Social Inconsistency

  • Thomas N. P. A. Brouwer (University of Leeds)


Though the social world is real and objective, the way that social facts arise out of other facts is in an important way shaped by human thought, talk and behaviour. Building on recent work in social ontology, I describe a mechanism whereby this distinctive malleability of social facts, combined with the possibility of basic human error, makes it possible for a consistent physical reality to ground an inconsistent social reality. I explore various ways of resisting the prima facie case for social inconsistency. I conclude, however, that the prima facie case survives scrutiny, and draw out some of the ramifications.

How to Cite:

Brouwer, T. N., (2022) “Social Inconsistency”, Ergo an Open Access Journal of Philosophy 9: 2. doi:



Published on
28 Dec 2022
Peer Reviewed

our social arrangements can be suboptimal, and often are. They can be unjust. They can be inefficient. They can be imperspicuous. They can be unprincipled. In this paper, I discuss another way in which our social arrangements might be suboptimal: they might give rise to inconsistency.

The case for this is, in a very small nutshell, as follows. How people think, talk, and behave shapes the way that social facts come about. Put more technically, our thought, talk and behaviour determine the grounding principles by which social facts arise from other facts about the world. Because we can think, talk and behave in incoherent ways, it is possible for us to arrange those grounding principles badly, in such a way that a consistent physical world might end up grounding an inconsistent social world.

In what follows, I will set out that line of reasoning in a lot more detail, and put it under pressure in a number of ways. This turns out to open up a rich vein of questions about the metaphysics of the social world. What, exactly, is distinctive about how social facts come about, and should we be less surprised if such facts turn out to be logically ill-behaved than if, let’s say, the facts of geology do? What, precisely, is at stake in asking whether social facts are real? To what extent should our views on what’s possible in social reality be constrained by our practical and theoretical purposes in dealing with that reality? Some salient theoretical choice-points are revealed, and while I will argue that the case for inconsistency survives scrutiny, I will indicate to those committed to consistency where their best options lie.

Beyond its intrinsic interest, the question of social inconsistency has ramifications for other issues. If the social world is possibly inconsistent, this matters for how we reason about that world; any logic on which contradictions imply everything (e.g., classical logic) would be ill-suited to the task. Social inconsistency also matters for normative theory: it is very common for social facts to have a normative upshot—to give rise to reasons, obligations and permissions—and it is reasonable to worry that if the social facts are inconsistent, their normative demands might sometimes be unsatisfiable. The possibility of social inconsistency also has methodological consequences: for any phenomenon that is possibly in whole or part socially constructed, we cannot lay down as a methodological constraint that our best theory of it will be a consistent one. Since there is much ground to cover in discussing the question of social consistency itself, my discussion of these ramifications will be limited, but I will try to indicate what the important questions are and how we can get started on them.

I proceed as follows. I outline (§1.1) a certain way of thinking about social facts: a two-dimensional approach to social ontology. I set out (§1.2) the basic mechanics of one view in this spirit: Epstein’s anchoring-grounding model. After some clarifications (§2) concerning the notions of inconsistency and contradiction, I articulate (§3) how inconsistency in the social facts might arise. I question (§4) this prima facie case for social inconsistency in a number of ways. Finally, I discuss (§5, §6) some ramifications of accepting it.

1. The Metaphysics of Social Facts

1.1. The Nature of Social Facts

The social world is made up of such things as organisations, social groups, artefacts, conventions, practices and laws. It is metaphysically distinctive, because its nature is conditioned by human thought, talk and behaviour, in particular thought, talk and behaviour of a collective sort. Whereas the chemical facts, for example, are indifferent to what we think or expect of them, it seems the social facts wouldn’t be as they are if we didn’t think about them and react to them as we do. To label this pre-theoretic contrast, we might say that the social world is humanly malleable in a way that much of the world is not.

One desideratum for a theory of social reality is that it make sense of this malleability, and part of that task is to distinguish it from other ways in which we shape the world around us. We also change the world by building things in it, landscaping it, polluting it, and so forth. These are profound ways in which we can change things, but they seem metaphysically different. These are causal, rather than constitutive ways of determining the facts.1 While we have causal influence over many parts of reality, it seems distinctive to social reality that we in addition have constitutive influence over it.2 When, for instance, a state passes a law to expand the franchise to the over-15s, 17-year-old Jane is thereby changed into a voter—there is no causal chain that runs between these facts, just a direct relation of metaphysical determination. The state could not, however, change the value of pi to 3 by passing a law to that effect.3 Once we’ve noted this power that we have to constitutively shape the social world, the challenge is to explain how it works, and to do this in a way that respects its limits. Most social facts are not subject to arbitrary stipulation, either by individuals or by collectives. We cannot, for instance, dictate that there is not a recession on; social facts are clearly not so malleable that they are whatever we want them to be. There is an epistemic dimension to this too: our relation to the social facts is not such that we can know them just by inspecting our own attitudes toward them. By and large, we have to find out what the social facts are, often through pretty onerous research. One way to put this is that the social facts seem to enjoy some form of objectivity.4 Making sense of this objectivity, whatever precisely it amounts to, is a second desideratum for our theories of social facts, which has to be squared with the malleability desideratum.

This, it turns out, is somewhat tricky to do. Some theories of social facts, particular earlier ones, have sought to make sense of the malleability of social facts by characterizing them as dependent on, or even consisting in, our attitudes.5 Hayek, for instance, writes that “in the social sciences the things are what people think they are. Money is money, a word is a word, a cosmetic is a cosmetic, if and because somebody thinks they are” (Hayek 1943: 3). This has the virtue of giving us a very concrete proposal for how we constitutively shape the social facts: for a social fact p, we make it the case that p by believing that p. However, it seems to put us in a bad position to account for the objectivity of social facts: how would we not have automatic epistemic access to social facts, if they are as Hayek describes them here, and how would we not change them by changing our beliefs about them?

So prima facie, a mind-dependence thesis accounts for the malleability of social facts at the expense of their objectivity. But there is room to manoeuvre. Searle (1995), for instance, manages to strike something of a balance between mind-dependence and objectivity by making the mind-dependence indirect: the general conditions that have to be met for social facts to obtain are mind-dependent, but whether things meet those conditions is up to the world, and not up to us. Thus we may have privileged epistemic access to what it takes for someone to be a voter, for example, but no privileged epistemic access to whether Jane is a voter—we’d have to find out Jane’s age first.

Even with such manoeuvres, though, Searle seems badly placed to deal with certain features of the social world—for instance, how it can be possible for there to be a recession on in a society in which no-one has the concept of recession (Thomasson 2003a). For such and other reasons, various authors (Guala 2014; Epstein 2015; Torregno 2017) have argued that we should let go of the idea that social facts are mind-dependent. Even without ruling on that question directly, we can say the following: that it might be best not to tie our understanding of the malleability of the social world directly to mind-dependence. We had best start by approaching the combination of our desiderata at a slightly higher level of abstraction, so that we can decide at a later point in the story how much mind-dependence, if any, we should countenance.6 I am going to suggest that theories that do a good job of reconciling malleability and objectivity do so by discerning a two-dimensional structure in social reality. Let me explain what I mean by that.

1.2. Two-Dimensionalism about Social Reality

Suppose that Jimmy wants to go on the roller-coaster, but the sign says that one has to be at least five feet tall to go on. It’s a social fact that Jimmy can’t go on the roller-coaster. If we want to explain that social fact, the most immediate and natural thing to say is that it is because Jimmy is only 4’6’’. That is kind of striking: it seems we can account for a social fact by citing a plain old physical fact. And indeed, relative to that being the rule, Jimmy’s height is what is salient in explaining why he is roller-coaster-ineligible.

Jimmy may nevertheless demand more explanation. Why can’t he go on, just because he’s 4’6’’? Just telling Jimmy that that’s the rule is unhelpful. The rule could have been different—the cut-off point could have been 4 foot instead of 5 foot—so we can ask why it is as it is. The explanation might be, for example, that the person who runs the roller-coaster wrote ‘5 foot’ on the sign, and they have the authority to make the rules. That is a distinct but complementary explanation of Jimmy’s predicament. It tells us how and why the plain old physical fact that Jimmy is 4’6’’ has the bearing that it does on the social fact that Jimmy can’t go on the roller-coaster.

We can typically distinguish two complementary explanations of this sort with respect to social facts. One concerns how the world is, and one concerns conditions that we require of that world. I call this dual explanatory structure a two-dimensional one because it gives social facts a certain sophisticated modal profile. We can imagine the social world to have been different in two separate ways. On the one hand, the plain old physical facts that underpin and determine the social world could have been different. On the other hand, the conditions that we require those physical facts to fulfil could have been different. Jimmy could have gone on the roller-coaster if he’d been taller, and Jimmy could have gone on the roller-coaster if the rule had been different.

This allows for two distinct kinds of counterfactual reasoning about the social world, which we can usefully think of as involving different modal dimensions. We can hold our world fixed and consider what it would be like under the imposition of different conditions, or we could hold our conditions fixed and consider what different worlds would be like under the imposition of those conditions. As an analogy, think of the former as looking at the same planet through a range of different telescopes, and the latter as using the same telescope to look at a range of different planets.

This nicely reconciles the desiderata, albeit only at a certain level of abstraction, with many details left to fill in. One the one hand, this approach does justice to the objectivity of the social world: social facts are explained and determined by underlying physical facts, and are just as objective as them, not mere figments of our collective imagination. On the other hand, it does justice to the malleability of social reality: we human beings—since we’re the ones who put the conditions in place—are crucially involved in determining how the social world comes about.

As it will turn out, this two-dimensional structure is also what allows us to imagine a consistent physical world giving rise to an inconsistent social world. To make the case for that (and, later on, some cases against it) it will help to have somewhat more precise tools to work with—for that purpose I borrow Brian Epstein’s (2015; 2016a) anchoring-grounding model.

1.3. The Anchoring-Grounding Model

Two-dimensional models of social facts have been around for a while, if not always under that description. Searle’s (1995; 2010) theory of institutional facts arguably has this sort of structure, or an approximation of it, under its hood, though Searle himself did not draw out this aspect and sometimes glosses his theory in ways that go against it. The same goes for theories that either build upon or resemble Searle’s theory, such as Hindriks (2009), Thomasson (2003b) or Tuomela (2013). A more explicitly two-dimensional approach was proposed by Einheuser (2003; 2006)—she adapts a Kaplan-style semantics for context-sensitive terms to give a metaphysics of convention-dependent facts, with special attention given to modelling counterfactual reasoning about such facts. But the most developed two-dimensional theory is arguably Brian Epstein’s (2015; 2016a) anchoring-grounding model.7 On Epstein’s theory social facts are first of all associated with grounds, in the standard metaphysical sense: facts that underlie and metaphysically determine them.8 The grounds associated with a given social fact may themselves be social facts, but by looking at the grounds of those facts and so on and so forth, we eventually get to some non-social facts, because the social world is not metaphysically basic. The facts that can play the grounding role with respect to social facts are not of any specific type: what they look like depends on the kind of fact that’s being grounded. That something is a dollar bill is grounded in how and where it was manufactured (a fact about its history); that someone is popular is grounded in how other people regard them (a fact about attitudes) that Jimmy can’t go on the roller-coaster is grounded in his height (a fact about physical properties).

When we ask why some social facts are as they are, we are typically asking about their grounds, in the first instance. If it’s a fact that Fred is guilty of bicycle theft, that’s explained by his taking a bicycle that wasn’t his. The latter fact grounds the former. But as in Jimmy’s case, there are further questions to be asked. Why is that what it takes to be guilty of bicycle theft? Now we’re asking why the grounding relations run as they do. In Epstein’s theory the grounding principles that govern a given case are explained and determined by different facts which he calls anchors. In this particular case, the anchoring facts will be facts about the legislative process which led to some law governing theft being adopted. Anchors are often social facts themselves, and what they anchor are grounding principles of the general form ‘if A, then A grounds B’ (where B is some social fact). The facts that show up among the grounds and the anchors are not metaphysically different kinds of facts, and one and the same fact could feature in either role on occasion. But the explanatory roles are different.910 The grounding principles together compose what Epstein calls the frame (hence he also calls them ‘frame principles’). One could think of a frame simply as a set of all the grounding principles that are in force. Another, more model-theoretic and coarse-grained way of thinking of it is as a function from worlds (without social facts) to worlds (with social facts). The space of possible frames and the space of possible worlds-not-yet-endowed-with-social-facts (‘pre-social worlds’, for short) constitute the two dimensions of a two-dimensional model; we can think counterfactually about the social world in one way by holding the frame fixed and varying the world, and in another way by holding the world fixed and varying the frame. We could vary them in concert for a ‘diagonal’ counterfactual.11 For our purposes it is worth saying a little more about the kinds of facts that crop up in the anchoring role. When we have done the work of uncovering the grounding principles that govern some aspect of social reality, and we then ask why those grounding principles are as they are, a variety of facts could come into view.12

  • Facts concerning speech acts or inscriptions. The fact that a crime of robbery needs to involve a threat of violence is explained (in part) by the fact that a certain form of words defining the act of robbery appears in the law code.

  • Facts about the functional roles of social kinds or institutions. The fact that waving your arm can constitute a greeting but thinking happy thoughts cannot is explained (in part) by the fact that the latter is not outwardly perceptible and hence not apt to play the role of a greeting.

  • Facts about paradigm instances of a social kind. That dancing the Macarena involves such-and-such moves is explained by the appearance of those very moves in the 1996 music video.

  • Stable patterns in behaviour. The fact that deer musk-markings constitute territorial claims is explained by the fact that deer are stably disposed to react to them in a certain way.

What this is meant to get across is that a variety of anchoring facts of different kinds can simultaneously be involved in explaining why the social world works as it does. Unlike what one might want of, say, an explanation of why the laws of chemistry work as they do, it’s quite reasonable to think of social reality as being ‘cobbled together’ by the distributed operation of many such facts. There is not likely to be a neat, unified explanation on offer of the social world as a whole—rather, there is a patchwork of local explanations. And harking back to our earlier discussion of mind-dependence, we can see that within Epstein’s approach, it becomes a more local question whether either the grounds or the anchors relevant to a given social fact are facts of a mental nature—as opposed to Searle (1995), mind-dependence is not an architectural feature of Epstein’s model.

I will rely on Epstein’s theory to impose some clarity and precision on what follows. But it will be possible to recapture the gist of what follows in other approaches that admit of some two-dimensional structure.

2. Inconsistency

Before talking about social inconsistencies specifically, it will also be useful to say a few things in general about the notions of inconsistency and contradiction.

A contradiction, as it is most typically characterized in logic, is a sentence of the form A & ¬A. Sometimes it is also characterized as a sentence A that is both true and false; these respectively ‘syntactic’ and ‘semantic’ characterisations are coextensive in many logics, but I will opt for the former for definiteness. A set of sentences is said to be inconsistent when a contradiction can be derived from it, and an inconsistency within a set of sentences can be characterized as a minimal subset of sentences that renders the overall set inconsistent. In the most basic case, that would be a pair of sentences {’A’, ‘¬A’}, but more than two sentences may be involved (e.g., {’A → B’, ‘B → C’, ‘C → ¬A’}).13 Dialetheism is the position that contradictions are sometimes true: that there are dialetheias, truths of the form A & ¬A. The most important motivating cases for dialetheism have been logical ones, in particular semantic paradoxes like the Liar (“this sentence is false”). Since satisfactory consistent solutions to semantic paradoxes are hard to achieve, the view that we cannot solve them and must take their conclusions at face value has gained a foothold in the philosophy of logic over the last few decades.14 Dialetheism has also sometimes been defended concerning other types of cases, such as vagueness (Hyde 1997) or physical change (Priest 1987/2006). There has not, however, been much discussion of dialetheias in the social world.15 Dialetheism is considered a provocative view, for a few reasons. First, inconsistencies have a very distinctive property within many logics, including classical logic: they entail every other sentence, and thereby induce triviality (they ‘explode’). This means that defending dialetheism involves adopting a non-classical logic in which inconsistencies do not have this property—a ‘paraconsistent’ logic. A distaste for such logical complications might incline one against dialetheism. Second, there is a long tradition in philosophy, from Aristotle onward, of arguing that ruling out inconsistency is in some way or other a cornerstone of rationality: to admit its possibility would, for example, undermine rational thought, linguistic meaning, or meaningful disagreement. Some of these misgivings about dialetheism can be fairly easily addressed, whereas others lead to thorny dialectics. It’s not going to be fruitful to involve ourselves in these general debates here, so I just point the interested reader to Priest (1998) for an overview.

A terminological note before we move on. The approach to social ontology that I build on here—Epstein’s—proceeds in terms of facts. The notion of a fact that’s at work here is a metaphysically bland one: just the everyday notion of something being the case. But because the approach proceeds in these terms, in what follows I will frequently be speaking of contradictory or inconsistent facts, or inconsistencies in the facts. In doing so, I am not intending to carve out some new and exotic notion of inconsistency, one that is metaphysical rather than logical. Rather, it’s just a different way of expressing the notion of inconsistency characterized above. I will ask the reader to imagine, when I speak of inconsistencies in the social facts, that we have agreed upon some language in which to characterize the social facts, and that I’m saying that the true description of the social facts, in that language, contains sentences which entail a contradiction. Similarly I will sometimes speak of, for example, ‘the fact that ¬A’. Rather than invoking some metaphysical theory of negative facts, in those instances I am simply referring to a fact that, in our chosen language, would be expressed by the negation of a sentence that expresses the fact that A.

In the next section, I will state the prima facie case for the possibility of social inconsistency, a case which I will then subject to scrutiny.

3. Social Inconsistency

Using Epstein’s model, we can describe how a consistent set of physical facts can ground an inconsistent set of social facts. All that it takes is for us to anchor the wrong set of grounding principles: one that is inconsistency-permitting.

We can think of Epstein’s frames as functions that take us from pre-social worlds—that is, worlds that don’t yet have any social facts—to social worlds—worlds that do have social facts. An inconsistency-permitting frame f would then be one which, for some pre-social world w1, yields a social world w1+ at which, for some social state of affairs A, both A and ¬A are the case.

We can also think of frames (in a more fine-grained way) as sets of grounding principles {G1, G2, G3, …} that take us from a set of non-social facts (the grounds) to a set of social facts (the grounded facts). Let us then call a frame f inconsistency-permitting if its grounding principles are such that, for some set of grounding facts Γ and some social state of affairs A, f makes Γ ground both the fact that A and the fact that ¬A.

Such an inconsistency-permitting frame (whether considered as a set of grounding conditionals or as a function from worlds to worlds) need not itself be inconsistent in any sense.16 Nor does an inconsistency-permitting frame guarantee an inconsistent world. But should the right grounding facts come about, such a frame gives us an inconsistent social world, grounded in a consistent world. (In what follows, I will sometimes use ‘inconsistent frame’ as shorthand for ‘inconsistency-permitting frame’.)

Those are just definitions, of course. The real question is: can we actually have a frame that’s inconsistency-permitting? Well, for all that we have said so far, the answer seems to be “yes”.

Anchoring, as described above, happens in a piecemeal way, a principle here and a principle there. The grounding principles are not put in place by some unified mechanism, but by the socially and temporally distributed operation of a variety of anchoring mechanisms. This means that, unless there is some kind of large-scale and effective attempt at coordination going on within a population as things get anchored, there is no guarantee that a frame would not end up with some dropped stitches. And one of the things that might go wrong is that at some point, by some anchoring mechanism, a grounding principle is anchored that entails that Γ will ground A, and at some other point, another grounding principle is anchored that entails that Γ will ground ¬A.17 Discussing some examples might help to make the possibility more concrete. (Note that I am putting these forward as illustrations, not as evidence.)

Priest (1987/2006) offers a simple example case. He imagines a polity which passes a law which says that those who own a certain amount of property have the right to vote. It also passes a law which says that being a woman bars you from voting. At the time these laws are passed, we are asked to imagine, the possibility that a woman might own the requisite amount of property is so distant that this scenario doesn’t even occur to the law-makers. But history ticks on and at some point a woman does own that amount of property. She now, legally, has the right to vote and does not have the right to vote.18

Here, by passing the first law, the community has anchored a grounding principle which lets the fact that someone can vote be grounded by the fact that they own a certain amount of property. And by passing the second law, they have anchored a grounding principle which lets the fact that someone cannot vote be grounded by the fact that they are a woman. Then all it takes is that someone satisfies both conditions for it to be a fact that they can vote and a fact that they cannot vote.

Another example would be the venerable case of Euathlus and Protagoras: Protagoras has trained Euathlus in rhetoric, and the two have signed a contract which entitles Protagoras to a fee when Euathlus wins his first court case. But Euathlus tarries in getting his legal career underway, so Protagoras sues him for the fee.19 We can take their contract, here, to set up a grounding relation: if Euathlus wins, that grounds the fact that he owes Protagoras the fee. But if Euathlus wins, that would also ground the fact that he does not owe Protagoras the fee.20

The Priest case is made up—though not particularly far-fetched—and the Protagoras-Euathlus case is likely apocryphal, but purported actual examples have also been offered. Ted Cohen (1990) argued that the then-official rules of baseball counted a runner as both in and out (where being in entails not being out and being out entails not being in) if they arrived at first base at the same time as they, or the base, were tagged. Cohen claimed to have seen this very situation arise. Here, it’s the adoption of a set of rules for baseball that sets up grounding relations between on the one hand facts about the relative position of people and balls, and on the other hand social facts about who is in or out. Since the rule-makers failed to spot certain implications, they created a situation in which consistent physical facts could ground inconsistent social facts.21

Still, we might think: if social inconsistency is indeed possible, wouldn’t we expect to be faced with it a bit more often? There are, however, reasons for thinking that such cases might be rare or unobvious.

First, social inconsistencies might be more common were it not for the fact that we typically do not have a practical need to put in place grounding conditions for a social state of affairs A and also, independently, for the state of affairs that ¬A. Rather, we would specify grounds for A, and let ¬A be grounded by the absence of grounds for A. In such cases, no opportunity would arise for rendering the frame inconsistency-permitting. Nevertheless, in sufficiently complex situations it could easily happen that grounding conditions are specified for some state of affairs A and grounding conditions are specified for some state of affairs B, where as it happens, A entails C and B entails ¬C for some state of affairs C, and the grounding conditions for A and B are such that they could be jointly satisfied. Imagine, for instance, that a tax code specifies grounds for being a small business owner, a status which makes one liable for a certain kind of tax, and that it also specifies grounds for working in a protected industry, a status which makes one exempt from that same tax, and that they fail to rule out that someone might have both statuses.

Second, it is also plausible that the very cases where inconsistency-permitting principles are more likely to have been anchored are also cases where the inconsistencies in question are less likely to manifest. An inconsistency in a set of sentences can be ‘high-k’, where k is the cardinality of the smallest inconsistent subset of the set; informally, this means that an argument that derives the contradiction would have to have many (at least k) premises.22 We are presumably better at avoiding anchoring inconsistency-permitting sets of grounding principles the easier it is to spot how they would permit contradictions. So when a frame permits inconsistencies, they are likely to be towards the higher-k end of the scale. Because more grounding conditions would have to be independently satisfied, the actual realisation of the inconsistencies in question would be less probable. And when they do occur, they are also relatively more likely to lurk benignly in the background, since it would take more work to derive an explicit contradiction from them.

Thus, if social inconsistencies are possible, they may in practice be rare, or hard to spot.

In this section, I’ve given an argument that—given plausible assumptions about the metaphysics of social facts—inconsistencies in the social facts are possible. Though ‘argument’ is a bit of a generous term: all I’ve really done is point out that a two-dimensional approach to social ontology does not rule out that a consistent physical reality might ground an inconsistent social reality. The real debate begins here, for one might well agree that a two-dimensional social ontology gives this result, but think that we have philosophical reason to refine the theory so that it doesn’t give this result. The interesting philosophical question is whether any refinements of this sort are (a) workable and (b) well-motivated.

4. Cases against Inconsistency

Should we really think that social inconsistency is possible? Unless one is already comfortable with dialetheism, one’s first reaction is probably to regard it as an unwelcome theoretical result. It would be philosophically unsatisfying, however, to merely find some theoretical tweak that makes the issue go away. We would like the changes we propose to be motivated, which is to say: the theory should tell us why the prima facie case just presented doesn’t survive scrutiny.

In this section I discuss lines of reasoning that one might appeal to in order to exclude the possibility of social inconsistency. I treat some options briefly, because their limitations are fairly clear. I dwell on one option more extensively, because I think it offers the defender of consistency their best chance. But even that option, I argue, gives limited succour: it allows the committed consistentist to resist social inconsistency in a way that’s well-motivated by their own principles, but it does not have any suasive force for those not antecedently opinionated about the possibility of inconsistency. So at the end of the day the prima facie case stands, or so I’ll argue.

4.1. A Simple Argument

One quick way to argue against social dialetheias is to argue against the possibility of dialetheias generally, and then extend that to the social case. For instance, one could (a) argue that one ought to accept classical logic, (b) demonstrate that dialetheias entail everything on such a logic, (c) note that the social world is clearly not trivial, and then (d) modus tollens one’s way to rejecting social dialetheias. If one could conclusively argue that classical logic (or some other explosive logic) is correct, or that dialetheias are for some other reason disastrous, rejecting social dialetheias on this basis would be fair play. It would not be fruitful to enter that general debate here; I point the interested reader to Priest, Beall, and Armour-Garb (2004).

The main thing to note about this argument is that it doesn’t do anything to explain why the social world doesn’t permit dialetheias, or give us any handle on how the mechanisms that generate social facts prevent contradictions from arising. Insofar as the mechanism I’ve sketched above casts some measure of doubt on the impossibility of real contradictions, invoking the impossibility of real contradictions against it only begs the question. For that reason I’ll move on to lines of argument that engage more specifically with the metaphysics of the case.

4.2. Social Anti-Realism

One response to the prima facie case is to take it at face value, but conclude from it that we ought not accord social facts the same kind of serious ontological status as, say, chemical or astronomical facts. From dialetheism about social facts, we infer anti-realism about social facts, the thought being that contradictions are tolerable only if they aren’t real. In this section, I explore this idea. What I will argue is that while some form of anti-realism about the social may be viable, it is not ultimately an effective way to deal with social inconsistency.

Anti-realism comes in varieties, and I will first note one type of social anti-realism so that I can set it aside for present purposes. A line of thought one sometimes encounters is that social facts are mind-dependent, and therefore not real.23 As we’ve seen in §1.1 above, it is debatable whether we should regard social facts in general as mind-dependent, and as various authors have pointed out in recent years (Khalidi 2016; Mason 2020), the inference from mind-dependence to anti-realism is a questionable one. I will say this: if there is a sense of the term ‘real’ in which one can reasonably conclude that something is not real from the fact that it is mind-dependent, it will be a rather attenuated sense which will not placate those worrying about social inconsistency. Some (e.g., Fine 2001) have proposed reserving the status of reality for the metaphysically fundamental facts, and since what is mind-dependent is thereby not fundamental, this would indeed let us validly infer that mind-dependent social facts are not real. However, this would do no more than put social facts on the same plane of reality as, for example, chemical, biological and astronomical facts—all standardly regarded as not metaphysically fundamental—and to find an inconsistency in the social facts would then be the same, as far as badness goes, as finding one among any of those facts. So I set the kind of anti-realism that is motivated by the (supposed) mind-dependence of social facts aside, and focus on more full-blooded anti-realisms.24

This still leaves us with options; it will help to have a somewhat specific proposal to grapple with, so for concreteness’s sake let us explore a fictionalist proposal. We construe Epstein-style frames as capturing the rules of an elaborate and useful pretence, one that we all participate in every time we engage in social interactions. Organisations, artefacts, laws and the like do not really exist, but it serves our practical purposes to act as if they do, and this pretence is pervasive, in that there aren’t any occasions where we have cause to drop it.

Instead of grounding principles of the form ‘if A, then A grounds B’, we would take the frame to consist in principles of the form ‘if A, then in the fiction, B’.25 If the frame is inconsistency-permitting, then depending on what the non-social world that we take our cues from is like, we may end up with a fiction that is inconsistent.26 But—the thought would go—this is not metaphysically problematic. Fictions, like beliefs, utterances, and theories, can be inconsistent.

Fictionalism in general is a well-explored approach, and the social fictionalist would have resources to draw on in articulating the details of their view.27 There are challenges, though. Thomasson (2003a) mounts an argument against treating social facts as fictional, which turns on the idea that for us to be able to pretend that a thing has some property, the property in question must already be associated with instantiation conditions that we take the thing to not really satisfy. To pretend, for instance, that a tree-stump is a bear is to have in mind what it really takes to be a bear, and pretend that the tree-stump satisfies those conditions.28 Our social practices couldn’t be a pretence of this sort, Thomasson argues. When we treat an object as having some social property, we do not act as if some object satisfies that property’s instantiation conditions when really it doesn’t; though we might have stipulated the relevant instantiation conditions out of thin air, in typical cases the object really does satisfy them, and as a result, really does have the property.

I suspect the social fictionalist could thread the needle here, and perhaps insist that the conditions we get to put in place for social facts aren’t really instantiation conditions for properties at all, but really just conditions for certain pretence-behaviour involving that object to be apt, whatever that might precisely entail psychologically and behaviourally. More generally, they could draw upon the example of other fictionalisms, for instance mathematical fictionalism (Field 1980) in which there is likewise no contrast to be drawn between what fictionally satisfies the instantiation conditions of a mathematical property and what really does.

There are other more general challenges for the fictionalist (see Eklund 2019 for an overview) but I think that at the end of the day, a coherent social fictionalism could probably be articulated. And if one is specifically concerned not to countenance anything inconsistent in reality, it will do that trick. Let us suppose for the sake of argument that this is done; I suspect that the fictionalist would not thereby have achieved quite what they hoped.

Here is the real trouble, as I see it. Let’s say that one’s misgivings about contradictions relate to explosion and triviality. Note, now, that a social fiction in which everything is true is not really any better than a social reality in which everything is true. For even if the social world is a fiction, it is a fiction we continually participate in and which we cannot practically opt out of—the social world, fictional or otherwise, is the means by which we collectively get along together and get things done. If, in the presence of a contradiction, we are committed to a social fiction which is entirely trivial, that fiction would not serve its purpose.

One could deal with this problem by adopting a non-explosive logic, but that option was already available without going fictionalist. And although I won’t argue the details here, I think that other types of general misgivings about contradictions, whatever their merits,29 are also unlikely to be assuaged simply by going fictionalist about social dialetheias, for the same reason: the social world is such that, even it isn’t real, we are condemned to treat it for virtually all intents and purposes as if it is, warts and all. This type of trouble would thus generalize to any form of social anti-realism which, like fictionalism, manages to classify the social as unreal but leaves us to deal with it for practical purposes as if it is real. The motivation for anti-realism falls away: an anti-realist would still have to take some additional, different measures to assuage their worries about social inconsistency, and there is no reason to think that they are in any better place to implement those measures in virtue of having gone anti-realist.30

The alternative that this makes salient is a ‘revolutionary’ anti-realism about the social which tells us not only that the social is unreal but also that we ought to stop treating it as real. This would be an error theory about social reality: there are no social facts, and thus no inconsistent ones. Error theories have been defended about specific social discourses; for instance, eliminativists about race (e.g., Appiah 1995) argue that, as there are no such things as human races, claims about race are erroneous across the board. But as a general approach to social facts, there seems to be very limited mileage in a revolutionary error theory; assuming it could be implemented at all, it would thereby effectively deprive us of the social world altogether, and of all that it does for us. We would cut off our noses to spite our faces.31

Thus social anti-realism seems, if not necessarily unviable as a general approach in social ontology, not to be an effective response to the prima facie case for social inconsistency set out above.

4.3. Appeals to Institutional Functions

Theories in social ontology sometimes accord an important role to institutions, as a central type of social entity used to explain various other kinds of social entities and social facts. Let’s take this idea on board for the sake of argument. Institutions typically serve a purpose, and for a given institution to serve its purpose it has to be put together a certain way. Take the aforementioned institution of greeting people. There are many possible conventions for greeting people, but the possibilities are constrained by the function of greeting. Waving your hand or clicking your heels could be a way of greeting. Mentally adding 5 and 7 together is not a possible way of greeting, because it isn’t a public event and therefore simply won’t do the job. Thus there are functional constraints on what principles we can anchor to govern institutions.

Perhaps, then, functional constraints are what stop us from having an inconsistency-permitting frame. We just cannot anchor the grounding principles that determine the facts about some social institution in such a way that the facts about it could end up being inconsistent, for then the social institution wouldn’t be able to fulfil its purpose. So if the institution does exist and does serve its function, then it cannot have been so anchored.

This initial thought, however intuitive, faces challenges:

  1. There would have to be a suitably close connection between (a) the facts about some social institution being inconsistent and (b) the institution not fulfilling its function. Setting aside the idea that dialetheias as such are world-ending disasters, it is not obvious that they always entail bad consequences for institutional functioning. Some might have no consequences for institutional functioning; some might even be beneficial. When they do give rise to problems in the functioning of an institution, these might be malfunctions rather than a general failure to function.

  2. Because this strategy invokes the idea of an institution’s purpose, it only applies when institutions have purposes. But some (e.g., the hereditary nobility) might not, and it would be odd to hold that institutions can be inconsistent, but only when they serve no purpose.

  3. Suppose that, in some case, inconsistency would stop some actually existing institution i from fulfilling its function. Suppose that some anchoring facts (A1, A2, A3, …) are in place that would, on the face of it, anchor some inconsistency-permitting frame fi governing i. On the present proposal, the operative frame for i cannot be fiso some different frame fi* must instead be in place. But given that there typically would not be a unique way to make fi consistent, what is the frame fi*, and what facts determine that it is the operative frame?

Thus, while it is plausible that there are ways our institutions couldn’t have been, this idea is hard to shape into an explanation of the consistency of social facts.

4.4. Anchor Holism

We have assumed above that anchoring happens in a piecemeal way: we anchor a principle here, a principle there, and the frame is what we get by adding up everything that’s been anchored. This is why inconsistency seems easy: if we anchor principles without coordination, why expect the result to always make sense? But perhaps this is not how anchoring should be understood. Perhaps anchoring is not piecemeal; perhaps entire frames are put in place at once. Call this the holistic option.

We can articulate this further. Let’s say that there is a space of candidate frames f1, f2, f3, …, best viewed in this case as the space of functions from pre-social worlds to social worlds. What determines which of these is our frame is which one of these frames fits best with our overall anchoring-related behaviours. These candidate frames are not just any old list of principles, in the same way that a possible world isn’t just any old set of propositions. A candidate frame has to fit a certain formal profile, and that includes (we propose) not being inconsistency-permitting. As a result, whatever is anchored is guaranteed to force consistency upon the social world.

This is just a technical proposal for how to build consistency into the Epstein model; it needs motivating. But a reasonable motivation can be given. Picture a frame as a theoretical posit: something that an anthropologist studying the social interactions within some population (say) would hypothesize to explain the behaviour they observe. They observe what people say and do, and they propose a way of making all that add up to a social reality, one in which people are responding in reasonable ways to each other and to the situations they’re in. A frame, viewed in this way, has a particular kind of explanatory function: it is meant to rationalize people’s social behaviour.32 If this is the point of positing a frame, then it makes sense that frames might have to obey certain general conditions of coherence, for that could be part and parcel of their rationalizing explanatory role.33

We’re not quite there yet. If this is to constitute a response to the issue of social inconsistency, we need to argue specifically that one of the conditions that candidate frames must obey is consistency. To do that we must show that consistency contributes to making the explanation yielded by the frame an appropriately rationalizing one. Furthermore, we cannot merely argue that by inducing consistency, a frame is ceteris paribus a better explanation. That would only motivate attributing a consistency-inducing frame on the specific occasions where it would yield a better explanation. If consistency is a general requirement on candidate frames, it must be the case that, in each case where we might explain a population’s behaviour with an inconsistency-permitting frame, there is a consistency-inducing candidate frame that does a better job of explaining that behaviour.

In effect we need something like a dominance argument for consistency as a condition on frames, somewhat like the dominance arguments that formal epistemologists offer for probabilism as a condition on rational credences (e.g., Joyce 1998). But it is not clear that such an argument is available.

Imagine what a case might look like in which we would consider attributing an inconsistency-permitting frame to a population. Imagine we are documenting the social structures of population P, the inhabitants of an isolated village. For population P, imagine that grounding principles for social facts are typically put in place via the pronouncements of a local authority figure—call them the Reeve. The Reeve comes up with principles and declares them to the group, and the villagers treat these pronouncements as authoritative. Now distinguish two cases.

  1. The Reeve has put in place a principle to the effect that sheep are not for eating, and later, perhaps in a lapse of judgement, puts in place a principle to the effect that mutton is to be served on Fridays. This presents the villagers with a conundrum, and they end up quietly ignoring the second pronouncement, treating it as some kind of anomaly. If we as theorists now have to choose between attributing an inconsistent frame to P on which sheep are both for eating and not for eating and a consistent frame which excludes the mutton-on-Fridays principle, the choice is easy; the latter explains the villagers’ behaviour much better than the former.

  2. The Reeve makes many pronouncements, and often doesn’t remember what principles they have put in place. Fortunately, their principles often concern matters of little practical import. They put in place a principle to the effect that lambs born on a Sunday are noble creatures, and a principle to the effect that lambs born on a weekend are not noble creatures. Nothing practical follows from this either way, so villagers are happy to affirm the truth of both of these principles if prompted. In this case, it seems there is no explanatory advantage to be gained by attributing to P a consistent frame which edits out the inconsistency, as opposed to an inconsistent one that leaves it in. A consistent frame, in this case, would have both a lesser degree of fit with the apparent anchoring facts and less explanatory value with regard to villagers’ behaviour (specifically, their utterances).

It does not seem, then, that consistent frames dominate inconsistent ones when it comes to rationalizing behaviour; it is not such a universal panacea that we need to build it into the very idea of a frame.

The defender of consistency has a reply to offer here, though. For if our imaginary anthropologist is a classical logician, it may not seem to them that interpreting case (2) through the lens of an inconsistent frame is a good idea at all. If it is true in (2) that sheep are and are not noble creatures, then by explosion everything is true in the social world of (2), including many things that do not chime at all with the behaviour in the group. The same applies to any situation in which an inconsistent frame is applied. So by the lights of such an anthropologist, consistent frames do dominate inconsistent ones.

This leaves the defender of consistency in a somewhat ambiguous position. On the one hand, they now have a reasonable way to maintain their position against the prima facie case for social dialetheias given in §4. Rather than just denying the possibility of such things flat out, they can tell a story about anchoring which explains, on their principles, how it is that dialetheias fail to arise. On the other hand, this story only works given a background commitment to consistency, for only on those principles does it make dialectical sense to appeal to the principle of explosion in the dominance argument. Thus if we were trying to sway an audience that’s unopinionated about the possibility of inconsistency, and which perhaps has been given some reason to doubt consistency in the form of the argument of §3, this story would do no good.

Independently of how it bears on the issue of inconsistency, anchor holism is an interesting refinement of the Epstein model, and may be worth exploring further for its own sake. It is also possible that anchor holism can be paired with motivations other than the ‘interpretationist’ one I’ve sketched here, and that might provide other ways for the dialetheist and consistentist to skirmish. But I will leave the matter here.

5. Ramifications of Social Inconsistency

I have explored a number of avenues of arguing against the possibility of social dialetheias. Of these, anchor holism has the most mileage in it, and I recommend it to the consistentist. But ultimately, even anchor holism does not serve to undercut the prima facie argument for the possibility of social inconsistency given in §3.

And arguably, we should not find social inconsistency all that weird. We do not find it metaphysically puzzling when people’s thoughts, utterances or theories turn out to be inconsistent. It is unfortunate, perhaps, but it is simply what we should expect, given basic human fallibility. If I am right, social inconsistency is just the same kind of basic human fallibility, but writ large upon on the canvas of social reality, in a way made possible by the way that social reality depends on us and our behaviour.

If the social world does permit inconsistency, then there will be a certain amount of constructive theorizing to do in order to deal with the ramifications of this possibility. I will not undertake that constructive work here; I’ll limit myself to noting some of the tasks that I leave for another occasion.

  1. As noted, any form of non-trivial dialetheism requires a paraconsistent logic. There are, of course, well-developed logics of this sort that one can more or less take off the shelf. But there would nevertheless be work to do in figuring out how adopting certain sorts of paraconsistent logic would interact with other formal machinery that a two-dimensional social ontology requires, such as a multi-modal or multi-frame logic for regimenting modal and counterfactual reasoning about worlds and frames. Perhaps certain paraconsistent logics would fit more neatly than others in the overall toolkit of social ontology—this is something to be explored.

  2. Social inconsistencies would arguably have some metanormative ramifications. Social facts often matter to us because they have a normative upshot; they give rise to reasons, obligations, permissions, excuses, and so on. Imagine a state of affairs A such that A entails, for some individual x, that x ought to φ, and ¬A entails that x ought not to φ. In such a case, one might think, an inconsistency to the effect that A and ¬A would give rise to a normative dilemma; x both ought and ought not to φ. Some work would need to be done to explore these issues: would social dialetheism commit one, for instance, to rejecting the principle of ought-implies-can?34

  3. Inconsistencies in the social facts might also have ramifications for the metaphysics of social facts itself. In Epstein’s anchoring-grounding model, it is common for the facts that serve to anchor grounding principles to themselves be social facts. Thus, it seems in principle possible that, when such anchoring facts end up being inconsistent, we end up with inconsistencies regarding what grounding principles are in force, and inconsistencies in what facts are and are not grounded. Some work is needed to explore how this sort of possibility would complicate the view.

6. A Methodological Moral

I take myself, at this point, to have given a basic case for social inconsistency, and to have shown that it survives a certain amount of scrutiny. Let us then assume, for the sake of argument, that the social world might be inconsistent. What consequences would this have for how we should approach social philosophy more widely?

Many phenomena studied by philosophers are suspected of having some aspect of conventionality or social construction about them. Paradigm cases include human kinds like races and genders; social roles; artefacts, artworks and institutions. The more controversial cases include moral and epistemic norms; various kinds of scientific classifications; logical and mathematical facts. Depending on how debate about these phenomena turns out, the category of social facts may end up embracing quite a bit of the reality around us. Thus, the nature of social facts has a bearing on areas of philosophy well beyond social ontology proper.

In philosophy, we tend to apply certain constraints to our theorizing independently of whether the phenomena we deal with are socially constructed or not. Consistency is one of these. If our attempt to give a theory of some phenomenon ends in contradiction, we take this to be a strike against the theory. But where we have no general metaphysical grounds for thinking that the phenomenon we are studying will be consistent, inconsistency should not be an immediate disqualifying feature in our theories. In such cases, an inconsistent theory might be a bad theory, but it also might be a good theory about an inconsistent phenomenon. If I am right, and social inconsistency is a possibility, then this is something that has methodological import for all those areas of philosophy that deal with facts that are or may be social ones, like the ones listed above. Consistency, as a constraint on theorizing, ought to be in abeyance.

7. Conclusion

The social world is a peculiar bit of reality. It is real and objective, but it is shaped by our thought, talk and action. Because of this, the incoherence that we expect to occur in thought and language can be expected to percolate through into social reality, in a way that it doesn’t into other parts of reality. I have argued for the possibility of social inconsistency, and I’ve teased out some of the significance that this possibility would have. All this is somewhat preliminary. I expect there to be further lines of argument against social inconsistency that I haven’t considered here, and perhaps further lines of argument in favour of it. And as noted, much remains to be done in the way of constructive theorizing, if we are to take this possibility seriously.


I would like to thank an audience at Social Ontology 2018 in Boston, the attendees of the work-in-progress seminar at Leeds, and two anonymous referees for questions, comments and suggestions. I’d also like to thank Robbie Williams and Brian Epstein for discussion. The research leading to this article has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement no. 818633).


  1. Discussion of the contrast between causal and constitutive relations in social ontology can be found in Epstein (2016a) and Ásta (2015), although Epstein calls what I here call constitutive determination ‘ontological’ determination.
  2. It may be tempting here to try to contrast the social against the natural, but it may well be that some facts are both social and natural (Khalidi 2015), so this is not a clean contrast.
  3. That this malleability is distinctive of the social should not be taken to mean that it necessarily extends to all social facts. For some social facts, how they come about might not be up to us at all. Take for instance the fact that Alice and Bob are jointly attending to a jigsaw puzzle. This seems to be a social fact, but there may not be anything malleable about what it takes for Alice and Bob to be jointly attending—it could be entirely a matter of how human beings are wired, as a matter of neuropsychology.
  4. Searle (1995) characterizes this as ‘epistemic objectivity’, to be contrasted with ‘ontological objectivity’, which he does not think social facts enjoy; see also Thomasson (2003a). This use of ‘objectivity’ should not be confused with another sense of the term which may be salient in relation to social facts. Since Weber (1904) there has been a long-running debate about whether social-scientific enquiry could be objective as opposed to value-laden (see Reiss & Sprenger 2020 for an overview). That debate is more or less orthogonal to the issues discussed here.
  5. See Udehn (2001: chs. 3–4) for an overview and discussion of ‘psychologism’ about social facts in early social science. A later approach with comparable metaphysical implications would be the global social constructivism of Berger and Luckmann (1966).
  6. A further issue is whether claiming that the social world is mind-dependent would commit us to some form of anti-realism about social facts. I will postpone discussion of this to Section 4.2, where it becomes dialectically important.
  7. See Kocurek, Jerzak, and Rudolph (2020) for a recent elaboration and application of Einheuser’s approach.
  8. In brief: grounding is a cross-level relation between facts that underpins a distinctive kind of metaphysical explanation, one that is constitutive rather than causal. As a matter of definition, when some fact A grounds some fact B, A is more fundamental than B. Formally the grounding relation is typically thought to be asymmetric, irreflexive and transitive, though there is debate on this (cf. Rodriguez-Pereyra 2015). Some regard the grounding relation as an independent theoretical posit, while others take grounding talk to indifferently refer to various cross-level metaphysical relations such as composition, constitution and realisation (cf. Wilson 2014). These debates won’t bear on the argument here. Griffith (2017) specifically discusses the application of grounding theory in social ontology.
  9. It has been questioned (see Schaffer 2019; and Epstein 2019a in reply) whether anchoring and grounding are really different relations, and whether the anchors of some social fact are really just more grounds of that social fact. I do not think this issue affects the argument of this paper. On Schaffer’s view, the facts that put in place the grounding relations relevant to a social fact become additional grounds of that fact, namely “rule-setting” grounds—this saves him the need to introduce an additional type of relation. A view like Schaffer’s can mimic the behaviour of a two-dimensional theory up to a point, by partitioning the grounds of a given social fact into ‘rule-setting’ and ‘move-making’ grounds and separately consider varying the former and varying the latter. Epstein (2019a) argues that, among other things, this view cannot adequately model counterfactual claims about social matters. While the matter is a delicate one, it seems to me that Schaffer’s approach would struggle particularly with counterfactuals concerning situations in which some facts play both rule-setting and move-making roles, which would require him to simultaneously vary and hold fixed the same grounding facts. But the inconsistency-generating scenarios which we will consider here are not such situations, and more generally do not involve counterfactuals, so there is no obvious obstacle to recasting them in Schaffer’s model. (It is also worth noting that identifying anchoring as a grounding relation is not ipso facto incompatible with two-dimensionalism. One could recover two-dimensionality from Schaffer’s view by allowing counterfactual (im)possibilities into the model in which the grounding principles are varied, but the facts that would counterfactually ground those grounding principles are not accordingly adjusted. I suspect that Schaffer would not like that move, but it might be attractive to some.) See also Hawley (2019) and Mikkola (2019) for other critical takes on the grounding/anchoring distinction, and Epstein (2019b) in reply.
  10. Epstein’s anchoring-grounding model is iterative: the facts that ground social facts are often themselves social facts, and thus subject to a similar analysis. The facts that anchor grounding principles are also often social facts, so the same applies to them. In addition to this, Epstein thinks that anchoring relations (the relations between anchors and the grounding principles they anchor) are contingent on the anchoring schemata that are in force, which are themselves often contingent upon anchoring facts. So Epstein’s model can give rise to a complex array of determining relations. But social reality is very complex, so as long as these relations pull their explanatory weight, this is not something to hold against the model.
  11. There is room to manoeuvre with regard to the underlying model theory. Einheuser (2003; 2006) favours a theory modelled on Kaplan’s (1989) treatment of indexicals, one which evaluates sentences at a ‘carving’ and ‘substrate’ rather than at a context and world. Epstein (in conversation) favours a multi-frame model which treats frames as spaces of worlds, and evaluates sentences at a world in a frame. For current purposes the two approaches are interchangeable.
  12. I’m embroidering on Epstein (2014; 2015; 2016b) here; he is not necessarily committed to countenancing all these mechanisms.
  13. A qualification is in order. Some logics, notably subvaluational ones (Jaśkowski 1969; Hyde 1997) are set up precisely to block the derivation of an outright contradiction from an inconsistent set of sentences (by, e.g., rejecting adjunction) and secure paraconsistency by this route. So in these logics, one cannot strictly characterize an inconsistency as a set of sentences that entails a contradiction; rather, one would have to characterize them as sets of sentences that would entail contradictions in the presence of, e.g., adjunction.
  14. Prominent dialetheic treatments of the semantic paradoxes are Priest (1987/2006) and Beall (2009). Whereas Priest considers the semantic paradoxes to be one among a number of dialetheic phenomena, Beall argues that the possibility of real inconsistency is specific to the semantic realm.
  15. The case of legal contradictions is discussed in Priest (1987/2006: ch. 13), with some follow-up in Beall (2017) and Priest (2017). Just recently Bolton and Cull (2020) have argued for the possibility of social dialetheias by constructing, refining and defending a thought experiment involving a club of dialetheists adopting an inconsistent set of rules to govern their club. Cameron (2022) also sketches a metaphysics of social reality which would allow for inconsistency.
  16. If a frame is taken to be a set of conditionals, it’s hard for it to be inconsistent in the absence of anything that renders the antecedents of these conditionals true or false. It could happen if f includes some conditionals Γ → A, Γ → ¬A where Γ is equivalent to (a) some logical truth or (b) some other conditional included in f.
  17. Along the same lines, one could also make a case for social indeterminacy: for some logically complete pre-social world w1, a frame could be such that, for some Γ, it yields a social world w1+ such that, for some state of affairs A, it is indeterminate whether the fact that A or the fact that ¬A obtains. I do not explore this possibility here for reasons of space, but it is equally worthy of attention.
  18. Of course various construals of this case are available. First, Beall (2017), in response to this case, argues that legal claims are implicitly always within the scope of an ‘according to legal system L, p’ operator. If so the contradiction here would only be apparent, since ‘according to L, p,’ and ‘according to L, ¬p’ are not formal contradictories. Beall argues that there is no way, in these cases, to ‘detach’ p and ¬p from these operators, so no contradiction can be derived. I do not think that Beall is right, but if he is, it would make the law a bad source of examples (cf. Bolton & Cull 2020). Second, most legal systems have meta-principles that are designed to filter out conflicts between laws. The widely implemented lex posterior principle says that where laws conflict, the later law supersedes the earlier law, unless the earlier law has constitutional status. To make the example work in spite of this, we can stipulate that there is no lex posterior or similar principle in force in the situation imagined, or that the laws are promulgated simultaneously.
  19. The story is related by Aulus Gellius (177/1927: bk. V, ch. 10)
  20. By similar reasoning, Euathlus also does and does not owe the fee if he loses. But that is inessential: the case is meant here to illustrate that a contradiction in the social facts (here, concerning what is legally owed) is possible, not that it is inevitable.
  21. I’d like to thank Don Baxter for bringing this example to my attention. The 2010 edition of The Official Rules of Major League Baseball fixed the rule (Christensen 2016).
  22. See List (2014) for details on this k-measure, which he there uses as a measure of the ‘degree of inconsistency’ of a body of information.
  23. Although this line of reasoning is a familiar bit of philosophical folklore in social ontology, it is nevertheless not easy to find authors onto whom this inference can be firmly pinned. Mason (2020) cites Hayek (1943), Searle (2007) and Thomasson (2003b), but of those only Thomasson is an unambiguous example (and her anti-realist conclusions do not extend to the whole of social reality). But see also Khalidi (2016) for some examples of the inference from mind-dependence to anti-realism in general metaphysics.
  24. Mason (2016) discusses a number of different ways of qualifying social facts as not real.
  25. Or ‘if A, then in the fiction A grounds B’, if one wanted to build in fictional grounding structure.
  26. For this to work, the rules of the ‘in the fiction …’ operator should permit us to derive ‘in the fiction, A and ¬A’ from ‘in the fiction, A’ and ‘in the fiction, ¬A’. But this seems like a plausible way for such an operator to behave, assuming it’s the same fiction in both cases.
  27. See Eklund (2019) for an overview of fictionalist proposals. I have not been able to turn up examples of fictionalist approaches to social facts generally, but Logue (2022) argues for fictionalism about gender.
  28. The example is from Walton (1990: 37).
  29. Priest (1998) gives an overview of such general misgivings.
  30. Might there be something special about fictions, such that moving to fictionalism makes available ways of dealing with inconsistency that we wouldn’t otherwise have? There is a significant literature on inconsistent fictions which we might consult for inspiration, in which some argue for the possibility of inconsistency in fiction (e.g., Currie 1990; Priest 1997; Wildman & Folde 2017) and some against (e.g., Hanley 2004; Xhignesse 2016; 2020). Since this literature is mainly concerned with literary fictions, the upshot of these debates for fictionalism is not necessarily obvious, but the fictionalist could nevertheless find resources there. Besides the already-mentioned approach of moving to a non-explosive logic, another interesting approach to dealing with fictional inconsistency is to move to a subtler account of truth-in-fiction: following Xhignesse (2016; 2020), for instance, a social fictionalist might propose that, when inconsistencies appear to arise, they count as (in some sense) claimed in the social fiction but nevertheless not true according to it. On such a strategy, the essential move is to not straightforwardly identify what’s true in a fiction with all the things prima facie laid down for it, but instead identify it with some more well-behaved function of the latter. I think there is some mileage in such an approach. But here too, I do not think fictionalism is really an essential step in implementing it: this is the structure of the ‘holist’ strategy that I will explore in Section 4.4, which I will ultimately recommend to the consistentist as their best option.
  31. Error theory is little explored in social ontology. One approach which might be classed as error-theoretic is the ethnomethodological approach in anthropology/sociology (Garfinkel 1967). It holds that ethnographers are not warranted in treating their subjects’ utterances about social matters as evidence about social structures that the people in the community under investigation inhabit. Instead, they should merely take these utterances as evidence about how those people try to make some sense (successfully or unsuccessfully) of the social interactions that they engage in. The subjects may be optimistically assuming the existence of some meaning-giving social order in order to make it through the day, but whether anything coherent or interpersonally stable could be reconstructed out of their ideas of it is doubtful, and moot. Whether this approach is best interpreted as an error theory is, of course, debatable—one might also frame it as a kind of methodological scepticism or quietism about social structures. See Collin (1997: ch. 1) for a philosophical assessment of ethnomethodology.
  32. A good explanation of this sort needn’t make the behaviour perfectly rational, just as a charitable interpretation of a person’s utterances needn’t make them perfectly correct—it just needs to make them reasonable to the degree that we expect human beings to be.
  33. This way of motivating anchor holism is inspired by metasemantic ‘interpretationism’ in the style of Lewis (1974). Like Lewis, I do not take interpretationism to entail any form of anti-realism about the category of facts in question.
  34. Priest (1987/2006: ch. 13) discusses what becomes of deontic notions in a dialetheic setting. One option is to deny ought-implies-can; this is an option that some metaethicists have independently argued for (e.g., Sinnott-Armstrong 1984). A dialetheist would also have the option of maintaining ought-implies-can and counting violations of it as just more dialetheias.


1 Appiah, K. Anthony (1995). The Uncompleted Argument: DuBois and the Illusion of Race. In Linda A. Bell and David Blumenfeld (Eds.), Overcoming Racism and Sexism (59–78). Rowman & Littlefield.

2 Ásta (2015). Social Construction (published under “Ásta Sveinsdóttir”). Philosophy Compass, 10(12), 884–92.

3 Aulus Gellius (1927). Attic Nights, Volume I. John C. Rolfe (Ed. and Trans.). Loeb Classical Library 195. Harvard University Press. (Original work published 177)

4 Beall, Jc (2009). Spandrels of Truth. Clarendon Press.

5 Beall, Jc (2017). Do Inconsistent Laws Deliver Gluts? In Patrick H. Glenn and Lionel D. Smith (Eds.), Law and the New Logics (199–207). Cambridge University Press.

6 Berger, Peter L. and Thomas Luckmann (1966). The Social Construction of Reality. Doubleday.

7 Bolton, Emma and Matthew J. Cull (2020). Contradiction Club: Dialetheism and the Social World. Journal of Social Ontology, 5(2), 169–80.

8 Cameron, Ross P. (2022). Chains of Being: Infinite Regress, Circularity, and Metaphysical Explanation. Oxford University Press.

9 Christensen, Chris (2016). The Philosophy Professor and the Holy Book of Baseball. Philosophy Now, 115, 28–29.

10 Cohen, Ted (1990). There Are No Ties at First Base. Yale Review, 79(2), 314–22. Reprinted in Eric Bronson (Ed.), Baseball and Philosophy (73–86). Open Court Books. 2004.

11 Collin, Finn (1997). Social Reality. Routledge.

12 Currie, Gregory (1990). The Nature of Fiction. Cambridge University Press.

13 Einheuser, Iris (2003). Conventionalism. Doctoral dissertation, Massachusetts Institute of Technology.

14 Einheuser, Iris (2006). Counterconventional Conditionals. Journal of Philosophy, 127(3), 459–82.

15 Eklund, Matti (2019). Fictionalism. In Edward N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Winter 2019 ed.). Retrieved from

16 Epstein, Brian (2014). How Many Kinds of Glue Hold the Social World Together? In Mattia Gallotti and John Michael (Eds.), Perspectives on Social Ontology and Social Cognition (41–56). Springer.

17 Epstein, Brian (2015). The Ant Trap: Rebuilding the Foundations of the Social Sciences. Oxford University Press.

18 Epstein, Brian (2016a). A Framework for Social Ontology. Philosophy of the Social Sciences, 46(2), 147–67.

19 Epstein, Brian (2016b). Replies to Guala and Galotti. Journal of Social Ontology, 2(1), 159–72.

20 Epstein, Brian (2019a). Anchoring versus Grounding: Reply to Schaffer. Philosophy and Phenomenological Research, 99(3), 768–81.

21 Epstein, Brian (2019b). Replies to Hawley, Mikkola, and Hindriks. Inquiry, 62(2), 217–29.

22 Field, Hartry (1980). Science Without Numbers. Princeton University Press.

23 Fine, Kit (2001). The Question of Realism. Philosophers’ Imprint, 1(1), 1–30.

24 Garfinkel, Harold (1967). Studies in Ethnomethodology. Prentice-Hall.

25 Griffith, Aaron M. (2017). Social Construction: Big-G Grounding, Small-g Realization. Philosophical Studies, 175, 241–60.

26 Guala, Francesco (2014). On the Nature of Social Kinds. In Mattia Gallotti and John Michael (Eds.), Perspectives on Social Ontology and Social Cognition (57–68). Springer.

27 Hanley, Richard (2004). As Good as It Gets: Lewis on Truth in Fiction. Australasian Journal of Philosophy, 82(1), 112–28.

28 Hawley, Katherine (2019). Comments on Brian Epstein’s the Ant Trap. Inquiry, 62(2), 217–29.

29 Hayek, Friedrich (1943). The Facts of the Social Sciences. Ethics, 54(1), 1–13.

30 Hindriks, Frank (2009). Constitutive Rules, Language, and Ontology. Erkenntnis, 71(2), 253–75.

31 Hyde, Dominic (1997). From Heaps and Gaps to Heaps of Gluts. Mind, 106(424), 641–60.

32 Jaśkowski, Stanisław (1969). Propositional Calculus for Contradictory Deductive Systems. Studia Logica, 24, 144–57.

33 Joyce, James M. (1998). A Nonpragmatic Vindication of Probabilism. Philosophy of Science, 65(4), 575–603.

34 Kaplan, David (1989). Demonstratives. In Joseph Almog, John Perry, and Howard Wettstein (Eds.), Themes from Kaplan (481–563). Oxford University Press.

35 Khalidi, Muhammad A. (2015). Three Kinds of Social Kinds. Philosophy and Phenomenological Research, 90(1), 96–112.

36 Khalidi, Muhammad A. (2016). Mind-Dependent Kinds. Journal of Social Ontology, 2(2), 223–46.

37 Kocurek, Alexander W., Ethan Jerzak, and Rachel Etta Rudolph (2020). Against Conventional Wisdom. Philosophers’ Imprint, 20(22), 1–27.

38 Lewis, David (1974). Radical Interpretation. Synthese, 27(3/4), 331–44.

39 List, Christian (2014). When to Defer to Supermajority Testimony – and When Not. In Jennifer Lackey (Ed.), Essays in Collective Epistemology (240–49). Oxford University Press.

40 Logue, Heather A. (2022). “Gender Fictionalism”. Ergo 8: 28. DOI:

41 Mason, Rebecca (2016). The Metaphysics of Social Kinds. Philosophy Compass, 11(12), 841–50.

42 Mason, Rebecca (2020). Against Social Kind Anti-Realism. Metaphysics, 3(1), 55–67.

43 Mikkola, Mari (2019). Grounding and Anchoring: On the Structure of Brian Epstein’s Social Ontology. Inquiry, 62(2), 198–216.

44 Priest, Graham (1987/2006). In Contradiction. 2nd ed. Oxford University Press.

45 Priest, Graham (1997). Sylvan’s Box: A Short Story and Ten Morals. Notre Dame Journal of Formal Logic, 38(4), 573–82.

46 Priest, Graham (1998). What Is So Bad about Contradictions? Journal of Philosophy, 95(8), 410–26.

49 Priest, Graham (2017). Where Laws Conflict: An Application of the Method of Chunk and Permeate. In Patrick H. Glenn and Lionel D. Smith (Eds.), Law and the New Logics (168–80). Cambridge University Press.

50 Priest, Graham, Jc Beall, and Bradley Armour-Garb (Eds.) (2004). The Law of Non-Contradiction: New Philosophical Essays. Oxford University Press.

51 Reiss, Julian and Jan Sprenger (2020). Scientific Objectivity. In Edward N. Zalta (Ed.), The Stanford Encyclopedia of Philosophy (Winter 2020 ed.). Retrieved from

52 Rodriguez-Pereyra, Gonzalo (2015). Grounding is Not a Strict Order. Journal of the American Philosophical Association, 1(3), 517–34.

53 Schaffer, Jonathan (2019). Anchoring as Grounding: On Epstein’s The Ant Trap. Philosophy and Phenomenological Research, 99(3), 749–67.

54 Searle, John R. (1995). The Construction of Social Reality. The Free Press.

55 Searle, John R. (2007). Social Ontology and the Philosophy of Society. In Eric Margolis and Stephen Laurence (Eds.), Creations of the Mind (3–17). Oxford University Press.

56 Searle, John R. (2010). Making the Social World: The Structure of Human Civilisation. Oxford University Press.

57 Sinnott-Armstrong (1984). ‘Ought’ Conversationally Implies ‘Can’. The Philosophical Review, 93(2), 249–61.

58 Thomasson, Amy L. (2003a). Foundations for a Social Ontology. ProtoSociology, 18–19, 269–90.

59 Thomasson, Amy L. (2003b). Realism and Human Kinds. Philosophy and Phenomenological Research, 67(3), 580–609.

60 Torregno, Giuliano (2017). Institutional Externalism. Philosophy of the Social Sciences, 47(1), 67–85.

61 Tuomela, Raimo (2013). Social Ontology: Collective Intentionality and Group Agents. Oxford University Press.

62 Udehn, Lars (2001). Methodological Individualism: Background, History and Meaning. Routledge.

63 Walton, Kendall (1990). Mimesis as Make-Believe. Harvard University Press.

64 Weber, Max (1904). Die ‘Objektivität’ Sozialwissenschaftlicher und Sozialpolitischer Erkenntnis. Archiv für Sozialwissenschaft und Sozialpolitik, 19(1), 22–87.

65 Wildman, Nathan and Christian Folde (2017). Fiction Unlimited. The Journal of Aesthetics and Art Criticism, 75(1), 73–80.

66 Wilson, Jessica M. (2014). No Work for a Theory of Grounding. Inquiry, 57(5–6), 535–79.

67 Xhignesse, Michel-Antoine (2016). The Trouble with Poetic Licence. British Journal of Aesthetics, 56(2), 149–61.

68 Xhignesse, Michel-Antoine (2020). Exploding Stories and the Limits of Fiction. Philosophical Studies, 178(3), 675–92.