When Negative Means Two Things
Negative Reinforcement is bad and Positive Reinforcement is good, right?
Sometimes scientists use words which non-scientists may find confusing. I can’t think of a better example than the whole positive/negative reinforcement/punishment quadrant that behaviorists came up with.
When most people hear the word “negative” it means “bad” to them, and the same applies to “positive” meaning “good.”
But ah, no – to behavior scientists, “negative” means “subtract” and positive means “add.”
Who knew scientists could be so mathematical?
Behaviorists define “reinforcement” as something that increases a behavior.
So adding a desired motivator to increase a behavior is Positive Reinforcement.
Examples of Positive Reinforcement:
You get a compliment for wearing a nice dress, so you wear it again.
You eat candy and it tastes good, so you eat another.
You get a bonus on your paycheck for coming to work early, so you’re more likely to come to work early again.
(Actually, pay checks ought to be an example of Positive Reinforcement, but if they’re so disappointing that the employee needs more money and quits their job, they’re not. In that case, the stimulus wasn’t what was desired. Keep that in mind when training.)
In the present horse world, most common training methods often fall under Negative Reinforcement.
Negative Reinforcement as such isn’t “bad.” It simply means removing an aversive stimulus to increase behavior.
A horse training example would be giving a horse a slap on the butt (aversive stimulus) to get the horse to move forward (increasing behavior.)
Another example is putting back on the bit (pain in the mouth is aversive) until the horse stops (behavior) and the reins are released (removal of aversive.)
But negative reinforcement is trickier to analyze than it sounds.
When Negative Reinforcement Happens At The Same Time As Positive Reinforcement
For example, let’s say you want to sleep, but the light is on keeping you awake. So you turn it off, and because the aversive stimulus of the light is gone, leading you to turn it off again in the future, that is Negative Reinforcement.
But if you analyze the very same event, you could say that when you turned off the light, you received darkness, which was what you wanted, therefore the event was Positive Reinforcement.
In this way of looking at it, Positive Reinforcement can be the flip side of the same coin as Negative Reinforcement.
Whoa! Did I just say that Positive Reinforcement is the same thing as Negative Reinforcement?
No, I did not.
What I did say that in some situations, one behavior can both remove an aversive AND add a desired motivator.
Your alarm clock goes off (aversive) and you get out of bed to stop the ringing (behavior) so it most certainly is Negative Reinforcement. If you either slept through the alarm or ignored the ringing, then it’s not Negative Reinforcement. But if you are asleep and want to continue to sleep in silence, then your behavior of turning off the alarm so that you get silence is also Positive Reinforcement.
How this applies to horse training:
Let’s say I have a wild mustang in a corral and I want to use Positive Reinforcement by clicker training a horse to look at me (behavior) and then feeding the horse (desired motivator.)
But what if initially, the mustang does NOT desire food from me. What he desires is for me to go away and leave him alone?
I can use that motivator paired with the clicker so that if when I approach, the horse looks at me for a moment (the behavior I’m shaping) and I click and remove myself. This situation would be both Positive Reinforcement (horse gets what he wants when he looks at me) and Negative Reinforcement (I remove my physical pressure the moment the horse looks at me.)
When would the two NOT be the same?
I think we have to look at the overall picture to see when the two aren’t the same.
If I take a horse simply standing in one place (horse in stasis) and tap the horse’s side with a whip, the horse may move over (the behavior I’m shaping) and if I stop tapping (the removal of the aversive), that’s Negative Reinforcement. However, in this case the horse is very aware that I was the agent of the tapping. I’m not really “giving” or “adding” a desired motivator–the horse is simply returning to what it was.
In fact, if my tapping is too extreme, the horse can read the above scenario as Positive Punishment. Before the horse moves over, he may consider me as an aggressor and wonder what he did and quickly consider escape. If the action causes the horse to avoid the repetition of the tapping, it’s actually punishment, not reinforcement. I know that there can be a hair’s width of difference, but there is a difference. And how the horse perceives the difference has everything to do with whether the behavior is reinforced or if the horse has been punished for standing still.
If the next time I approach the horse to tap his side with the whip and he starts dancing around and pulls away, guess what? I punished the horse previously and Negative Reinforcement did not occur.
Neither did Positive Reinforcement because the horse associated ME and the WHIP with the tapping (aversive), not his own behavior of moving over which stopped the aversive.
But if the next time I approach the horse and tap his side and he simply moves over, then Reinforcement has occurred. If I can get to the point with the horse that I barely touch the horse’s side–or even just gesture the touch–and the horse moves over, then the tapping (aversive) has shifted to a simple cue.
Now, here’s the neat thing: I can teach the tapping gesture to be the cue without letting it become aversive at all. I can teach the cue using only Positive Reinforcement and I can do that faster and more thoroughly than teaching it with Negative Reinforcement.
More on that later….