Hunting for Maxwell's Demon in the Wild

Abstract: The apparent paradox of Maxwell's demon motivated the development of information thermodynamics and, more recently, engineering advances enabling the creation of nanoscale information engines. From these advances, it is now understood that nanoscale machines like the molecular motors within cells can in principle operate as Maxwell demons. This motivates the question: does information help power molecular motors? Answering this would seemingly require simultaneous measurement of all system degrees of freedom, which is generally intractable in single-molecule experiments. To overcome this limitation, we derive a statistical estimator to infer both the direction and magnitude of subsystem heat flows, and thus to determine whether -- and how strongly -- a motor operates as a Maxwell demon. The estimator uses only trajectory measurements for a single degree of freedom. We demonstrate the estimator by applying it to simulations of an experimental realization of an information engine and a kinesin molecular motor. Our results show that kinesin transitions to a Maxwell-demon mechanism in the presence of nonequilibrium noise, with a corresponding increase in velocity consistent with experiments. These findings suggest that molecular motors may have evolved to leverage active fluctuations within cells.