
Imagine a rat in a cage. This is a special cage (called, in fact, a “Skinner box”) that has a bar or pedal on one wall that, when pressed, causes a little mechanism to release a food pellet into the cage. The rat is bouncing around the cage, doing whatever it is rats do, when he accidentally presses the bar and -- hey, presto! -- a food pellet falls into the cage! The operant is the behavior just prior to the reinforcer, which is the food pellet, of course. In no time at all, the rat is furiously peddling away at the bar, hoarding his pile of pellets in the corner of the cage.
A behavior followed by a reinforcing stimulus results in an increased probability of that behavior occurring in the future.
What if you don’t give the rat any more pellets? Apparently, he’s no fool, and after a few futile attempts, he stops his bar-pressing behavior. This is called extinction of the operant behavior.
A behavior no longer followed by the reinforcing stimulus results in a decreased probability of that behavior occurring in the future.
Now, if you were to turn the pellet machine back on, so that pressing the bar again provides the rat with pellets, the behavior of bar-pushing will “pop” right back into existence, much more quickly than it took for the rat to learn the behavior the first time. This is because the return of the reinforcer takes place in the context of a reinforcement history that goes all the way back to the very first time the rat was reinforced for pushing on the bar!