Quantum theory cannot consistently describe the use of itself*

Can quantum theory be used to consistently describe an agent who herself uses the theory? If quantum theory was a universally valid theory, applicable to arbitrarily complex systems, then this should clearly be the case. To study the question, we consider a setup consisting of four computers that take the role of agents. They are programmed with the rules of quantum theory, which they use to derive statements about the outcome of particular measurements, as well as about the statements that the other computers derived. We then show that the computers sometimes arrive at contradictory statements. This indicates that quantum theory cannot be extrapolated to complex systems, at least not in a straightforward manner.

This talk is based on Frauchiger and Renner, "Quantum theory cannot consistenlty describe the use of itself," Nature Comm. 9 (2018).