(Also known as Generative Learning) A learning strategy that involves questioning the assumptions upon which particular questions or challenges are based (See also Adaptive learning/Single-loop learning).

In contrast to single-loop learning which involves using knowledge to solve specific problems based on existing assumptions and often based on what has worked in the past, double-loop learning goes a step further and questions existing assumptions in order to create new insights. For example, take the problem 'how do we prevent earthquakes from killing people?'. The single-loop answer would be to learn how earthquakes happen and try to predict them in order to be prepared. The double-loop answer would question our notion of 'earthquake' and might conclude that earthquakes do not kill people, falling buildings do.

Adjusting underlying assumptions and developing new mental models as a result of organisational learning. See also Single-Loop Learning, Mental Models.