Afshar's experiment was recently proposed as a violation of the principle of complementarity. Acknowledging the novelty brought by the experiment, we argue that it allows a refinement of this principle, with the establishment of two points: (1) the possibility of modifying the "type" of phenomenon (wave or particle) without changing the quantum-mechanical state, and (2) the recognition that the type of phenomenon, associated to a detected quantum, refers to a specific region traversed by the quantum object. The first point is explored in Mach-Zehnder interferometers, with polarization devices and in Unruh's double setup. The second point emphasizes that a phenomenon can be corpuscular with respect one region and undulatory with respect to another. Afshar's originality lies in his proposal of a way of ascertaining both simultaneously. Restriction (1) turns out having a negligible effect in Afshar's experiment.
Philosophy of physics; Foundations of quantum mechanics; Complementarity; Wave-particle duality; Quantum-mechanical state; Mach-Zehnder interferometer; Retrodiction; Afshar's experiment; Unruh's experiment
