In 1969, Marvin Minsky and Seymour Papert published Perceptrons, a mathematical analysis stressing what single-layer networks cannot represent—the classic example is XOR—and questioning the approach’s scalability.
The cultural impact was huge: funding and attention shifted toward symbolic AI. Historians note the book did not “kill” multilayer nets, but it did chill investor and academic climate around connectionism.
Rehabilitation came in the 1980s with backpropagation and, decisively, after 2012. Reading Perceptrons today shows how a technical result can reorder an entire research agenda.
