Ambiguity in the branching process of Many-Worlds Theories

Quantum mechanics correctly predicts measurement results and is therefore considered a useful physical theory. However, in a recently proposed thought experiment by D. Frauchiger and R. Renner, agents use quantum theory to describe how other agents use the theory, and contradictory statements appear.

Among the most popular interpretations of quantum theory that try to circumvent such consistency problems are the many-worlds theories. They provide a collapse-free model of the physical world: The wave function of the universe evolves unitarily in time, and every time a measurement happens, the global state splits into branches. Each branch is taken to correspond to an eigenstate of the measured observable, and in that sense, all measurement outcomes happen, but each one in a different world.

In this talk, we present a theorem stating that none of the existing many-worlds theories can give a consistent model of a physical process. Our argument is independent of commonly raised objections to many-worlds theories such as the preferred basis problem or the meaning of probability. In connection with the Frauchiger-Renner thought experiment, we show that the branching process leads to an ambiguity when judging the reality of worlds.