Connectionism, today defined as an approach in the fields of artificial intelligence, cognitive psychology, cognitive science and philosophy of mind which models mental or behavioral phenomena with networks of simple units¹⁾, is not a theory in frames of behaviorism, it preceded and influenced behaviorist school. Although it is today used in different contexts (mostly referring to neural networks and artificial neural networks that have not emerged until second half of the 20th century) it has origins dating as far back as Greek philosopher Aristotele, who claimed that memory is composed of simple elements connected in a variety of ways²⁾.

Connectionism was later introduced by Herbert Spencer, William James and his student Edward Thorndike in the very beginning of the 20th century. Connectionism was then based on principles of associationism which claimed that³⁾:

• Mental elements or ideas become associated with one another through experience
• Experience consists of such things as spatial and temporal contiguity and (dis)similarity of ideas
• Complex ideas are composed and can be reduced to a set of simple ideas
• Simple ideas are sensations
• Simple additive rules are sufficient to predict complex ideas

But connectionism expands this ideas of associationism by introducing distributed representations or supervised learning⁴⁾ and should not be confused with associationism. Although he is considered one of the first true connectionist, William James' student Edward Thorndike also made the move towards behaviorist ideas.

In order to explain some questions concerning learning, Thorndike introduced two laws of learning⁵⁾. The first law is the law of exercise or use or frequency, which states that stimulus-response (S-R) associations are strengthened through repetition or weakened through lack of repetition. His second law, law of effect, states that the consequence or outcome of a situation-response event can strengthen or weaken the connection between situation and response. If an event is followed by a reinforcing stimulus, the connection will be strengthened and vice versa. This laws have set the basic principles of behaviorist stimulus-response views on learning.

Based on this laws Thorndike considered that all learning is incremental and not insightful, which he tried to prove at the very end of 19th century through experiments first with chickens and later with cats and dogs. In one experiment he placed a hungry cat inside a puzzle box, which had a mechanism that would open the doors of the box every time a string would be pulled or a button pushed. Behavior which was “rewarded” by opening the doors was slowly increased and learning occurred. Thorndike also believed that transfer of learning (application to new situations) occurs only because of previously encountered situations and that intelligence is just a consequence of learned connections.

Connectionism was at its time considered a general theory of learning for both humans and animals. Thorndike's ideas which could well be applied for learning are the idea that rewards promote learning and that repetition enhances learning. in his book on learning of mathematics⁶⁾, he suggested problem should children are expected to solve and learn from should be realistic

Thorndike tried to apply this to learning mathematics⁷⁾, spelling and reading ⁸⁾, measurement of intelligence ⁹⁾ and adult learning ¹⁰⁾ mostly through his laws of learning. Thorndike was one of the pioneers of active learning, proposing children should learn by themselves rather than being thought.

• Connectionism, stimulus-response, S-R, networks of simple units, associationism, supervised learning, law of exercise or use or frequency, law of effect, incremental learning
• Herbert Spencer, William James, Edward Thorndike

Thorndike tried to prove that all forms of thoughts and behaviors can be explained through repetition and reward, without need for introducing any unobservable internal states, yet this is today generally considered incorrect. Connectionism was in the first decades of 20th century succeeded by behaviorism, but Thorndike's experiments also inspired gestalt psychology.

TIP: Connectionism (E. Thorndike).

Reinemeyer, E. Edward Lee Thorndike. Muskingum University. May 1999.

Medler, David A. A Brief History of Connectionism. Neural Computing Surveys, 1(2), p18-72. 1998.

Thorndike, E. Educational Psychology: The Psychology of Learning. New York: Teachers College Press. 1913.

Thorndike, E. The Psychology of Arithmetic. New York: Macmillan. 1922.

Thorndike, E. The Fundamentals of Learning. New York: Teachers College Press. 1932.

Thorndike, E. et al. Adult Learning. New York: Macmillan . 1928.