Her original attempt to save the kantian category of causality, by on the one hand retaining the universal validity of the pure concept of causality, while on the other hand accepting, with Bohr and Heisenberg, the definitive character of statistical predictions. Her attempt consists in showing that once a result has been actually obtained, it is possible, by working backwards, to reconstitute, retrospectively and completely, the causal chain which has necessarily produced such a result.