1. The basic Wigner’s friend setup

• Imagine two agents:

  • Friend: inside a lab performing a quantum measurement on a system (say, a qubit in superposition).

  • Wigner: outside the lab, treating the entire lab (including the friend and the system) as a quantum system.

• Step 1: The friend measures the system. From the friend’s perspective, the system “collapses” into a definite outcome.

• Step 2: Wigner, who hasn’t interacted yet, can describe the entire lab+friend system as still being in a superposition of all possible outcomes.

The paradoxical question arises: How can the system be both collapsed (for the friend) and still in superposition (for Wigner)?

---

2. The twist with extended Wigner’s friend experiments

Modern versions involve multiple agents and correlated measurements to test assumptions like:

• Observer-independent facts: Can there be a single set of facts everyone agrees on?

• Locality or non-contextuality: Can measurements be independent of distant events?

These setups can, in principle, produce statistical results that challenge certain interpretations of QM that assume observer-independent facts.