2-3 September 2010, Workshop on Advances in Algorithmic Game Theory
Location: CWI Amsterdam, The Netherlands
Costs: free

Algorithmic game theory is a rather recent research field that lies at the intersection of economics, computer science and mathematics. Its origins in the early 1990′s were largely motivated by the wealth of new applications that came into existence with the emergence of the Internet. The field pursues mathematical studies of games with a particular focus on computational and algorithmic issues. These studies are interdisciplinary in flavor and often demand for a combination of methodologies and techniques from the areas of optimization, algorithms and game theory. Algorithmic game theory contributed successfully to the understanding of many fundamental games in recent years and has become a highly active research field.

The two-day workshop will be hosted at the Center for Mathematics and Computer Science (CWI) in Amsterdam and will take place September 2-3, 2010. There will be four keynote talks and about 12 contributed talks.
Please register before August 16, 2010. For more information, see http://www.cwi.nl/~schaefer/agt10/

6-17 December 2010, Logic Summer School, Canberra, Australia

As computers become more powerful, their ability to perform complicated reasoning tasks increases. In order to harness their power, we need to understand the reasoning they can do, and how they may do it more efficiently. This understanding begins with logic.

The Logic Summer School comprises a blend of practical and theoretical short courses on aspects of pure and applied logic taught by international and national experts. The school provides a unique learning experience for all participants, backed up with state-of-the-art computational science facilities at the Australian National University.

For more information, see http://lss.cecs.anu.edu.au/.

Workshop on Decision Theory

Wednesday 29th September

Detailed Programme TBA (starts 10:00, ends 19:00)

Venue: T206, Lakatos Building, Portugal Street

Speakers

James Joyce (Michigan, Ann Arbor), A defense of imprecise probabilities

David Etlin (Leuven), Vague desire

Richard Bradley (LSE) Counterfactual reasoning and causal decision theory

Jason Alexander (LSE) Why the Angels Cannot Choose 

Christian List and Franz Dietrich (LSE) Conditionalisation unifie

The Faculty of Computer Science of the Free University of Bozen-Bolzano (FUB), Italy, offers an opening for 18 positions for its 3-year PhD program. 9 of the positions are fully funded by a PhD studentship. 3 of the PhD positions with studentship are offered by the KRDB Research Centre for Knowledge and Data.

The research activities in the KRDB research centre require good knowledge of Logic and of Foundations of Databases, and some knowledge of Artificial Intelligence and of Knowledge Representation. Good knowledge of English is also preferred.

The application deadline is Oct. 18, 2010 (arrival date of documents, electronic application is not possible). For information about the PhD program, the studentship, and the application, please visit http://www.unibz.it/en/public/research/phd/prospectivePhdstudents.html.
The KRDB Research Centre invites applicants to get in touch with the research group as soon as possible, in order to have a better understanding of the possible research activities in which the applicants may be involved. For more information, see the KRDB website at http://www.inf.unibz.it/krdb/ or send an email to prof. Diego Calvanese at calvanese@inf.unibz.it. To get in touch with the current PhD students, see https://www.inf.unibz.it/phd/.

Logic and Games Seminar
Friday, September 3, 2010  4:15 pm   Room 4419, GC
Dr. Willemien Kets   (Santa Fe Institute (visiting scholar Stern NYU))
Finite belief hierarchies in games

Abstract. The decision-theoretic approach to game theory requires players to have beliefs about all relevant uncertainty, including
beliefs about other players’ beliefs, and so on, which naturally leads to infinite hierarchies of beliefs. However, players may not be able
to “reason all the way.” This paper constructs finite belief hierarchies by assuming that players’ language is too coarse to reason
about higher-order events. This gives a class of type structures which form a generalization of the standard Harsanyi structures.

The workshop featured the following speakers and talks:

David Christensen (Brown) “Rational Reflection”
The main question of the paper concerned the relationship between what it is rational for one to believe and what it is rational for one to believe about what it is rational for one to believe. Christensen approached the issue mainly in the context of graded belief.  He formulated a rational reflection principle: Cr (A/Pr(A)=n)=n, where Cr stands for the agent’s credences, and Pr stands for the credences that would be maximally rational for someone in that agent’s epistemic situation. Christensen showed how this prima facie plausible principle leads to some puzzling results and explored a number of different reactions to the puzzling cases.

Peter Milne (Stirling) “Belief, Degrees of Belief, and Assertion”
On the basis of an in-depth discussion of the role of assertion and the relationship between assertion and belief, Milne went on to discuss two puzzling issues: The first concerned the assertion of indicative conditionals.   Much recent research in the psychology of reasoning supports two theses: that people ascribe to the indicative conditional the so-called “defective truth-table” in which an indicative conditional with false antecedent is deemed neither true nor false; and that people assign as probability to an indicative conditional, the (conditional) probability of consequent conditional upon antecedent.  The second issue Milne discussed concerned the relationship between assertion and degrees of belief. Certain well known cases suggest that a high degree of belief is neither necessary nor sufficient for a willingness to assert.

Brian Weatherson (Rutgers/Arche) “Rational Belief and Rational Action”
Weatherson discussed the principle, defended by Jeremy Fantl and Matthew McGrath, that if S’s belief that p is justified, then it is permissible for S to use p as a premise in practical reasoning. Weatherson argued that the principle is false. In cases where the agent has some rational beliefs and some irrational beliefs, it might be that justification is insufficient to ground action. Weatherson went on to discuss a novel suggestion regarding the relationship between rational beliefs and rational degrees of belief.

Alan Hájek (ANU) “Staying Regular”
Hájek first reviewed several arguments for regularity as a norm of rationality. There are two ways an agent could violate this norm: by assigning probability zero to some doxastic possibility, and by failing to assign probability altogether to some doxastic possibility. Williamson and Easwaran have argued forcefully that the former kind of violation may be unavoidable. Hájek’s discussion focused on violations of the second kind. Both kinds of violations of regularity have serious consequences for traditional Bayesian epistemology, in particular conditional probability, conditionalization, probabilistic independence and decision theory.

Carl Hoefer (Barcelona) “Connecting Objective and Subjective probability:  How to Justify the Principal Principle”
Hoefer focused on David Lewis’s Principal Principle and the claim that “Truth is to belief as agreement with objective chance is to degree of belief.”
The plausibility and defensibility of these theses depend, clearly, on how we understand objective chance.  The paper introduced an account of objective chance: a Humean/reductive account closely related to Lewis’ own.  One of the key virtues of this account of chance is that it allows us to demonstrate that the PP is indeed a requirement of rationality. Hoefer also argued for a second thesis on which agreement with actual frequencies, rather than chances, serves as the core virtue for degrees of belief.

Branden Fitelson (Berkeley) & Kenny Easwaran (USC) “Partial Belief, Full Belief, and Accuracy-Dominance”
The paper had two main aims: (1) to make some (cautionary) remarks about the set-up and interpretation of some recent accuracy-dominance based arguments for probabilism (with respect to partial beliefs), and (2) to discuss some interesting (formal and informal) analogies (and disanalogies) between partial belief and full belief, when it comes to the phenomenon of accuracy-dominance. In particular, Fitelson and Easwaran discussed the case of an extremal agent (who can only assign degrees of belief 0 or 1) and showed that there are non-probabilistic extremal functions that are not even weakly dominated by any probabilistic extremal function (using the Brier score).

Philip Ebert & Martin Smith

At the Faculty of Computer Science, Institute of Artificial Intelligence, the position Chair (W2) of Computational Logic is to be filled at the earliest possible date. The successful candidate is required to represent the area mentioned above in research and teaching.

The task in teaching consists of lectures (in English) in Computational Logic in the international master program in Computational Logic, lectures (in German) in Computer Science in the other programs of the faculty or for other faculties as well as the supervision of bachelor-, master- and PhD-students.

In research, applicants are expected to have made novel contributions to the logical foundations of declarative programming, semantic technologies, constraint and satisfiability problems, verification, intelligent autonomous agents, cognitive systems, the integration of virtual and physical systems, machine learning or natural language processing, their implementation in running systems as well as their application. International publications and contacts as well as the participation in research projects in one or several of the above-mentioned subareas are expected.

The applicants must fulfill the employment qualification requirements of the Higher Education Act of Saxony (namely of § 58 SächsHSG). Applications from women are particularly welcome. The same applies to disabled people. (For information please phone +49 351 463 38340.)

Please send your application until October 15, 2010 (stamped arrival date of the university central mail service applies) to: TU Dresden, Dekan der Fakultät Informatik, Prof. Dr. rer. nat. Andreas Pfitzmann, 01062 Dresden, Germany. Your application should contain the usual documents (CV, transcripts and certificates, lists of scientific publications, teaching activities and projects funded by third parties, teaching evaluation reports).

Opening the conference, Hans Kamp (University of Stuttgart) tackled the issue of the relation between mental representations and meaning of utterances and discourse. Successful communication is based on similarity between the mental representations of the utterer and of the audience, i.e. on a shared understanding of the relation between words and thoughts. However, Hans elaborated on the crucial fact that, in order to obtain successful communication, background information has to be commonly understood and represented as well. Paul Egre (Institute Jean Nicod, Paris) reported on work on vagueness led in collaboration with Pablo Cobreros (Navarra), David Ripley (IJN) and Robert van Rooij (ILLC Amsterdam). The semantic framework put forth by Paul accounts for the principle that, if x is P, then y is also P if y is similar enough to x. To account for the principle, classical truth nicely interacts with two further notions of consequence defined in terms of the classical one. The semantic framework provides a solution of the sorites paradox and its relation to paraconsistent truth was discussed. In his contribution, Wilfried Sieg (Carnegie Mellon University) investigated the inner workings of the mathematical mind by looking at the relation between mathematical arguments and rigorous proofs. The analysis was tackled from the points of view of (abstract) axiomatic method, on the one hand, and (concrete) proof construction as directed by proof theory, on the other, providing a rich picture that touched on computer and cognitive science alike. Rohit Parikh (City University of New York) talk on “Behavior and Belief” offered a new perspective on realistic formal models for epistemic reasoning. The desideratum is a notion of belief that does not entail logical omniscience and that can accommodate, among other things, the inconsistencies and incompleteness characterizing human epistemic performance. Rather than tweak existing systems of epistemic and doxastic logic, Rohit proposes to adopt a “Ramseyian” approach, deriving belief from observed behavior and agents’ perferences. The talk explored three ways in which agents revise their belief ascriptions: through observed actions, heard sentences and performed inferences. The multi-agent case was also tackled in the talk. Questions pointed out possible asymmetries between behavior-based attributions and actual mental states, also due to the complexity of interpretation of behavior as in case, for instance, of the conjunction fallacy. Brian Skyrms (UC Irvine and Stanford University) presented new work on “The concepts of information and deception in signaling games.” Adding probabilities to standard Lewisian signaling games (and leaving open its intended interpretation), Brian analyzed of the informational content of signals according to the intuition that a signal is informative insofar as it moves the probability of states. After discussing examples, the talk went on to cover deception. Contrary to Kant’s claim that universal deception is self-contradictory or practically impossible, Brian showed there are signaling games in which all players saying ‘half-truths’ (interpreted as partial pooling of states in one signal) constitutes an equilibrium. Questions revolved around the measure used for the informativity of signals and the relationship between the dynamics of signaling and information flow in networks. Oliver Schulte (Simon Fraser) gave a talk on his recent research on relational (as opposed to single-table) data. To account for relational data tables, Oliver uses parametrized Bayes nets where nodes are construed as functions. After describing the hurdles faced by knowledge-based  models constructed on such machinery, and explained how Markov logic networks can overcome them, Oliver described a very fast and accurate new algorithm for learning causal graphs in such models. In his dense philosophical talk, James Woodward (University of Pittsburgh) investigated the relationship between two accounts of causation, dubbed “geometrical” (think collisions, pushing etc.) and “difference-making” (comparative, counterfactual-based). The connectedness necessary to the GM account is not necessary for DM, and it is also not sufficient (a physical connection is not enough). Experiments about perception of causal link were also reported.

Stephan Hartmann (University of Tilburg) opened the second day with a provoking Bayesian account of confirmation and reduction. Stephan offered a Bayesian formalization of Schaffner-Nagel reductionism, arguing that the Bayesian network framework helps represent relations of conditional independence and hence simplify calculations relative to confirmative evidence. Some of the numerous questions were concerned with how the model deals with confirmation of bridge laws and with the relation between Schaffner-Nagel’s philosophical account and its Bayesian formalization. Branden Fitelson (UC Berkely, from Fall: Rutgers University), reporting on joint work with Jim Hawthorne (University of Oklahoma), attacked the problem of irrelevant conjunction: if some evidence confirms a hypothesis, then the same evidence confirms also the same hypothesis in conjunction with an arbitrary X, which, however, need not lend plausibility to the hypothesis H. Branden’s proposed solution is to consider X irrelevant if it is independent of the evidence given the hypothesis. Questions brought up qualitative alternative solutions based on non-monotonic logic, and the problematic case in which both H and X entail the evidence. After lunch, the audience were treated to Hannes Leitgeb’s (Bristol University, from Fall: Ludwig-Maximilians University Munich) “peace project,” aiming at the unification of qualitative and probabilistic beliefs. Recognizing that both kinds of belief serve important purposes in epistemology, Hannes provided axioms for both types, bridge principles linking them together and a representation theorem. The theorem validates the left-to-right direction of the “Lockean thesis” that p is qualitatively believed iff its probability is higher than a given threshold, while the other direction can be (partially) obtained by adding further axiomatic constraints on qualitative belief. Applications are of course wide-ranging, from the lottery paradox to a probabilistic analysis of Duhem’s problem, from Ramsey test and the logic of conditionals, to the possibility of re-uniting logical and probabilistic techniques in accounts of inductive reasoning. Jim Joyce (University of Michigan) defended the use of imprecise probabilities in inference and decision-making. Jim reviews a counterexample to the idea of imprecise credal states due to Roger White, undermining White’s conclusion against imprecise probability and positively arguing about a solution to the counterexample based on Seidenfeld and Wasserman’s notion of dilation. Jim then went on discussing, against White, the implication of imprecise credal states for decision-making. Teddy Seidenfeld (Carnegie Mellon University) reviewed three approaches on how a decision maker might frame her probability space: (i) assess precise probabilities for the random variable of interest only; (ii) choose “wisely” the set of events to which precise probabilities can be assigned; (iii) bite the bullet and choose an incoherent prior, then use Bayesian learning to update to a less incoherent posterior, minimizing loss. Teddy’s talk was as usual rich in theoretical implications. E.g, the analysis of (ii) revealed that de Finetti’s coherence conditions do not entail that the decision-maker have determinate probabilities over the entire space, but only over a subset determined by the decision problem at hand. Wolfgang Spohn (University of Konstanz) concluded the line-up of invited talks with a tutorial on ranking theory—Wolfgang’s theory of degrees of belief that avoids some of the pitfalls of Bayesianism (for instance, the lottery paradox). The talk highlighted applications of ranking theory in philosophy of science, for instance analyzing the relation between Bayesian nets and ranking nets.

The two day celebration ended with a round table discussion on formal epistemology moderated by Horacio Arló Costa and featuring panelist from various geographical and academic backgrounds: philosophers from South (Eleonora Cresto, University of Buenos Aires) and North America (Frederick Eberhardt, Washington University, Kishida Kohei, University of Pittsburgh, Jeff Helzner, Columbia University, Giacomo Sillari, University of Pennsylvania, Mike Titelbaum, Wisconsin-Madison) as well as philosophers and computer scientists from Europe (Eric Pacuit, Tilburg University, Choh Man Teng and Greg Wheeler, CENTRIA, Lisbon).

Giacomo Sillari

Date and time: Monday, 23 August,  16:00-17:30

Location: Room 353, Xinzhai Building

Speaker: Prof. Sven Ove Hansson (Department of Philosophy and the History of Technology, Royal Institute of Technology, Sweden.)

Title: Cognitive realism in belief revision: Representing a finite mind

Abstract:  Since a human mind cannot deal directly with infinite structures, cognitively realistic models of belief change should operate on belief states that have a finite representation. The standard AGM model cannot easily be reconciled with that requirement. Different ways to achieve a finite representation are discussed, in particular: finite language, belief bases, and specified meet contraction. Formal results that compare and connect the different approaches are presented.

Second Call for Registration

Arch ‘Logic or Logics?’ Mini-course and Workshop

27 September – 1 October, 2010, University of St Andrews

The Arch ‘Logic or Logics?’ Mini-course and Workshop are organised by the members of the AHRC funded Foundations of Logical Consequence project. The Mini-Course is intended for graduate students and younger researchers (postdocs and junior faculty) working on related topics. The aim is to provide intensive graduate-level instruction on the latest thinking about pluralism and revision in logic. Topics will include the revision of logic debate, logical pluralism vs. absolutism (or monism), and combining logics. The week will conclude with a Workshop dedicated to contemporary research on the same theme.

The speakers for the Mini-course are:
       – JC Beall (University of Connecticut)
       – Carlos Caleiro (Instituto Superior Tcnico, Portugal)
       – Joo Marcos (DIMAp / UFRN, Brazil)
       – Graham Priest (University of Melbourne/CUNY)
       – Greg Restall (University of Melbourne)
       – Gillian Russell (Washington University, St Louis)
       – Johan van Benthem (University of Amsterdam

We invite interested parties to register here: http://tinyurl.com/2wtxkp6
For the full programme see the event website here: http://www.standrews.ac.uk/~arche/events/event?id=398
Any further inquiries should be directed to arche@st-andrews.ac.uk
Messages to the list are archived at http://listserv.liv.ac.uk/archives/philos-l.html.
Prolonged discussions should be moved to chora: enrol via http://listserv.liv.ac.uk/archives/chora.html.
Other philosophical resources on the Web can be found at http://www.liv.ac.uk/pal.