gardenfors-triviality-theorem

Summary

If a reasoning system is allowed to draw "what would have happened if" conclusions, then it is mathematically impossible to update its knowledge in a way that is both fully rational (never wasteful, never contradictory) and consistent with the standard way we evaluate if-then statements. In practice this means the system must deliberately relax one of those principles when counterfactual reasoning enters the picture, rather than expecting a single update rule to satisfy everything at once.