My research centers around formal epistemology and philosophy of science, causal modeling, the metaphysics of causation, causal paradox, conditionals, Bayesian reasoning and Bayesian networks, foundations of probability, as well as algorithmical aspects of classical and philosophical logic. I have a background in logic, computational linguistics, automata theory, databases, object oriented programming, and cognitive linguistics. More recently, I have started working on the application of Bayesian epistemology and causal inference methods to problems of risk assessment in pharmacological research.
In my work on cause–effect relations I am investigating different approaches towards modeling causation in a formal manner (utilizing probability and graph theory), the communicative aspects, and the cognitive and computational foundations thereof. Coming from a descriptive angle I am especially interested in approaches building on counterfactual (what-if-things-would-have-been-different) and interventionist (wiggling-causes-to-make-a-difference) intuitions. Two related analytic lines of thought stick out: While David Lewis (1973) bases his causal analysis on the evaluation of counterfactual conditionals in a possible-worlds semantics, computer scientist Judea Pearl (2000/2009) criticizes the unhandiness of this approach and turns to methods from statistics and probability theory. He analyzes intuitions about causal relations in close connection with laboratory practice and extends the probability theorist's toolbox of Bayesian networks as structures of belief propagation by introducing the concepts of functional-deterministic causal mechanisms and (hypothetical) external interventions.
In my own analysis I am investigating causal models with entangled (extensionally overlapping or identical) variables: To be able to incorporate such variables I am augmenting the definition of a causal model by the addition of intensional (epistemic) contours transferring knowledge deterministically, non-directionally, and instantaneously. Causal Knowledge Patterns thus conceived allow for the integration of intensional markers and for causal inference from heterogeneous graphs once the Markov assumption is relativized suitably. I have applied these ideas to various problems in causal reasoning: Paradoxes in causal decision theory (Newcomb's problem and the prisoners' dilemma), problems in linguistics and identity theory (Cambridge change variables), modeling problems in foundations of physics (quantum entanglement and the EPR paradox), as well as the problem of mental causation.
On the philosophy of science side I am also applying the framework to the concepts of explanation and (Bayesian) confirmation. Furthermore, moving into philosophy of language I have used causal Bayes nets and Pearl's interventions in a formal approach to the semantics of (indicative and subjunctive) conditionals (also in the nested variant). Find all my posts on talks and events related to causation and statistical methods here.
One common thread beneath all my research interests is the application/applicability of computational methods. I have implemented and visualized algorithms in classical logic, but I am also interested in foundational methodology (recursive functions, Turing machines, and automata theory) as well as meta-questions: What does the Turing test tell us about modern virtual reality? How can the foundations of Artificial Intelligence inform us about the morality of artificial agents? My posts on logic algorithms can be found here, while many of my posts related to computation and Artificial Intelligence are tightly connected to causation and statistical methods, too.