Sunday, June 19, 2011

Training perspectives

Everyone who trains a dog knows the old joke "The only thing two dog trainers can agree on is the third one is wrong". Stupid but true. Dog Trainers come in all flavors. Some are so sugary sweet you want to puke, while others are so bitter you want to puke as well. But every once in a while you get those that are just right. Those that do not train a specific "Method" but rather read the dog and the situation and use what is appropriate. These are the trainers I prefer to work with and learn from.

Everyone thinks I am against e-collars. I'm not. There are valuable uses for them when used correctly. The problem is that most handlers and even so-called E-collar experts do not know what they are doing. Some even propose that E-collars are simply a form of positive re-enforcement. WTF? Are you kidding me?

Obviously, there is a lack of understanding of scientific terminology of +R. So lets go over this to eliminate any confusion out there. "The Principles of Learning and Behavior" by Domjan is a text book used by Universities to instruct the principles by which animals (people and dog included) learn. Classical conditioning, operant conditioning, association is association whether it is intentionally manipulated or occurs spontaneously by chance. In dog training, both occur and sometimes those spontaneous learning moments are significant, both good as well as detrimental.

Positive Reinforcement occurs when a response produces an appetitive stimulus. An example of this in search dog training; the dog is exposed to a target odor, the dog exhibits scenting behavior to the odor such as smelling or nosing the target then the behavior is marked with a conditioned reinforcer and dog receives a primary reinforcer. This should increase the dog exhibiting scenting behavior again.

Negative reinforcement prevents or removes an aversive stimulus resulting in an increase in the behavior. An example of this in search dog training; an aversive stimulus can be applied to the dog until it performs a desired behavior, which ends the aversive stimuli. An example of this is when a dog fails to offer a passive final alert, such as a sit or down at target odor. The dog is then presented with target odor and jerked or shocked until a sit or down is offered. The dog increases the passive alert behavior to avoid or stop the application of the stimulus. However, one concern is whether the dog also now has an association via classical conditioning of target odor tied to the unpleasant event of being jerked or shocked. If so, this association may result in a decrease in the dog’s reliability to hunt and locate this target.

Positive punishment is the application of an aversive stimulus in conjunction with the target behavior which will decrease the likelihood of the behavior occurring again in the future. Relevant example to scent training: the dog engages in inappropriate pursuit of an undesired odor, such as that of another animal. The dog is given a verbal or physical correction which results in the dog being less likely to engage in the inappropriate behavior again.

Timing is essential. If the dog associates the punishment with any non-target behavior being performed at the time of the punishment, he is also less likely to perform that behavior. Therefore, if the dog was actually scenting a target odor and the handler misinterpreted the behavior or the target odor dissemination, it may decrease the dog’s likelihood to engage in the actual search behavior.

Negative punishment, also referred to as omission training by Domjan, is the removal an appetitive stimulus or the opportunity to acquire an appetitive stimulus resulting in a decreased likelihood of the behavior occurring again. Many dog trainers are now incorrectly replacing the term negative punishment with extinction. Extinction results from non-reinforcement of a behavior that has a history of receiving reinforcement (Domjan, 2003) therefore it is not an accurate replacement for negative punishment. Relevant to scent dog training: removing the dog’s opportunity to obtain the reward due to undesirable behavior such as failing to offer an alert or focusing on other scents beside the target odor. The handler simply removes the dog from the training situation into a kennel and does not allow the dog the opportunity to obtain any type of reward. Often, the dog is placed within view of the training area and allowed to watch subsequent teammates using that dog’s toy reward to reward their dogs at the target odor. Eventually the handler takes the dog out of the kennel and attempts the training scenario again. If the dog goes to the correct target and offers a final response, the previously withheld reward is given. The probability is decreased for the future occurrence of the incorrect behavior of ignoring the target scent by removing the chance to obtain the reward.

So, if we go by scientificly validated learning paradigms, e-collars are positive punishment. However, the bottom line, make no mistake, is that E-collars are NOT appetitive, meaning they are not pleasant. Shock is unpleasant. How many people out there enjoy getting a rug shock? Primates (chimps, apes, and us) are notorious for exploiting pleasant experiences, hence drug addicts. If Shock was pleasant, you would see thousands of people hiding in alleys, paying someone to shock them or ordering them online and carrying them around shocking themselves. I have not seen that, have you?

Now......sometimes unpleasant training experiences are needed. Dogs that chase animals, dogs that run away.......there are a variety of VALID uses for an E-collar.

Training a SAR dog IMHO is not one of those because of the principles of learning, Associative Learning to be exact.

Associative learning is used in scent dog training to establish associations between conditioned reinforcer and the discriminative stimulus, the specific target odor. This association results eventually in a trained final response paired with the target odor. With HRD dogs especially, final response issues is one of the major stumbling block for handlers. Perhaps final response reliability problems may be the result of commands which have been unintentionally paired with stimulus perceived by the dog as painful, fear inducing, or aversive. How does this occur? Training a sit or down with compulsion (jerk, pop, shock) and then pairing this aversily trained response with a target odor. The baggage of , "Oh I have to sit, or I am going to get shocked" is now paired with target odor. Or for those who use "compliance to command" with E-collars, they are directly shocking their dog AT the target odor. Sit (shock) Sit (shock)........How does this make any kind of rational sense? let alone, it flies in the face of everything we have learned scientifically about learning and performance, but what the hec, Science is useless right?

This will not change the way these trainers train or the way they think. Fine by me, just please do not bring those dogs to look for me if I am ever lost. I want a traditionally positively trained search dog, one who sees their target odor almost as an addiction because everything associated with that odor has always been fun, great, a blast. Please do not bring the dog who might figure out today that since his owner does not know where the odor is, he cannot be shocked, therefore, OMG I can do anything I want to, and I don't have to find that stupid target odor where I get "stimulated".

No, please bring the happy goofy search dog that thinks the "toy thief" is the greatest thing in the world.

No comments:

Post a Comment