The epistemic domain is evaluative. It contains normative facts: you should think Trump is a toss-up to win re-election; reasons: your reason to believe Stephen Miller’s couldn’t write a traffic ticket; evaluative claims: knowing Donald Jr. was too ignorant to engage in criminal conspiracy is better than merely believing he was; normative properties: the property of being justified in your belief that all of this is simultaneously seriously troubling and wildly entertaining; and so on.
Other evaluative domains include the moral, aesthetic, instrumental, and the etiquettical. It’s increasingly common to think there are interesting ways in which evaluative domains interact. In particular, it’s increasingly common to think the epistemic domain sometimes interacts with other evaluative domains. I find it helpful to think of proponents of views in this area as comprising three non-mutually exclusive, probably not exhaustive, sorts: effecters, encroachers, and eliminators.
Effecters think a non-epistemic domain sometimes has an effect on the epistemic domain. One version of this idea is that although (say) one’s moral reasons aren’t ever themselves epistemic reasons, they can have an effect on one’s epistemic reasons. Encroachers think a non-epistemic domain sometimes moves over to or intrudes on the epistemic domain. One version of this idea is that (say) one’s pragmatic reasons are themselves sometimes one’s reasons for belief; pragmatic reasons can constitute epistemic reasons. Eliminators think the epistemic domain can be eliminated in favor of a non-epistemic domain. One version of this idea is that (say) epistemic reasons are instrumental reasons. Hence we can eliminate the idea of a distinctive class of epistemic reasons.
In previous work I’ve been by turns an effecter, an encroacher, and an eliminator. This has always puzzled me. Granted, the views might not be mutually exclusive; but why don’t I settle on one? Putting aside more compelling autobiographical explanations, I have a new hypothesis: it’s because I don’t have any kind of grip on what distinguishes evaluative domains one from another. In particular: What distinguishes the epistemic domain from non-epistemic domains?
This is a question about how to think about boundariesbetween evaluative domains, between evaluative types. Contrast two approaches. The first thinks about evaluative boundaries as mind-independent features of the evaluative world. Our activity is then understood in terms of the metaphors of discovery, perception, or carving evaluative reality at its joints; call this evaluative type realism. The second approach thinks about evaluative boundaries as mind-dependent features of the evaluative world. Our activity is then understood in terms of the metaphors of construction, projection, or carving evaluative reality — but, crucially, not at any joints! Call this evaluative type anti-realism.
Don’t confuse these approaches with metanormative realism and anti-realism. Metanormative and evaluative type (anti-) realism target separable questions. Metanormative realists can accept evaluative type anti-realism and metanormative anti-realists can accept evaluative type realism. (The latter view would admittedly be an odd duck.) Still, the views share a lot in common; we’re on familiar ground. I’m tempted by evaluative type anti-realism.
Evaluative type realism faces analogues of familiar worries, among which are Mackie’s worries about disagreementandqueerness. Assume metanormative realism is true. This ensures we’re not mixing up intuitions about Mackie’s worries targeting thatrealist view with how those worries apply in the present case, with respect toevaluative type realism.
First, disagreement. In the present context, the Mackian thought is that there is widespread, systematic, disagreement over the boundaries of evaluative domains. For instance, some philosophers think the epistemic domain includes what others insist belongs to morality, or to prudence. This systematic disagreement presumably militates in favor of thinking we’re not in the business of tracing lines antecedently laid down in evaluative reality.
Next, queerness. Mackian queerness comes in two sorts: metaphysical and epistemic. Focus on the former. In the present context, the Mackian thought is that evaluative reality’s having natural joints would be a metaphysically odd feature for evaluative reality to have. Why odd? Because even if you think value is mind-independent, it’s odd to go on to think that that value comes with edges, or specific boundaries at which it makes sense to say one type of value ends and another begins. Even among mind-independently real entities, this just doesn’t seem to be how boundaries between types work. An analogy can help.
I’m an animal realist: animals’ existence doesn’t depend on any mindly facts. I bet you’re an animal realist too. But it’d be quite odd to go on to accept animal type realism, the view that typesof animals are a mind-independent feature of reality we aim to accurately represent in our thinking about the animal kingdom. That is clearly false, and it’s not in any way motivated by our shared commonsensical animal realism. There are no metaphysically real divisions among different types of animals that exist independently of our mindly activity of categorization thereof. This is not to say that there are no divisions it’s more or less ‘natural’ to draw among animals. I’m not inclined to group toads with gazelles, or E. coli with sparrows. But that there are boundaries we’re more or less inclined to accept is not probative with respect to the claim that there are mind-independently real boundaries we’re aiming to trace, and to which our inclinations are responsive. I agree gazelles go more naturally with deer than with toads. That does not involve any commitment on my part to some mind-independently real category C such that deer and gazelles are necessarily C and toads are not. Indeed: ‘real’ mind-independent boundaries between types of animals would be metaphysically quite queer: what sort of thing would such a boundary comprise? Would it be a bit of DNA (no), a shape (no), environment (no), some shared ability (no)? There just isn’t any mind-independently real relation or quality for the specific boundaries between different types of animals to be. Of course, we can posit that nevertheless there aresuch boundaries. But then such boundaries are, metaphysically speaking, queer. Instead, we should think that boundaries between types of animals are like boundaries between nation-states: they exist, but only because and to the extent we say so. Such borders depend on us.
The same goes for value. Values (we’re supposing) exist independently of any mindly activity. But this is neither here nor there when it comes to the question of whether ‘real’ mind-independent boundaries between kinds of value exist. There are no metaphysically real boundaries between different types of value. This is not to say that there are no divisions it’s more or less ‘natural’ to draw among values. I’m not inclined to group great art with great persons. But that there are boundaries we’re more or less inclined to accept is not probative with respect to the claim that there are real boundaries we’re aiming to trace, and to which our inclinations are responsive. I agree kindness goes more naturally with compassion than with truth. That does not involve any commitment on my part to some mind-independently real category of ‘moral value’ such that kindness and compassion are necessarily moral values and truth is not. Indeed: ‘real’ mind-independent boundaries between types of value would be metaphysically quite queer: what sort of thing would such a boundary comprise? We should think that boundaries between types of value are like boundaries between types of animals and between nation-states: they exist, but only because and to the extent we say so. Such borders depend on us.
There are ways for the realist to respond. Judging from the literature, the likely reply will be either that (i) there’s a single object (e.g., belief) or collection of objects (e.g., doxastic attitudes) such that these and only these objects are the appropriate objects of a type of evaluation (e.g., epistemic) or that (ii) there’s a single value (e.g., truth) or collection of values (e.g., knowledge, truth, understanding) such that these and only these values uniquely determine a type of evaluation (e.g., epistemic). Such replies shift the question without answering it. We shall now want to know why theseattitudes are the exclusive domain of thisevaluation. (Equivalently: what distinguishes the doxasticfrom the non-doxastic?) We can certainly stipulate that epistemic evaluation is evaluation that turns on a connection to truth, or some other plausibly epistemic value, such as ‘knowledge’ or ‘understanding’. But such stipulation is question-begging in the present context, since we shall want to know why ‘truth’ but not ‘promoting my desires’ is an epistemicvalue. (A variation on this theme is the idea that particular domains involve certain reactive attitudes, but not others. You can see why this too is running in place.)
I think these considerations militate in favor of an anti-realist approach. I don’t have anything like the space to go into detail here, so I’ll just say something suggestive and RIP inbox. We should be functionalistsabout types of evaluation. Evaluative types are functional types, defined by what they’re in the business of doing. The central idea is familiar from a range of other disputes in and outside philosophy. Biological functionalism: hearts are forpumping blood. Mental state functionalism: beliefs are forserving as maps to our environments. Evaluative type functionalism: epistemic evaluations are forsome purpose; moral evaluations are forsomething else. Evaluative types are defined by their roles. I have views about what those roles are, at least in the epistemic case, but I’m already over-limit.
I agree with the thrust here. The idea that Western Europeans, round about the 1600s, somehow “discovered” a mind-independently existing boundary between other types of reasons and a special domain of reasons called “moral reasons” is kind of silly; it’s hard to see how that could legitimately be described as a story of discovery. Even after that point in intellectual history, folks like Hume still refuse to treat “prudence” and “morality” separately, as we might be inclined to do, and I think it’s just prima facie absurd that Hume lacked epistemic access to some mind-independent truth about boundaries.
Any thoughts on how this position interacts with the debate over all-things-considered judgments? (i.e. Tiffany, Baker, etc)?
Nick,
As I understand Baker, Tiffany, et. al., they’re interested in the question of whether there’s a single ought that (authoritatively) collects the others. I’m with them both on a ‘no’: we should be normative pluralists of the sort Baker and Tiffany both describe. This means that one way of attempting to trace the lines around different evaluative domains, viz. by way of their authoritative relations (morality trumps prudence, prudence trumps etiquette, etc.) to each other, won’t work. Still, it’s possible to think there are ways to outline evaluative domains without saying anything about authority in this way. For instance, setting aside the questions of whether there’s an authoritative ought and what it’s like, you ask what the shape of epistemic or moral oughts are like — what their appropriate domain of application is. I don’t think that latter question has any realist sort of answer. So I think Derek’s position (I’m thinking of his Skepticism about Ought Simpliciter) goes some of the way toward where I’d like to go. Maybe he’d be happy to go the rest of the way too?
Hi Nate,
Cool post. I agree that if one denies that there’s an all-things-considered ought, there is pressure to start doubting that standard evaluative domains carve nature at the joints. Here’s one way of thinking about this: if you deny that there is any normative authority, you have to think that all normativity is formal (as Tristram McPherson puts it). To be formally normative is just to be normative in the sense that of being a standard that things can succeed at meeting or fail to meet. Formally normative standards are cheap. It is easy to stipulate them into existence. This is what happens when I tell everyone that we’re going to adopt certain house rules for playing Monopoly, for example.
So now let’s say A says that pragmatic considerations can’t be epistemic reasons, and B says they can. An evaluative domain that only contained evidential reasons would be formally normative, and an evaluative domain that included both evidential and pragmatic reasons would also be formally normative. So both are advocating for formally normative standards.
Now if one of these domains possessed some degree of authority regarding what to believe that the other didn’t, it would make sense to think that that domain is the real epistemic domain. But if all we believe in is formal normativity, there is no authority, and neither of these two domains is more authoritative than the other. So both have equal claim to being the real epistemic domain. In short, there’s some pressure to reinterpret intra-domain disputes as cases where people are simply advocating for different ways of grouping reasons, values, etc. (After all, both domains are equally real and equally normative–since formal normativity is the only normativity there is.)
So there’s definitely that. On the other hand, there do seem to be some natural divisions despite this. The most obvious one would be the legal domain, at least if positivists are correct. Assuming positivism, we have a set of evaluative facts that reduce to a specific set of sociological facts. That would seem to give a joint-carving way of specifying the legal. More generally, let’s say that we discover that those normative facts we intuitively identify as moral don’t admit of naturalistic reduction, but those we intuitively identify as legal facts or facts of etiquette do. That again would seem to point to some joint-carving difference.
Admittedly, the way we’re carving joints seems to be about metaphysical nature, rather than their status as normative. But I’m not sure at this point how to state the feeling.
Finally, this is probably a stupid question, but what’s wrong with those standard taxonomies you see in biology when it comes to joint-carving distinctions among animals? I.e., dogs, wolves, and hyenas all belong to the family of canidae, etc? Are there problems with taking these classifications as marking real, mind-independent distinctions?
Derek,
Thanks for this! I agree that the denial of an ATC ought puts pressure on one way of drawing the boundaries, and for exactly the reasons you describe. But that leaves open the possibility that there are other ways of drawing the boundaries in realisty ways that don’t appeal to authority. My anti-realist says there aren’t. The legal case is a nice one.
If the positivists are correct, then one set of evaluative facts reduces to some sociological facts. But I take it their claim isn’t just that this one set of facts (‘legal facts’) is in principle reducible to another set of facts (‘sociological facts’). It’s that the legal domain is properly understood as having its boundaries *set by* or *grounded in* the sociological domain. It’s a claim about priority. And if we show up on the scene and claim that some facts in the legal domain are not grounded in the sociological domain, the positivist will think we’re making a mistake. What kind of mistake? I think: unless we’re talking past each other, not a mistake that should be realistically construed. I see no reason to think the positivists so understood have found some ‘joints’ along which they’ve carved the legal domain. They’ve just offered us one specification among many possible specifications. It may be a really attractive specification of what the legal domain comprises, but it’s not more attractive because it’s more metaphysically prestigious. If we disagree with them, it’s not because we disagree over what the boundaries of the legal domain *really* are.
This goes to the question about biological taxonomy. I don’t mean to deny that it’s possible to mark mind-independently real distinctions using such taxonomies. (It’s equally possible to mark mind-independently real distinctions in the way the positivist does…) Family-level classifications, e.g., canidae, are actually I think a bad case, because they’re sometimes quite messy. Clades are better: they’re simply defined as groups whose members have a single ancestor. Sure, clades really exist. They exist in the same sense that any set with specified membership conditions exists. And there are mind-independent facts about which clades you and I belong to, etc., that exist in the same way. But I don’t think by identifying either the idea of a clade or some particular clade we’ve thereby identified a special ‘joint’ at which to carve up the animal world. It’s just a super clean way of phylogenetic investigation. And we are (or better: someone is) interested in it. Compare: the set the contains all and only those facts about which credences it’s useful to have. Or: the set that contains all and only those facts about agential well-being.
You’ll recognize this is just reheated rhetoric from, e.g., Blackburn, Price, et. al’s ‘pragmatism’ or ‘functional pluralism’. Another way to put the point is that whatever your other metanormative commitments, we should be functional pluralists about the boundaries between evaluative domains. (As an aside: I’m interested in this view because I think once you pair it with a certain metanormative view about the source of normative authority, the typical objections to that metanormative view get a lot less worrying…)