Search

With two further studies, we aim to address concerns raised by our reviewers at the journal Mind & Language. First, combining the results from studies 1 and 3 on the one hand, and Study 2 (i.e., the study discussed under IB. Baseline in the previous preregistration document) on the other, we see a reversal from “must” (preferred in Study 2) to “was supposed to” (preferred in studies 1 and 3). We claim (a) this reversal effect is explained by the impossibility present in studies 1 and 3 but not in Study 2; and (b) that (a) is in turn explained by MIC. However, there might be other ways of explaining the reversal effect. The most obvious difference between Study 2 and studies 1 and 3 is that in the former (i.e., Study 2), the choice is between “is supposed to” and “must”, whereas in the latter (i.e., studies 1 and 3), it is between “was supposed to” and “must”. It is possible that “was supposed to” would be preferred to “must” across the board, if the options remained constant across studies. If that were the case, there would be no real reversal effect to speak of, and MIC could contribute nothing to explaining the only remaining effect, which, by hypothesis, is that “was supposed to” is preferred to “must” irrespective of impossibility, but “is supposed to” is not. Thus, we need to show that in situations such as Adam’s in our story, there is a clear preference for “was supposed to” over “must” in cases of impossibility, but not when impossibility is removed. We already showed the preference for “was supposed to” in cases of impossibility in studies 1 and 3. The first new study addresses the issue of the preference for “must” over “was supposed to” when the impossibility is removed—discussed in detail in Section I below (Baseline Proper). This reversal is predicted by MIC but is not currently predicted by any alternative theories, or at any rate, none that we are aware of. There is a second issue, however, regarding the structure of this prediction. MIC predicts the preference pattern described above in virtue of the general assumption we call No contradiction in the main text of our manuscript: all else being equal, speakers will resist endorsing contradictory propositions. Now, what follows directly from MIC is, uniquely, that “must but cannot” expresses a contradictory proposition; given this and No contradiction, the larger prediction of MIC is that “must but cannot” should be seen as a bad sentence. However, none of studies 1 to 4 provide clear evidence for “must but cannot” being perceived as bad (in fact, Study 4 may be seen as providing evidence against it). Here is why. What we see in studies 1 and 3 is that “was supposed to” is preferred to “must” as the better description of Adam’s situation. That the former is chosen as a better description does not entail that the latter is excluded as a bad description, since the two options are not mutually exclusive: it may be true that Adam was supposed to go on the expedition and it may also be true that he must still do it. Therefore, to further probe whether MIC is on the right track not only in terms of predicting the reversal effect, but also in terms of explaining why it happens, we need further information about how participants judge the “must” sentence, in particular, those participants who choose “was supposed to” as the better description. The second new study addresses this issue—discussed in detail in Section 2 below (Graded Follow-Up).

In our paper, we put forward a formal semantic theory of the interaction between deontic and circumstantial modal verbs. We refer to our model as MIC (or “Must Implies Can”). To cut a long story short, according to MIC, sentences of the form “S must φ, but (s)he cannot” express a logical contradiction. We shall refer to such complex sentences as M-sentences. The general method of our studies is the following. Each participant is asked to a read a story that involves: (a) the imposition of an obligation on an agent and (b) a subsequent event that makes the performance of the obligatory action impossible. Then, the participant is asked to evaluate the appropriateness of an M-sentence. M-sentences are contrasted with sentences of the form “S was supposed to φ, but (s)he cannot”. We shall refer to these sentences as displaced sentences, as they temporally displace the obligation, pushing it back in the past. One of our main predictions is that participants will opt for the displaced alternative.

Follow up study to earlier preregistration

International audience; An open question in the semantics of modality is what relations there are among different modal flavours. In this article, we consider the thorny issue of whether ascribing to an agent the obligation to φ implies that it is possible for the agent to φ. Traditionally, this issue has been interpreted as whether ‘ought’ implies ‘can’. But another linguistic interpretation is available as well, namely, whether ‘must’ implies ‘can’ (MIC). We show that ‘must’ does imply ‘can’ via a convergent argument. The first strand of the argument is theoretical: it consists in proving MIC from a well- established theory of modality in natural language, i.e., that proposed by Kratzer. The second strand is empirical: we present novel acceptability judgment studies showing that MIC predicts and explains the linguistic behaviour of native English speakers.

An open question in the semantics of modality is what relations there are among different modal flavours. In this article, we consider the thorny issue of whether ascribing to an agent the obligation to φ implies that it is possible for the agent to φ. Traditionally, this issue has been interpreted as whether ‘ought’ implies ‘can’. But another linguistic interpretation is available as well, namely, whether ‘must’ implies ‘can’ (MIC). We show that ‘must’ does imply ‘can’ via a convergent argument. The first strand of the argument is theoretical: it consists in proving MIC from a well- established theory of modality in natural language, i.e., that proposed by Kratzer. The second strand is empirical: we present novel acceptability judgment studies showing that MIC predicts and explains the linguistic behaviour of native English speakers.

Can two people disagree without having conflicting beliefs? The philosophical literature is unevenly divided on this question. The majority, default, view (henceforth “Majority View”) is that disagreement is always grounded in a tension between beliefs, while the minority view (henceforth “Minority View”) is that disagreement is possible in virtue of a tension between other attitudes such as desires, intentions, emotions, or plans. Both sides agree that there are apparent cases of non-belief disagreement. However, according to the Majority View, such an apparent disagreement is merely derivative; that it is always an expression of disagreement in the underlying beliefs. In the study, participants are asked to read a story describing two individuals with differing desires under circumstances where it is hard to ascribe to them differing beliefs about what these desires should be. We aim to establish whether participants readily apply the term “disagreement” to the relation between the two individuals. In a study involving 300 participants, we manipulate two variables to forestall alternative explanations of our predicted results, which we take to be a refutation of the idea (attributable to the Majority View) that it is intuitively true that individuals can only disagree if there is a tension between their beliefs. The full description is provided in the preregistration file.