There are a lot of comparisons between the verbs runs and knows (and their cognates) that seem relevant to the contextualism debates.
First, whether a certain kind of activity counts as running depends on features of the context, very broadly construed. For instance, (1) is not inconsistent.
(1) At the last turn in the 400m final, Harry stopped running and jogged the rest of the way.
That’s despite the fact that (2) could sound true even if the way Jack moved on his way to work was exactly the same as what he did down the straight in that 400m.
(2) Today, Harry ran to work.
In other words, whether a certain kind of jogging counts as running seems to depend on features of context, very broadly construed.
The repeated reference here to broad construal is because it isn’t clear whether the best explanation of (1) and (2) is by something analogous to contextualism in epistemology, or something analogous to subject-sensitive invariantism. (These are not the only options of course.) The contextualist about runs says that it is the speaker’s context that determines the boundary between running and something less energetic, the subject-sensitive invariantist says that it is the context of the subject.
At first glance, (1) and (2) look more suitable to the SSI account. Another reason for thinking this is that it’s unlikely there is a syntactic place attached to runs that marks how energetic the motion must be for it to count as running. Since SSI doesn’t need this variable place, but (according to some theories) contextalism does, SSI looks more plausible.
So (1) and (2) are meant to be analogous to high-stakes/low-stakes examples that (allegedly) support contextualism/SSI about knowledge. There is another way in which runs is like knows. In both cases contrast effects, either through explicit mention or the use of contrastive stress, can decrease the range of cases that fall under the verb.
Assume, as was the intended background in (2), that Harry normally flies to work on his broomstick, but today started his exercise regime of jogging to work. He jogs reasonably briskly, but he isn’t going to break any landspeed records. In a discourse-initial setting, the appropriate answer to (3)
(3) Did HARRY run to work today?
is Yes. On the other hand, the appropriate answer to (4)
(4) Did Harry really run to work today, or only jog?
is He only jogged. And possibly, though it might depend a lot on the features of the conversation, the most appropriate, or natural answer to (5)
(5) Did Harry RUN to work today?
is No, he jogged.
The same effects seem to arise with knowledge.
Assume Harry believes that his boss dislikes him because (a) Harry knows that his boss dislikes all the other workers and (b) Harry can’t see any reason why he’d get specially favourable treatment. Moreover, Harry is right – his boss does dislike him, though Harry has no direct evidence of this.
I think the appropriate answer to (6)
(6) Does HARRY know his boss dislikes him?
(7) Does Harry know HIS BOSS dislikes him?
is Yes in each case. But when knows is explicitly contrasted with something else, as in (8)
(8) Does Harry really know his boss dislikes him, or only believe it?
the most appropriate answer is He only believes it, with perhaps an additional comment that the belief is indeed true. And even some kinds of stress can get the same result, though as with (5) here it depends a lot on the context and a lot on the particular type of stress.
(9) Does Harry KNOW that his boss dislikes him?
It’s not too hard to get oneself into the frame of mind where the appropriate answer to that question is No.
This kind of data is less obviously explicable on the SSI model, because the subject has not changed. There are three options on the table, both in the case of runs and in the case of knows.
First, we could adopt the kind of contextualist options currently on the market. (With, presumably, some work to show how they apply to runs.)
Second, we could say that focus, stress, contrast etc have truth-conditional effects, so the appropriate answers in all these cases are the true answers, but this is because of phonologically (or typographically) encoded semantic features, rather than the kind of contextual effects currently posited.
Third, we could say that changing stress patterns can’t change the true answers to these questions, though they can change the speaker meaning of the questions, which in turn changes the appropriateness of various answers. This move is available in both the runs and knows cases.
The main point here is just to stress analogies rather than to promote one or other of these lines. A subsidiary point is to point out the importance of keeping effects of changing expectations (as in (1) and (2)) apart from effects of changing contrasts. It’s perfectly consistent to think that the background (whether it’s a 400m race or a jog to work) affects the truth of running-ascriptions, but changing contrast sets does not. (I’m not necessarily endorsing this view, I’m just saying it’s consistent, and the arguments against it should be sensitive to this kind of distinction.)
But what I really want to stress are the analogies, which seem much closer than between knows and, say flat or tall. The obvious way in which runs is closer to knows is that it’s a verb not an adjective. It also is hard to find natural ways of making the salient and intended running-standard explicit, just like it is hard to find natural ways of makiing the salient and intended knowing-standard explicit.
Having said all that, there is one very important disanalogy between runs and knows. (I’m grateful here to Zoltán Szabó for pointing this out.) Runs is an event verb, and knows is a stative verb. Hence it would not be at all surprising if they differed in some crucial respects. This is not to say they must differ in the kinds of respects being discussed here, just that we should be very wary of drawing too many analogies. The analogies do seem closer than with flat or tall, but they are not at all exact, so it’s more than possible that we shouldn’t make the moves in the contextualist debates about runs as in contextualist debates about knows.