Time involved for my records isn't that much - or let me put it a different way. I reuse the time I *used* to spend in dead ends trying to figure out why I was getting what I was getting now planning my session and observing the dog.
It has become so much simpler and I am able in great detail to describe *exactly* to another person what behavior the dog is doing.
I vividly recall one of the first days with Bob and Marian, after our initial session with chickens, where he asked us to describe pecking behavior. And the comments were that the chicken appeared distracted, a bit nervous, that it didn't peck much, that it hopped on the floor once. Maybe that a chicken wanted to be alpha <g> The chicken pecked a lot at things other than what it was supposed to peck with. All sound like good observations?
And then he pointed out that none of these were able to be measured to provide progress detail. Chicken is distracted? Did your partner think the chicken was distracted? How can you tell if the chicken is less or more distracted than another chicken? Now what if I tell you that the chicken looked at the door for two seconds and physically oriented to the door while it was open? Now can you exactly see the behavior and see how to measure improvement? And believe me, the vast majority of behavior that detriorates, the animal is telling you strongly before you get there that it is not sure of the behavior - in your terms, the dog just lost the behavior.
So now let's go back and look at notes - and make those notes as specific as for planning purposes and only for human purposes, what are you trying to accomplish? Maybe the goal is Maggie barking at the fence only from the porch? Maybe Maggie barking for ten seconds and no longer? Maybe Maggie alerting, no time constraint, but able to reorient toward the house from the fence? These are all different criteria but do become very important in how you are going to address the response of the dog.
And then I label in my planning notes, for only a few steps, since the dog may have better ideas <g>, what I think the first steps are.
If I am clicking, I try to make sure that I pinpont what I *think* I need to click so it fits in with where I want reinforcement to occur and that time-wise, the dog is spending a lot of time doing things related to what I want him to do and less time involved in things that are not part of the task - so the example of Maggie and the fence, I would look at different criteria and ask myself which of the ways involves Maggie for the greatest amount of time doing the *right* things.
All of which comes to the point that practice makes perfect and I rarely spend more than five minutes doing this - but if you don't practice, it takes more time up front planning but less confusion time for the dog. Or as Saint Bob says, *never* back up.
Hmmm, there are ad hoc session and actual specific response sessions. And I don't use a clicker for the majority of them. An ad hoc session may be what Carol does a lot of - you know that sometime during the next 30 minutes, the dog might be faced with a choice you have been working on. A bird explodes under their face, they start to lunge, check themselves, don't pull, you use a clicker word or click. IOW, you are *always* looking for the opportunity to reinforce your dog for reducing what you don't want and increasing what you want.
So for my dogs, who like to fence run, I may let them run and then interrupt their behavior. We practice it so they practice reducing their own arousal - learning self control. Or one of my dogs is bad about jumping on kids, so we practice sitting when we meet kids. But somewhere in my master list for that dog, I have listed things that I would like to change about that dog, including laying down at my feet when I eat. In these cases, if I haven't listed what I want the dog to do, I fail to recognize it when I do get it and so the dog never changes behavior.
But I don't count these as training sessions, I call these "honey, we'll both be happier if you'll do *this*" sessions. I don't keep records on these but I *do* right a behavior list so I reward correct behavior and don't inadvertantly reward incorrect behavior.
And then there are real training sessions, generally only five minutes long or so. With a very specific criteria that is one step in a series toward an overal trained response - like teaching a bow. Or, for my older dog, teaching the specific response of meeting a child and sitting close to, but not touching, facing me instead of the child, to allow petting. This was a trained response, rather than a "caught" response. This response involved paying kids in the neighborhood to follow my directions, short sessions in different places, different kids, different devices (water, bicycles, skateboards, hats, bags). Each session might present several criteria shift - different places in the neighborhood as well as different children - or the same criteria - the same corner but different children. This dog was the one able to maintain low reinforcement rate and able to generalize rapidly. This dog did have fairly detailed records.
The *only* reason for a record is
1) to improve your ability to keep records important to you
2) minimize "eddy" time when actually working with the dog when the dog is confused and you are getting consistent behavior but you don't understand how you got there or how to get out of it
I am currently keeping detailed records on only a couple of things for each dog - for the things I am requiring exact timing of reinforcement or actually clicking on. I have a "cute trick" for each dog that I am working on, a performance related activity, and an activity that is the same for each dog to compare rates and my skills in training.
No, the performance related behaviors are just pinpointing for the dog within a range of "sloppy" responses, which exact response is a better choice. In this case, they are non-discrete behaviors and deal with faster, tighter, closer, further. They are behaviors I have stopped clicking on (put on variable reward) but have placed back on continuous each time the dog meets the new, higher criteria.
Umm, sorry but my dog is better at 101 things than I am since I don't remember what they've done (and they do), and fail to click when they think they should be clicked and click when I shouldn't.
So, no, this is changing a sit you've been happy with for three years to a different sit, or for my own purpose going to a known response and actually measuring how fast the dog sits, what the rate of reinforcement I am capable of, improving my timing and watching how the response of the dog changes to improved timing.
Oh, and format may change between people. I like neat notebooks, then found I wouldn't keep sloppy notes in a notebook so changed to scraps of napkin and hensratches, shorthand, marks on pieces of paper, whatever. And I review the session and translate hatch marks into total responses, correct responses, rate, time elapsed (and so time per response, sometimes the only measure of improvement). Do I want to continue - am I where I think I should be? Do I break the session and transfer to by permanent book because I'm seeing something I didn't expect and need to think about it.
But for a spontaneous session, like what? If the goal is a straighter sit, and it's spontaneous, then I would point out that's like telling me your chicken is less distracted. Would I be able to look at your dog and tell that you got a straighter sit? Looked straighter to you but the same to me? So I never practice a straighter sit without something to measure.
A faster sit? FME, my dogs are better able to judge time than I am - although other trainers may be better. So if I am improving time, there is my stopwatch in my hand and that makes it not spontaneous.
But a spontaneous session may be going out to the mailbox and rattling it and having the dogs charge toward the window and bark for - count to yourself the seconds, observe how long before the dogs stop barking and get their feet off the sill - two behaviors, two dogs. I may file it in my brain to write down later that day or the next day, when I think about it. I may forget it. If I just recall to write it down 80% of the time, I still see a trend of behavior. I am also able to note if I must politely beg to disagree with people that if I "praise" my dog while they barking at me at the mailbox, it does or does not increase the behavior in that dog.
Oh, no, if you're taking any notes at all, that's great. And notes must be specific to what encourages you to do it more - so lots of notes suppress that behavior <g> Notes should also improve your skill at observing behavior or modifying behavior - and so improve your skill. All other things aside, my stand-by phrase is, if in doubt, ask the dog.
Great, so there are really only three things to keep in mind - timing, criteria, and rate of reinforcement. And lo, the greatest of all is get the behavior. Criteria - what *exactly* do you want the dog to do. Where are you going to click? How are you going to measure improvement? Time, distance, correct/incorrect? Is that measurement able to be observed by another human for judging whether you clicked the correct response by the dog? How are you going to establish the environment to get the behavior? How are you going to raise the criteria - which ones and how soon? How are you going to tell if the dog is responding to the change in criteria indicating that your criteria change was too much or what you thought he knew wasn't really what he knew?
Your records should provide this type of information.
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