focus on each other and tune up the activity they do together. Because they work with bits of behavior, some precision can be achieved.
It takes thought and planning to make speed trials work well. Usually, both trainer and animal enjoy the sessions. Both trainer and animal develop increased precision, and the sessions are generally emotionally satisfying for both, also.
Speed trials give both trainer and animal experience with success, which is very gratifying and motivating to both. The resulting motivation easily carries over to other, more complex training exercises.
--Carol Whitney and Bob Bailey
Clicking on the "Original Message*" link will take you directly to the original message under the following conditions:
1) you are a member of the group in which the message was posted
2) you are already signed in to Yahoo
3) the Yahoo server on which the message is located is operating
If you are not signed in to Yahoo at the time you click on "Original Message*," Yahoo will prompt you to sign in, but then, it may, obstreperously, lose track of where you came from, which nullifies the effect of the link you clicked on to see the original message.
To get around that little problem, after you sign in to Yahoo, you may use your browser's Back button to return to the page where you clicked on "Original Message*," and click there again.
That procedure should take you to the original message within the group.
However, if Yahoo happens to be working on the server where the message is located, you will not gain access to the message. Just try again later.
Afterwards, if you wish to return to Coherent Dog again, you can again use your Back button, or, if you are using a tabbed browser, you may go through these procedures by opening different tabs for each different page.
Last modified on
Sunday, 26-Oct-2008 09:44:49 PDT
It is now
Wednesday, 26-Apr-2017 06:37:15 PDT