Abstract We lack viable explanations of how brain functions emerge from collective activities of neurons in networks. We recorded field potentials from many local networks in the human cerebral cortex during a wide variety of brain functions. The network dynamics showed that each local cortical network produced fluctuating attractor states. The state trajectories continuously stretched and contracted during all brain functions, leaving no stable patterns. Different local networks all produced this dynamic, despite different architectures. Single trial stimuli and tasks modified the stretching and contractions. These modified fluctuations cross-correlated among particular networks during specific brain functions. Spontaneous states, rest, sensory, motor and cognitive states all emerged from this dynamic. Its mathematical structure provides a general explanation of cortical dynamics that can be tested experimentally. This universal dynamic is a simple functional organizing principle for brain functions at the mm 3 scale that is distinct from existing frameworks.