Edward Thorndike - Connectionism

Connectionism

Thorndike was a pioneer not only in behaviorism and in studying learning, but also in using animals in psychology experiments. Thorndike was able to create a theory of learning based on his research with animals. His doctoral dissertation, “Animal Intelligence: An Experimental Study of the Associative Processes in Animals”, was the first in psychology where the subjects were nonhumans. Thorndike was interested in whether animals could learn tasks through imitation or observation. To test this, Thorndike created puzzle boxes. The puzzle boxes were approximately 20 inches long, 15 inches wide, and 12 inches tall. Each box had a door that was pulled open by a weight attached to a string that ran over a pulley and was attached to the door. The string attached to the door would lead to a lever or button inside the box. When the animal pressed the bar or pulled the lever the string attached to the door would cause the weight to lift and the door to open. Thorndike’s puzzle boxes were arranged so that the animal would be required to perform a certain response (pulling a lever or pushing a button), while he measured the amount of time it took them to escape. Once the animal had performed the desired response they were allowed to escape and were also given a reward, usually food. Thorndike primarily used cats in his puzzle boxes. When the cats were put into the cages they would wander restlessly and meow, but they did not know how to escape. Eventually, the cats would step on the switch on the floor by chance, and the door would open. To see if the cats could learn through observation he had them observe other animals escaping from the box. He would then compare the times of those who got to observe others escaping with those who did not, and found that there was no difference in their rate of learning. Thorndike saw the same results with other animals, and he observed that there was no improvement even when he placed the animals’ paws on the correct levers, buttons, or bar. These failures led him to fall back on a trial and error explanation of learning. He found that after accidentally stepping on the switch once, they would press the switch faster in each succeeding trial inside the puzzle box. By observing and recording the animals’ escapes and escape times, Thorndike was able to graph the times it took for the animals in each trial to escape, which eventually resulted in a learning curve. In Thorndike’s learning curve the animals had difficulty escaping at first, but eventually “caught on” and escaped faster and faster with each successive puzzle box trial, until they eventually leveled off. The quickened rate of escape results in the s-shape of the learning curve. The learning curve also suggested that different species learned in the same way but at different speeds. From his research with puzzle boxes, Thorndike was able to create his own theory of learning. The puzzle box experiments were motivated in part by Thorndike's dislike for statements that animals made use of extraordinary faculties such as insight in their problem solving: "In the first place, most of the books do not give us a psychology, but rather a eulogy of animals. They have all been about animal intelligence, never about animal stupidity."

Thorndike meant to distinguish clearly whether or not cats escaping from puzzle boxes were using insight. Thorndike's instruments in answering this question were learning curves revealed by plotting the time it took for an animal to escape the box each time it was in the box. He reasoned that if the animals were showing insight, then their time to escape would suddenly drop to a negligible period, which would also be shown in the learning curve as an abrupt drop; while animals using a more ordinary method of trial and error would show gradual curves. His finding was that cats consistently showed gradual learning.

Thorndike interpreted the findings in terms of associations. He asserted that the connection between the box and the motions the cat used to escape was strengthened by each escape. A similar, though radically reworked idea was taken up by B. F. Skinner in his formulation of operant conditioning. The associative analysis went on to figure largely in behavioral work through mid-century, and is now evident in some modern work in behavior. Thorndike supported Dewey's functionalism and added a stimulus-response component and renamed it connectionist. His theory became an educational requirement for the next fifty years.