Category: Theory, Terms & Abbreviations ¤
  Author: Elizabeth Naime ¤
   Title: R+, R-, P+, P- ¤

LisaSaxtn@aol.com wrote:

> I know that the definitions of this stuff always get me.  However, I believe
> the ear pinch is negative punishment.  The REMOVAL (negative) of the PUNISHER
> (ear pinch) encourages repetition of the BEHAVIOR (dumb bell holding)

NOPE.  Back to basics:  a REINFORCEMENT (or reinforcer) is something that
causes the behavior to happen more often.  a PUNISHMENT (or punisher) is
something that causes the behavior to happen less often (and in some
cases to dissapear altogether).  for POSITIVE and NEGATIVE think
mathmatically, as you have: any positive is something added, negative is
something removed.  Sooo...

the ear pinch is a negative reinforcement.  The behavior of taking the
dumbell removes (subtracts) the pinch;  this increases the behavior of
holding the dumbell.

I don't think I am nitpicking.  I think the behaviorist's definitions of
REINFORCEMENT and PUNISMENT are capable of fundamentally changing the way
we think of training.  We WANT to assign values to everything --
reinforcement is good, punishment is bad, for instance.  We want to
moralize -- the dog refuses the dumbell because it is bad, or willful, or
spiteful.  We want the dog to "work for us, not for food", as if the
Puritan work ethic applied to dogs.

Behaviorism looks only at what happens, not at what should happen.  We can
make some good guesses about what will reinforce or punish, but ultimately
we have to look at what actually happened.  And I think we can add our own
ideas of "why" and what the dog might be thinking -- as long as we
remember that what actually happens is more important than what we think
should happen!  Reinforcement and punishment exist only in relationship to
the behaviors they affect.  

> The only way I am keeping this stuff straight in my head is instead of
> reading negative and positive, I read plus and minus.  Therefore negative

This is a good way of remembering plus and minus!

A lot of people think in terms of unpleasant stuff vs. pleasant stuff.  I
find this an important distinction myself -- there are things I won't do
for what I consider insufficent reason, i.e. I won't ear pinch a retrieve
because a retrieve just isn't "worth" the discomfort to my dog.  It's a
personal line that you should draw for yourself.  I think behaviorism
keeps this distinction when it talks about "aversives" -- aversives in the
lab are usually shocks, in the real world I think they are anything
painful or uncomfortable or annoying or irritating:  "anything the subject
will work to avoid".  +P and -R both require the use of aversives.  +R and
-P do not use aversives, at least not as I've tried to describe them here
(physical aversives).  Notice also that in -R it is necessary to apply the
aversive before the dog has done anything wrong, and before it has the
opportunity to do anything right.  If the subject is properly prepared,
you can set the situation up so that the correct response is likely to be
offered very quickly, but the aversive MUST be experienced at first or
learning by negative reinforcement will not happen.

Assuming that each of these methods works, here's why they work:
click and treat for taking the dumbell:  +R
time out for dropping the dumbell,
eat a cheesburger in front of dog for dropping the dumbell: -P
    (losing the cheesburger or the attention DECREASES the dropping
     behavior)
ear pinch: -R   (the release of the pinch reinforces taking the dumbell)
collar correction when dumbell is dropped: +P (adding the correction
decreases the dropping behavior)

Clear as mud?  <grin>
Elizabeth