ACT-R, which stands for Adaptive Control of Thought-Rational, is a cognitive architecture developed over decades by John R. Anderson and collaborators at Carnegie Mellon University. A cognitive architecture is a fixed, general theory of the mechanisms underlying human thinking, expressed precisely enough to run as a computer program and produce the same behavior, timing, and errors that people do. The defining statement of the modern version is the 2004 paper “An Integrated Theory of the Mind,” published in Psychological Review by Anderson and colleagues, which is the source cited here.
ACT-R splits the mind into interacting modules: a declarative memory of facts, a procedural memory of production rules, perceptual modules for vision and hearing, and motor modules for action, all coordinated through small buffers. Knowledge of facts is stored as chunks in declarative memory, each with an activation level that rises with use and recency and predicts how quickly and reliably it can be recalled. Knowledge of how to do things is stored as production rules that fire when their conditions match the current buffers. A notable claim of the 2004 paper is that these modules correspond to specific regions of the brain, letting ACT-R models be compared against brain-imaging data, not just behavior.
Alongside Soar, ACT-R is one of the two most influential cognitive architectures, widely used in psychology, human-computer interaction, and intelligent tutoring systems that model how students learn.
Why business readers should care: ACT-R is a rigorous, testable account of human cognition that has powered tutoring software and human-factors models. It represents the symbolic, mechanistic tradition of explaining intelligence, a useful contrast to the learned, statistical models that dominate AI today.