The law of effect
Era 1 · The foundations (1763–1942) › Neuroscience and the psychology of learning
- Lineage
- Acting in the world
- Kind
- Theory
Who
Columbia University
What happened
Edward Thorndike publishes the dissertation Animal Intelligence: An Experimental Study of the Associative Processes in Animals, with experiments in which cats learn to escape from puzzle boxes.
Why it mattered
The law of effect: behaviors followed by a satisfying consequence become more likely. It is the conceptual root of reinforcement learning.
The full story
In the late nineteenth century, animal intelligence was studied through anecdotes: a dog that opened a door, a cat that seemed to reason. Edward Thorndike, a doctoral student in New York, wanted to measure it instead. Thorndike built puzzle boxes, cages that opened only if the animal pressed a lever or pulled a string, and put hungry cats inside them with food in sight outside.
At first the cats tried everything: clawing, pushing, biting the bars. Sooner or later they hit on the mechanism by chance and got out. With each repetition they took a little less time, with no moment of sudden insight. Thorndike explained this without appealing to ideas or reasoning: the animal forms an association between the situation and an impulse to act, and the satisfying consequence strengthens it.
The dissertation, published in 1898 as a monograph supplement to the Psychological Review, is regarded as one of the most influential works of the first half-century of experimental psychology. In 1911 Thorndike included it in a book and stated the idea as the law of effect: responses that produce a satisfying effect in a situation become more likely in that same situation.
Forty years later, in 1938, B. F. Skinner turned it into operant conditioning. In their textbook on reinforcement learning, Richard Sutton and Andrew Barto trace to the law of effect the idea of learning by trial and error from consequences, which is the idea the algorithms of that field follow.
Leads to
Temporal differences (August 1988)
Inherited, with no proven direct link
The law of effect is the conceptual forerunner, with no direct line between the authors
Threads that pass through here
- From reward to alignment (step 1 of 12)
Sources
Original source
- Animal Intelligence: An Experimental Study of the Associative Processes in Animals (Psychological Review, monograph supplement no. 8; Internet Archive) (opens in a new tab). Edward L. Thorndike, 1898. public domain
- Reinforcement Learning: An Introduction, 1st edition, MIT Press. Section "History of Reinforcement Learning" of the online edition (opens in a new tab). Richard S. Sutton and Andrew G. Barto, 1998
Scholarship and context
How to cite
“The law of effect.” The History of AI, revised November 13, 2026. https://thehistoryofai.com/en/milestone/thorndike-law-of-effect
This entry was last reviewed on November 13, 2026.