Foundations of Belief and Decision Making
A Rackham Interdisciplinary Working Group
about.
FBDM is an interdisciplinary working group at the University of Michigan, Ann Arbor. We meet regularly throughout the semester to discuss work by faculty, graduate students, or visitors working on models of belief or decision making, broadly construed. Our members come from a number of disciplines, including philosophy, economics, political science, statistics, computer science, and others. All events are free and open to the public. See the schedule for upcoming talks. If you'd like to get on our mailing list or give a talk at an upcoming meeting see the contact page for more details.
schedule.
The Probabilistic Foundations of Rational Learning
Simon Huttegger (UCI)
The Probabilistic Foundations of Rational Learning
- Speaker:Simon Huttegger (UCI)
- Date:March 14, 2017
- Location: TBA
Title: TBA
Kevin Blackwell (Michigan) and Daniel Drucker (Michigan)
Title: TBA
- Speaker: Kevin Blackwell (Michigan) and Daniel Drucker (Michigan)
- Date: April 11, 2017
- Location: TBA
Title: TBD
Sarah Moss (Michigan)
Title: TBD
- Speaker: Sarah Moss (Michigan)
- Date: April 18, 2017
- Location: TBD
Entropy and Indifference
Anubav Vasudevan (University of Chicago)
Entropy and Indifference
- Speaker: Anubav Vasudevan (University of Chicago)
- Date: Thursday February 16, 4pm
- Location: Angell Hall 3222
Theoretical Statistics is the Theory of Applied Statistics
Andrew Gelman (Columbia University)
Theoretical Statistics is the Theory of Applied Statistics: How to Think About What We Do
- Speaker: Andrew Gelman (Columbia University)
- Date: Friday February 10, 3pm
- Location: Kuenzel Room, Michigan Union.
A Tale of Two Expected Accuracy Maximization Arguments
J. Dmitri Gallow (Pittsburgh)
A Tale of Two Expected Accuracy Maximization Arguments
Abstract: Greaves and Wallace (2007) have provided an argument for the belief-revision norm conditionalization which goes roughly as follows: if you stand to learn that one of a partition of propositions is true, and you wish to adopt an actionable strategy for revising your degrees of belief in response to evidence which maximizes the expected accuracy of your posterior degrees of belief, you could do no better than to plan to update by conditionalization. Leitgeb and Pettigrew (2010) have offered an importantly different argument for conditionalization, which goes roughly as follows: upon receiving the evidence E, if you wish to maximize the accuracy of your posterior degrees of belief amongst those possibilities compatible with your new evidence, then you could do no better than to update by conditionalization.
Greaves and Wallace therefore presuppose the norm that you should maximize the expected accuracy of your update strategy, while Leitgeb and Pettigrew presuppose the norm that you should maximize the expected accuracy of your posterior degrees of belief. While these two norms agree when your potential evidence forms a partition, they disagree when your potential evidence fails to form a partition. In the talk, I will argue that, in those cases, we have reason to follow the norm of Leitgeb and Pettigrew; and I draw out some consequences of this position for some other debates in epistemology.
- Speaker: J. Dmitri Gallow
- Date: November 6, 2015
- Location: Tanner
contact.
Our current coordinators are:
Sara Aronowitz (skaron at umich dot edu)
Boris Babić (bbabic at umich dot edu)
If you would like to present a paper or be added to the mailing list, feel free to contact either of us. We accept work in progress from students and faculty in any department.
2016 conference on decisions, games, and logic.
In July 8-10th, 2016, FBDM will host the 9th annual workshop on Decisions, Games, and Logic (DGL2016). More information coming soon!