Category: Theory, Terms & Abbreviations, Crossing Over ¤
  Author: Carol Dunster ¤
   Title: Command vs Cue and Other Definitions ¤

I can see that you are a thinking trainer with a strong background in
more traditional obedience. This is good and now I would like to give
you some terminology that way it is generally used when using Operant
Conditioning (OC) and clicker training. This will help our
communications so we can talk and understand each other and,
hopefully, training better!

>I good way to combine these two approaches is to give a simple command when
>you see your dog about to get distracted by a potentially aggressive
>situation. 

When we are training with OC, we make a distinction between "command"
and "cue". A command is an order, a "do this or else" statement. The
"or else" might be quite mild, but inherent in the command based
training is that we will *make* the dog do the command. With clicker
training we are using "cues". A cue is when you let the dog know that
it will be reinforced for doing the behavior that is cued and only
that behavior. It is our job to find a reinforcer (or patterned
history of reinforcers) that is strong enough to motivate the dog to
want to do the behavior. 

Keep in mind that in some ways it is very like traditional obedience.
We start in a low distraction area with simple behaviors and work
until they become a habit. (I say "sit" the dogs sits, it does not
stop to decide if sitting is more fun than standing or sniffing.) Then
we work in a higher distraction environment until the dog is reliable.

Using this method, it would be quite likely and useful that the dog
could be trained to sit on cue reliably enough that he would do it
when he sights another dog (before he is really wound up) and get a
c/t or jackpot for doing this. It would work well if you train the dog
to sit reliably first, pick a quiet place and have a friend with a dog
under control, so you can pick the distance that your dog will
comfortably work and gradually decrease it.

> When my mix was younger and more inclined to nerves and
>bristling, when I took him for a walk and he encountered something arousing I
>would tell him to sit or lie down.  If he obeyed he was praised and often
>treated for it.  If he did not, I reinforced the command in the usual manner.

When we talk of reinforcers, we have a specific meaning. A reinforcer
is a consequence of the dog's behavior that will cause the dog to
*repeat* the behavior. If the dog obeys, it is reinforced. If the dog
does not sit when cued, the trainer has to think about why - is the
trainer asking too much from the dog? Does the dog understand what is
wanted? Does the trainer need to back up a step, lower  the criteria
and reinforce more? The trainer may be asking the dog to sit in too
distracting a situation and will have to back up to a more calm place
and retrain a bit. So, if the dog does not sit, you do not want to
reinforce it (which should technically encourage the dog to *not* sit
when asked). These distinctions can be very difficult for a
traditional trainer to grasp - particularly if you have been a
successful trainer... <gr>

You would have used a correction, which functions as a punisher. A
punisher is a result of a behavior that causes the behavior to be
diminished (we sometimes call it a diminisher). It is difficult to
time a punisher to clearly communicate to the dog *what* is to be
diminished. (In this case the action of not sitting when cued.) It
might also be difficult for the dog to understand it is being punished
for a lack of action. They eventually get it (with much trial and
error), but they lose some trust in their handlers and learn not to do
anything without permission. Some trainer like that. Most of us are
here because we want a joyful, fast learning dog that trusts us and
works with us. If a trainer gets this using more traditional methods
(and they can with a good dog and excellent timing), they might get
even more even faster with clicker training.

> In either case I paid little mind to the distraction itself.  If he was
>corrected for the down, I would then praise him when he was in the down
>position.  No matter which path was taken (obedience and praise, or
>disobedience, correction and praise) I would inevitably get my dog's
>attention, and then I would have the attention to praise by throwing in a
>complementary "good watch!"  

This is a method that would work for many people and you are obviously
someone with good timing and understanding of how to use it in the
correction based (or command based) training. However, when using the
clicker, you would pay some attention to where the distraction is and
train far enough away that your dog will still respond without being
corrected. Then you c/t (click, then treat) when the dog looks at you,
or offers other acceptable behaviors (including the sit), until the
dog is offering whatever behavior you choose (with no cue or command!)
regularly. Then you tell the dog it is sitting when it is about to sit
for the c/t anyway. Pretty soon the dog figures out that it will get a
c/t when it hears "sit" and does it... Then it has learned the cue.

You would go a bit closer and start over - c/t when the dog sits (no
cue) until the dog is more interested in sitting than lunging at the
other dog. Then add the cue back in and, when the dog is reliable, go
a bit closer or work in a different place. This sounds round about and
slow, but actually, as the dog figures this out, it will learn more
quickly and be more interested in you as the handler than the
distraction. The dog soon will learn that distractions are a chance to
get wonderful goodies and will become overall less aggressive than if
the distractions are a reason to get corrected...

>Positive reinforcement was a very powerful tool for him to overcome his
>tendencies.  If you must "punish" (no, that's not technically the right term,
>is it?) then be prepared to praise after.  In obedience it is called a
>"correction" -- and the dog should certainly be told when his behavior has
>been suitably "corrected" and is now appropriate, so he will know how to
>respond better the next time. 

I think I explained how this works above. If you have any questions,
please email me.

> In either case remain calm and in control.
> Commands are commands, corrections are corrections.  Give a command as a
>command and not as a threat -- in other words, in as normal a tone as you can
>manage. 

Actually in clicker training, commands are a threat (you will *have*
to do this) and are not used. Corrections are punishers (diminish
behaviors) and positive reinforcers are used to motivate the dog, with
a conditioned reinforcer (the click, which is conditioned to be
reinforcing to the dog with the treats) to communicate more clearly
exactly what is desired.

>Let the correction be swift, effective, but by no means punishing
> (in other words, get the dog down or sitting but don't shake  his ruff doing
>it).  

By their very nature, corrections tend to be punishing (ie diminishing
behaviors), however I commend you on wanting to avoid abusive ones...
I love forward to continuing our conversation on this subject!

Carol D.
Patchwork English Setters and 
Carwyn Silky Terriers