I’ll start this post with a point about Williamson’s style. In reading Williamson I have found his philosophical style to be terse. He does not provide an abundance of signposts. This requires the reader to take extra time to see what he is arguing for and how the current argument is connected to both what has come before it and what is to come after it. You might say Williamson is famous for pure explication and analysis. At the same time, working through the density of his ideas is very rewarding. When reading Williamson closely I find myself discovering several additional layers to his ideas. This is true of most philosophy, but it finds heightened expression in Williamson as compared to, say, Robert Audi who takes pains to be clear and keep his reader tracking with his arguments. That being said, I’ll dive into section 7.1 of The Philosophy of Philosophy [a shameless signpost :-)].
The big idea in 7.1 is that the thesis of Evidence Neutrality (EN) is false. EN is an initially attractive thesis. It indicates that debate over a hypothesis does not have to be resolved prior to identifying what will be recognized as evidence in the debate. For instance, pro-life and pro-choice adherents debate the hypothesis abortion is morally wrong (h). This debate, as evidenced by the history of the debate, persists in a state of disagreement. However, if EN is true, both sides of the debate can agree that, for example, (p) the neurological-status of the embryo at week 7 is evidence for hypothesis h. Both sides can agree that p constitutes evidence without first having to resolve their differences over h. Both sides can point at the neurological capacities of the embryo at week 7 in debating h. This does not preclude the two sides interpreting the same evidence differently: a pro-life advocate might argue that at 7 weeks the embryo can feel pain, yet a pro-choice advocate might respond that the higher brain functions are not yet developed so it is unlikely that the embryo can feel pain. Both sides agree that the embryo is evidence, but they disagree how the evidence bears on the hypothesis. Despite the attractiveness of this thesis Williamson argues against it:
Having good evidence for a belief does not require being able to persuade all comers, however strange their views, that you have such good evidence. No human beliefs pass that test. Even in principle, we cannot always decide which propositions constitute evidence prior to deciding the main philosophical issue; sometimes the latter is properly implicated in the former. (2007: 212)
For Williamson EN forces one to psychologize evidence. Different people with different theoretical commitments will be committed to saying conflicting things about whether something is evidence. Evidence is not uncontentiously decidable. For Williamson all evidence is true; evidence consists only of facts, and a fact is a true proposition. The only way to agree on the facts, when two parties hold conflicting theoretical commitments, is to retreat to psychological claims (e.g., that I believe or that I am inclined to believe the Gettier proposition is true). Both sides can agree that they are inclined to believe the Gettier proposition without agreeing that the Gettier proposition is true. Such claims may be uncontentiously decidable, but they create a gap—arguing from a psychological premise (a belief) to an epistemological conclusion (a truth). Now I will discuss some concerns I have with Williamson’s argument against EN.
First, it seems Williamson has either begged the question against EN or set the bar too high for what constitutes neutral evidence. If evidence is factive, and within most domains of intellectual inquiry facts (truths) are contestable, then, ceteris paribus, everyone will not agree what constitutes evidence (what really constitute the true propositions). However, starting with a fallibilist account of evidence might accommodate the failures of EN ‘in practice’. Assuming a factive account of evidence from the start and analyzing EN against this yardstick is bound to result in the inability of EN to pass the test of ‘uncontentious decidability’. In other words, the stance Williamson takes on the nature of evidence dooms EN from the start. Regarding EN ‘in principle’, Williamson’s definition of evidence sets up problems for EN. The example Williamson uses involves two theories that entail two propositions:
- T = Every math theorem is evidence.
- T* = No math theorem is evidence.
Whether a math theorem is evidence is not uncontentiously decidable. When proponents of T debate proponents of T* proponents of T* will be committed to saying a theorem is not evidence whereas proponents of T will be committed to saying a theorem is evidence. It seems the two sides will need to resolve their theoretical differences prior to agreeing what constitutes evidence.
Evidence is decidable if there is a method for determining whether the evidence is included in a theory (i.e., a set of well-formed formulas closed under logical consequence). Evidence is ‘uncontentiously decidable’ if both sides can agree that a proposition is admitted as evidence, but whether a proposition is admitted as evidence depends on the theory under which it is admitted. This is accurate when, to borrow a term from Brian Weatherson, EN is interpreted under an ‘alethic’ reading . If EN is read as agreement whether a piece of evidence is true, then theoretical commitments will produce contentious differences in cases where theories entail propositions that are the negation of each other, as in Williamson’s example. But if EN is given an ‘apodeictic’ reading, then it must be decidable whether a piece of evidence bears on the proposition under scrutiny.
Williamson and Weatherson prefer the ‘alethic’ reading because it matches their assumption about evidence, as consisting only of true propositions. Against this, a great deal of evidence can be agreed to as relevant in the debate without having to initially persuade the other party of the truth of the evidence in relation to the hypothesis at issue. Having the bar of EN set at agreeing over the truth of propositions prevents many debates from ever getting off the ground. What is at issue may not get resolved under an ‘apodeictic’ reading of EN, but with the debate off the ground and the evidence agreed to in play the ability to make the other party accept a proposition as true depends on skill of argumentation. It depends on showing showing how one’s conclusion drawn from one’s interpretation of the evidence gets the other party something else they value. In such a case, the two parties are attempting to unseat enough propositions in the other’s theory set as to undermine each other’s theoretical commitments. This makes the dialectic more fluid and dynamic, as adjustments are made to both theory and evidence.
 Weatherson: 30-32 (http://brian.weatherson.org/ENw.pdf).