Research

Francis Jeffry Pelletier Dept. Philosophy, Dept. Linguistics Simon Fraser University

Research

Publications

Courses

Research Links

Current Research My current research centers on topics in Language and Logic. I bring to these studies a strong inter-disciplinary perspective that unites elements of linguistics, artificial intelligence, philosophy, and cognitive science. I am particularly interested in the issue of ’Äúinference’Äù: theoretically, as defined by Logic; semantically, as it manifests itself in natural language; computationally, as it might be implemented in automated reasoning systems, agent communication methods, and knowledge representation schemes; and psychologically, as it is exemplified by ordinary people’Äôs reasoning. I also have a research interest in the history of this notion, as it plays out in such figures as Plato, Aristotle, Frege, Russell, Jaskowski, and Gentzen.	Research Support Funding from these groups supports graduate students and the research associated with the projects listed here. NSERC Research Grant Program SSHRC Canada Research Chair Pgm. Canadian Foundation for Innovation B.C. Knowledge Development Fund Simon Fraser University Other Connections I attend the Computational Logic research group meetings, with an interest in many areas of logic and computation. I have research collaborations with scholars at many different universities. Some of these linkages can be found here (as well as links to some friends, whose work I find interesting).
Issues in Generics There are two types of ’Äúgeneric statements’Äù: one concerns reference to ’Äòkinds’Äô, as in The Dodo is extinct and the other concerns statements that are ’Äògenerally true’Äô, such as The Lion has a mane. Concerning the first type there are many interesting issues both of the philosophical sort (What are kinds? What is reference to kinds anyway?) and the linguistic sort (How is such reference achieved in different languages? What is the correct relation between syntax and semantics in this realm?). The second type is especially interesting because these statements can be true even in the face of ’Äúexceptions’Äù (not all lions have manes). In turn, this raises questions about the nature of truth, about the correct semantic description of this feature, and about how people actually use these ’Äúgeneral truths’Äù in their reasoning about the world. Such statements figure prominently in computational accounts of knowledge representation, and I believe that an adequate account of the philosophical, psychological, and linguistic issues can provide a good framework from which to approach the computational issues. Students who are interested in working on this topic from any of these points of view are encouraged to contact me.
Issues with Mass Terms Mass terms are such as ’Äòwater’Äô and are contrasted with count terms such as ’Äòperson’Äô. Intuitively, mass terms designate ’Äústuff’Äù whereas count terms designate ’Äúthings.’Äù There are many syntactic, semantic, and pragmatic features that have been proposed to mark this distinction, but I think none of them work correctly. There is a linguistic issue about the correct syntactic-semantic account that should be given to this distinction, and there are also many philosophical issues that are related to it. In the latter category we find such metaphysical topics as the relation between things and the stuff that comprises them, and as well there are issues about the ultimate nature of reality’Äîis it thing-like or stuff-like? This topic is related to the notion of a sortal predicate, which is somehow supposed to limn the underlying nature of human thought. In turn this raises questions concerning what an adequate ’Äúsortal logic’Äù might be. Furthermore, mass terms are often used in generic statements (of both types), and so they too give rise to all the topics mentioned above about generics. Students interested in doing research on the logical, linguistic or philosophical issues raised by mass and sortal terms are encouraged to contact me. Vagueness Vagueness comes in many different forms, or so it is usually held. One type is realistic’Äîthe vagueness is traced to there being some item of reality which is ’Äúvague’Äù. Another type is epistemic’Äîwhere the vagueness is traced to a lack of knowledge by someone who is evaluating some (completely precise) feature of reality. And still a third type is claimed to be linguistic’Äîthe vagueness is traced to some feature of language (or other representational system). Some theorists allege there is but one type of vagueness, ’Äúreally’Äù, and that the other types are somehow an illusion or to be defined in terms of the one basic type. Each type of vagueness suggests a particular sort of underlying logic that accurately captures its basic presuppositions. One direction of my research has been to characterize these differing logics from a formal point of view. But the resulting logics predict different formulas to be theorems. To determine whether one or another of these logics is the one actually employed by ordinary people in their reasoning about vagueness, I am interested in conducting experiments employing these different logics. Students who are interested in psychological investigation are asked to see me about work in this area.
Compositionality The Principle of Semantic Compositionality is the view that the meaning of a syntactically complex expression is a function of the meanings of its simpler parts. This Principle is debated in Computational and Formal Linguistics and Semantics and in Linguistics as a whole. It is also related to wider disputes concerning compositionality and holism in many subareas of Cognitive Science, such as whether there can be thoughts of complexes that are not completely determined by thoughts of the parts of the complex and how the parts are combined. Speaking loosely, compositionality is the opposite principle to the claim that ’Äúthe whole is greater than the sum of its parts’Äù , and is sometimes called The Principle of Contextuality and sometimes The Principle of Holism. There are many subtleties involved with these general principles, even just in linguistic semantics, where my work has mostly taken place. And once one moves into theories in the social or cognitive sciences, further intricacies arise. I am interested in many aspects of the debate: from linguistic theories and natural language examples that allegedly violate compositionality to philosophical discussions about the dispute between Holism and Atomism to the logical question concerning whether all theories can be given a compositional account. Students interested in any aspect of this issue are encouraged to see me about possible research topics.
Semantic Tableaux A semantic tableaux method for a given logical system attempts to provide an efficient method by which formulas can be evaluated for semantic validity. I have been working on such methods for classes of logics that are thought to be of use in Artificial Intelligence: modal logics, many-valued logics, and fuzzy logic. Work on this topic is ongoing, and students with an interest in formal-logical issues (or in the relation between formal topics and artificial agents) are encouraged to contact me about research in this area.
Non-Monotonic Reasoning and Belief Revision When a reasoner recognizes that some newly-acquired information causes a conflict with a set of currently-held beliefs, there may be different ways of resolving that conflict, and these different ways correspond to abandoning (or at least calling into question) different ones of the currently-held beliefs. The process according to which this is performed is called belief revision. One issue that arises under this characterization concerns the principles by which such a decision is made, i.e., the decision to abandon belief i rather than belief j, so that the new information can be incorporated into a consistent situational model. Non-monotonic reasoning occurs when an argument from known premises to a conclusion x can be overturned by the incorporation of further premises (even though none of the previous premises are given up). Both non-monotonic reasoning and belief revision are crucial parts of everyday reasoning, scientific reasoning, and even (artificial) agent communication. My research in this direction has been to characterize how people perform these sorts of tasks, and to use these results to inform the accounts of these processes that have been proposed in the Artificial Intelligence literature. While I have no current projects in this area (except as indicated under ’Äúgenerics’Äù above), most of my publications on belief revision are on line, and if students find some topics in this area to be intriguing, I would be interested in directing some further research.