Extinction is the term for the disappearance of an association with an involuntary response or a behavior which was learned earlier.
An example of extinction would be when a horse paws the ground when tied and after a few sessions of being tied with no one untying him or giving him attention, he gradually quits pawing. At that point, the pawing has become “extinct.” (At least we hope so.)
It’s very important to understand how extinction works because in horse training, there will invariably be unwanted behaviors you need to stop.
Extinction in Classical Conditioning
Almost everyone is familiar with Classical Conditioning if they have heard of Pavlov’s Dog.
Essentially, Pavlov rang a bell before he fed a dog and then discovered that the dog salivated as soon as he heard the bell. This is known as an association between a natural stimulus (food) and a neutral stimulus (bell) to cause an involuntary response (salivating).
In classical conditioning, extinction occurs when the bell is rung over and over, but the dog doesn’t get fed. Eventually, the dog will quit salivating whenever it hears the bell.
Horses often quickly make the association between a neutral stimulus (a mailbox along the road) with something they fear (a loud passing truck). One way to extinguish that association would be for the trainer to make sure that the horse encounters many, many mailboxes when there is no truck present. As you can imagine, that might take awhile.
Another Classical Conditioned response might be the horse tensing up in a particular corner of the arena because a rider yanked on the bit in that spot. Extinction might require the rider spend quite a bit of time with the horse in that spot on a loose rein.
Because Extinction may be difficult, the best way to deal with unwanted conditioned associations is to try to not let them form in the first place.
For example, when I ride on the road past our farm, I have to count the telephone poles so that I don’t turn to go home at the same one each time; otherwise my horse’s heart rate increases and he goes faster when he approaches the pole where we turn.
Extinction in Operant Conditioning
Operant Conditioning is when a subject (horse/student/child/spouse or significant other) learns to do a voluntary behavior (conditioned response) after a neutral stimulus (any kind of cue) in anticipation of a reward.
Operant Conditioning generally is more efficient, more effective, and more controllable than Classical Conditioning.
The reinforcement schedule often determines how well the subject learns the behavior, and how resistant it is to extinction.
If a gambler wins something EVERY time she pulls a slot machine lever, that is a Continuous Reinforcement Schedule.
If a gambler wins something only a few random times for every time she pulls a slot machine lever, that is a Partial Reinforcement Schedule.
Subjects learn a behavior faster with a Continuous Reinforcement Schedule (if my non-gambling aunt doesn’t win something in the first two pulls of a slot machine lever, she leaves).
Behavior learned under a Partial Reinforcement Schedule is harder to extinguish, a phenomenon which explains gambling addiction and a fact that all casinos use in their favor.
A horse can quickly learn to buck under a rider if, the first two or three times he’s ridden, the rider falls off when he bucks. (Continuous Reinforcement Schedule) This learned behavior gets doubly hard to extinguish if the rider falls off later every third or fourth time the horse bucks. (Partial Reinforcement Schedule)
Of special note to horse trainers is that context–anything in the environment such as location, lights, time of day, or presence of any objects– can influence how hard it is to extinguish a behavior.
Simply, if a horse stops pawing the ground while he’s tied up at the fence, that behavior might not be extinguished when he’s tied up in the trailer.
Therefore, the only way to test Extinction is to proof it in several settings.
Problems with Extinction
The process of Extinction can take a long time to complete.
For example, I haven’t fed my cats anything from the refrigerator for at least nine months, but no matter where they’re sleeping in the house, they still show up when they hear the nearly silent opening of the refrigerator door.
Moreover, during Extinction, one or more Extinction Bursts are likely to occur. An extinction burst is a dramatic increase in the behavior linked to the neutral stimulus. (The dog REALLY salivates when he hears the bell–more than when he was getting fed.)
The first week I stopped feeding them anything from the refrigerator, my cats meowed and demanded my attention every time I even got near the refrigerator. I tried using Differential Attention which is increasing my attention for good behavior and giving no attention for bad. (Petting the cat when he’s by his dry food dish and trying to ignore him when he sinks his claws into my foot when I’m digging in the frig.) It didn’t help much.
Extinction Bursts can get pretty radical (Extinction-induced Aggression), especially when dealing with teenage mules who are under the assumption that once they’ve learned something, a deal is a deal.
Also, Extinction doesn’t mean the association is gone forever.
Most classical conditioned associations are, to some degree, learned for life. I’m sure that when my cat is 15 years old, he’ll still wake up when I open that refrigerator door. Scientists now believe that a behavior is never truly extinguished, but that learning a new behavior simply replaces it.
This is good news because it means that a horse can be taught an alternate behavior to replace the unwanted behavior.
When I’m trying to retrain a horse who has been “spoiled” or “ruined” (in other words, who has learned behaviors which need to be extinguished), if I teach all new behaviors on the off side (right side) of the horse instead of the near side (left side), usually the old, unwanted behaviors are quickly extinguished. In fact, I’m usually surprised by how fast the old behaviors are “overwritten” by the new, simply by switching sides.
Much more on that later.