Story of Psychology - Morton Hunt [173]
For a week both rats get the same training every day. By the end of the week the first rat knows the route perfectly and races through the maze, making no mistakes; the second rat still makes as many errors as ever. But finally the second rat gets a food reward at the end of its run—and, remarkably, on the next trial makes no errors. It learned as much from one rewarded trial as the other rat learned in a week. The experiment demonstrates the operation of two principles: reward produces learning, exemplified by the first rat’s behavior; and lacking reward, there may be latent learning, exemplified by the second rat’s behavior. (Learning takes place in some sense when there is no reward but becomes activated as soon as a reward is associated with the “right” behavior.)
What has this got to do with human behavior? Any teacher can tell you. A child practicing drawing or any other skill may make little progress until the teacher has a moment to say something encouraging or complimentary; then, suddenly, the child shows improvement. Similarly, a novice at flying may make a dozen bumpy landings, finally “grease one in” half by accident, winning praise from his instructor, and from then on make landings as if he had at last “got the idea.”
—One at a time, a number of rats are put in the start box of a simple T-shaped maze. At the end of the right-hand branch is a white door behind which is a bit of cheese; at the end of the left-hand branch is a black door behind which is a metal grid floor that gives the rat’s feet a mild but unpleasant shock. The rats learn, after a while, to turn right and push through the white door. But once they’ve learned, the experimenter switches the situation. Now the white door and food are at the end of the left branch, the black door and electrified grid at the end of the right branch. The rats turn right, get shocked, and soon learn to turn to the left.
Once again the diabolic experimenter reverses things, but now the rats learn almost immediately; they have come to associate reward and punishment with the colors of the doors, not their direction. The experiment revalidates Pavlov’s principle of discrimination, the determination of the rewarding cue in a two-cue situation.
Does this apply to humans? Of course. A novice at gardening gets only a meager crop of tomatoes but sees that his neighbor, who plants a different variety in a sunnier location, gets a bumper crop. The novice tries the neighbor’s variety the next year; still no luck. He realizes that the number of hours of sunshine must be the critical factor, fells some trees to get more sunshine, and is successful.
—Another T-maze in which rats learn to turn to the right. This time there is no punishment for choosing the left branch but merely a lack of reward. Some rats are lucky; they find a reward every time they choose the right side. Others are unlucky; they find food there only once every four times. The unlucky rats learn far more slowly than the lucky rats to choose the right-hand branch. The experiment demonstrates that partial reinforcement is less effective in learning than is continual reinforcement.
But now the experimenter changes things; there is no reward at either branch for either group. What happens? Oddly, the rats who had previously been lucky lose their conditioning rapidly and begin to alternate their choices, while the ones who were rewarded only every fourth time continue to choose the right branch for a long while. The experiment has demonstrated the partial reinforcement effect: the higher the creatures’ expectations, the more disruptive a change in the situation; with lower expectations, their learned behavior is more stable when change occurs.
A human analogue: A highly efficient employee has had a generous raise every year; in a year of poor income for the company, he gets only a modest raise, loses his drive, starts taking longer lunches, leaves promptly at 5:00 p.m., and calls in sick