Imagine being on a train, and being told by the conductor that the train is late ‘because of delays’. This is an annoying and (at least for British people) very familiar phenomenon. The conductor is pretending to offer an explanation of the train’s lateness, but what they provide is just another description of that lateness. In proper explanations there should be some distance between the explanans, which does the explaining, and the explanandum, the thing explained. The conductor’s failed attempt at explanation lacks this feature, which I call ‘explanatory distance’.
I was inspired to work on this issue for a number of reasons. Like most people, I often encounter cases like this in everyday life, such as on trains or at the doctor’s surgery, and I have always found them odd and puzzling. I have also been working on emergence, and some of the issues I was writing on hinged on this question about when two sentences or facts are too close for one to explain the other. I realized that many philosophical debates have explanatory distance at their heart. For example, explanatory distance is central to recent conversations about Humean laws and explanation, which focus on the idea that Humean laws are too close to the phenomena they must explain for genuine explanation. Another relevant area is grounding. I’ve long been troubled by some cases of purported grounding explanation, like Socrates’ singleton set being grounded in and so explained by Socrates, or the fact that some paint is red being grounded in and so explained by the fact that the paint is scarlet. There is something intuitively off about these as explanations, and part of what is not-quite-right about them is that they seem to lack explanatory distance.
There are also some problem cases that pull both ways, in that there are interpretations on which they lack proper distance and interpretations on which they do not. A good example is Molière’s dormitive virtue explanation, where a doctor ‘explains’ that opium reliably induces sleep because it has a dormitive virtue. Molière intended this to be a joke, a laughably bad attempt at explanation, and part of why it appears to be so bad is that it seems to lack explanatory distance. But when we look more closely, this case is very similar to explanations permitted by contemporary theories of grounding and of dispositions, where facts about patterns in events are explained by and grounded in facts about dispositions. If so, then perhaps this dormitive virtue explanation isn’t so bad after all.
It might seem that this issue of explanatory distance is resolved by the idea that nothing can explain itself, which is a standard feature of most accounts of explanation. But, as I argue in the article, there are cases that at least appear to meet the irreflexivity requirement but that still lack proper distance. Accordingly, this issue deserved an independent treatment.
In the article, I consider some accounts of explanatory distance that seem promising but turn out not to work, including a modal account, a hyperintensional account, and some accounts based on recent work on identity. Instead, I propose that explanatory distance is connected to dependence. This is motivated by ‘backing models’ of explanation, which encode the rough idea that explanations report on and are supported by dependence relations, such as causation. I propose that if the explanans gives information on which whatever is described in the explanandum depends, then there is enough distance for explanation (though, of course, the explanation might fail for other reasons). This makes sense of what went wrong in the train case—there is no dependence between the delay and the lateness, and so there is not enough distance here for explanation. Whereas if the conductor described a cause of the lateness, such as a crash earlier in the day or even the classic ‘leaves on the line’, then the distance requirement would be satisfied.
I leave open questions about the nature of dependence and what forms of dependence can play this explanation-supporting role, so as not to attempt to settle core disputes in philosophy of science about causal versus non-causal explanation or explanatory realism versus its alternatives. But overall I favour a kind of pluralism about explanation, and hence about dependence, according to which there are many different kinds of explanation, backed by different kinds of dependence, and context determines which dependence relations and hence which kinds of explanation are salient and appropriate. I think this does a good job of making sense of the diversity of our explanatory practices, and also my own lingering sense of unease about certain explanations that do meet the distance requirement. For instance, if the fact that some paint is red genuinely depends on the fact that it is scarlet, then this explanation will be fine with respect to distance; but it might also be totally inappropriate in most contexts, which goes some way to account for its impression of oddness.
One interesting feature of this proposal is that it reflects the fact that often puzzles and questions about explanation point towards puzzles and questions about metaphysics. For instance, think back to the dormitive virtue case. If we endorse an ontology of dispositions and a metaphysics of grounding such that the disposition—the dormitive virtue—grounds the patterns in opium’s behaviour, then on this interpretation the dormitive virtue explanation displays proper distance. If there is no disposition and no grounding, however, then the attempt at explanation lacks distance. The puzzle about this attempt at explanation is generated by a puzzle about the background metaphysics, and to resolve it, we have to think about metaphysics as well as about explanation.
This proposal has upshots that are relevant to some recent debates in philosophy. One is that it offers resources for thinking about attempts at explanation where the explanans and explanandum are very far apart, such as if I try to explain the coldness of my tea by citing the density of a distant planet. These are (at least apparently) too distant for dependence, and so too distant for explanation. Anna-Sofia Maurin is currently writing on this topic, with regard to some broader issues about metaphysical explanation. Another interesting feature is that the distance requirement accommodates cases where parts depend on wholes, and so features of wholes may explain features of the parts, as in priority monism, and perhaps also in Humean explanations. Erica Shumener has a terrific article about partial explanation and Humean laws, recently published in the BJPS, that everyone interested in this topic should read.
Moving forward, I’m continuing to develop this picture of explanation as giving information about different kinds of dependence relations. This is funded by a broader interest in connections between explanation and metaphysics. Often facts about explanation are used as a guide to facts about metaphysics, but typically that work has been based on some presumptions about explanation that I don’t share, including explanatory realism. I want to see what routes from explanation to metaphysics might remain once explanatory realism is off the table, and articulating this picture of explanations as backed by different kinds of dependence relations is central to that work. I’m also interested in the challenges raised by explanations of social phenomena for realist pictures of explanation, and of inquiry more generally (see here, here, and here).