What have we learned about learning from accidents? Post-disasters reflections

International audienceThe disasters of the past years in different high risk industries (e.g. aviation, offshore, nuclear) push for a moment of reflexivity about learning from accidents. In the aftermath of these events, one wonders whether learning from accidents remains a viable endeavour for companies and states or whether recurring technological disasters such as these seriously and definitely undermine any attempt to prove the feasibility of learning. Progress has certainly been made in the past, but apparently not enough so to be able to reach the highest safety levels, even in systems with dedicated resources. As a result of the current situation, some have been able to argue that 'we don't learn about disasters'. Although appealing and right, this is a very generic statement. There are many studies addressing aspects of learning from accidents which are in a position to bring insights about the drawbacks of learning. But this wealth of research is also part of the problem. When one wants to step back and to look broadly at the topic, to understand the reason why 'we don't learn', one is left with a fragmented scientific literature covering a very large spectrum of interests and views on the subject. This paper tackles this problem by first designing a framework to organise the diversity of studies and second, by extracting four lessons on learning from accidents, putting together for this purpose works in psychology, sociology and political science