In this paper, I argue against what I call “natural realism” about social kinds, the view according to which social categories have natural boundaries, independent of our thought. First, I draw a distinction between two different types of entity realism, one being about the existence of the entity, “ontological realism”, and the other one being about the direct mind-independence of the entity, “natural realism”. After endorsing ontological realism, I present the natural realist argument according to which there would be certain social kinds, such as economic recessions and racism, even if we had no clue about their existence and nature. I claim that the argument fails insofar it mistakes the single instances for the kinds themselves. I then argue against natural realism by showing how the vagueness characterizing the boundaries of social kinds puts the realist in front of a dilemma: either she accepts the vagueness of these boundaries as ontic, a metaphysically problematic thesis, or she accepts that most or all social kinds remain unknown, an epistemologically problematic thesis. Thus, I argue that antirealism, according to which social kinds are constructed, is a better alternative to the realist account.
In recent years, social ontology witnessed a new trend emerging, according to which many - if not all - social kinds are not the product of our mind. This realist turn is revolutionary insofar as it takes a break from a dominant tradition that considers social kinds as the constructed categories par excellence. What I set out to argue is that, while this view has the merit of showing that the nature of social phenomena does not solely depend on our mental states, this type of realism does not hold for social kinds. Here is how the work is structured.
1. I first draw a distinction between two different types of realism, one being about the existence of the entity, “ontological realism”, and the other one being about the direct mind-independence of the entity, “natural realism”. I endorse the former type of realism and show how most social ontologists are realist in this sense too.
2. I then present natural realism, according to which social kinds may have ‘natural boundaries’, independent of what we think of them, and highlight the novelty of the view, which is in contrast with the more orthodox antirealist and constructivist stance in social ontology.
3. I proceed by raising an objection to the common realist argument according to which there would be certain social kinds, such as economic recession and racism, even if we had no clue about their existence and nature. I argue that this argument fails insofar it takes the single instances for the kinds themselves, since there is a difference between the way in which we draw the contours of social objects – the kind - and the social objects themselves – the instances of the kind.
4. Finally, I give an argument against natural realism that is based on the vague boundaries that characterize social kinds, and which puts the realist in front of a dilemma: either she accepts the vagueness of these boundaries as ontic, a metaphysically problematic thesis, or she accepts that most or all the social kinds remain unknown, an epistemologically problematic thesis. Thus, I argue that antirealism, according to which social kinds are constructed, is a better alternative to the realist account.
Among the numerous polysemic words belonging to the philosophical jargon, ‘realism’ (in)famously stands out as some of the trickiest.1 Rather than engaging in the quixotic endeavor of elucidating the disparate uses of the term, I set out to explain the only two senses in which realism about social kinds is at issue here. Showing what these views amount to will consequently clarify what their respective antirealist stances stand for, given that they simply consist in the negation of the realist theses. My goal is to defend one of these kinds of realism, that I call ‘ontological realism’, and reject the other one, that I call ‘natural realism’.
Let us begin then with ontological realism. If you are realist about something in this sense, you are committed to the existence of that thing. This is, for example, how the term is employed when it comes to the debate on universals: a realist about universals believes in their existence. Clearly, being a realist about a certain entity does not entail to be a realist about another one: for instance, one can be a realist about mathematical objects, but not about persons. Rather than ‘antirealism’, the view denying realism about a certain entity is usually called ‘eliminativism’ or ‘nominalism’.
If we define ontological realism about social kinds this way, that there are not many proponents of
this view, at least among contemporary authors, but there are some prominent eliminativists about
specific social kinds: for example, Kwame Anthony Appiah ([
Here’s a couple of arguments in favor of ontological realism. One resides in the prominent role they play in both ordinary and scientific life, which gives at least prima facie reason to accept their existence. The other argument is that, assuming we construct social kinds, something being created is good evidence that that thing exists (or existed): knowing that someone built a house is good evidence that that house exists or existed at some point! Clearly, there is a huge difference between buildings and social kinds, but what we are interested in it is not the nature of the thing, but its existence.
Now, two objections can be raised at this point. Let us start with the one according to which it is false that we are committed to the existence of everything we create: by writing a novel, we do not thereby bring into existence the protagonists of our story. There are two replies to be given to this point. First, there are respectable realist views about fictional entities, so it cannot be taken for granted that such things do not exist.3 However, I am not here to argue for the existence of Anna Karenina, so let us grant that fictional entities do not exist: there is still something we are bringing into existence when we write novels, whether this is a string of words telling the story or a representation of the characters that the readers form in their mind. Someone who is a constructivist about social kinds may be therefore a realist too, and she may argue that what we bring into existence when we create social kinds are certain concepts.
Another objection one can raise is that, while social kinds exist, they do in a peculiar manner, given
that there are different ways in which something may exist: it seems obvious that social kinds exist in
a different way than buildings do, and that this would impugn their reality. Social kinds do exist, but in
a different sense than, say, physical ordinary objects do. Thus, it makes sense to be antirealist, although
in this weaker sense, about social kinds. This objection hinges on the meta-ontological principle that
there are various meanings of ‘existence’, a view that Morton White [
Ryle famously argues in favor of it by claiming that it would be ridiculous to say in the same sentence that “there exist prime numbers and Wednesdays and public opinions and navies”.4
On a closer look, Ryle’s objection is weaker than it seems. The fact that certain sentences asserting
the existence of wildly different kinds of entities exist sound absurd do not entail that they are also
meaningless, but only that they are unusual for our linguistic conventions. And, as Peter van Inwagen
[
Antirealism, or constructivism, holds that social kinds are constructed: John Searle has been the
most prominent supporter of this view in the analytic tradition of social ontology.6 According to Searle
[
However, in recent years, several philosophers disagreed with this longstanding tenet of social ontology theories. Amie Thomasson is the pioneer of this new approach and argues that while intentionality plays a role in the genesis of the social world, it does not always do so in the direct way Searle describes:
The possibility is often overlooked, however, that there may be entities (and kinds of entities) depending
on mental states of various kinds, without their depending on any beliefs about them (or about that kind
itself). Some social kinds such as racism, superstition, etc., do depend on the existence of certain sets of
4 “It is perfectly proper to say, in one logical tone of voice, that there exist minds and to say, in another logical tone of voice, that there exist
bodies. But these expressions do not indicate two different species of existence, for ‘existence’ is not a generic word like ‘coloured’ or ‘sexed’.
They indicate two different senses of ‘exist’, somewhat as ‘rising’ has different senses in ‘the tide is rising’, ‘hopes are rising’, and ‘the average
age of death is rising’. A man would be thought to be making a poor joke who said that three things are now rising, namely the tide, hopes
and the average age of death. It would be just as good or bad a joke to say that there exist prime numbers and Wednesdays and public opinions
and navies; or that there exist both minds and bodies.” [
beliefs and intentional behaviors, but may exist without the existence of any beliefs that are themselves about racism, superstition, etc. Large scale social facts or those involving statistical generalizations over social facts (e.g., the fact that we are in a recession, or that 70% of Americans support the death penalty) similarly depend on certain mental states (regarding money and financial transactions, regarding the death penalty), but not on any intentional states regarding recessions or statistical generalizations. [18] This account of social kinds has two significant consequences, one epistemological and one metaphysical.9 From an epistemological point of view, it grants that social sciences are capable of genuine discoveries about the world. If all social kinds and, more generally, entities were self-referential as Searle argues, then they would be epistemically transparent, and the only kind of discovery possible would be for people outside a certain social group, but never within that social group (e.g. ethnographers finding out what are norms of a specific tribe they are not member of). Unlike Searle, Thomasson is therefore able to “explain the possibility of unknown social kinds awaiting discovery by the social sciences” [19].10
From a metaphysical point of view, it means that, as Thomasson puts it [18], certain social kinds have “natural boundaries”, meaning that the kind “is not merely a division artificially imposed on the world by human concepts”. According to Thomasson [18], this is the case because a kind has natural boundaries if and only if it meets both what she calls the ‘Ignorance Principle”, according to which it is possible that the boundaries remain unknown to anyone, and what she calls the “Error Principle”, according to which any belief about these boundaries may turn out to be massively wrong.11 I call this view about social kinds “natural realism”.
Jonathan Friedman [21] subsequently raised a similar objection and Muhammad Ali Khalidi [22] built on Thomasson’s argument for his tripartition of social kinds, borrowing her examples of economic recession and racism as social kinds that do not depend on our direct intentional states. Francesco Guala [23] takes a step further by arguing that no social kind constitutively depends on our propositional attitude towards them: “dependence on collective propositional attitudes directed toward the kind itself is neither necessary nor sufficient for an institutional kind to exist”. Guala uses the expression “institutional kinds” because he is following Searle’s terminology here, and he is referring to those kinds that are apparently directly dependent on collective attitude, such as football fan or owner of government bond, and he is already assuming that kinds like inflation and unemployment do not depend on our attitude toward them. Thus, Guala’s thesis applies to all sorts of social kinds. Collectively setting up conditions for a kind may play the role of helping to coordinate the actions of social actors, but what ultimately constitutes the social kind is a certain set of behavior and beliefs that are not directed toward the kind itself.
The realist move relies on the following observation: many – if not all - social kinds are what they are regardless of what we think of them. Business cycles bear causal properties regardless of our beliefs about them. Racism bears causal properties regardless of our beliefs about them. And for Guala, even money is what it is regardless of what we think money is. Of course, our beliefs about social kinds might change society, since when we are equipped with economical and sociological expertise and 9 As the quote shows, the account covers also social entities more generally, but of course I will focus on what she says about social kinds. 10 Guala [20] argues that Thomasson is mistakenly taking Searle’s view and, more generally, collective acceptance views of social kinds as implying “infallibilism”, since Searle actually points out that people can be wrong about the conditions defining the nature of a kind (for example, that something is money only if it is “backed by gold”). While Thomasson may have downplayed Searle’s epistemic caution, her point that the collective acceptance view ignores certain kinds of social kinds still holds. 11 This view that certain social kinds have natural boundaries seems at times at clash with some passages from her other paper from 2003, where she writes: “Some may be tempted to the view that, as social phenomena, such things don’t really exist until we have the concepts for them. But the point here is precisely that, while some social concepts (such as money) require intentional states involving that concept in order for things of that kind to exist, others (such as recessions) do not. The idea that recessions and racism do not exist until we have concepts for them, on reflection, has no more plausibility than the idea that electrons did not exist until we developed the concept for them. Something falling under the concept of “electron” could (and did) exist long before scientists discovered them, and beliefs and practices that fall under the concept of racism could (and did) exist long before the concept or word was known” [19]. See also [18], where she talks about economists stipulating the conditions for what counts as a recession. However, recall that according to her view social kinds are not concepts, so the consistency of the view is not threatened. knowledge, governments and citizens are able to take steps to improve the economy or help stopping racial discrimination. However, while these beliefs about the kinds may help us change the social world, our intentionality would still not be what constitutively makes up the social kinds.
Although I find merit in this view, to the extent that it sheds light on the limits of an overly intellectualistic theory of the social world, I argue that the realist argument fails insofar as it falls victim of conflating the instances with the kinds they belong to: there is in fact a difference between on the one hand the objects that populate the world and on the other hand the way these objects are grouped together. This difference does not hinge on any peculiar metaphysical characterization of kindhood, but on a minimal account of what kinds are, namely partitions of entities according to one or more properties that their instances share. This partition can be “natural”, as realists claim, or “artificial”, as antirealists claim, but it nevertheless remains a division of entities according to certain features that characterize them.
If we look at Thomasson’s examples [19], they are not social kinds, but social objects: “[…] a given economic state can be a recession even if no one thinks of it, and even if no one regards anything as a recession or any conditions as sufficient for counting as recession. […] something or someone can be racist without anyone regarding anything as racist – racism clearly existed long before anyone took any activity or pattern of behavior to be racist”. Arguably, economic recessions are here seen as events, and something being racist is seen as either an action or a disposition, but none of them is a social kind.
Guala’s argument [20] relies on the same classificatory mistake, as the way he describes what characterizes money suggests that he is referring to social objects, not to kinds: “the kind money ultimately is nothing but the set of actions, and the related set of expectations”. Just as Thomasson, Guala is actually referring to the instances of the kind rather the kind money itself. My objection only shows that the realist argument fails, but it is not enough to show that natural realism is false, since there might be other reasons why it is true. Thus, let us move on to the next section, where I provide an argument in favor of antirealism.
Let us assume that social kinds are real, and so that they do have “natural boundaries” that do not directly depend on any mental state in order to exist. Now, let us take into account the social kinds we have seen so far: economic recession, racism, and money. Arguably, all these kinds are marked by boundaries that are vague, given that there is no sharp demarcation between what is an instance of that kind and what is not: while there are some clear-cut cases of economic recessions, racism, and money, for others it is unclear whether they count as such or not. A glance at the news suffices to find plenty of examples of how these kinds have vague boundaries, witnessed by debates about whether the current decline in economic activity should be regarded as a recession or not,12 whether cryptocurrency is money or not,13 or whether certain symbols are racist or not.14
Why is vagueness an issue for the realist? The reason is that, since for the realist the kinds are not intentionally constructed, the vagueness of the boundary must be ontic. However, the problem is that it is highly controversial that there is such thing as de re vagueness, as Gareth Evans’s seminal paper [24] on the topic shows. The natural realist has therefore this extra metaphysical burden to carry that the antirealist does not have because she treats the vagueness of the boundaries as uncontroversial de dicto vagueness, so in conceptual or semantic terms.
Still, the realist may bite the bullet and accept that there are such vague boundaries, but that they belong to our concepts, and not to the real social kinds: if we deal with vague boundaries in the social sciences, it means that social kinds, with their natural boundaries, are still awaiting discovery. This way, the realist does not have to worry about entering the metaphysical minefield of ontic vagueness. However, the problem with this way out is that there is a high epistemological price to pay, given that 12 https://www.afr.com/policy/economy/are-we-still-in-a-recession-or-not-economists-disagree-20201202-p56k07. 13 https://www.nytimes.com/2021/04/26/business/economy/fed-digital-currency.html. 14 https://www.theguardian.com/cities/2018/sep/26/statue-wars-what-should-we-do-with-troublesome-monuments. it implies that most kinds in the social sciences are not real kinds, and this runs against the initial motivations of a realist account.15
The existence of vague boundaries puts the realist in front of a dilemma. Either she accepts ontic vagueness, incurring in a metaphysical cost, or she accepts that no kind is investigated by the social sciences, incurring in an epistemological cost. Thus, I argue that the realist account should be abandoned in favor of the antirealist one, according to which social kinds are constructed.16 Notice that the fact that we come up with social kinds does not entail that anything goes, since there are empirical constraints on how we represent the social world, and which make some categories better tools than others.17 For example, defining an economic recession as a period of time during which the national GDP decreases for only a couple of weeks would not make any explanatory or inductive sense. Of course, the antirealist account needs to be further metaphysically polished, as it must address how we exactly construct social categories through intentionality, but its core thesis remains the best account of social kinds.
In this paper, I argued that, while ontological realism about social kinds is true, natural realism is false. The natural realist argument fails because it mistakes the instances for the kinds, and the existence of vague boundaries gives us reasons to believe that there is no social kind with natural boundaries. Thus, I suggested that, although it needs metaphysical refinement, the antirealist account is a better alternative to realism.
15 Interestingly, the only social kinds with no vague boundary are those to be found in the legal domain, but a realist would not accept them as having natural boundaries because they are fiat boundaries. 16 Of course, not only social kinds are vague, but other kinds, such as for example biological kinds. I believe my argument holds against analogous forms of natural realism concerning these kinds, but I am here concerned only with social kinds. 17 And some claim that there are normative constraints too (see for example the ‘ameliorative project’ of social kinds put forward by Haslanger [15]). [15] Haslanger, Sally (2012), Resisting Reality: Social Construction and Social Critique, Oxford:
Oxford University Press. [16] Ásta (2018), Categories We Live By: The Construction of Sex, Gender, Race, and other Social
Categories, New York: Oxford University Press. [17] Mallon, Ron (2016), The Construction of Human Kinds, New York: Oxford University Press. [18] Thomasson, Amie (2003a), “Realism and Human Kinds”, Philosophy and Phenomenological
Research 67 (3): 580-609. [19] Thomasson, Amie (2003n), “Foundations for a Social Ontology”, ProtoSociology: An
International Journal of Interdisciplinary Research 18 - 19: 269 - 290. [20] Guala, Francesco (2010), “Infallibilism and Human Kinds”, Philosophy of the Social Sciences 40 (2): 244-264. [21] Friedman, Jonathan (2006), “Comment on Searle’s ‘Social Ontology’: The Reality of the
Imaginary and the Cunning of the Non-Intentional,” Anthropological Theory 6 (1): 70 - 80. [22] Khalidi, Muhammad Ali (2015), “Three Kinds of Social Kinds,” Philosophy and
Phenomenological Research 90 (1): 96-112. [23] Guala, Francesco (2014), “On the Nature of Social Kinds,” in Mattia Gallotti and John Michaels (eds.), Perspectives on Social Ontology and Social Cognition, Dordrecht: Springer Netherlands, pp. 57-68. [24] Evans, Gareth (1978), “Can There Be Vague Objects?”, Analysis 38: 208.