The Machine That Is Man

In the AI Story

The article appeared at a historically specific moment: early public excitement about artificial intelligence was cresting; the first wave of AI optimism (Newell, Simon, Minsky, McCarthy) had produced ambitious claims about impending machine consciousness; the cognitive revolution was displacing behaviorism in academic psychology; and Skinner was nearing the end of his career, defending his framework against an intellectual environment increasingly hostile to it.

Skinner's argument in the article was not that machines could or could not think. It was that the question is incorrectly formulated. The concept of thinking, he argued, is not a scientific concept but a folk-psychological label applied to certain behavioral outcomes. Whether we apply the label to a machine depends on our willingness to extend folk vocabulary to new organisms, not on any feature of the machine itself. The more productive question — whether the machine's behavior can be analyzed through the contingencies that produce it — applies equally to humans and machines and does not require the label of thinking at all.

The Skinner volume treats this article as prophetic in a way Skinner himself did not fully anticipate. In 1969, the machines that might provoke the question of whether they think were distant possibilities. By 2026, such machines exist and interact with hundreds of millions of users daily. The machines were built on operant principles (RLHF), implement operant contingencies on their users, and produce behavioral effects that the science of behavior predicted half a century before the technology existed. Skinner's insistence that the analysis of contingencies would resolve the mysteries surrounding both thinking men and thinking machines has proven correct in its most important respect: the contingencies are specifiable, the mechanisms are identifiable, and the modifications are engineerable — regardless of whether either the human or the machine is described as thinking.

Origin

B.F. Skinner, "The Machine That Is Man," Psychology Today 2, no. 11 (April 1969): 22-25, 60-63. The article was published concurrently with Contingencies of Reinforcement: A Theoretical Analysis (Appleton-Century-Crofts, 1969), which contained the same argument in expanded form.

Key Ideas

The question is mis-posed. Whether machines think depends on whether thinking is a real phenomenon or a folk-psychological label.

Mental vocabulary is explanatorily empty. Applied to either organisms or machines, terms like thinking redescribe behavioral outcomes rather than explaining them.

Contingency analysis is the productive frame. The real scientific question is which contingencies produce which behaviors, and this question applies uniformly to any responding system.

The article anticipated the AI moment. Skinner's claim that contingency analysis would resolve mysteries of thinking machines proved accurate in its most important respect.