Winter Semester 16/17

Roland Poellinger is giving a course on causal reasoning in Bayesian networks at FAIR, the Forum for Artificial Intelligence Research in Stellenbosch, South Africa, 28-30 November, 2016. The forum with the topic “Causality: Bayesian Networks as Platform for Knowledge Representation in Science and Philosophy” is organized by the Center for Artifical Intelligence Research (CAIR) and the University of Pretoria.


In 1913 Bertrand Russell famously refused to pursue any analysis of the concept of causa-tion, because he believed that the law of causality […] is a relic of a bygone age, surviving, like the monarchy, only because it is erroneously supposed to do no, harm (ON THE NOTION OF CAUSE). With the arrival of formal methods in Philosophy, authors as David Lewis or Judea Pearl have picked up loose ends and started re-constructing the concept close to intuition. Nevertheless, there still seems to be enough room for dispute. In his seminal book CAUSALITY (2009) Pearl expresses his astonishment about the current state of affairs: We are witnessing one of the most bizarre circles in the history of science: Causality in search of a language and, simultaneously, the language of causality in search of its meaning. The two-day course “De/Encoding Cause and Effect” will focus on the relationship between probability, correlation, and causation, on the Bayes net interventionist account of causality and the limits of causal graphs, and on how a Bayesian would confirm causal hypotheses by amalgamating evidence.


1: Decoding Cause and Effect

Propensity, probability, regularity, correlation, counterfactuals, and mechanisms

2: Causal Graphs and Interventions

Causes as difference-makers, Bayesian networks and interventions, inferred causation and algorithmic aspects, type vs. token and the actual cause, modularity (and criticism)

3: Causal Paradox and the Concept of Event

Causal decision theory, Newcomb’s problem and prisoners’ dilemma, the concept of event and Cambridge change variables, quantum mechanics and locality

4: Causal Hypotheses and Evidence Amalgamation

Bayesian epistemology, Bayesian confirmation, evidence and evidence amalgamation, application: encoding the Hill guidelines for the assessment of causality in pharmacology

More information

Please find more information on FAIR 2016 and Roland Poellinger's two-day course on the forum's website here:

Starting in April 2016, Roland Poellinger will teach topics from “Formal Epistemology” (co-teaching with Barbara Osimani).


In this seminar we will discuss selected problems and questions in formal epistemology. Topics include:

  • coherence
  • confirmation
  • learning theory
  • the problem of old evidence
  • explanatory power
  • evidential vs. causal decision theory and Newcomb's paradox
  • updating on conditionals
  • computability and the Church-Turing thesis
  • ranking functions
  • the Duhem-Quine problem

A reading list will be made available in the first session.


Students of LMU Munich can sign up for this course through LMU’s LSF system and earn ECTS credit points for participation plus essay.

formprop poster 2016

Im Wintersemester 2016/17 veranstaltet das Studienbüro Statistik zusammen mit dem Center for Mathematical Philosophy (MCMP) der LMU erneut das Propädeutikum "Auf Du und Du mit Statistik und Co." als Einführung in formal(isiert)es Denken und empirisches Argumentieren.

Idee und Motivation

Formale Techniken und Argumentationen besitzen auch in den Sozial-, Geistes- und Wirtschaftswissenschaften eine große, und immer weiter wachsende, Bedeutung, stellen aber für viele, darauf nicht so gut vorbereitete Studierende eine sehr große Hürde dar. Im Rahmen eines fachübergreifenden Propädeutikums sollen nichtmathematikaffine Studierende der LMU sanft in die Formalisierung eingeführt und mit wesentlichen Techniken (wieder) vertraut gemacht werden.

Vortrag "Zusammenhänge präzisieren im Modell" (Roland Poellinger)

Roland Poellinger geht im fünften Modul des Kurses auf die formale Modellierung von Zusammenhängen ein: Jedem präzisen Argument geht die Entscheidung voraus, welche Eigenschaften welcher Dinge für das Argument wirklich relevant sind. Indem die strukturelle Information aus konkreten Zusammenhängen herausdestilliert wird, entsteht ein abstraktes Modell, mit dem formal argumentiert werden kann. Im fünften Block des Propädeutikums beschäftigen wir uns mit verschiedenen Techniken des Modellierens und stellen uns die Frage nach dem Verhältnis von Informationsverlust und Erkenntnisgewinn.

Weitere Informationen

Zeitplan und Inhalte finden sich auf der Homepage der Veranstaltung hier.