This is what we do.

Making a Bayesian model for an environmental decision making problem requires multiple components that must be coupled to each other. Our research projects may concentrate on detailed modelling of one of these components or on the integration of all of them.

• Valuation. This component defines the objective of the decision making problem by expressing how desirable different future states of nature are. For example, the future catch of fish might be valued based on the amount of energy the catch provides compared to the energy needed for obtaining that catch. The mathematical representation of the valuation is called the utility function.
• Decisions. What are the alternative actions we are willing to consider in order to reach the objective? A rational decision-maker would choose the action with the highest expected utility - the main task of the decision analysis is to find out what these expected utilities are.
• System dynamics and uncertainty. How will the alternative actions change the future state of nature we value?  This causal dependency is always uncertain: at the time of choosing the action, we cannot have empirical evidence in the form of observed data from the future. Instead, we need theory that predicts the outcomes of our decisions in the form of a probability distribution. This probability distribution together with the utility function then define the expected utilities of the alternative actions. Coming up with this predictive probability distribution is one of the major research problems - we need to describe the current understanding of the system dynamics and of the current state using probability statements. The probabilistic formulation of this knowledge is the fundamental idea behind the Bayesian approach: once knowledge is phrased as a probability statement, then it can be consistently processed, integrated, and updated using the mathematical logic of probability theory.

A few examples of our research themes: