The Pragmatics of Pragmatic Encroachment

Practical stakes do not impact knowledge

In my day job, I’m an academic philosopher. I write philosophy papers for journals that other philosophers will read. Non-philosophers don’t read philosophy journals, and even philosophers don’t read as many philosophy articles as they could. So I’m starting a mini-project on this blog to summarize the main ideas from my written articles in a way that anyone can understand. Perhaps this will encourage you to take a closer look at the article itself. But if I do my job right, a short post like this will be enough for you to understand what I’m up to and why it’s cool. I’ll be going through my CV in chronological order, starting with my first article: “The Pragmatics of Pragmatic Encroachment.”

---

Some of the most famous thought experiments in contemporary epistemology are the “bank cases.” Compare the following two scenarios:

Low Stakes: Keith is driving home with his wife on Friday evening. They have a check to deposit at the bank. Looking in, they see a very long line at the bank that they would have to wait in. Keith remembers having gone to the bank before on Saturday mornings. So he says to his wife, “Let’s just come back tomorrow. I know the bank is open on Saturday mornings.”

High Stakes: Everything is the same as in Low Stakes. The only difference is that it is very very important that they deposit the check. If the check is undeposited, then their rent check to their landlord will bounce, and they might get evicted. But again, seeing the long line, Keith says to his wife, “Let’s just come back tomorrow. I know the bank is open on Saturday mornings.”

Most philosophers, when they encounter these two cases, have thought that Keith is saying something true in Low Stakes, but false in High Stakes.¹ That is, in Low Stakes, Keith does know that the bank will be open on Saturday morning. But in High Stakes, he does not know that the bank will be open on Saturday morning. This is somewhat puzzling! All of the “traditional epistemic factors” are the same in the two cases. Keith has the same belief in both cases. In both cases that belief is true. It is based on the same evidence (his memory), and his memory is equally reliable in both cases. The only difference between the two cases is practical. In one case, it matters a lot whether Keith’s belief is true. In the other, it doesn’t matter at all. So this seems to imply that whether or not Keith knows doesn’t just depend on traditional epistemic factors like what evidence he has. Knowledge also depends on his practical stakes. In other words, whether you know isn’t just a matter of how likely it is that you’re right. It’s also a matter of how much it matters that you’re right. But that sounds a bit weird.

There are three responses that have been offered to this puzzle. One response, the contextualist response, says that the word ‘know’ means different things in different contexts. Sometimes the word ‘know’ means (something like) ‘knows according to very high standards.’ And sometimes the word ‘know’ means (something like) ‘knows according to lower, every day standards.’ So the contextualist can say that the word ‘know’ means two different things in the two cases. In Low Stakes, ‘know’ means ‘knows according to lower, every day standards.’ And Keith does know according to every day standards that the bank will be open. So what he says is true. In High Stakes, on the other hand, ‘know’ means ‘knows according to very high standards.’ But Keith doesn’t know that the bank will be open according to very high standards. His evidence is good but not completely infallible. So in High Stakes, Keith says something different, and that different thing isn’t true.

A second response, the pragmatist response, says that the word ‘know’ doesn’t change its meaning. Rather, knowledge just has a pragmatic condition on it. According to a pragmatist, in order to know that some claim is true, you must (a) have good evidence that it’s true, and (b) it must also be rational to treat it as true when reasoning about what to do.² Condition (a) is satisfied in both Low Stakes and High Stakes. But (b) is only satisfied in Low Stakes, and not in High Stakes. That’s why Keith knows the bank will be open in Low Stakes, but not in High Stakes.

The third response, sometimes called “purist invariantism,” simply denies the “intuitive data” about the two cases. Keith knows the bank will be open in both Low Stakes and High Stakes. So there’s no need to explain the difference between the two cases, because there is no difference between the two cases.

Now, even for the purist invariantist, there’s something to explain here. Why have so many philosophers thought those two cases should be treated differently, even though they aren’t fundamentally different? The most common answer to this puzzle is to appeal to the notion of implication. Keith says something true in both cases when he says that he knows the bank is open. But in the High Stakes case, he implies something false, and the fact that he implies something false makes his claim suspect.

This seems like a promising line of argument for the purist invariantist. But it leaves a lot unanswered. What is being implied in the High Stakes case that isn’t being implied in the Low Stakes case? And why is it being implied?

I’m a purist invariantist. In my paper, “The Pragmatics of Pragmatic Encroachment,”³ I offer a new solution to these problems.

Most purist invariantists base their account on contextualism. According to the contextualist, “I know that the bank will be open” can mean two different things: “I know-according-to-high-standards that the bank will be open” or “I know-according-to-lower-every-day-standards that the bank will be open.” According to most purist invariantists, “I know that the bank will be open” only says the second thing: “I know-according-to-lower-every-day-standards that the bank will be open.” But when the stakes are high, saying “I know-according-to-lower-every-day-standards that the bank will be open” implies “I know-according-to-high-standards that the bank will be open.”

There are some very well-known problems with this account. So I do things a little differently. I base my account on pragmatism, not contextualism.

According to the pragmatist, “I know that the bank will be open” means “(a) I have good evidence that the bank will be open and (b) When I’m reasoning about what to do, I can assume that the bank will be open.” But according to my account, “I know that the bank will be open means “I have good evidence that the bank will be open,” and it implies “When I’m reasoning about what to do, I can assume that the bank will be open.” That implication is true when the stakes are low, but not when the stakes are high. That explains our different reaction in the two cases.

This saves purist invariantism from the challenge posed by the bank cases, and it avoids the problems that the contextualist-based account has.

That’s the basic idea. For more details about how this is all supposed to work, see the paper.

So why care? Two reasons. First, this debate is part of the larger debate about the nature of knowledge. What is knowledge? What does it mean to know something? The contextualist, pragmatist, and purist invariantist have different answers to that question. So if you care about what knowledge is, then this is something you should care about.

That’s a bit abstract; here’s a sexier reason you might care. In recent years, it has been common to say that moral stakes make a difference to knowledge, not just practical stakes. According to this view, if you have a belief that is true and supported by excellent evidence but is morally wrong, then you don’t know it’s true. The common examples of this involve beliefs that are racist or sexist. If you believe a racial stereotype, and that is true and well-supported by evidence, you still can’t know that it’s true, because it’s racist. My account implies that this is wrong. You can know that a racial stereotype is true (if, I hasten to add, it is true and you have very good evidence that it’s true. Both of those conditions are usually not met!). But! Even if you do know that the stereotype is true, it will still be wrong to say that you know that this stereotype is true. That’s because saying you know the stereotype is true would typically imply that it’s permissible to treat people differently on the basis of race, and that’s not permissible.

Well, that last bit isn’t in the paper, and I wouldn’t insist on it. But the only way to get people to care about anything these days is via a culture war hook, so there’s your culture war hook.

Either way, I hope you found this interesting!

1. I emphasize that this is the reaction of most philosophers because philosophers aren’t actually a representative sample of people, and there’s some evidence that the reactions of non-philosophers to this case are a bit more complicated and diverse than philosophers have said. But among philosophers, these reactions to the two cases are the non-negotiable “intuitive data” that must be explained.

2. These are both necessary conditions on knowing, but they’re certainly not jointly sufficient. And I’m being hand-wavey on purpose about what “good evidence” consists in. Substitute in your favorite account.

3. The title is kind of a pun. The phenomenon where practical stakes make a difference to whether or not you now is known in epistemology as “pragmatic encroachment.” And the phenomenon where words imply more than they strictly say is known in the philosophy of language as “pragmatics” (as opposed to “semantics). So this is a pragmatic account of pragmatic encroachment.