How to make environmental models better in supporting social learning? A critical review of promising tools

Social learning, defined as a convergence in the perspectives of stakeholders on a complex (environmental) problem and its possible solutions, is considered an important mechanism in developing integrated solutions requiring broad societal support and concerted action. Quantitative environmental models can support social learning of stakeholders by providing a platform for communication and integration and by allowing exploration of the consequences of different choices. However, in many integrated assessment projects that combine quantitative modelling with stakeholder participation, the models fail to play a significant supporting role in social learning. Two major reasons for this failure are: (1) stakeholder perspectives are inadequately integrated into the model, and (2) insufficient iterations are made between model outcomes and stakeholder choices. In recent years, a number of software tools have become available that may help to remedy these shortcomings. For instance, with tools that link conceptual models with quantitative models the integration of stakeholder perspectives could be made more efficient. Other tools may make the feedback between model outcomes and stakeholder choices more efficient, for example, interactive visualisation or scanning tools. However, our analysis of the state-of-the-art reveals that most of these tools have not reached the stage of a fully functional version and so far none have been evaluated in real cases with real stakeholders. The major bottleneck in the interaction between stakeholders and quantitative models appears to be model complexity, and the tools discussed in this paper do not adequately address this problem. We conclude therefore, that if the aim is to better support social learning of stakeholders, the models used in integrated assessment must be simplified and the participatory processes intensified. For integrated assessment modellers this means that much more emphasis should be placed on the investigation of options to reduce model complexity.