Projects in Previous Years

2nd Semester 2014/15: Introduction to STIT Logic

Roberto Ciuni
If you are interested in this project, please contact the instructor by email.
This project introduces students to STIT logic (the logic of 'seeing to it that') for reasoning about individual agency, the interaction of many agents and group agency. STIT logic originated in the early Nineties with a series of papers by Belnap, Perloff, Xu and Horty, and it later developed a rich framework for the logical foundations of multi-agent systems and strategic reasoning. STIT logic extends the language of propositional logic with many (agent-relative) modal operators for agency, the so-called 'stit operators'. The semantics of stit operators incorporates interesting aspects of normal-form games from Game Theory. For instance, in STIT the interaction of many agents obeys an 'axiom of independence' that expresses a 'no action-profile gaps' condition, while the construction of the choices of a group obeys conditions of 'superadditivity' (the combined choice of disjoint groups is more powerful than the choices of the groups in question) and 'monotonicity' (a group achieves at least everything its members achieve, and possibly more). These features motivate the current applications of STIT logic to consequentialist oughts and group agency.

The project will focus on these applications, it will explore the limits of current proposals and potential for refinements. The overall aim of the project is to provide students with an overview of STIT logics, its application to multi-agent systems and its connections with basic game-theoretical notions, so that students can incorporate reasoning about 'seeing to it that' in their own research, if they so desire.

The course will be a mixture of introductory lectures by the instructor and group discussion. The introductory lectures (first four meetings) will introduce the basic STIT setting for individual agency, the deontic extension provided by Horty, the connections with normal-form games, and the extension of STIT with group agency operators. In the remaining four meetings, student will participate to discussion on features of STIT that will be met in the introductory lectures. This will also be a chance for individuating and exploring the topic of their written reports.
Familiarity with basics of modal logic, deontic logic and game theory will be helpful. Although this course does not assume knowledge of specific aspects of multi-modal logics, some familiarity with product logics would be helpful too. This course will be self-contained, and will not assume familiarity with the research area of multi-agent systems (MAS).
Students will be evaluated by a combination of written homework sets, and a short written report. The written report should provide either a survey of some of the literature with a new perspective or some (small) original result. Students are expected to actively participate in all class discussions.