Santorio on being neither able nor unable
Some ability statements sound wrong when affirmed but also when denied. Santorio (2024) proposes a new semantics that's built around this observation.
Suppose Ava is a mediocre dart player, and it's her turn. In this context, people often reject (1):
It's obviously possible that Ava gets lucky and hits the bullseye. But ability seems to require more than mere possibility of success. A common idea, which Santorio endorses, is that ability comes with a no-luck condition, something like this:
But here's a puzzle. The negation of (1) also looks problematic:
(2) seems to convey that Ava is guaranteed to fail if she tried.
To make sense of this, Santorio suggests that the no-luck condition isn't in the asserted content of (1), but in a "not-at-issue" component that is preserved under negation. To a first approximation, his semantics goes as follows.
I'll pause here to register some reservations.
I don't agree that ability reports generally come with a no-luck requirement. For example, imagine we're not interested in the dart game, but in how far Ava can throw the dart. You think she can't even reach the wall. Pedro thinks she can reach it, but only the lowest parts. I think she is stronger: she can hit every point on the wall, including the bullseye. In this context, (1) is true. More generally, I think ability reports like (1) have two readings, a strong reading and a weak reading. Only the strong reading comes with a no-luck condition. I've argued for this at some length in Schwarz (2020), which Santorio seems to have read, but he's evidently not convinced.
Here's another example from my paper, which I think brings out the two readings more clearly. Suppose Maisy doesn't know the combination to a safe, and has little time to try different combinations. In this situation, (3) is true in one sense and not true in another:
To bring out the stronger reading, imagine we need access to the safe's contents. "Is anyone able to open the safe?" you ask, in a hurry. In this context, (3) doesn't seem true. On the other hand, Maisy can obviously dial the right combination. It is in her power to open the safe. Focusing on this, (3) seems true.
Now, importantly, (3) isn't just untrue on its strong reading, it's false – meaning that its negation is true. "Is anyone able to open the safe?" you ask, in a hurry. Maisy: "Sorry, I'm not [able to open the safe]. I don't know the combination." On its weak reading, however, where (3) is true, its negation is false. It's in Maisy's power to open the safe. So it's not true that she is unable to open the safe.
I'd say the same about (1) and (2). (1) has a weak reading on which it is true, and a strong reading on which it is not. On the weak reading, the negation (2) is true. On the strong reading, it is not. There is nothing non-classical here that would call for a not-at-issue content that's preserved under negation. Santorio creates the appearance of non-classicality by presenting (1) in a context where the strong reading is salient and (2) in a context where the weak reading is salient.
Let's move on. There are other, more convincing cases in which an ability report and its negation both seem problematic. Santorio considers the case of a baby safely strapped into a carrier. Here, (4) is obviously not true, but (5) also sounds bad:
The badness of (5) is puzzling. The baby is guaranteed not to fall, no matter what it tries to do. Why, then, does (5) sound odd?
Santorio suggests the following diagnosis: that the baby won't fall is due to how its strapped into the carrier; it's not due to properties of the baby. When we talk about whether S has or lacks the ability to φ, we presuppose that whether S φs depends on intrinsic properties of S.
In fact, Santorio suggests that whether S φs should depend only on intrinsic properties of S, together with the laws of nature. To a second approximation, his semantics looks like this:
The revised not-at-issue component subsumes the no-luck condition in (S2): in normal cases, if it's a matter of luck whether S φs then their φing is not determined by their intrinsic properties and the laws.
To motivate the idea that S's success at φing should not depend on external circumstances at all, Santorio offers the following case:
Magical dart: Ava is bad at darts, but Camille, a magician, is going to cast a spell that will make her next throw hit the bullseye.
Ava is guaranteed to hit on her next throw, but Santorio intuits that (1) still isn't true: it's not true that Ava is able to hit the bullseye on her next throw. (S3) explains why: her success depends on the presence of the magician.
My account in Schwarz (2020) gives a different explanation of why (1) might still seem problematic. I suggest that the no-luck condition that's built into the strong reading of 'able' is epistemic. My account predicts that (1) is unambiguously true if Ava knows of Camille's plan. This seems right to me. (S3) gets this version of the case wrong. But it has much more serious flaws.
I am, right now, able to throw my coffee cup into the air. But I'm not throwing it. Do my intrinsic properties, together with the laws, entail that I don't throw it? If so, (S3) gets this completely normal case wrong. Let's suppose that we can avoid this result, perhaps by restricting the "intrinsic properties" or by insisting that I only have the ability to throw the cup in a moment, not right now. (S3) still gets the case wrong. For surely my intrinsic properties, together with the laws, don't entail that I do throw the cup.
The problem is that we've dropped the "if she tries" condition when moving from (S1) to (S3). We should have said this:
That's still not enough. If you only knew my intrinsic properties and the physical laws, you couldn't tell if I would succeed to throw my cup if I tried. It depends, among other things, on whether the cup is in reach, whether I'm in a suitable gravitational environment, and so on. Let's fix this as well:
Santorio doesn't make either of these repairs. He doesn't like an if S tries condition because (like me) he wants to allow for ability statements in which the subject is not an agent. He doesn't want to add the external circumstances because (a) he then couldn't account for the supposed untruth of (1) in Magical Dart, and (b) he couldn't explain what's wrong with (5) in the baby carrier scenario. He does, however, note that (S3) doesn't work. What he suggests is this:
The relevant background propositions, Santorio says, often include counterfactual assumptions about what the agent tries to do. We could get the right result in the coffee cup scenario if we assume that the background propositions also include facts about the presence of the cup etc.
Unfortunately, (S6) makes no predictions about whether 'S can φ' is true in a given situation, except in the unusual case where S's intrinsic features and the laws settle whether S φs. Is a mediocre dart player able to hit the bullseye on her next throw? It depends on the conversational context. If the background includes the proposition that she hits, the answer is yes. Am I able to throw my cup? Am I able to keep typing on my computer? It depends on the conversational context. If the background doesn't include facts about the presence of my cup and computer, the answer is no. Am I able to jump over the moon? Yes, if the background includes the counterfactual assumption that I jump over the moon.
The account is far too permissive. Tight constraints should be imposed on what's part of the background. In general, the only false propositions that can be supplied should be propositions about the subject's intentions. If there's no cup around, I'm not able to throw a cup (right now). We can't supply the false proposition that a cup is around. On the other hand, true propositions about the circumstances generally must be part of the background. If my cup is right here and all other aspects of the situation are normal, it is unambigously true I can throw the cup. We can't fail to supply the proposition that the cup is here.
If we fix these problems with (S6), we're back at (S5). We've lost the explanation of what's wrong with (5) in the baby carrier case.
So what is wrong with (5)?
What's wrong, I suspect, is that falling isn't something people actively do. It's not something one could sensibly intend to do.
Take another example mentioned by Santorio:
This, too, is odd. I suspect it's odd because winning isn't an activity. Suppose the lottery is a raffle-style lottery with physical tickets. Ava is about to draw a ticket. Compare (6) with (7).
(7) is like (1) and (3). I think it's true on one reading, and false on another. (6) is different. It's odd.
Some more examples:
These are all odd, and so is their negation.
I'm not sure if "activity" is the right term for what matters. In some sense, of course, falling is an activity. And we want to allow for statements like (14), in which the prejacent isn't much of an activity in an intuitive sense:
I'm also not sure how to formally model the oddness in question. It doesn't feel like presupposition failure.
And I'm not sure why (6) gets better, as Santorio points out, if we add that only people whose first name begins with 'A' can win.
So cases like these raise a nice puzzle. But Santorio's answer doesn't look right. His account entails that (7) and (8) are fine, and that the others are either true or untrue depending on context.