On the validation of risk analysis—A commentary

Aven and Heide (2009) [1] provided interesting views on the reliability and validation of risk analysis. The four validation criteria presented are contrasted with modelling features related to the relative frequency—based and Bayesian approaches to risk analysis. In this commentary I would like to bring forth some issues on validation that partly confirm and partly suggest changes in the interpretation of the introduced validation criteria—especially, in the context of low probability–high consequence systems. The mental model of an expert in assessing probabilities is argued to be a key notion in understanding the validation of a risk analysis.