An e-mail exchange between Karen Pryor and Richard Orser:
Richard: If you wouldn't mind, I would like to ask a question: If I am reinforcing several behaviors (by clicking/treat) e.g., sit, sit/stay, & come, I am making these behaviors more likely to occur when I am around the puppy (10 weeks old).
Karen Pryor: Yes.
Richard: But, then I have to further shape each behavior to occur only when I present a certain hand or verbal signal (a different one for each behavior).
Karen Pryor: Yes.
Richard: OK, finally to the question: Is it better to complete one behavior at a time right through to the behavior being elicited by the signal or to work on all simultaneously (let's say one behavior per training session, but in parallel).
Karen Pryor: Good question! Here's my answer. When you are SHAPING behavior you can work on all kinds of different behaviors in the same training session and in the same day, in parallel or alternating.
However, when you are buckling down to add a CUE you should work on just that one behavior and that one cue, in a given training session. And it might be just as well to stick with just that behavior for as long as it takes for the dog to learn that cue. When the cue seems secure, then you can go back to training lots of different things more or less at once.
The cue might be built very rapidly, with complete retention, in just a few clicks; I once saw a four month old Labrador learn its first four behaviors and four cues all in one session. Usually however some repeat experiences are needed, to truly establish a discrimination. For example you may have to reshape correct response to the cue in various new environments—on wet grass, among crowds, or whatever. I would want to work just on that cue and its behavior, during a given session, each time I raised criteria in this fashion.
I think the issue here is that when you are shaping behavior, any new effort might be rewarded, so you can be very free and click all kinds of behaviors, knowing that they will all increase over time, with each click, even if sometimes one kind of thing gets clicked and sometimes another.
When you are establishing a cue, however, you are actually training a discrimination. "Do X when you see/hear cue for X, and at no other time." So you have to be on a fixed schedule of 1:1 or you will risk confusion. Thus, during that period of establishing a cue, it might be best to work on one behavior (and its cue) only. If you want to change behaviors and don't want to stop training, break the session in half. Finish the series with a bit of play. Then move to another part of the room or the house to start work on a different behavior or cue.
In practice I am finding one exception. If you are training a pair of behaviors, that is, two exactly opposite behaviors, it is very efficient to train the matching cues in exact alternation. For example, bark/silence bark/silence. I use one hand signal (pointing at the dog's nose) for "bark" and another (a "stop!" signal with the palm of the hand facing the dog) for silence. Cue and click both the bark AND the silence in rapid alternation. This guarantees the very high rate of reinforcement that is so conducive to rapid learning, and you get two cues for the price of one.
In any case, as you have probably found already with your puppy, once the dog has learned the cues for three behaviors, subsequent cues can sometimes be acquired in just a few trials, since the dog "gets the concept" that cues open the door to specific paths to reinforcement.
Richard: Now that I'm going, I'll venture another query. Are there standard, more effective hand signals that would be good for me to adopt, for example because that's what we'll use when we start Puppy Kindergarten Training Classes?
Karen Pryor: There are standard hand signals in advanced obedience training (utility) but I doubt that anyone thinks about them in puppy kindergarten You are free to develop and use your own hand signals. Then if you need to change a signal you can always add a new one by just giving it before you give the old signal, then fading the old signal.
Richard: Is it more effective to use silent hand signals or verbal commands, or *both* together?
Karen Pryor: Dogs seem to learn hand signals and body movements more easily than spoken words. Most puppy trainers however seem to use voice commands rather than hand signals; maybe they are easier to teach to the people! If I were to make a recommendation, I'd start with a hand signal for each behavior and trick, and then I'd add the word by using it just before the hand signal until the dog starts to recognize it; then I'd use one or the other, depending on what was convenient. It's nice if your dog can respond equally well to both a spoken cue and a silent gesture. Giving both at once doesn't make the cue any stronger, if the animal understands both. (It might SEEM more effective in cases where the animal really only understands one, but the handler doesn't know which one.)