Recall that both Knowledge Sensitivism (SK) and Assertion Sensitivism (SA) aim to capture the Shiftiness Intuition, i.e. the intuitive stakes sensitivity of proper assertability. Both accounts take it, contra the view defended by this thesis, that the felt variation concerns epistemic propriety. Recall, also, that, for all that has been said so far in this thesis, the SK, SA and KNA explanations of the data are on a par. Now, if that is the case, bringing in theoretical considerations is the way forward in settling the issue. In what follows, I will try to do just that; I will point to some theoretical advantages that speak in favour of my view over sensitivism of both sorts.
Let us begin with Knowledge Sensitivism. Recall that this view keeps the epistemic standard for proper assertion fixed – usually knowledge – and argues that the degree of warrant necessary for attributing/meeting it varies with practical context. Several positions defended in the literature belong here: contextualists claim that features of the attributor’s context affect the truth-‐conditions of knowledge ascriptions (e.g. DeRose 2002), while champions of pragmatic encroachment argue that having knowledge itself is affected by the subject’s practical situation (e.g. Hawthorne 2004).
According to champions of SK, due to change in practical context, as opposed to ASPIRIN 1, in ASPIRIN 2 one fails to know the target proposition and, therefore, is not in a position to assert it.
Now, crucially, note that according to contextualists themselves, invariantism is the default position; we need to be argued out of it: “we seem, if anything, to be ‘intuitive invariantists’”. According to Stewart Cohen (1999, 78), for instance, “many resist [the contextualist] thesis — some fiercely. Moreover, those who do accept the thesis, generally do so only as a result of being convinced by philosophical reflection”. Considerable amounts of ink have been spilled on pointing out theoretical and empirical difficulties for both contextualism and pragmatic encroachment.86 I will not rehearse these arguments here, in the interest of space. Also, I take it that, since the central concern of this thesis is with the normativity of assertion, out of the two sensitivist theories, it makes sense to focus the discussion on the one that actually proposes a competing norm for this speech act, i.e. assertion sensitivism.
What I will do, then, is only discuss two problems for knowledge sensitivism which speak in favour of combining a knowledge norm with
a classical invariantist account in general, and, in particular, favour the account defended in this thesis.
6.1.1 Two for Invariantism
1. The ‘Whose Stakes? Dilemma’: Recall that we have identified two positions under the knowledge sensitivist umbrella: contextualism about knowledge attributions and pragmatic encroachment. For the former, the relevant stakes are those of the attributor, while for the latter, the subject’s practical interest is what makes the difference. Now, as it so happens, the question regarding whose stakes matter will quickly lead SK into what I will call the ‘Whose Stakes? Dilemma’: if shiftiness depends on attributor’s stakes, the view notably divorces what it is to know from what it is to properly assert; (e.g. Hawthorne 2004). That is because, while the relevant stakes for knowledge attribution will be those of the attributor, the relevant stakes for proper assertability will lie with the subject.
To see why this is unfortunate, note that this puts the contextualist in the awkward position to have to accept statements of the form: ‘Louise knows that p. KNA is true and so if Louise knows that p, she may (epistemically) assert that p. However, Louise ought (epistemically) not assert that p’.
On the other hand, if what matters are the stakes associated with the subject, SK fails to account for the independently plausible function of assertion of generating testimonial knowledge. After all, if the two parties, hearer and speaker, do not share stakes, knowledge transmission can fail. This will happen in cases where the stakes of the hearer are higher than those of the speaker. Whenever that is the case, the corresponding assertions, although epistemically proper, will fail to fulfil their epistemic function, in spite of the otherwise friendly environment. Furthermore, this difficulty also comes with an important theoretical burden on the shoulders of the SK defender: misfit with all extant accounts of testimonial knowledge, according to which whether knowledge gets generated by testimony is independent of pragmatic factors.87
Now, one move in the direction of escaping the stakes dilemma that, at least at first glance, looks fairly promising, is the more recent contextualist suggestion that it’s the conversational purpose which determines whose stakes are relevant and this can vary between the attributer and the subject. John Greco (2010), for instance, has notably
defended a very flexible view on which even the interests of a third party can be the ones that matter. Will this kind of manoeuvre help the contextualist here? Alas, the answer is ‘no’. Here is why: think of DeRose in the high stakes bank case again: everything stays fixed, except that, after DeRose denies knowledge to himself, his wife asks: ‘How about calling Stew? He also has an account here, maybe he knows whether the bank is open on Saturdays”. Now, say that, as matter of fact, Stew has the exact same warrant as DeRose to believe that the bank is open on Saturdays: he’s been there two weeks ago. Say, also, that DeRose is aware of this. It looks as though, then, the natural thing to answer would be ‘No, he doesn’t know either’. Given the purpose of the conversation, the stakes are definitely high. But now imagine that, at the same time, Stew is in a low stakes bank case. Surely, it is perfectly fine for him to assert that the bank will be open on Saturday. Again, the contextualist divorces knowledge attribution from assertability.
2. Parsimony: SK champions are in need of explanations involving overriding norms anyway, for explaining further data. Given this, the account defended here is the most parsimonious.
First, in situations in which the stakes of the hearer are higher, 1) pragmatic encroachment needs to employ some explanation involving overriding anyway, in order to account for impropriety, while 2) the explanation offered by contextualism does not seem to stand up to further linguistic scrutiny. To see this, take Jessica Brown’s AFFAIR case again; let us start with champions of pragmatic encroachment; first of all, what these philosophers will have to say about this case is that, given his low stakes, Friend does indeed know, and is therefore in a perfectly fine position to assert. If you are still not convinced, imagine a variation of the case where the knowledgeable individual is but an indifferent neighbour. Still, even so, it looks as if she should not assert it to poor Husband’s face for want of sufficiently strong evidence.
Now, according to the functionalist picture I have put forth, this is but a straightforward case where the prudential function takes precedence over the epistemic one. Of course, there is nothing keeping pragmatic encroachers from explaining this case in a similar way. Note, however, that insofar as an account of overriding is employed anyway, it is not clear why we should not prefer a uniform, independently motivated picture like the one proposed here. After all, the alternative is a mixed account that still owes us an explanation as to why some of the cases put forth are cases of overriding, while others are cases where practical stakes affect whether you know.
Let us then move on to contextualism. According to this view, what is going on in AFFAIR is that the relevant context to measure the
propriety of Friend’s assertion against is one that is also partly ‘infected’ by Husband’s stakes, which explains the epistemic impropriety of the assertion.
Recall, though, that contra contextualism, my explanation in terms of prudential considerations stepping in and overriding the epistemic ones seems to also be supported by further linguistic considerations: surely, […] Friend […] has enough warrant to be pretty sure that she might be having an affair. Still, it still looks like Friend would better abstain from asserting the latter in the presence of Husband also (McGlynn 2014, 126).
Furthermore, second, SK champions will also be unable to explain urgency situations without appealing to overriding. Here is why: note that, roughly speaking, SK offers a directly proportional scheme for the association between stakes and knowledge/assertability. That is, the higher the stakes, the more warrant seems to be needed according to these views for knowledge, and therefore assertability, to be in place. Think, however, of Williamson’s TRAIN case: here, the relationship between stakes/urgency and assertability seems to work the other way around, that is, it looks as if assertability varies inversely proportionally with stakes/urgency. The more urgent it is for you to get to your destination, the lower the amount of warrant I need for making the corresponding assertion.
As far as I can see, there is nothing in the SK scheme that enables them to explain this phenomenon. As such, again, it looks as if they will need an explanation in terms of overriding here and so are bound to sacrifice on two counts: parsimony and the general motivation for the view.
6.2 WAMs
For several people who like classical invariantism about knowledge attributions, the move from variation in assertability with stakes to contextualism or pragmatic encroachment seems rushed. As such, these authors venture to account for the Shiftiness Intuition under a classical invariantist umbrella by arguing for the context-‐sensitivity of proper assertability. This move has become known in the literature as a Warranted Assertability Maneuver, or WAM for short.
Now, there are two extant ways of being a WAM-‐er: one can, on the one hand, hold the epistemic norm of assertion fixed – say, defend KNA – and argue that the source of variability pertains to what is pragmatically conveyed by the assertion in question rather than by