The following web page is from the late Dr. Cathy Toft. Cathy was one of the brightest people I've ever met. We shared an online-only relationship for about a decade before her life was cut short by cancer, ironically a few months before we were supposed to meet face-to-face for the first time.  This page is the one I referred the most people to, and I requested and received permission from her sister Nancee to duplicate it here because it's not clear how long Cathy's pages will be maintained. The Toft family can revoke this privilege at any time, and I will remove the page if they so request.There is a lot of other good information on her page, you can use this link to go there until it goes away.
Read this page very carefully, it explains the words used by dog trainers who use operant conditioning, nearly all of them. A trainer who claims to be R+/P- only cannot use a choke chain because a choke chain is P+. A trainer who claims to be rewards-only based probably (but not certainly) does not know what they are talking about because they are probably using P-. This lexicon does not favor one means of training over another; it merely provides a common terminology so trainers can concisely explain their philosophies. In my opinion anyone paid to train dogs who is not intimately familiar with these terms is a charlatan.

Aside from the comments at the bottom Cathy's descriptions are completely neutral.

R+ Dog Training

 

A brief lexicon of behavioral terms used under this training method:


TERM

DEFINITION

Reinforcement

Anything that strengthens a behavior

Punishment

Anything that suppresses a behavior

Positive

Something added to influence a behavior

Negative

Something taken away to influence a behavior

 

Application of these terms:

TRAINING APPLICATION

WHAT HAPPENS

To do this:

Positive Reinforcement  R+

Something is added that strengthens a behavior

the something is good or desired by the subject

Negative Reinforcement R-

Something is taken away that strengthens a behavior

the something is bad or aversive to the subject

Positive Punishment     P+

Something is added that suppresses a behavior

the something is bad or aversive to the subject

Negative Punishment    P-

Something is taken away that suppresses a behavior

the something is good or desired by the subject

Extinction                    E

A behavior goes away because it is not reinforced

nothing happens to the subject, good or bad

 


Summary of dog training realms:

TERM

DEFINITION

"Traditional" training 

Works primarily in the R-, P+ realm

Positive traditional training*

Mixes rewards and lures with R-, P+

So-called "clicker" training

Works primarily in the R+, P-, E realm

Crossover trainer

PTT or a traditional trainer who switched to clicker training

Catherine Toft 2007


Examples -- applying operant training:

 



Training the sit

Training the recall

Positive reinforcement R+

Dog sits (either on cue or in free shaping with no cue)
Trainer then gives the dog a treat

Dog comes (cue or no cue)
Trainer then gives the dog a treat

R+ with event marker

The dog sits (cue or no cue).
The instant the dog's fanny hits the ground,
the trainer clicks.
Trainer then gives the dog a treat.

Trainer cues the dog to come.
The instant the dog turns toward the trainer,
the trainer clicks.**
When the dog comes in, it gets a treat.

Luring (non-operant)

The trainer pulls out a treat and
holds it high above the dog's head.
In trying to reach the treat, the dog rocks back and sits.
The dog then gets to eat the treat

The dog is running away or chasing a cat.
The trainer gets out a treat and yells
"Lassie, Cookie!!"
Dog comes in and gets the treat (you hope).

Negative punishment P-

Trainer cues the dog to sit, but the dog stays standing.
Trainer leaves the room with treats in hand.
Dog loses the opportunity to be rewarded.

Trainer cues the dog to come.  Dog still chases cat.
Trainer catches dog and removes the dog from the presence of the cat and removes its ability to run and play.

Negative reinforcement* R-/P+

Trainer commands the dog to sit.
If the dog does not quickly comply, the trainer jerks sharply and upward on the leash.  The choke or pinch collar tightens and the dog feels uncomfortable. The force of the jerk upward makes sitting an easy choice for the dog.
The pressure is then relieved. 

Dog is running away or chasing a cat.
Trainer commands the dog to come.
If the dog does not comply the trainer closes
the button on the flexilead or steps on the long line.
The dog experiences a sharp jerk and turns toward the handler, relieving the pressure from the collar and leash.
The handler then praises the dog as it comes in.**

Physical "modelling"

(non-operant)

Trainer pushes dog's behind downward.
Under this force, the dog sits.
Trainer rewards or praises dog.

Dog is out running after a cat.
Trainer reels the dog in with the long line,
saying "come" as she does so, then rewards or praises.

 

*In practice there is no distinction between negative reinforcement and positive punishment, which is not the case for R+ and P-.  In  another example, the dog is pulling on the leash.  As the dog pulls, the trainer pops the leash sharply and repeatedly for as long as the dog is pulling, causing the dog discomfort while it is pulling.  As soon as the dog stops pulling and thus starts to walk on a loose leash, the trainer stops the jerking.  Positive punishment happened when the discomfort of the jerking and tightening the collar suppressed the pulling.  Negative reinforcement happened when taking away the jerking strengthened the behavior of walking on a loose leash. 

 

** In each of these examples, there is a secondary or conditioned reinforcer.  In the R+ example, the clicker (or other event marker)  is a conditioned reinforcer meaning that food or play is coming.  In the R- example, the praise is a conditioned reinforcer meaning that no discomfort will be applied while the praise is occurring.

 

Other notes:

1.  The word "positive" is unfortunately used in two completely distinct senses in animal training.  The first use comes directly from plain English usage, that is, "positive" means something good or desirable.  The second use is a formal, strict definition in operant behavior analysis and means something "added to".  If that something added is meant to suppress a behavior, as in positive punishment, then it is anything but good or desirable

.

2.  "Positive" dog training can therefore mean a variety of things.  Just because someone says they are "positive" does not mean that aversives are not also used.  In fact, "positive" traditional training is a particular mix of the use of "rewards" or things that the dog desires as lures or mood changers (and not as operant conditioned reinforcers) and corrections.  Traditional dog training does use operant conditioning, but specifically that using aversives or corrections, such as leash pops, verbal reprimand, intimidation, or electronic stimuli.  In other words, traditional dog training is a mixture of operant and non-operant methods, and of positive and aversive elements.

 

If food or play is used in a non-operant fashion, i.e., as a lure, it is extremely difficult to fade out.  For this reason, the mix of reward and punishment is ensured, because the operant conditioning methods of R-/P+ are required to strengthen desired behaviors.

 

However, the price that this blend of methods exacts is high.  The dog may perform well and reliably, but the dog's attitude is often not the same as a dog trained with R+/P-.  In fact, many breeds of dogs and many individuals in any breed maybe passed off as untrainable with R-/P+ methods,  written off as "stubborn", "stupid", "willful", "has own agenda", "soft", and so on. 

 

Karen Pryor has written eloquently on this topic.  Her article, Poisoning the Cue [Teaching Dogs, 2002, vol.1(1)], is the only place I know of where the concepts of mixing positive and negative "discriminative stimuli" are explained.