Behavioral Terminology:
What are they talking about?
You hear dog trainers tossing around training terms that
don’t make sense to you? Here’s a look into decoding the lingo.
The definitions are generally accepted among behaviorists
(which is where this terminology is standardized): Whether a given act is
reinforcement or punishment is defined by what the dog does in the future.
Doesn't matter what anyone else thinks of it, it's the dog who decides what it
is.
Reinforcement is anything that increases the chances that
the dog will repeat the desired behavior in the future. Punishment is anything
that decreases the chances that the dog will repeat the desired behavior in the
future. Positive and negative attached to these categories is used in the
mathematical sense of adding or subtracting something. Positive reinforcement =
giving (adding) something that increases the chances the dog will repeat the
behavior in the future (reinforcement) Negative reinforcement = taking away
(subtracting or removing) something that increases the chances the dog will
repeat the behavior in the future (reinforcement) Positive punishment = giving
(adding removing) something that decreases the chances the dog will repeat the
behavior in the future (punishment)
Negative punishment = taking away (subtracting or negative)
something that decreases the chances the dog will repeat the behavior in the
future (punishment) What makes this terminology handy is that if used
correctly, it's a convenient shorthand that expresses certain concepts about
how certain types of learning occurs. And if used correctly, it reminds the
user that it is the dog who decides what is desirable or undesirable. Whether
something is desirable or not is not determined by imagining how the dog feels,
it's determined by observing the dog’s behavior (ideally in over at least
several trials). There are some things that most dogs find desirable (ie, raw
liver) and some things that most dogs find undesirable (the scent of
citronella)--but "most dogs" doesn't count for anything with this
model. This last idea is one that's difficult for many novice (and some
not-so-novice) trainers to really grasp.
Sometimes it's easier to see through example. For instance,
my boisterous puppy loves being wrestled down and ground into the dirt. Any
time a dog can catch him and roll him, he enthusiastically comes back for more.
This pup is a rough player--but he clearly shows by his actions that he
considers being rolled positive reinforcement. And the rougher the rolling the better.
On the other hand, some dogs hate to be rolled. If they are given what the
first pup would consider a very minor pin, they immediately start avoiding the
dog that pinned them. For those dogs, being rolled is a positive
punishment--they were given something that decreased their behavior of playing
with the dog that rolled them.
For an example of negative punishment, here's the simplest
way to teach kenneled dogs to sit as you open the kennel gate. If the dog is
just dying to come out and play with the human approaching, they often hurl
themselves at the kennel gate, barking, jumping and generally acting like a
wild thing. If the human backs up every time the dog touches the kennel gate,
they are (temporarily) taking away the chance for the dog to come out and play.
This is negative punishment--the removal of the chance to get the goodie
(continuing forward when the dog sits is positive reinforcement). The act of
jumping at the kennel gate is decreased. For an example of negative
reinforcement, look at the way the remote collar is used by many trainers (and
very effective it is, too). First the dog is taught a correct response, let's
say a recall. When the trainer is sure that the dog understands that coming
when called is a good thing to do, the trainer puts the remote collar on the
dog, adjusted to be noticeable to the dog but not agonizing (the dog should
turn his head or otherwise mildly alert when it feels the stimulation). The
trainer calls the dog and whenever the dog deviates from moving continuously
towards the trainer, the trainer stims the dog. The dog "turns off"
the sensation by continuing to move towards the handler. The act of coming when
called is increased. If someone understands this terminology, someone can also
get a leg up on figuring out the nuts'n'bolts of a certain behavior simply by
describing a behavior, the consequences that are immediately contingent on that
behavior and the dog's future behavior.
For instance: Max and Teddy are both bugging me for a taste
of the orange I am peeling. I hand each of them an orange peel as their noses
touch my arms for the umpteenth time. Max grabs the peel, gets a weird look on
his face and spits it out. Max then goes and lays down on his bed. Max doesn't
nose my arm again while I'm sitting eating my orange. Teddy grabs the peel and
eagerly chews it up. Teddy then noses my arm again, harder and longer until I
give her another orange peel. By the time I'm out of orange peel, Teddy is
smacking my arms with her face and her feet. Max was given a positive
punishment (I added something--the orange peel--and it decreased the behavior
that occurred immediately before I gave it to him). Teddy was given a positive
reinforcement (I added something--the orange peel--and it increased the
behavior that occurred immediately before I gave it to her). Reward and
punishment is determined by the dog's perception, not the trainer.
