Suppose you have a body of evidence. Maybe you’re a detective and you’ve got hours of interviews and forensic evidence at your disposal, or maybe you’re a paleontologist with lots of fossils and lessons from comparative anatomy. Now you want to know if the evidence “favors,” in any way, some hypothesis. Does the forensic evidence support the view that the butler did it? Or does it maybe rule out the butler’s involvement? Or is it completely silent on the matter?
There’s a view in epistemology called the Uniqueness Thesis, where a given body of evidence can only support one attitude towards any given hypothesis. So if I “ask” the evidence if the butler did it, then according to the Uniqueness Thesis, the evidence can give me, at most, one definitive answer, and no others. It can support either (a) believing that the butler did it, (b) believing that the butler did not do it, or (c) suspending judgment about the butler’s involvement. But it cannot support more than one of those options. The evidence can’t “say” that I’m free to choose between (a) or (c) but not (b), for example.
The UT is the opposite of Permissivism, which says that, actually yes, evidence can occasionally permit/license/warrant a choice between attitudes. Maybe you look at the fossil evidence and conclude that T. rex was a scavenger, but another paleontologist consults the same evidence and concludes that the king of the dinosaurs was a predator. The UT would say that at least one of you must be mistaken; Permissivism allows for the possibility that the evidence supports you both.
Two things to note here. First, it’s not as if Permissivism says you can both be right in the sense that both of your views are true; Permissivism doesn’t imply that T. rex was both a scavenger and a predator. It just says that the evidence can “point” in different directions, and it’s not irrational to go with one direction over the other.
Second, in the summary I’m giving here, I’m trampling over some obvious questions in a way that’s familiar to the UT vs. Permissivism literature. There’s the issue of what evidence is, for instance, an issue I nodded to in a previous post. On some understandings of evidence, people can share evidence, but on others, they cannot. If evidence can be stuff in the world (like crime scene photos), then sure, people can share evidence, but if evidence is always something in a person’s head, then they can’t. So the traditional way of framing the UT question starts on the wrong foot, depending on your view what evidence is.
I’ll put my cards on the table here: I’ve always been a big Uniqueness guy. But I think the background picture had a lot of unexamined assumptions in it, assumptions that probably don’t stand up to scrutiny. I think part of it has to do with how I imagined “evidential support” to work.
As I said in the post on evidence metaphors, we use a lot of physical, causal metaphors to understand the role of evidence in our cognitive lives. If you think that the butler’s fingerprint on the bloody candlestick makes it more likely that the butler did it, it’s hard not to think of the evidence as somehow doing something to the hypothesis that the butler did it. The evidence somehow imparts likelihood with normative force, like the transmission of an electrical current or a causal oomph.
And you know, the thing about causation is, it ain’t real. A lot of philosophers once took the relation between cause and effect to be a necessary one. In order for the flip of the light switch to be the cause of the light turning on, the flip must somehow necessitate the light turning on. If the flip of the switch is the cause, then it would be impossible (without further background shenanigans like power outages) for the switch to flip without the light turning on. After Hume (really, after Berkeley, since it was Berkeley who argued that none of our ideas reveal any causal powers in nature), many philosophers doubt whether there really are any necessary connections in nature like this. Maybe cause and effect is really just constant conjunction; there’s no such thing as necessitating powers.
I suspect that our physical, causal metaphors for evidence carry over this necessitating idea. Evidence makes some hypotheses more likely than others, and in fact, it couldn’t not do that. The butler’s fingerprint on the candlestick not only makes it more likely that he did it, it necessitates the higher likelihood that he did it. Unless there are some background shenanigans (like the private chef framing the butler), the evidence couldn’t exist without raising the likelihood that the butler did it.
This picture fits more comfortably, to me anyway, with the Uniqueness Thesis than it does with Permissivism. I’m not sure why. It’s hard to spell this out and it’s hard to see it clearly. But I think it has to do with the old-fashioned simplicity of UT mirrored by the physical-causal metaphors. In the old picture of causation, causation is a relation between exactly two things: a cause and an effect. UT suggests that evidential support is also a two-place relation, between evidence and a single attitude (related by, I don’t know, the making-likely relation). But once you relax that picture — once epistemic support becomes more complicated than an intellectual oomph — UT starts to look a lot less plausible.
It’s hard to say what all this amounts to, but I think I can say that I’m a lot less of a Uniqueness guy than I once was.
Coming soon: a blog post about the relationship between UT and infallibilism!