Introduction
Chapter 4: What’s the Point of Knowledge-how?
2. The Function of KNOWS-THAT
In this section, I lay out the central elements of Craig’s account of KNOWS, and argue that his account of KNOWS supports the claim that knowledge is the norm of assertion.
Craig starts with our practical needs. Each of us has a need for true information about our environment, in order to allow us to bring our practical projects off successfully. But our access to information about the environment is limited, and information is
distributed between agents. It would be helpful for agents to be able to tap the information possessed by others as and when it becomes pertinent. This gives us a practical need: the need to pool information between agents. Craig’s hypothesis is that this need is addressed by KNOWS, positing the following metasemantic function:
Pooling information: the function of KNOWS is to help us to pool information between different agents.
One way to understand this need is by considering the situation of an inquirer,123
someone who has an open question of practical significance, and faces a number of different agents who might be in a position to resolve that question for her.124 What an
agent in this kind of situation needs is a concept that can help her to assess potential informants on her question. Craig’s central hypothesis is that KNOWS addresses our need to pool information by providing us with a state which provides a standard on potential informants (Craig, 1990, p. 11).
Craig primarily focuses on how KNOWS can facilitate informant choice, but he actually considers a couple of other mechanisms by which KNOWS helps us to pool information:
i. An agent who is among various potential informants for some inquirer can apply the concept KNOWS to herself in order to determine whether she is in a
sufficiently good epistemic position to assert (Craig, 1990, pp. 63–5);
ii. An agent who is neither inquirer nor informant, can KNOWS in order to
recommend someone else as a good informant (Craig, 1990, pp. 82–97); To these we might also add:
iii. After testimony has taken place, agents can use positive knowledge ascriptions to praise the speaker, and negative knowledge ascriptions to censure the speaker. These positive and negative ascriptions provide social pressure to drive up the general epistemic standards for testimony (Reynolds, 2002, 2008).
I think that the best gloss of Craig’s functional story is that KNOWS helps us to address our need to pool information between agents by picking out an epistemic standard on information-provision. This means that the claim that the metasemantic function of
123 On the distinction between the situation of inquirers and examinars, see (Williams, 1973, p. 149;
Craig, 1990, p. 19).
KNOWS is to pool information gives us a gloss on the semantic function of knowledge — that knowledge is associated with the functional property of being a standard on good informants.
This gloss on the functional properties of knowledge is closely related to the knowledge-norm of assertion.125 Supporters of the knowledge-norm of assertion endorse
the following claim (Williamson, 2000, Chapter 11; DeRose, 2002; Hawthorne, 2004; Turri, 2011):
KNA: One must: assert that p, only if one knows that p126
This norm says that an agent’s assertion is permissible only if she knows what she says. If a speaker says that something which is false, unjustified or not known, then there is something inappropriate about that assertion. Crucially, KNA makes a claim about
epistemic permission: an ignorant or false assertion can still be morally or prudentially permissible, and a knowledgeable assertion can still be morally or prudentially
impermissible.127
One might have thought that KNA and Pooling Information were in competition. It is true that the examples of knowledge ascriptions used to illustrate KNA and the pooling view are importantly different: Craig focuses on the perspective of the inquirer, considering prospective knowledge-ascriptions which are used to flag good informants, whereas
supporters of KNA focus on the perspective of the examiner, considering retrospective uses of knowledge-ascriptions to assess whether an assertion fulfilled an epistemic rule. These families of examples do demonstrate two different kinds of pragmatic functions of
knowledge-ascriptions: a flagging use, and a evaluating use (McGrath, 2015). However, if
125 A number of authors have also suggested that Craig’s account supports the knowledge norm of
practical reasoning. (Greco, 2008, 2012; Hannon, 2013; McKenna, 2013, 2014).
126 The knowledge-norm of assertion is formulated in several different ways: as an imperative or
must claim (Williamson, 2000, pp. 241–3), as a claim about appropriateness (J. Brown, 2008b), or as a claim about permissibility (Turri, 2011). There is also debate about whether the norm should be formulated as a sufficiency claim (J. Brown, 2008b, 2012), and how we should understand the notion of assertion (E. Fricker, 2015). For the purposes of my discussion of epistemic norms, the differences in formulation will not be significant, and I will focus on the weaker necessity direction of the norm.
127 From this point on, I will use unqualified claims about permission to refer to epistemic
we look a little more carefully, it should become clear that KNA and Pooling Information are not in competition.128 For one thing, the two views are about different things: Pooling Information concerns the function of the concept KNOWS-THAT, whereas KNA is a view about the functional properties associated with the state of knowledge-that. Pooling Information is a claim about metasemantic functions and KNA is a claim about semantic functions. Not only are these two views not in conflict: endorsing KNA provides a plausible way to unify the various ways in which we can use knowledge ascriptions to facilitate the pooling of information. One way to put the point is that KNOWS facilitates the pooling of information in part because it picks out a state which is the norm of assertion: KNA is a semantic function which naturally supports the metasemantic function posited by Pooling Information.129