Demarcating the Social World with Hume

Abstract

Where lies the boundary between the natural and social worlds? For the local constructionist, who wants to say that whilst global constructionism is false, nonetheless there remains a domain of socially constructed phenomena, there is going to be a demarcation question. In this paper I explore two initially plausible accounts of the boundary, based on mind-dependence and constructive mechanisms, and show that each is bound to fail. After further rejecting an explanatory account drawn from the work of Ásta, I look at Hume’s account of the artificial to develop a distinctly Humean account of the boundary, improving it with a necessity condition to deal with potentially pernicious counter-examples, and suggesting that it provides our best answer to the local constructionist’s demarcation question.

Introduction

For the moderate social constructionist, who denies the excesses of global constructionism, and endorses local constructionism about certain domains, such as gender, race, money, schools, etc., it looks as if there is going to be a serious question about how to demarcate those kinds that are socially constructed, i.e., what is a part of the social world, from those that are a part of the natural world.¹ In this paper I will develop a Humean demarcation criterion for dividing the social from the natural worlds. I begin with two initially plausible attempts to delineate between the social and natural worlds, but show that there are clear problems that cut against both accounts. Next, I discuss an interesting explanatory account drawn from Ásta’s work, suggesting that it unfortunately will not work in any of the ways I spell it out. I then turn to Hume to give us the beginnings of a better account. Developing this account by adding a necessity condition, I suggest that such a Humean account offers the correct demarcation between the social and the natural, both in the cases which caused problems for our initial accounts and further potential counter-examples.

Throughout I ask about the distinction between the natural world and the social world. This is, strictly speaking, somewhat misleading, though I keep the terminology for stylistic purposes. What I am really interested in is the distinction between social and natural kinds. That is, what is it about money that makes it a social, rather than a natural kind? What is it about tiger that makes it a natural rather than a social kind?²

Mind-Dependence Won’t Cut the Mustard, Nor the Conceptual Space

One initially plausible stab at an answer one might offer to these questions is the following naïve mind-dependence account: social kinds are mind-dependent, whilst natural kinds are mind-independent.³ The thought here is simply that social phenomena are dependent for their existence on minded creatures such as ourselves. The universe would simply not have social phenomena unless there were creatures with minds to construct such phenomena. This seems to nicely capture the local constructionist’s intuitions that tigers are natural features of the world whilst money is a social feature. Moreover, in addition to its initial plausibility, I take it that this kind of position has implicitly been lying in the background of much discussion of the social world, both within analytic social ontology and elsewhere.

Unfortunately, as has been discussed elsewhere, mind-dependence will not do, as it looks like there are some mind-dependent kinds which are natural. These kinds seem to be mind-dependent, but in some mundane way which means that we are nonetheless inclined to treat them as natural:

What makes this requirement hard to specify is partly that it is usually agreed that some sorts of mind-to-world determination (or world-on-mind dependence) are metaphysically unmysterious and unproblematic. One challenge is, then, to divide these mundane dependencies from others. This challenge is especially pressing given social sciences including psychology, anthropology and sociology where the objects of study look to be minds, or to essentially involve minds. (Mallon 2016, 140).

[M]undane kinds of dependence on the mental are to be ignored when realism is characterized as mind-independence … We should, however, acknowledge that it is not straightforward to say which kind of dependence are mundane. (Jenkins 2005, 199)

What kind of kinds are problematic in this way? I will discuss three examples: psychological kinds, super-heavy elements, and poodles.⁴ Psychological kinds, such as Alzheimer’s disease (to take Khalidi’s 2013, 143–4 example) seem to be an issue because they appear to necessitate a mind for their existence. If one does not have minds, one does not have Alzheimer’s. Yet it would seem odd for us to suggest that Alzheimer’s disease is socially constructed. Meanwhile, super-heavy elements such as oganesson, which have only been observed in laboratory conditions due to the intentional experimentation of scientists such as Yuri Oganessian (Oganessian et al. 2006) seem to be mind-dependent, as they, as far as we can tell, would not have existed without human minds deciding to perform particular experiments. Again, however, it would seem odd for us to categorise the 118^th element of the periodic table as a social, rather than a natural kind. Finally, poodles, a dog breed created by humans intentionally selecting dogs with desired features to reproduce, would, on a naïve mind-dependence account of the demarcation between the social and natural worlds, come out as a social kind. After all, poodles only came into existence thanks to human intervention on the basis of particular human desires. This however seems odd: poodles are not a social kind.⁵

The naïve mind-dependence theorist might attempt to resist these sorts of examples in a couple of ways, not least by biting the bullet and claiming that, yes, things like poodles and oganesson are social kinds. However, I take it that this is hardly the most satisfying bullet to bite, and I suggest we should like some independent reason to do so beyond that it saves the naïve mind-dependence account. A more promising response from the naïve mind-dependence theorist might be to say that it is misleading to claim that these examples are really mind-dependent. That is, we might think that the oganesson example is better described as a particular mind-dependent set of social practices (scientific research) revealing features of a mind-independent part of reality. On such a description, the kind oganesson was out there waiting to be found, and its properties discovered, but scientists had to manipulate aspects of nature in order to discover those properties. According to this line of argument, oganesson is not mind-dependent, rather, only our discovery of oganesson was mind-dependent.⁶

How successful is this kind of response to the above examples? To my mind, it fails for a couple of reasons. First, it seems to posit that natural kinds exist prior to the existence of any member of that kind. That is, no examples of oganesson existed prior to their creation in a lab, and yet the naïve mind-dependence theorist seems to need to claim that the kind oganesson already existed and we could go out and discover its properties. This seems to get the dependence relation the wrong way round: plausibly kinds are grounded in their members, rather than the members of a kind being grounded in the kind. In both the cases of oganesson and poodles, the members of the kind are themselves grounded in intentional action: actual token oganesson particles and actual token poodles would not exist but for the intentional actions of certain humans. This dependence upon the mental is precisely what causes them to be trouble for the naïve mind-dependence theorist.⁷

Further, whilst there might be something to be said for the claim that there was a ‘gap’ in the periodic table at atomic number 118, and discovering the properties of ‘whatever kind is constituted by atoms with atomic number 118’ this sort of response really struggles to rule out poodles. To claim that there is a kind poodle out there, prior to the construction of poodles through an intentional breeding process, and this kind was ‘discovered’ through that breeding process seems to both commit one to a rather strong version of Platonism about kinds and a distorted picture of what was going on when the breeding programme took place.⁸

It looks, therefore, like a mere naïve mind-dependence criterion is simply going to fail to capture the boundary between the social and the natural—it is far too inclusive on the side of the social, treating what should be natural kinds as social kinds. Here is one way that one might make it more restrictive, raising the barrier for entry into the social world: suggest that a kind X is a social kind iff necessarily, X is mind-dependent, and a natural kind otherwise. That is to say, some phenomenon is social if and only if in every possible world where the phenomenon occurs, that phenomenon is dependent on some mind.

This modified version of the mind-dependence criterion is better than the original on two of the three examples we raised: oganesson and poodles. Super-heavy elements, such as oganesson could, in some possible world, have occurred without human (or other beings’) intentional action. As such, it is not the case that they are necessarily mind-dependent, and therefore our new account treats them (correctly) as natural, and not social kinds. Poodles, meanwhile, could just have happened to evolve into the breed we know today via natural selection, rather than the artificial selection that actually occurred. Moreover, there is a possible world in which lightning strikes a primordial ooze and creates a number of exact copies of actually existing poodles. These swamp creatures then would appear to be poodles, but are not mind-dependent in any sense. As such, poodles are not necessarily mind-dependent, and are therefore (correctly) treated as a natural kind by our new criterion.

However, the new mind-dependence criterion fails to deal with the psychological kind objection. After all, it looks like, necessarily, Alzheimer’s disease is dependent on there being minds. Without minds, there are no psychological kinds. It looks, therefore, as if despite enabling us to get around a couple of the problematic cases, our updated mind-dependence criterion simply will not do.⁹

A Disjunctive Account: Just List Off the Ways of Making the Social World Dammit!

One might remark that there is a fairly obvious answer to these questions waiting in the sidelines here: social kinds are all and only those kinds which are socially constructed, whilst all the other kinds are natural. As plausible (and probably correct) as this answer is, it isn’t particularly informative. We need a little more detail on what ‘social construction’ is. But we have been studying social ontology for some time now, and we are beginning to get some idea of how social construction works. Maybe we can use this fact to our advantage, by listing off the ways that parts of the world get socially constructed according to our best theory or theories of the social world—perhaps including collective intentions assigning status functions, collective willing, performativity, conferral, and so on.¹⁰ Then, we say that whatever is constructed as a result of these constructive mechanisms is socially constructed, and everything else is natural. So, our theory might look something like this:

Socially constructive mechanisms: M₁ … M_n

Phenomena constructed by M₁ … M_n: S₁ … S_n

The social world, on this account, is all and only those things S₁ … S_n.¹¹

This method initially looks quite good—supposing gender is performative for a moment, and Judith Butler’s notion of performativity is one of M₁ … M_n, then it looks as if gender comes out as a part of the social world (see Butler 1990). Meanwhile, glacial troughs, the U-shaped valleys formed by the process of glaciation, come out as a part of the natural world, as glaciation (the movement of a glacier down a slope, scouring out a valley, before melting and retreating) is plausibly not a part of M₁ … M_n.

However, there is a problem lying in the background here involving how to pick the mechanisms that should go into M₁ … M_n. Why is it that we should include performativity as a part of M₁ … M_n, and not glaciation? What principled reason can we give to suggest that one of these two mechanisms a socially constructive mechanism, and the other not? We cannot simply appeal to the mind-dependency of one mechanism and the mind-independency of the other mechanism, as it looks like this is going to fall into the traps that were sprung for our initial stab at the problem—psychological kinds, dog breeds, and lab-created compounds all look like they are the result of mind-dependent mechanisms. Take for instance, the poodle, which was created via artificial selection—humans intentionally selecting those dogs with curly hair, intelligence, and webbed feet for breeding with dogs with similar phenotypes. It looks like artificial selection of this sort, given a mind-independence criterion, ought to be included in M₁ … M_n, and therefore that poodles are a part of S₁ … S_n and that they are therefore social. Meanwhile, oganesson, an element of atomic number 118, which has only ever been observed in a laboratory thanks to the intentional bombardment of californium-249 with calcium-48 looks like it will come out as a social kind, as the mechanism by which it was created (intentional bombardment) is mind-dependent. It therefore looks like the initial problem that we attempted to avoid by moving to look at the particular mechanisms of social construction merely appears at a higher level.

Will the move to necessary mind-dependence help us here? Potentially—suppose we move to a model that looks like this:

Socially constructive mechanisms: M₁ … M_n

Phenomena constructed necessarily by M₁ … M_n: S₁ … S_n

Here, poodles do not come out as socially constructed, as it is possible that the mechanism that gave rise to poodles was not artificial selection, but natural selection, or even a swamp-man (swamp-dog?) style lightning strike hitting primordial ooze. Meanwhile, in the case of oganesson, it seems entirely possible that sort of cosmic accident or supernova could end up bombarding californium-249 with calcium-48 entirely mind-independently.

But still, psychological kinds cause a problem for such accounts. Any mechanism that could possibly bring about instances of Alzheimer’s disease is going to require the existence of a mind, or else one simply won’t have an instance of Alzheimer’s. As such, all such mechanisms will be members of M₁ … M_n, and therefore Alzheimer’s is a member of S₁ … S_n—wrongly counted as a social kind. So, it looks as if the move to specifying the boundary between the natural and the social in terms of the particular mechanisms involved is simply going to run into the same problems as our initial stab at setting forth a boundary. This said, whichever of these two attempts one prefers, whether the naïve mind-dependence or listing account, it looks as if putting a modal spin on the purported boundary rules out certain problem cases, even if this isn’t enough alone to escape worries about psychological kinds.

An Explanatory Account

A quite radically different suggestion might be drawn from Ásta’s remarks in her 2018 book Categories We Live By:

Here is my methodological suggestion: one should consider in what kinds of explanations the property occurs. If it functions in explanations of various social facts and it isn’t playing the explanatory role of various physical or natural facts, but rather, some ‘nearby’ property is doing so, that should give us reason to think that the property in question is a social property and hence conferred. (Ásta 2018, 71)¹²

Now I do not think that Ásta is advocating the above as a way of demarcating the social from the natural. But there is an account that might be offered lying in the background here which I think is worth exploring—that which is social features in the explanations of social facts. Such an account would dispense with a metaphysical criterion for demarcating the bounds of the social, such as mind-dependence, or some constructive mechanisms, and instead posit an explanatory criterion. The thought here is that the popularity of Pokémon, being social, will explain a number of social facts: the rising stock price of the Pokémon company, the development of school regulations to deal with a student black market in Pokémon cards, references to Pokémon in popular media, and so on. The popularity of Pokémon will presumably not, however, explain any natural facts.

One question that immediately arises for this kind of account is that of as yet undiscovered phenomena that would be required for explanation in the social sciences (think of ‘recession’ prior to the recognition of recessions by social scientists). Unless one thinks that recessions did not exist, or were natural phenomena prior to their recognition by economists, we need to change the account. However, there is a simple fix: we have to have a ‘In a completed science of the social’ clause in our spelling out of this account.

As such, let us spell out the account. There are options available here, which I list in order of decreasing strength:

A1: X is social iff it only features in the explanation of social facts, and not natural facts (in a complete science).

A2: X is social iff most explanations it features in are explanations of social, and not natural facts (in a complete science).

A3: X is social iff the most salient explanation(s) with respect to X are explanations of social facts (in a complete science).

A4: X is social iff it figures in at least one explanation of a social fact (in a complete science).

However, each of these looks problematic. It strikes me that A4 is obviously over-inclusive, given that the path of the River Severn figures in an explanation for the locations of human settlements across the United Kingdom county of Gloucestershire—natural geological and geographical facts are going to play a role in explaining many social phenomena. Meanwhile, it may also be under-inclusive, given that there might be features of our social world that are simply not required by our completed science.¹³

A3, meanwhile, seems odd: what is it for some explanation to be the most salient with respect to X? Perhaps we might say it is those explanations that feature the essential features of the phenomenon in question—but even here, nitrous oxide looks like it is going to turn out to be social. Any complete explanation of the club culture surrounding laughing gas will need to make reference to the fundamental chemical properties of N₂O, if only to explain the behaviour of those taking the drug.

A2, meanwhile, seems likely to run into questions of how to individuate explanations, and moreover, supposing that the number of explanations relevant to an area of interest is proportional to the complexity of that area, what counts as social and what counts as natural is going to be dependent on the relative complexity of the social as compared to the natural. If the social is significantly more complex than the natural, it looks like we are going to see a number of phenomena that should come out as natural being counted as social by the theory—simply because those phenomena require more explanations.

Finally, A1 looks like it is going to wrongly treat the socio-economic demands that led to the breeding of poodles as not being social facts. That poodles have webbed feet is to be explained in terms of an intentional breeding programme for that phenotype, which was driven by a particular demand for duck-hunting dogs in Germany in the middle of the last millennium. However, this explanation of a feature of a natural kind should not lead us to conclude that the particular economic climate for dog breeders in question was a not a social fact.

Here is another concern you might have about these accounts in general: they look worryingly circular. In order to work out whether something is social, on this kind of account, we need to already have a notion of the social in mind, in order to determine whether that which it explains is social. We need to be able to replace the term social on the right side of these biconditionals with something else if we are to find them persuasive as accounts of the demarcation of the social from the natural. We might update our accounts so as to replace explanations of social facts with ‘explanations in the social sciences’, but this simply pushes our question back a step—now we have to answer the question as to the remit of the social sciences without reference to the social. I am unconvinced that this is a productive strategy.

A Humean Boundary

Interestingly, I think that Hume offers us a better demarcation criterion. Hume distinguishes three senses of the term ‘natural’ in his Enquiries:

Natural may be opposed, either to what is unusual, miraculous, or artificial. In the former two senses, justice and property are undoubtedly natural, but as they suppose reason, forethought, design, and a social union and confederacy among men, perhaps that epithet cannot, strictly, in the last sense, be applied to them. (Hume 1975, 307: EPM App. 3.9n2; SBN 258n2)

He gives more detail on the two of these senses in his Treatise:

To avoid giving offence, I must here observe, that when I deny justice to be a natural virtue, I make use of the word natural, only as oppos’d to artificial. In another sense of the word; as no principle of the human mind is more natural than a sense of virtue; so no virtue is more natural than justice. Mankind is an inventive species; and where an invention is obvious and absolutely necessary, it may as properly said to be natural as any thing that proceeds from original principles, without the intervention of thought or reflexion. Tho’ the rules of justice be artificial, they are not arbitrary. Nor is the expression improper to call them Laws of Nature; if by natural we understand what is common to any species, or even if we confine it to mean what is inseparable from the species. (Hume 1978, 484: T 3.2.1.19; SBN 484)¹⁴

From these texts we might spell out what Hume has in mind when he calls something natural or not natural:

Natural₁ vs Artificial: That which is not a product of human thought or reflection vs that which is a product of human thought or reflection.

Natural₂ vs Arbitrary: That which is obvious and necessary vs that which is not obvious and necessary.

Natural₃ vs Supernatural: That which is not miraculous vs that which is miraculous.

A few examples may help to illuminate these distinctions. Man-made shelters are natural₂ and natural₃ but, for Hume, not natural₁. Meanwhile, posters of Carly Rae Jepsen are neither natural₁ nor natural₂, but are natural₃,¹⁵ and the planet Mars is natural₁ and natural₂ and natural₃. For Hume, justice is like a man-made shelter: a necessary and obvious invention of humankind (so natural₂) developed for humankind’s benefit produced by human thought or reflection (so not natural₁). This is to be contrasted with other aspects of human life, which are not products of thought or reflection, but are rather natural₁. For instance, the ‘natural appetite betwixt the sexes’ (Hume 1978, 486: T 3.2.2.4, SBN 486), is, for Hume, natural₁, being not a product of human thought or reflection, but rather, that which conditions much human thought and reflection. All of the above examples are Natural₃.

Let us insert here a quite commonplace understanding of thought and reflection that distinguishes such capacities from brute instincts such as hunger or sexual desire. What is required is consideration, over and above simple urges that take place without the agent thinking about what is occurring.¹⁶ Then, following Hume, on natural₁ we can thus distinguish between the natural and social worlds by offering the criterion that X is a social construct iff it is dependent on human thought or reflection. Call this the Humean account. Here we have a mind-dependence account of the delineation between the social and natural worlds, where it is the type of mind-dependence which matters. Hume gives us an account of the boundary between the natural and the social in terms of natural₁ that has a nice feature: it suggests that psychological kinds are features of the natural, and not the social world. Alzheimer’s is not a part of the social world, because it is not a product of thought or reflection, but are rather simply natural₁. It may condition thought and reflection, but it is not produced by human thought or reflection.¹⁷

Of course, this conception of the boundary between the natural and the social falls victim to the dog breed and heavy element examples raised above—on this Humean account, poodles and oganesson are socially constructed, as they are the products of human thought or reflection. Thankfully, our above discussion provides us with a nice modification of the account to deal with such problems: we just have to add a necessity condition. On this modified Humean account, something is socially constructed iff it is necessarily dependent on thought or reflection.¹⁸ Thus, given that poodles could have arisen through natural selection without human interference, or indeed any thought or reflection at all, in the case of swamp-dog, they are not socially constructed, even if, as is the case, their instantiations in the actual world were artificially selected for. Meanwhile, given that oganesson could have been created without the intervention of human thought or reflection via cosmic accident, the modified Humean account treats oganesson as natural.

Potential Trouble

One might suggest, following the recent anti-individualist trend in social ontology (see e.g., Epstein 2015) that the Humean account (in both its modified and unmodified versions) is too individualistic to fully capture our social world. After all, if it is necessary that thought or reflection produce some phenomenon in order for that phenomenon to count as social, we seem to be ruling out a number of potential phenomena from the social world which don’t appear to be solely and directly dependent on thought or reflection.

For instance, it looks like large swathes of what we would normally take to be aspects of the social world are ruled out. Take US dollars—as Epstein puts it,

What does being a dollar supervene on? Among the things it supervenes on are the properties of pieces of paper. Given that the constitutive rule for dollars is what it is, for something to be a dollar requires that it be printed in green ink, on a particular kind of paper. Moreover, it supervenes in part on the properties of the Bureau of Engraving and Printing. (Epstein 2015, 58).

So, US dollars are not solely dependent on human thought or reflection. Meanwhile, it looks as if recessions might also cause trouble:

 … a given economic state can be a recession even if no one thinks it is, and even if no one regards anything as a recession or any conditions as sufficient for counting as a recession. The concept of a recession is recent—the first recorded use of the term was in a 1929 article in the Economist—but there could have been recessions long before then. (Thomasson 2003, 276)

In such cases, thought or reflection cannot be that upon which the recession is dependent, at least directly, as no one is aware that a recession is occurring.¹⁹

But here is the catch—the Humean account I have been developing does not claim that social phenomena require sole and direct dependence. For instance, it is perfectly compatible with both the Humean and modified Humean accounts that some phenomenon X is partly dependent on factors other than thought or reflection (indeed, in most cases we would expect this to be so) or that X is only indirectly dependent on thought or reflection, and that nonetheless, X is social. In the case of the dollar, the Humean is happy to admit that dollars depend on their physical properties and are nonetheless social, so long as it is the case that they are also dependent on thought or reflection. Similarly, the Humean can help themselves to the notion of indirect dependence. A recession, for instance, is ‘two or more consecutive quarters (a period of three months) of contraction in national GDP’ (HM Treasury 2010). It may well be the case that a recession could occur without anyone thinking about or reflecting upon recessions. However, the recession itself is dependent on several social features which are themselves dependent on thought or reflection—including nations and particular monetary systems. Recessions, then, are dependent on nations, which are in turn dependent on thought and reflection. They are therefore indirectly dependent on thought and reflection—but this is nonetheless a kind of dependence that is acceptable to the Humean.²⁰

Fictional Kinds

Khalidi has suggested that mind-dependence is a ‘red herring’ when it comes to this and related issues, as mind-dependence accounts of social kinds ‘assimilate social and psychological kinds … to fictional kinds such as unicorns and wizardry’ (Khalidi 2015, 109). We have already discussed how psychological kinds are distinguished from social kinds, so let us now turn to what Khalidi calls ‘fictional kinds’. Khalidi’s thought here is that unicorns and wizardry are (a) mind-dependent, and (b) kinds, and thus any mind-dependence account of the social is going to wrongly treat them as social kinds.

There is a question here as to whether unicorns and wizardry really constitute kinds—there are not, and never have been any actual instances of either after all. One response that one could make would therefore be to just suggest that these mind-dependence accounts should be read as having an implicit ‘for their existence’ clause. Thus, we might put the modified Humean account as follows:

Something is socially constructed iff it is necessarily dependent on thought or reflection for its existence.

Here, given that wizardry and unicorns do not and have never existed, we might think that they fail to meet the criteria of even the naïve mind-dependence account and so won’t count as social. Kinds are supposed to be groupings of aspects of the world—the actual world. Wizards and unicorns are not and have never been a part of the actual world and thus there is no kind wizard nor kind unicorn. Now, were one in the possible world where there are in fact unicorns, wizards, witches and covens, one legitimately might ask what kind of kinds unicorns, wizards, witches, and covens are.²¹ But from the perspective of that world, none of these kinds are fictional, and indeed, we can simply apply the modified Humean account in order to determine whether such kinds are natural or social. From the perspective of that world, unicorns are natural (even if in that world they were created by a witch or wizard, there is a possible world in which unicorns evolved absent minded interference, and a world featuring swamp-unicorns), witches and wizards are similarly natural (there is a possible world in which there are swamp-witches and swamp-wizards), and covens are social (the gathering of witches is not a coven unless there are particular thoughts and reflections regarding the rituals and rites involved on the part of the witches).

From the perspective of our world, all such kinds are fictional. They have no, and have not had any, actual members. Nonetheless, the tools we now have enable us to talk about fictional social kinds and fictional natural kinds, that is, distinguishing between the natural and social kinds that exist in a fiction. If someone wrote a book set in the possible world described above, which featured witches, wizards, unicorns, and covens, we can simply apply the modified Humean account from the perspective of someone in that possible world.

Conclusion

The Modified Humean theory of the boundary between the social and natural worlds provides the local constructionist with a clear boundary that respects ordinary intuitions regarding where that boundary should lie. Not only is it extensionally accurate, it develops and refines an intuition that the social world is in some way dependent on human (or other creatures’) activity. Insofar as we are local constructionists, I suggest we adopt the modified Humean account of the boundary between social and natural worlds.

Acknowledgements

I would like to thank Sally Haslanger, Esa Díaz-León, Jennifer Saul, Robert Stern, Rosa Vince, Kayleigh Doherty, Alana Wilde, James Lewis, Robbie Morgan, Rory Wilson, and an anonymous reviewer at Philosophical Papers, along with audiences at Massachusetts Institute of Technology and the University of Sheffield for their helpful comments and questions on this piece.

Notes

1 I will not argue for local constructionism here, but rather assume it for the purposes of exploring this problem. Nonetheless I maintain that it is the most plausible picture of our world.

2 Note here that I am talking about kinds rather than categories—that is, I am not interested in whether our ideas about what makes something money or a tiger are socially constructed, but rather whether money and tigers are themselves socially constructed. See Hacking (1999, 9–16) and Haslanger (2012, Ch. 3) for more on why we should distinguish the construction of our classificatory tools from those kinds that we are attempting to classify.

3 I have a fairly inclusive view on the kind of dependence at stake here. Whatever kind of dependence is in question, whether constitutive dependence (as grounding or as anchoring), or causal dependence (see Haslanger 2003, 317–18 and Epstein 2015, Ch. 6), in what follows I will assume that any of these sorts of dependence will do for something to count as dependent on something else.

4 For a rich and important discussion of these sorts of issues in a somewhat different context, see Rosen 1994.

5 One might, if one has an austere notion of natural kinds that does not allow for breeds of dog to be natural kinds, complain that the poodle is not robust enough a grouping to count as a kind proper. However, it is enough that my reader thinks that on the local constructionist picture it would be odd to count the poodle as a part of the social world, whether one thinks it is a mere grouping or a kind proper. If one is still unsatisfied, note that the species canis lupus familiaris, or the domestic dog, itself provides a problematic kind. Canis lupus familiaris only exists thanks to human mindedness, in the form of domestication. But we should be extremely unwilling to count this species as a social kind—species seem to be paradigmatic natural kinds.

6 Thank you to an anonymous reviewer at Philosophical Papers for pushing me on this point.

7 A small point of clarification: at least in the case of poodles, the continued existence of any given poodle is not dependent on (human) mindedness. However, all actual (and past) poodles would not exist had there not been an intentionally guided breeding programme in the past. It is in this sense that their existence is mind-dependent.

8 What about Alzheimer’s? Here the naïve mind-dependence theorist might respond to the putatively problematic example by saying that we can distinguish Alzheimer’s as a cognitive condition from Alzheimer’s as a neurological condition. One might think, for instance, that Alzheimer’s as a cognitive condition is mind-dependent (and therefore social) whereas Alzheimer’s as a neurological condition is dependent only on the human nervous system (and therefore not social). However, I worry that this commits the naïve mind-dependence theorist to a form of mind-body dualism: after all, on any materialist picture of the world, the human mind is (at least partially) constituted by the central nervous system, and as such Alzheimers as a neurological condition is dependent on a part of the human mind. If this fails to convince, the opponent of the naïve mind-dependence theorist might simply swap in depression as an example—a disease that is defined purely in terms of its cognitive symptoms and likely does not have a common underlying neurological basis (see, for instance, the depressive disorders in detailed American Psychiatric Association 2013).

9 Esa Díaz-León has suggested to me an alternate attempt to restrict the mind-dependence account that would exclude non-constitutive mind-dependence. So, we might suggest a version of the account along these lines that runs as follows: a kind X is social iff X is constitutively mind-dependent, and natural otherwise. This seems to get us the right answer in poodle and super heavy element cases, as they are not constitutively, but merely causally mind-dependent and therefore come out as natural kinds. However, it still looks as if psychological kinds are going to be a problem for this type of account, as it looks as if many, if not all, psychological kinds are going to be constitutively dependent on minds. Moreover, we might also worry that such an account is also too exclusive, ruling out features of our social world that are merely causally mind-dependent, but also social. One might, for instance, think that desire paths are a social kind, but that nonetheless, desire paths are not constitutively mind-dependent.

10 See, e.g., Gilbert 1989, Butler 1990, Searle 1995, and Ásta 2018.

11 I suspect that something like this pluralist view is in the background for Díaz-León (2019). Of course, both there and in her 2015 paper she argues that certain kinds of social construction are more apt for particular (debunking) projects, but this is different from a claim that only (say) constitutive social constructions are social constructions, and only they count for the purposes of demarcating the social world.

12 One might suggest a reading of Ásta which suggests that X is social property iff X is conferred, and use this claim to somehow demarcate the social world. Note however, that being a recession was a social property prior to the recognition of recessions by economists, and therefore was a social property but not conferred, thus the above biconditional cannot hold. Moreover, causal social construction seems to be hard to fit into this sort of picture. Plausibly, conferral is sufficient but not necessary for sociality. For more on conferring and conferralism see Ásta 2018.

13 There may be social phenomena that are causal dead-ends, for instance, that don’t serve to help in the explanation of any further facts.

14 I admit that it is possible that, whilst I run together natural as opposed to unusual (as in the Enquiries) and natural as opposed to the arbitrary (as in the Treatise) in the category natural₃ in what follows, it may be that Hume developed four senses of nature. In any case, for the purposes of this paper, it is largely besides the point.

15 That Carly Rae Jepsen posters are not natural₂ is a point of some contention among those who are great fans of her album Emotion. However, I take it that their sense of what is obvious and necessary to humankind has been somewhat misled by the catchiness of ‘Call Me Maybe’, which despite their protestations, is in fact natural₃.

16 I freely admit that this may not what Hume had in mind, but I am (for the most part) only really interested in getting the best account of the boundary between the natural and the social, and not in being particularly faithful to Hume.

17 It might be objected that there may be instances of psychological kinds that are or were dependent on human thought or reflection. Dementia, in addition to being caused by (among other things) genetic factors, can be caused by excessive alcohol consumption (and is therefore dependent on thought and reflection given how alcohol is distributed in our society). It thus seems to, on the Humean account, be social, rather than natural. However, thankfully, the move to a modified Humean account featuring a necessity condition (see below) rules out such potential counterexamples, as all such examples could have not been dependent on human thought and reflection: not only could the dementia have arisen as a result of say, genetic, causes, we can imagine a case where excessive alcohol is imbibed without thought and reflection being required.

18 The absence of ‘human’ in the definition is a further friendly amendment to the Humean account—we should hardly like to rule out non-human thought and reflection as possibly giving rise to social phenomena.

19 Recessions and other social phenomena that do not require recognition on the part of minds for their existence demonstrate that utilizing John Searle’s account of the boundary between the social and the natural will not do. If Searle’s account is simply that x is a social kind iff our attitudes towards that x are partly consitutive of x, then it looks like, given the absence of such attitudes, recessions are wrongly counted as not social (Searle 1995, 33). See Thomasson 2003 and Khalidi 2015 for more on this.

20 Here is another potential counterexample: swamp piano. Suppose lightning strikes a swamp of primordial ooze, and instantaneously creates what appears to be a working (and perfectly tuned) piano. As such, one might suggest that pianos are not necessarily dependent on thought or reflection, given that one could be dependent on merely lightning and ooze, and that therefore the modified Humean account wrongly treats pianos as a natural, rather than a social kind. However, I don’t think that this description quite captures exactly what is going on in this case. Take, by way of explanation, the flat-topped rock that sits in the woods behind my house, whose shape was created by erosion. What is odd is that it is an exact physical match of a rock that sits in my back garden and serves as the table where I take tea and eat barbecue in the summer. I suggest that the rock in my back garden is a table, whilst the rock in the woods is not. There is a sense in which we could call the rock in the woods a table—certainly, I could take tea on either the rock in my garden or on the one in the woods. However, I suggest that it is only the investment by minded creatures of the rock with the status of a table that makes the rock a table in either case. That is, tables are dependent on thought and reflection because (in Searle’s terms) a status function is required to make a flat, elevated surface into a table. The modified Humean account therefore correctly treats tables as a social kind, along with other such artifactual kinds. What does this mean for the swamp piano case? Well, I suggest that what is created in the lightning strike is not a piano. Just as a flat-topped rock created by lightning strike is not a table until given an appropriate status by minded creatures, so too the lightning-made ivories and hammers in a wooden case is not a piano proper until endowed with an appropriate status via thought or reflection by minded creatures.

21 A coven is a gathering of witches who meet to perform certain rituals and rites.