Schedule (tentative), scroll down for abstracts.

Wednesday, March 23

9:00-9:50 Andy Egan (Philosophy, Rutgers), What I Probably Should Have Said About Epistemic Modals.

9:50-10:10 Comment: Malte Willer (Philosophy, UChicago)

10:10-10:30 Q&A

10:40-11:30 Angelika Kratzer (Linguistics, UMass), Evidential moods.

11:30-12:00 Q&A

Lunch break: 12:00-14:00

14:00-14:50 Kai von Fintel (Linguistics, MIT) & Thony Gillies (Philosophy, Rutgers), Still going strong: The semantics and pragmatics of epistemic must.

14:50-15:10 Comment: Dan Lassiter (Linguistics, Stanford)

15:10-15:30 Q&A

15:40-16:30 Craige Roberts (Linguistics, OSU) & Stewart Shapiro (Philosophy, OSU), Logical omniscience and the sense of epistemic modals.

For slides, click.

16:30-16:50 Comment: Rich Thomason (Philosophy, Michigan)

16:50-17:10 Q&A

Thursday, March 24

9:00-9:50 Giorgio Sbardolini (Philosophy, OSU), An Aristotelian theory of modality.

9:50-10:20 Q&A

10:30-11:20 Øystein Linnebo (Philosophy, Oslo) & Stewart Shapiro (Philosophy, OSU), Potential Infinity: a modal account.

11:20-11:40 Comment: Ethan Brauer (Philosophy, OSU)

11:40-12:00 Q&A

Lunch break: 12:00-14:00

14:00-14:50 Stefan Kaufmann (Linguistics, UConn), Guesses, Suppositions and Updates.

14:50-15:20 Q&A


  • Jefferson Barlew (Linguistics, OSU)
  • Ethan Brauer (Philosophy, OSU)
  • Janice Dowell (Philosophy, Syracuse)
  • Michael Glanzberg (Philosophy, Northwestern)
  • Hans Kamp (Institute for Natural Language Processing, Stuttgart)
  • Teresa Kouri (Philosophy, OSU)
  • Dan Lassiter (Linguistics, Stanford)
  • Eric Snyder (Linguistics, OSU)
  • Rich Thomason (Philosophy, Michigan)
  • Malte Willer (Philosophy, UChicago)


  • A. Egan (Rutgers), What I Probably Should Have Said About Epistemic Modals.

I argued in (2005,2007) for a de se relativist account of epistemic modals, primarily based on arguments from eavesdropper’s assessments of the truth-values of epistemic modal claims taking place in the conversations they’re eavesdropping on. While I still believe that the view I advocated there is approximately correct, there are various errors of substance and presentation in those papers that I now regret. My main goal in this paper is to state a better version of the same sort of de se relativist view, state it more clearly, and make a better argument for it. Under this heading, I will aim to (a) make clear that, and how, my preferred sort of relativist view is Kratzer-continuous, in a sense to be explained below, (b) make clear(er) the role of the different moving parts of the view, including the commitments about the theoretical role of semantic content, and (c) offer a better version of the sort of eavesdropping argument I made in (2005,2007). My second goal is to make explicit the points at which this sort of de se relativist view differs from the sort of relativism John MacFarlane (2009) advocates, and identify what I take to be the central selling points of a de se relativist view of epistemic modals over MacFarlanian relativism.

  • K. von Fintel (MIT) and T. Gillies (Rutgers), Still going strong: The semantics and pragmatics of epistemic must.

In our paper “Must … stay … strong!” (von Fintel & Gillies 2010, henceforth MSS), we set out to slay a dragon, or rather what we called The Mantra: that epistemic “must” has a modal force weaker than expected from standard modal logic. The Mantra is supported by intuitive judgments that a speaker who utters “It must be raining” is signaling a certain lack of confidence in the conclusion that it is raining. In MSS, we argue that this is not a semantic fact about “must”. Rather “must” means something like “it follows from the evidence that”, which is semantically strong in two ways: “must” is a strong necessity modal and its modal base is a realistic one. This conspires to ensure the entailment from “must p” to “p”. We argue that “must”, in addition, carries a signal that the conclusion is not settled directly by a privileged subset of the evidence (roughly, trustworthy observations and reports). This evidential signal, sometimes together with a “the lady doth protest too much” implicature, can give rise to pragmatic inferences about the speaker’s level of confidence. In MSS, we suggested that it would be nice to derive the evidential signal as a conversational implicature, but concluded that we couldn’t work that out and then argued that it must be a presupposition.

In the mean time, several scholars have taken up the cause of the Mantra (Lassiter, Giannakidou & Mari, for example) and some have proposed a different derivation of the evidential signal (Mandelkern, for example). In our talk, we will discuss these proposals and try to reach an adjudication.

  • S. Kaufmann (UConn), Guesses, Suppositions and Updates.

A number of independent recent proposals on epistemic modals and conditionals show a convergence of ideas on the relationship between the truth conditions and the epistemic properties of a certain kind of modal meaning. Specifically, this concerns proposals on English ‘will’ (Cariani and Santorio), counterfactual conditionals (M. Schulz), and conditionals in general (Kaufmann). What these proposals share in common, roughly speaking and glossing over important differences in implementation, is that (i) they give truth conditions not in terms of quantification over modal-base worlds, but in terms of a single modal-base world, and (ii) the selection of this modal-base world is arbitrary in some precise sense. My goal in this talk is to pull these strands together into a general formal framework, with some immediate payoff as well as some promising avenues for future development.

The framework employs the “Bernoulli models” of van Fraassen (1976; see also Stalnaker and Jeffrey 1994, Kaufmann 2009, 2015). Its main attraction in the area of conditionals lies its ability to deliver a probabilistic analysis of simple and compounded conditionals that is not subject to Lewis-style triviality problems. The intriguing part in view of the recent proposals is that this kind of approach may not only (and not even primarily) be applicable to conditionals per se, but rather to the epistemic modal which surfaces as ‘will’ in indicatives and as ‘would’ in counterfactuals. The idea that all these expressions are cut from the same cloth is not new in linguistics, and merely a shift of emphasis in terms of the formal analysis, but interesting nonetheless for its potential to provide a general framework for modeling the logical and epistemic profiles of modal and conditional sentences. To illustrate, I will outline concrete ideas about (some subset of) belief dynamics, modal subordination, conditionals and unconditionals.

  • A. Kratzer (UMass), Evidential moods.

My talk will be about moods in the evidential family: the indicative and the German reportative subjunctive (Konjunktiv I). Evidential moods, and moods more generally, are usually seen as dependent or ‘selected’ by certain verbs that embed sentential complements. This is a mistake, I think. I will argue that evidential moods are not selected by attitude verbs, they are responsible for CREATING the modal semantics of attitude verbs.

  • Ø. Linnebo (Oslo) and S. Shapiro (OSU), Potential Infinity: a modal account.

Beginning with Aristotle, almost every major philosopher and mathematician before the nineteenth century rejected the notion of the actual infinite. They all argued that the only sensible notion is that of potential infinity. The list includes some of the greatest mathematical minds ever. Due to Georg Cantor’s influence, the situation is almost the opposite nowadays (with some intuitionists as notable exceptions). The received view is that the notion of a merely potential infinity is dubious: it can only be understood if there is an actual infinity that underlies it.

After a sketch of some of the history, our aim is to analyze the notion of potential infinity, in modal terms, and to assess its scientific merits. This leads to a number of more specific questions. Perhaps the most pressing of these is whether the conception of potential infinity can be explicated in a way that is both interesting and substantially di§erent from the now-dominant conception of actual infinity. One might suspect that, when metaphors and loose talk give way to precise definitions, the apparent differences will evaporate.

As we will explain, however, a number of differences still remain. Some of the most interesting and surprising differences concern consequences that one’s conception of infinity has for higher-order logic. Another important question concerns the relation between potential infinity and mathematical intuitionism. We show that potential infinity is not inextricably tied to intuitionistic logic. There are interesting explications of potential inifnity that underwrite classical logic, while still differing in important ways from actual infinity. However, we also find that on some more stringent explications, potential infinity does indeed lead to intuitionistic logic.

  • C. Roberts (OSU) and S. Shapiro (OSU), Logical omniscience and the sense of epistemic modals.

Suppose that in 1990, a mathematician said that

(1) In view of what is known, Fermat’s Last Theorem might be false.

The intuitive verdict is that, in that context (or as assessed from that context), (1) is true. No proof of Fermat’s last theorem known then—by anyone.

On virtually all extant accounts of epistemic modals, however, that utterance of (1) comes out false. The various accounts have it that we first identity an epistemic (or evidential) base from the context. An utterance in the form “it might be that P” is true just in case the prejacent P is consistent with that base. Since our utterer was a mathematician, we can assume that the base includes the axioms of Dedekind-Peano arithmetic (and the rudiments of elliptical function theory). The fact is that the negation of Fermat’s Last Theorem is not consistent with that epistemic base—and never was. That is what Wiles showed (in 1994). We thus confront the matter of logical omniscience. The problem is that modeling epistemic possibility

in terms of quantification over possible worlds presupposes that we know what worlds are possible. But the (sad) fact is that we do not know what is and what is not possible. We did know in 1990 what (1) means. But what we did not know, then, is that Fermat’s Last Theorem is true, and hence that Fermat’s Last Theorem is not false in any possible world. All that was known in 1990 was that if true, Fermat’s Last Theorem was true in all possible worlds, and that if false, Fermat’s Last Theorem was false in all possible worlds.

Here is one way to put it: We knew, then and now, the “sense” of “Fermat’s Last Theorem is false”, but we didn’t yet know its “referent”: the set of possible worlds in which it’s true. Hence, giving the meaning of an epistemic modal statement like (1) in terms of possible worlds gives merely its Bedeutung. We’re after its Sinn.

We consider a number of possibilities that have been suggested for the problem of logical omniscience and epistemic modality, including structured propositions, hyperintensionality, and impossible worlds, and argue that none of these is adequate to the task. We take a Lewisian approach to the problem, asking the question What would the Fregean sinn of a modal proposition do?, and what can we use to model that behavior? We take the desiderata for an adequate account to include that it (a) make use of an epistemic or evidential modal base (Kratzer 1981, von Fintel & Gillies 2010, Roberts 2015); (b) permit finer distinctions between the meanings of modal statements than are possible with a simple set-of-worlds approach; (c) yield a sense that’s intuitively graspable by human agents: something we can reason with, relate to our beliefs and desires, etc.; and yet (d) yield a sense that is public, shared by other competent speakers (the externalist criterion); and (e) include a satisfying account of what it is for evidence to be available for drawing the required inferences, a notion which Stalnaker (1991,1999) and Parikh (2008) argue is crucial for understanding the problem of logical omniscience. Finally, (f) ideally, the resulting account should work smoothly with truth conditional semantics, wherein propositions are sets of possible worlds. That is, the bedeutung of a modal statement should be the proposition expressed, a set of possible worlds; and its sinn should determine that bedeutung, in keeping with Frege’s characterization of those notions.

In this work in progress, we offer a first sketch of an account, wherein the sense of a modal statement is captured in a Discourse Representation Structure. Satisfying contextual presuppositions, like that of a Kratzerian modal base, was the original raison d’etre of DRSes (Kamp 1981, Heim 1982; for modals Roberts 1989, Frank 1997) (criterion a). Like structured propositions, such structures are finer-grained than sets of possible worlds (criterion b), but are arguably better candidates for graspable, public information states than possible worlds (criteria c and d), and their use in reasoning has been fairly well-explored (Kamp & Reyle 1993, Sauer 1993, Kamp, Genabith & Reyle 2011), making them suitable to serve as the basis for natural deduction and abductive inference. We offer an account of how such abstract representations of the sense of modal statements give us the means of explaining what it is for contextual information to be evident and available for drawing inferences, satisfying criterion (e). And DRSes can be verified in possible worlds (Kamp 1981, Kamp & Reyle 1993), so that they are compatible with the usual truth conditional semantics (criterion f). Hence, just as we now know the sense of the next President of the United States, but do not yet know its referent, so in 1990 we knew the sense of (1), but we could not yet say what worlds made it true precisely because we didn’t yet know what worlds were possible.

  • G. Sbardolini (OSU), An Aristotelian theory of modality.

I present an account of the distribution of epistemic readings for modal auxiliaries in the present tense, with evidence from English and Italian, based on Krifka’s (1998) algebraic semantics for telicity. Since the telic/atelic distinction is said to date back to Aristotle (like pretty much everything else), this is an Aristotelian theory. The basic observation is that epistemic modal bases are compatible only with atelic complements, which may be syntactically diverse, i.e. including simple stative predications, habituals or progressives. Furthermore, present tense epistemic modality is restricted to an interpretation of the time at which the complement is realized as overlapping the evaluation time of the modal (which, in the unembedded contexts I’ll be concerned with, is utterance time). This account falls out naturally from the evidential restrictions associated with the use of epistemic modals, assuming that speakers systematically cast aside evidential claims about events that have not yet taken place at the time of utterance, in favor of non-epistemic readings on which the event is understood as, say, required or permitted to come about in the near future.