This is a post by Teresa Marques (University of Barcelona).
Imagine the following scenario: My husband and I go to the supermarket. When we get home, we bring the shopping bags from the car to the kitchen and as we start to put things away in the fridge, I notice that the eggs are missing. I wonder if we left them in the car. And so, I ask my husband for the car keys, to which he replies:
– The keys might be on the table.
He says ‘might’ because he is not completely certain if they are on the table, in his backpack, or in the pocket of his jacket. Note that my husband does not know that he left the keys on the table, otherwise he would have just said “the keys are on the table”.
The sentence he uses is what people call a modal sentence. We use modal sentences like this all the time. When we use them, we are not asserting that it is metaphysically possible that the keys are on the table. Obviously, there is a possible world very close to our own where that is the case! Because of that, it would be silly of us to assert something that everyone already knows to be true. That sentence is, more specifically, an epistemic possibility modal sentence. It says what is possibly true given certain information that is available to him in the context (situation or occasion) where he spoke.
Epistemic modal sentences characterize epistemic states. The canonical interpretation of epistemic modal claims is contextualist. Given a sentence like that in (1) below, when we use the epistemic possibility modal, we convey what is possible given relevant information available to me or to us when we speak, that is, at our context of utterance:
(1) The keys might be on the table.
In my recently published paper, “Falsity and Retractions: New Experimental Work on Epistemic Modals”, I discuss and report experimental studies on people’s intuitions about the truth-conditions associated with an assertion of sentences like (1), and about the conditions in which the retraction of such assertions would be required.
Why does this need testing? In recent decades, some philosophers have challenged the standard view of epistemic modal sentences. On the standard view, my husband’s assertion that the car keys might be on the table is true (as made on that occasion) just in case it is possible that the keys are on the table given the information that was available then – that is, at the context of utterance. Moreover, his assertion of that sentence is correct only if the asserted sentence is true in those same conditions. If what he said had not been true, then it would have to be impossible that the car keys were on the table, given what was known then. For instance, if we had both known that he had either left the keys in the car or in his backpack, then we would know that it would not be possible for the keys to be on the table. And if what he said were not true, then his assertion would be incorrect.
Now, some semantic relativists challenge this because of alleged shared intuitions. Suppose that we are so busy putting things away that we temporarily forget the question of where the eggs are. A couple of hours later, I look at the table and see that the car keys are not there. I proceed to tell my husband, “The keys are not on the table”. This is a new conversational context, one where we can assess his assertion from two hours ago.
What do you think he should say?
– (A) Should he say, “I guess I was wrong, what I said was false. I take it back”? Is he required to retract his earlier statement?
– (B) Or can he say instead, “I see that the keys are not on the table, but I didn’t say they were there. I said that they might have been there.”
Relativists, in particular assessment-relativists, claim that the right view is (A): my husband is required to retract what he said, and that what he said two hours ago is false, because the non-modal sentence, “the keys are on the table” is false now.
What is assessment-relativism, then? Roughly and briefly, it is a metasemantic view first advocated by John MacFarlane about the relativity of our assertions to both contexts of utterance (or use) and contexts of assessment. This departs from standard contextualist views that admit for variations in the truth-conditions of a given sentence with respect to various parameters of the context in which it is used (speaker, audience, place, time, available information, etc.) What is distinctive of relativism is the pragmatic difference in the conditions for the retraction of assertions, and not merely the relativity of the truth of interpreted sentences to various parameters beyond possible worlds.
No one in this literature says that one should not aim to assert what is true at the context of utterance. But not everyone agrees that one ought to retract a correctly made assertion. I.e., not everyone agrees that an assertion’s correctness, or “accuracy”, is itself relative. Unless the conditions of required retractions are sufficiently well supported by speakers’ intuitions, as studied for instance through cross-linguistic empirical studies with many subjects, then assessment-relativism is unmotivated. This is the evidence I sought in the two studies reported in the article.
Joshua Knobe and Seth Yalcin had previously run some tests to check what people (North-American native English speakers) thought about similar cases. Their results did not confirm relativist intuitions regarding falsity judgments. But, in my view, their tests didn’t ask the right retraction question, because they did not ask whether people think that a speaker is required to retract, which is necessary to motivate assessment relativism in the first place! I tested people’s opinion with this stronger formulation, both with native English speakers in the USA, and with native Spanish speakers in Spain.
My results confirmed that people are not inclined to judge that the assertions like the example in (1) above are false when people have information that falsifies the related non-modal sentence (“the keys are on the table”). Indeed, and against relativist predictions, both English speakers and Spanish speakers disagreed, or tended to disagree, that the epistemic possibility modal claim is false, and likewise disagreed, or tended to disagree, that the speaker is required to retract his past assertion. Interestingly, speakers diverged in their judgments about when a retraction is appropriate and when it is required. Marcus Kneer has in the meanwhile reached similar results.
Why does this matter? Because it undermines the very motivation to introduce these additional parameters: contexts of assessment.
To read more about this, check out my article, “Falsity and retraction: New experimental data on epistemic modals” which appeared this year in Retraction Matters: New Developments in the Philosophy of Language. The preprint version of the article can be freely accessed here.
Teresa Marques is an Associate Professor (Profesora Agregada) at the University of Barcelona. She is a philosopher of language, and has interests in metaethics, epistemology, and feminist and social philosophy. In recent years, she has been working on disagreement and retraction, conceptual engineering, evaluative language, and the expression of emotions in language.
Open Philosophy 