Metaphysics of knowledge

DOI: 10.4324/9780415249126-P066-1
Version: v1,  Published online: 2017
Retrieved October 18, 2018, from

2. The ground of knowledge

Related to the question of the metaphysical nature of knowledge is the question of the metaphysical ground of knowledge. Some metaphysicians (Fine 2001) take grounding to be a relation of metaphysical explanation, whereas others (Schaffer 2012) view grounding as a worldly relation that may back such metaphysical explanation. Also, some metaphysicians (Fine 2001) consider grounding a distinctive metaphysical relation that obtains when some goings-on are said to hold in virtue of some other goings-on. In their view, grounding serves to illuminate metaphysical dependence between grounded facts and grounding facts. However, it can be argued (Wilson 2014) that grounding is unsuited to do this work, and that appeal to specific metaphysical, grounding relations suffices, such as functional realization or the determination relation. An example of functional realization is the implementation of functionally characterized software by particular computing hardware, and an example of the determination relation is the way the determinable property red is related to its determinate properties crimson, scarlet and other specific shades of red.

A good starting point is to ask the question of which facts constitute the supervenience base for epistemic facts (see Supervenience). With very few exceptions, for example Lehrer (1997), epistemologists have embraced the claim that epistemic properties (E) supervene on non-epistemic properties (N), such as natural, descriptive or physical properties. The relevant type of supervenience is typically that of strong, individual supervenience, stated in terms of metaphysical necessity. Consider the following formulation:

Epistemic Supervenience (ES)

Necessarily, if an agent S has epistemic property E, then S has some non-epistemic property N such that, necessarily, any agent S* with N also has E.

In general, a set of properties A supervenes on a set of properties B just in case nothing can differ in respect of its A-properties without also differing in respect of its B-properties. Strictly speaking, the supervenience relation is reflexive, transitive and non-symmetric, yet is best understood in the epistemic domain as holding asymmetrically since non-epistemic properties do not plausibly supervene on epistemic properties. The question is now what N might be if E is knowledge. Obviously, since knowledge is factive, and most of an agent's knowledge pertains to the external world, the supervenience base for most knowledge will include all kinds of features of the external world. Moreover, if content externalism is true, then the contents of most, if not all, states of knowledge will be wide – that is, individuated by environmental facts beyond the agent in question. Given that states are individuated by their contents, it follows that the supervenience base for knowledge will also include such individuating environmental facts. Since neither of these dependencies is distinctively epistemic, we can instead ask the question of what N might be if E is that which turns true beliefs into knowledge. Following Plantinga (1993), this epistemic dimension of knowledge is called 'warrant'. Warrant should not be equated with propositional justification – that is, with the justification an agent has available for a belief. First, for a true belief to count as knowledge the justification the agent has for the belief must also be the reason for which the agent holds that belief. In other words, the agent's belief must be properly based so as to be justified. Proper basing is what converts propositional justification into doxastic justification, where such basing is typically understood in causal terms. Second, even a justified true belief may fall short of knowledge either through epistemic luck as in Gettier cases or through other types of epistemic defeat. For example, an otherwise justified belief that p may be defeated as a result of falsely believing that the cognitive process that produced the belief that p is malfunctioning.

An important distinction in epistemology is that between internalism – that is, the view that what justifies a belief must be reflectively accessible to the agent – and externalism – that is, the view that such reflective access to justifiers is unnecessary (see internalism and externalism in epistemology). Epistemologists belonging to both the internalist and externalist camps have tended to confine the supervenience base for warrant, or at least doxastic justification, to individualistic facts of the agent – that is, to features residing inside the bodily boundaries of the agent. For example, reliabilists, for example Goldman (1986), have restricted the cognitive processes that produce justified (or warranted) beliefs to those that occur inside the body of the agent in question. Likewise, evidentialist mentalists, for example Conee and Feldman (2004), maintain that the factors that confer justification on beliefs consist solely of mental states and processes of the agent in question. However, with the advent of social epistemology such epistemic individualism has come under critical scrutiny. Whether or not an agent’s belief is justified is often heavily dependent on the epistemic credentials of cognitive states, processes or abilities of other agents. Ordinary testimonial beliefs are a case in point in that the justification of a hearer’s belief owes much to the epistemic virtues of the speaker (see Testimony). For more details, see for example Goldberg (2010) and Kallestrup and Pritchard (2012).

Whatever the supervenience base N is for a particular epistemic property E, proponents of (ES) must also explain what metaphysical grounding relation between E and N makes (ES) true. After all, (ES) merely says that a pattern of variation holds between E and N. Assuming (ES) is not to be regarded as an inexplicably brute fact, an explanatory account invoking some grounding relation is owed of why these sets of properties co-vary.

One potential metaphysical grounding of E in N is the relation of determination (Funkhouser 2006; Wilson 2009). Following Williamson (2000), perhaps knowing that p is related to seeing that p as a determinable to its determinate, in much the same way red is related to scarlet. Propositional seeing is a specific way, amongst others, of knowing such that necessarily if an agent sees that p then the agent knows that p, but it is possible for the agent to know that p without seeing that p. The agent could instead hear, or deduce, or introspect that p. The determination dimensions are the ways in which propositional seeing, hearing, deducing or introspecting specify knowledge by way of differing from one another in respect of knowledge. Because knowledge shares these dimensions of variation with these more specific ways of knowing, the latter are fairly demanding statuses, requiring not only true belief but also whatever modal profile knowledge may have, including safety or sensitivity. See Turri (2010) and French (2012) for details. Importantly, the thought would then be that knowing supervenes on propositional seeing, hearing, deducing and introspecting, because (in the explanatory sense) the determinable knowing is grounded in these more determinate states. Perhaps the latter are still partially epistemic states in which case they themselves are classified as determinables relative to more determinate states, and so on until the ordering bottoms out in super-determinate states which are genuinely non-epistemic.

Citing this article:
Kallestrup, Jesper. The ground of knowledge. Metaphysics of knowledge, 2017, doi:10.4324/9780415249126-P066-1. Routledge Encyclopedia of Philosophy, Taylor and Francis,
Copyright © 1998-2018 Routledge.

Related Articles