1 Introduction

It is currently common to conceive of the classic methodological individualism–holism debate in level terms. In traditional terms, the debate is about the proper focus of social scientific explanations and involves two basic positions: methodological individualism, which states that only individualist explanations should be offered in the social sciences, and methodological holism, which maintains that both individualist and holist explanations should be advanced. When the debate is understood in level terms, it concerns the proper level of social scientific explanations. Methodological individualism is then the view that only individual-level explanations should be put forward, whereas methodological holism is the position that both individual-level and social-level explanations should be provided.

In this paper, I argue that the general methodological individualism–holism debate is not apt to be characterized in level terms. Moreover, I discuss the consequences of this finding for the ongoing dispute.

I proceed as follows. In Sect. 2, I outline the methodological individualism–holism debate in more detail. In Sect. 3, I explicate the basic idea of levels and the conditions that must be met in order to phrase the individualism–holism dispute in level terms. In Sects. 4 and 5, I discuss the two dominant ways in which to underwrite the level formulation, namely by appeal to a supervenience or an emergence account of the social world as levelled. In Sect. 6, I show that, roughly since the 1950s, methodological individualists have mostly adopted inclusive notions of individualist explanations that, relative to the supervenience and emergence accounts, fail to qualify as individual-level explanations. For this reason, I contend, the level formulation is inadequate as a characterization of the general individualism–holism debate. Subsequently, I reject two objections to my finding. In Sect. 7, I discard the contention that inclusive notions of individualist explanations should be dismissed from the debate. In Sect. 8, I rebut the claim that the level conception is defensible if based on an alternative account of the social world as levelled. In Sect. 9, I discuss the consequences of my rejection of the level formulation for the ongoing methodological individualism–holism dispute and then I wrap up in Sect. 10.

2 The methodological individualism–holism debate

The methodological individualism–holism debate is about the proper focus of social scientific explanations. The dispute comes in different versions. In what follows, I present its classic version: it is this version formulated in level terms that is the focus of the present paper.Footnote 1

The classic discussion is premised on a fixed distinction between two types of explanations (explanantia): individualist and holist explanations. Individualist explanations solely describe individuals, their actions, beliefs, desires, etc. They are exemplified by an explanation of why Anna baked a cake which states that Susan and John wanted her to, and by an explanation of why the crime rate had gone up which states that individuals a, b, c, etc. had lost their jobs and felt frustrated about having little money and no job opportunities. In contrast, holist explanations only describe social phenomena such as social entities (universities, churches, etc.), social processes (revolutions, economic growth, etc.), statistical properties of groups (the literacy rate, the suicide rate, etc. in a group), properties of social networks (their density, cohesion, etc.), and beliefs, desires, and other mental properties ascribed to groups (i.e. a government’s desire to stay in power). Holist explanations are illustrated by the claim that the government wanted to lower the national minimum wage in reply to the question of why the unions protested, and by the claim that there had been a rise in unemployment in reply to the question of why the crime rate had gone up. For now, this very rough characterization of individualist and holist explanations will do. Later I discuss different ways in which further to spell out the notion of individualist explanations.

The basic positions in the debate disagree about the extent to which the two types of explanations should be employed in the social sciences. Methodological individualists maintain that only individualist explanations should be put forward. In their view, holist explanations should be dispensed with. They are opposed by methodological holists who hold that sometimes it is in order to advance individualist explanations, other times holist explanations. In other words, both kinds of explanations are indispensable. There is also a stronger form of methodological holism according to which holist explanations alone should be provided; individualist explanations should be dispensed with. However, as this position has few, if any, current defenders, I disregarded it in the subsequent discussion. I concentrate exclusively on what it takes to maintain that the classic debate, as it plays itself out between methodological individualists and moderate methodological holists, is apt to be phrased in level terms.

3 The basic idea of levels and the level conception

The notion of levels is widely employed in philosophy and many sciences. More precisely, it is variously maintained that there are levels of explanation, ontology, description, science, analysis, etc. Among these, ontological levels are often regarded as the most basic in the sense that other other kinds of levels, including explanatory ones, are derivative upon them. Below I first offer a brief sketch of the received conception of ontological and derivative explanatory levels.Footnote 2 Then I show how the level formulation of the classic individualism–holism debate amounts to the claim that the dispute is really about explanatory levels that are derivative upon ontological levels.

The basic idea of ontological levels is that a set of phenomena, such as entities, properties, or substances, may be organized into higher and lower levels. Accordingly, phenomena at some level are lower-level phenomena relative to the level(s) of phenomena above them (if any), and higher-level phenomena relative to the level(s) of phenomena below them (if any). The inter-level relation, i.e. what makes it the case that some phenomena are at a higher or lower level than others, may be variously spelled out (Craver 2007:172ff). For example, Craver discusses three types of inter-level relations. The first cashes out the relation as one of causation as illustrated by the claim that some phenomena are at a higher level than others in virtue of their control over the activities of these other phenomena. The second explicates the inter-level relation as one of relative size: some phenomena are said to be at a higher level than others because they are bigger in size than them. The third takes the inter-level relation to be one of composition as exemplified by the view that some phenomena are at higher level than others since they spatially contain these other phenomena.

Moving on to explanatory levels, the standard view is that a set of explanations in the sense of explanantia may likewise be arranged into higher and lower levels. In this perspective, an explanation describes one or several phenomena at the same level. Accordingly, explanations are at the same level if they describe phenomena that belong to the same ontological level. That is, assume that explanation 1 describes phenomenon x, that explanation 2 describes phenomenon y, and that phenomenon x and y are same-level phenomena. In that case, explanation 1 and 2 constitute same-level explanations because they point to same-level phenomena. Moreover, same-level explanations are at higher or lower levels than other explanations depending on whether they refer to phenomena at higher or lower ontological levels than these other explanations. Thus, slightly changing the example offered a second ago, suppose that phenomenon x is at the level above, rather than at the same level as, phenomenon y. If so, then explanation 1 qualifies as a higher-level explanation relative to explanation 2 since explanation 1 refers to a phenomenon (viz. x) at a higher level than that mentioned by explanation 2 (viz. y). In this fashion, explanatory levels are derivative upon ontological levels: the level of particular explanations is determined by the level of the phenomena they describe.

The level formulation of the classic individualism–holism debate may be characterized as the application of the standard view of ontological and derivative explanatory levels to this dispute. The conception assumes that individualist and holist explanations qualify as lower individual-level and higher social-level explanations respectively. Moreover, it takes it that these explanatory levels are ontologically derivative and, by implication, the level conception presupposes that the social world may be organized into two levels. More precisely, the ontological assumption underlying the level conception is that individualistic phenomena (individuals, individualistic properties, or the like) and social phenomena (social entities, social properties, or the like) form two adjacent levels: individualistic phenomena are lower individual-level phenomena and social phenomena are higher social-level phenomena. This assumption, in turn, underwrites the postulation of corresponding lower individual-level and higher social-level explanations: lower individual-level explanations describe lower individual-level phenomena whereas higher social-level explanations describe higher social-level phenomena. Finally, and as already noted, these considerations link up to the classic individualism–holism debate via the claim that individualist and holist explanations qualify as lower individual-level explanations and higher social-level explanations respectively.

These points may usefully be summarized by saying that, in the contemporary debate, the level conception presupposes that two conditions may be met:

  1. 1.

    The ontological level condition: individuals, individualistic properties, or the like, are lower individual-level phenomena and social phenomena (social entities, social properties, or the like) are higher social-level phenomena.

  2. 2.

    The explanatory level condition: Individualist explanations (explanantia) qualify as lower individual-level explanations since they only describe lower individual-level phenomena (they don’t describe phenomena belonging to other levels) and holist explanations (explanantia) qualify as higher social-level explanations because they solely describe higher social-level phenomena (they don’t describe phenomena belonging to other levels).

Insofar as both these conditions are met, the individualism–holism debate is aptly characterized in level terms, that is, as turning on whether only individual-level explanations should be offered in the social sciences (the view of methodological individualists), or whether sometimes individual-level explanations, other times social-level explanations should be advanced (the position of methodological holists).

In the next two sections, I outline the two most influential attempts to show that the conditions may be fulfilled: one bases itself on the notion of supervenience, the other on that of emergence. In this process, I do not critically discuss various aspects of the two accounts or go into ways in which the accounts may be further clarified or refined. My sole aim is to bring out how they purport to establish that the classic individualism–holism debate is aptly characterized in level terms.

4 The supervenience-based level conception

The notion of supervenience gained currency in philosophy of mind around the early 1970s. In the 1980s, it began to make its appearance in the methodological individualism–holism debate: level accounts of the social world by appeal to supervenience were put forward while being accompanied by the level conception of the dispute.Footnote 3

Supervenience accounts of the social world are concerned with the properties of social entities, processes, and individuals.Footnote 4 More precisely, they draw a distinction between individualistic and social properties. There is some variation as to what is considered individualistic properties. For instance, these are sometimes specified as individuals’ properties that do not presuppose the existence of social entities and processes (Kincaid 1996, p. 146). Or, they are more narrowly identified with individuals’ bodily movements and mental states (Currie 1984, p. 347). More consensus obtains when it comes to the circumscription of social properties. These are the properties of social entities, processes, etc. as well as individuals’ social role properties, that is, their social roles and social role actions as exemplified by being a nurse, a CEO, voting, and firing.

Based on this distinction, supervenience accounts state that social properties supervene exclusively on individualistic ones. This means, roughly, that once all the individualistic properties are fixed, the social properties are fixed too. As such, there can be no change at the level of social properties unless there is also a change at the level of individualistic properties.Footnote 5 By way of illustration, consider individuals with various beliefs, practices, etc. who form a constellation with the social property of being a sports club. According to supervenience accounts, this property is a supervenient social property: it is fixed once the individualistic properties are fixed, and the group of individuals cannot cease to be a sports club without a change in the individualistic properties that subvene the property.

Supervenience accounts of the social world continue by claiming that the supervenience relation should be viewed as an inter-level relation: because social properties supervene exclusively on individualistic properties, social properties are higher social-level properties relative to individualistic properties that are lower individual-level properties. This granted the ontological level condition is met.

What then about the explanatory level condition? In the view of supervenience accounts, individualist explanations solely describe individuals and their individualistic properties. Since individualistic properties are lower individual-level properties, these explanations qualify as lower individual-level explanations. Further, holist explanations only describe the social properties of social entities, processes, and individuals. Consequently, because social properties are higher social-level properties, holist explanations qualify as higher social-level explanations. In this manner, the explanatory level condition is fulfilled too. By implication, the general methodological individualism–holism debate is apt to be phrased in level terms: it concerns whether individual-level explanations alone should be offered in the social sciences (the view of methodological individualists), or whether it is sometimes in order to provide individual-level explanations, other times social-level explanations (the position of methodological holists).

The supervenience picture of the social world has recently been described as the consensus view among philosophers (Epstein 2009, 2015:33ff). Thus, not surprisingly, the supervenience-based level formulation of the individualism–holism debate is particularly popular among philosophers. Moreover, it may be noted, it has primarily been promoted by proponents of methodological holism.

5 The emergence-based level conception

The first comprehensive discussions of the notion of emergence go back to the tradition of British Emergentism that came into existence around the 1850s (McLaughlin 1992). In the context of the individualism–holism debate, though, it was not until the 1970s that the idea of emergence really began to make its appearance. Around this time, Roy Bhaskar founded the social scientific school of Critical Realism (Bhaskar 1978[1975], 1998[1979]). Following Bhaskar’s lead, critical realists offer a level picture of the social world by appeal to emergence while using this picture as basis for their level conception of the individualism–holism debate. As Dave Elder-Vass has presented a particularly clear account along these lines, I base the following discussion on his writings (Elder-Vass 2007, 2010).Footnote 6,Footnote 7

The starting point of Elder-Vass’ account is that social entities are composed of individuals standing in various relations to each other (see Elder-Vass 2007, p. 31).Footnote 8 Considered as parts of social entities, individuals have various individualistic properties defined as the properties that they also possess in isolation or as elements of an unstructured collection of parts. From the perspective of their being wholes, social entities have various social properties.

Elder-Vass explains that there are two types of social properties: resultant and emergent. Resultant social properties are properties of social entities that their parts, viz. individuals, also have in isolation or as parts of an unstructured collection. As such, they are a function of the aggregation (simple addition) of individuals’ individualistic properties (Elder-Vass 2010, pp. 17–18). These properties are exemplified by the ability to scream as ascribed to a kindergarten: children also have this property independently of being, at that moment, parts of the kindergarten. Emergent social properties are not possessed by individuals considered in isolation or as parts of an unstructured collection. Social entities have these properties in virtue of individuals (with their individualistic properties) standing, at that moment, in certain relations to each other. Some emergent social properties are ascribed to social entities qua wholes as illustrated by a firm’s power to adopt a new sales strategy. Others are properties of the whole that are exercised by individuals as exemplified by individuals’ social role properties (see Elder-Vass 2007, p. 32).

Elder-Vass holds that the relations of aggregation and emergence are both inter-level relations: because social properties either result from the aggregation of individuals’ individualistic properties or are emergent from suitably interrelated individuals with their individualistic properties, social properties are higher social-level properties and individualistic properties are lower individual-level properties. Hereby, the ontological level condition is met. Elder-Vass stresses in particular the significance of emergence: only emergent social properties are novel properties compared to those found at the individual level. This being the case, I refer to his account as an emergence picture of the social world.

Elder-Vass continues by making it clear that individualist explanations only describe individuals and their individualistic properties. Accordingly, since individualistic properties are lower individual-level properties, these explanations qualify as lower individual-level explanations. In contrast, holist explanations solely describe social entities and their resultant or emergent social properties.Footnote 9 Because resultant and emergent social properties are higher social-level properties, holist explanations qualify as higher social-level explanations. Thus, the explanatory level condition is fulfilled too. It follows that the general individualism–holism debate is apt to be characterized in levels terms: it turns on whether to use individual-level explanations only (the methodological individualist position), or whether to use both individual- and social-level explanations (the methodological holist stance).

Critical Realism is currently a highly influential school of social theorizing. Hence, it is primarily among social scientists—Critical Realists and their followers—that the emergence account of the social world serves as ground for the level conception of the individualism–holism debate. In this dispute, critical realists defend the holist position. Thus, similarly to the supervenience-based level formulation, the emergence-based level formulation tends to go hand-in-hand with the defense of methodological holism.

6 A rejection of both the supervenience- and emergence-based level conception

The level formulation requires that both the ontological and the explanatory level condition be met. This means that in order to dispute the level conception, it suffices to demonstrate that one of these conditions fails. For the purposes of this paper, I assume that the supervenience and emergence accounts of the social world each satisfy the ontological level condition. Hence, I set to a side the questions of which of these rival accounts is preferable, what objections may be raised against them, and the like. Instead, I concentrate exclusively on establishing that, if these accounts are granted, the explanatory level condition cannot be met.

Lars Udehn remarks that “[t]here has been in the history of methodological individualism a development from extreme to less extreme versions of methodological individualism” (Udehn 2002, p. 498). In line with this observation, I first show that, since the 1950s, most methodological individualists have defended one or several inclusive notions of individualist explanations. On that basis, I argue that because these inclusive explanations do not qualify as individual-level explanations, the level formulation is inadequate as a characterization of the general debate. Or differently put, the level conception should be rejected as a characterization of the general individualism–holism debate since it fails to encompass the widespread inclusive individualist positions.

A good place to start is by returning to the rough characterization of individualist explanations presented in Sect. 2. According to it, individualist explanations solely describe individuals, their actions, beliefs, desires, etc. This specification raises various questions including what types of descriptions of individuals and their actions are permissible, and how the “etc.” should be spelled out. Among participants in the individualism–holism debate, there are different views on this matter. As noted a second ago, methodological individualists have tended, over time, to favor increasingly broader notions of individualist explanations in the sense of allowing these explanations to avail themselves of more and more types of description. I now illustrate this trend by considering two specific issues as to how the notion of individualist explanations should be further cashed out.

The first issue concerns whether individualist explanations, in their description of individuals and their actions, are also permitted to use social role descriptions, that is, descriptions of individuals’ roles (like “nurse” or “politician”) and role actions (like “firing” or “voting”). In reply, the narrow conception maintains that social role descriptions are not allowed to figure in individualist explanations, whereas the inclusive notion holds that it is acceptable to include them. Since the 1950s at least, methodological individualists have mostly adopted the inclusive notion. For instance, Watkins, a key protagonist of methodological individualism in the 1950s, provides examples of individualist explanations that refer to individuals’ social roles (Watkins 1952, 1957). More generally, Stephen Lukes, in his classic 1968 paper, observes that this inclusive view of individualist explanations is widespread among individualists (Lukes 1968, p. 125). An inclusive explanation in this sense is illustrated by the claim “the prime minister had just resigned” in response to the question of why the main television stations featured interviews with all the leading politicians in the country.

The other issue revolves around whether the “etc.” in the characterization of individualist explanations should be unpacked to include selected aspects of individuals’ context of action. That is, should individualist explanations be allowed to describe not only individuals, their actions, and their reasons behind the actions, but also selected aspects of the circumstances in which their actions takes place? A narrow conception answers this question in the negative, whereas an inclusive notion singles out one or several types of factors that may be described as part of individuals’ context of action.

Thus, one inclusive conception contends that individualist explanations may mention material factors as part of a description of individuals’ context of action. That is, the explanations may specify the physical environment, material artefacts, material resources, etc. as these constrained or facilitated individuals’ actions. As Joseph Agassi observed already in 1960, there is a long tradition among methodological individualists for permitting individualist explanations to refer to individuals’ material circumstances of action (Agassi 1960, see also his 1975). In some of his writings, Watkins is a representative of this trend too when he defines individualist explanations as ones that may describe individuals’ “physical resources and environment” (Watkins 1957, p. 106). These inclusive explanations are exemplified by stating, in response to the question of why the race had been cancelled that, as the heavy rainfall continued (a material factor), the runners had first decided to seek shelter and then determined that it would be too risky to continue.

Another inclusive conception goes one step further. It holds that individualist explanations may additionally mention institutions as part of a description of individuals’ context of action. For the present purposes, it suffices to note that institutions are, among other things, exemplified by states, schools, and universities, and that the explanations may refer to them by way of institutional terms like “state,” “school,” and “university.” Like material factors, these institutions may only be described from the perspective of how they (partly) set the scene for individuals’ actions by constraining or facilitating these. The first explicit statement of this position is often credited to Popper (see, e.g., Agassi 1960). However, his remarks on the matter are rather sketchy, so the view has first and foremost been elaborated by Agassi and Ian Jarvie (Agassi 1960, 1975; Jarvie 1972).Footnote 10 Methodological individualists who endorse the use of these inclusive individualist explanations are commonly referred to as institutional individualists. Today, institutional individualism is widely endorsed. As Lars Udehn remarks, it “is the dominating version of methodological individualism in political science and in the new institutional economics” (Udehn 2001, p. 348). A simple example of an individualist explanation along these lines might be the claim “in dictatorship country z (the institutional context of action), individuals a, b, and c had managed to convince other people that they would all be better off if they overturned the dictator” in response to the question of why a revolt had broken out there.

The discussion so far has pointed to different ways in which methodological individualists have opted for inclusive notions of individualist explanations, namely by allowing these also to describe individuals’ social roles and role actions, their material context of action, and/or their institutional context of action. On this basis, it may now be asked whether explanations along these lines qualify as individual-level explanations insofar as the supervenience and emergence pictures of the social world are each granted. Inclusive individualists do typically not talk about ontological and explanatory levels and hence they do not address this question. Nevertheless, it is fairly easy to answer.

To see this, note first that according to both level accounts of the social world, individuals’ social role properties are social-level properties. By implication, an inclusive notion of individualist explanations that permits the description of social roles hereby allow reference to social-level properties. Further, from the perspective of the supervenience and emergence pictures, the material factors that are part of individuals’ circumstances of action likely belong to a level or levels different from the individual one. Proponents of these pictures often hold that below the individual level, there is a biological level, followed by a chemical level and then a physical level. This being the case, it is reasonable to think that material context factors should be regarded as belonging to one or several of these additional levels. For the present purposes, it does not matter at what other level(s) they fit in. Thus, I shall simplify by saying that they belong to a lower material level. The important point is that an inclusive conception, which takes it that individualist explanations may describe individuals’ material circumstances of action, allows reference to properties that are not at the individual level. Finally, by the lights of both ontological accounts, institutions that are part of individuals’ context of action are straightforward social-level properties (i.e. being a university, say, is a property of a constellation of individuals). Thus, an inclusive notion that allows individualist explanations to describe institutional context factors licenses reference to properties at the social level.

Because inclusive notions allow individualist explanations to refer not only to individuals with their individual-level properties but also to social-level properties (social roles and/or institutions) and/or material-level properties (material factors), these explanations do not qualify as individual-level explanations. They go beyond the individual level: they are multi-level rather than single-level explanations.

The fact that, since the 1950s, most methodological individualists have defended notions of individualist explanations that are inclusive in one or several of these senses means that it is incorrect to maintain that individualist explanations generally qualify as individual-level explanations. A view along these latter lines fails to acknowledge—and encompass—the inclusive individualist positions in the debate. Assuming that the supervenience and emergence pictures each meet the ontological level condition, the explanatory level condition is not fulfilled due to the existence of these inclusive notions of individualist explanations. By implication, the level formulation of the individualism–holism debate is inadequate as a characterization of the general debate: the debate is not properly described as turning on whether to employ individual-level or both individual- and social-level explanations in the social sciences.

The problem with both the supervenience- and emergence-based level conception, then, is that individualist explanations are equated with narrow notions of individualist explanations that only permit the description of individuals with their individual-level properties. Since individualist explanations thus conceived qualify as individual-level explanations, it is wrongly taken that the explanatory level condition is satisfied by individualist explanations in general. Once the existence of inclusive notions of individualist explanations is recognized, it becomes clear that the general individualism–holism debate is not apt to be understood in level terms: the formulation fails, to repeat, to capture the inclusive individualist positions that have been advocated by most methodological individualists roughly since the 1950s.

7 Why inclusive individualist positions may not be dismissed from the debate

One obvious way in which to dispute the rejection of the level conception is by maintaining that inclusive individualist positions should be dismissed from the individualism–holism debate: they should not be allowed as participants in the discussion. In that case, it would be perfectly in order to disregard these positions and they would not constitute an impediment to conceiving of the general debate in level terms. I now examine what I take to be the two most significant arguments in support of this approach.

The first purports to show that inclusive individualist positions fail to be truly individualist. In a widely cited paper, Geoffrey Hogdson offers an argument along these lines (Hodgson 2007). Though it targets a different notion of individualist explanations than considered so far, it is worth examining as the claims may be modified to apply to the present discussion.

Hogdson’s argument is directed against methodological individualists who define individualist explanations as ones in terms of individuals and their interrelations (ibid.220). He begins by pointing out that social structures are nothing but sets of relations between individuals. This being the case, he contends,

[t]he broader version, where explanations are supposed to be in terms of individuals and relations between them, turns out to be equivalent to the proposition that explanations of social phenomena should be in terms of both individuals and social structures. There is nothing wrong with this proposition. But why call it methodological individualism, when structures and individuals are given equal status as elements in the explanation? (ibid.223 - my italics).

In response, assume for the sake of argument that social structures are indeed sets of relations between individuals. Even so, Hogdson’s argument may be opposed on the ground that explanations in terms of individuals and their interrelations only contain descriptions of individuals and their interrelations. As such, these explanations do not, say, use institutional terms like “government” or “university” to refer to institutions which Hogdson regards as a special case of social structures (ibid. 220). Instead, the explanations refer to social structures, like governments and universities, by way of descriptions of the interrelated individuals who make up, or are part of, the social structures. The explanations then state how these individuals acted in ways that brought about the event described in the explanandum. This fact, viz. that the explanations only contain descriptions of individuals and their interrelation and no institutional terms like “government” or “university,” makes it reasonable to regard positions that endorse the exclusive use of these explanations as truly individualist stances. Or so methodological individualists of this stripe may argue.

For the purposes of the present discussion, what matters is that Hogdson’s argument may be amended so that it both stirs free of this objection and potentially applies to the inclusive notions of individualist explanations examined in the last section. To this end, it may be changed into the claim that these notions do not deserve to be called individualist since, as elements in explanations, descriptions of individuals are given equal or less status compared to descriptions of non-individual factors. Is that right? Well, methodological individualists who allow individualist explanations to refer to individuals’ social role properties take “descriptions of individuals” to include descriptions of their social role properties. Thus, they would deny that their explanations hereby describe non-individual factors. With respect to inclusive notions that permit individualist explanations to describe individuals’ material and/or institutional context of action, inclusive methodological individualists are likely to acknowledge that these descriptions refer to non-individual factors. Still, they may plausibly claim that, as elements in these explanations, descriptions of individuals are given higher status: The explanations are organized around descriptions of individuals’ actions in the sense that the explanations describe these actions (and possibly the reasons behind them) and only permit the description of material and institutional factors if these formed part of individuals’ context of action and constrained or facilitated their actions. In light of these considerations, the modified version of Hodgson’s argument should be rejected too: inclusive individualist positions qualify as truly individualist.

The second argument contends that inclusive individualist positions should be dismissed from the debate because they obliterate the distinction between methodological individualism and holism. For instance, Udehn gives expression to this sort of view when he remarks that institutional individualism exemplifies “a move, which makes it difficult to continue talking about individualism and holism as opposite doctrines” (Udehn 2001, p. 348).

In reply, consider first the inclusive notion, which allows individualist explanations to refer to individuals’ social role properties. This way of defining individualist explanations does certainly not wipe out their distinctiveness from holist explanations. It means that individualist explanations describe individuals, their actions, etc. including individuals’ role properties whereas holist explanations describe social entities, social processes, statistical properties of groups of individuals, properties of social networks, actions and mental states as ascribed to social entities, etc. Next, turn to the inclusive conception that permits individualist explanations to mention individuals’ material context of action. This move does not render individualist explanations more like holist ones: the latter are rarely, if ever, associated with the description of material factors. Lastly, contemplate the inclusive notion that also countenances the use of institutional terms like “nation,” “hospital,” etc. to describe individuals’ context of action. No doubt, the concern with institutions is a hallmark of holist explanations. Thus, by permitting the employment of institutional terms in individualist explanations, the gap between individualist and holist explanations is reduced. Still, the distinction is far from obliterated. The inclusive notion only permits the description of institutions as part of an account of individuals’ context of action. As such, it remains the exclusive domain of holist explanations to describe institutions only, that is, without any mentioning of individuals and their actions too. An example of such a standard holist explanation might state that the government (an institution) had decided to lower the minimum wage in response to the question of why the unions protested. Further, it remains the case that only holist explanations may describe other social phenomena than institutions such as social processes (i.e. a revolution and economic growth), the statistical properties of groups of individuals (i.e. the crime rate and literacy rate), and properties of social networks (i.e. their density or cohesion). Examples of some of these types of standard holist explanations are an explanation of why the crime had gone up which states that there was an increase in unemployment, or an explanation of why there had been an increase in poverty by stating that it was an effect of the revolution. In this fashion, inclusive notions do not eradicate the distinction between individualist and holist explanations and, by implication, they do not undermine methodological individualism and holism as opposite doctrines. Thus, on a more general note, it may be concluded that there is no basis for holding that individualist positions, which endorse one or several of the inclusive notions of individualist explanations, should be dismissed from the methodological individualism–holism debate.Footnote 11

8 Why it doesn’t help to adopt an alternative ontological level account

Another way in which to salvage the level conception of the individualism–holism debate might be to argue that the supervenience and emergence accounts of the social world should be replaced by an alternative level picture. Relative to a different account that meets the ontological level condition, it might be held, it is possible to satisfy the explanatory level condition.Footnote 12 In this section, I demonstrate that, whatever alternative level account is proposed, this line of defense must be found wanting.

In order to assess whether an alternative level picture of the social world may underwrite the level formulation, it is not necessary to go into any specific suggestions of this sort. Rather, the following more general considerations will do. One option is that the alternative ontological account contends that at least one, but possibly more, of the following phenomena belong to a different level than the individual one: social role properties, material factors, and institutions. Relative to such an account, at least one, but possibly more, of the inclusive notions that allow individualist explanations to describe these phenomena hereby promote the employment of individualist explanations that fail to qualify as individual-level explanations (the explanations do not describe individual-level phenomena only). This being the case, individualist explanations do not in general qualify as individual-level explanations and the explanatory level condition fails.

The other option is that the alternative account stirs free of this problem by holding that social role properties, material factors, and institutions all fit in at the individual level. As a result, inclusive notions of individualist explanations that permit the description of these phenomena do indeed qualify as individual-level explanations (they describe individual-level phenomena only). Yet, at the very least, this sort of proposal runs into difficulties due to its categorization of institutions as individual-level phenomena. It is not clear on what ground institutions may be regarded as individual- rather than social-level phenomena, and what sort of relationship between the individual and social level is compatible with institutions being thus conceived. Further and crucially, the explanatory level condition fails but for a different reason than considered so far. Relative to this account, standard holist explanations that solely describe institutions fail to qualify as social-level explanations. Instead, these explanations qualify as individual-level explanations since institutions are individual-level phenomena by the lights of such an account.

The upshot of these considerations is that however an alternative level picture that meets the ontological level condition is spelled out, it is not going to help in the sense of rendering it possible to meet the explanatory level condition. Thus, relative to a supervenience, an emergence, or any other level account of the social world, the general individualism–holism debate is not apt to be understood in level-terms.Footnote 13

9 Consequences for the ongoing debate

The conclusion that the level formulation is inadequate as a general characterization of the individualism–holism debate obviously means that this way of conceiving of the general dispute should be dropped. Yet, there are also other consequences worth noticing.

One is that the finding opens up for a greater concern with inclusive individualist positions. The level formulation invites two ways of proceeding that involve a failure properly to engage with them. It is instructive briefly to examine these ways and then show how the rejection of the level conception means that there is no longer any basis for proceeding in these manners.

The first natural way of advancing in light of the level formulation is simply to pass over inclusive individualist positions. The positions do not even come into view because the inclusive individualist explanations fail to qualify as individual-level explanations. The assumption that the explanatory level condition may be met stands in the way of noticing them. The second way of proceeding is to misconstrue inclusive individualist positions and then dismiss them. More precisely, inclusive individualist explanations are interpreted as individual-level explanations that describe individual-level phenomena. Accordingly, the positions are ascribed the view that any phenomena referred to by inclusive individualist explanations should be classified as individual-level phenomena (otherwise the explanations fail to refer to individual-level phenomena only). Here is how Brian Epstein describes this expansion of what counts as individual-level phenomena:

This is a common move in social theory: we bend what is meant by “individualistic.” When we speak of the “individualistic level,” we don’t really mean the individualistic level. We mean individual-sized things, whatever they are. Or we mean the microscopic level, the level of whatever parts society is made up of. Or we mean anything that causally interacts with individuals. In short […] we slide from individuals to individual-territories, or else from individuals to things that are not in any respect individualistic (Epstein 2015, p. 48 – italics and bold in the original).

As Epstein also implies, the range of phenomena that are classified as individual-level ones is hereby enlarged beyond what may reasonably be regarded as belonging to this level. Therefore, these implausibly broad conceptions of individual-level phenomena are rejected together with the inclusive notions of individualist explanations that go hand-in-hand with them.Footnote 14

The reason why this line of reasoning involves a misinterpretation of inclusive individualist positions is that their advocates do not, explicitly or implicitly, maintain that individualist explanations are permitted to describe individual-level phenomena only. Likewise, they do not explicitly propose to inflate the category of individual-level phenomena and, pace Epstein, there is no good ground to see them as suggesting this implicitly. As amply illustrated by the foregoing discussion, the most natural reading of inclusive individualist positions is that they allow individualist explanations to describe other factors than individuals with their individualistic properties. Hence, inclusive notions should not be dismissed on the erroneous ground that they are tied in with unacceptably broad conceptions of what constitutes individual-level phenomena.

As noted earlier, it is above all methodological holists, who conceive of the individualism–holism debate in level terms. As such, they are prone to ignore or misconstrue individualist positions that adopt inclusive notions of individualist explanations. They tend to direct their arguments against those positions that subscribe to narrow conceptions of individualist explanations that qualify as individual-level explanations. However, because most current methodological individualists do not subscribe to these narrow notions, methodological holists are mainly combatting strawmen. At the same time, they leave inclusive individualist positions off the hook even though these are presently prevailing. Methodological holists should begin to engage with the inclusive methodological individualist positions; this will make for a much more interesting debate. Most notably, they should critically examine the reasons offered by inclusive methodological individualists in support of their position. Moreover, from the 1980s and onwards, methodological holists have presented a variety of arguments in support of their position [see Zahle (2016) for an overview]. As already intimated, these rely mostly on narrow conceptions of individualist explanations. Methodological holists should also examine to what extent they think these arguments may be put forward if inclusive notions of individualist explanations are presupposed.

The rejection of the level conception of the general individualism–holism dispute also raises the issue of the extent to which reference to levels should continue to figure in the characterization of narrow varieties of the debate. That is, the previous discussion does not show that the level formulation is inadequate when it comes to narrow notions of individualist explanations (that most methodological individualists have not endorsed since the 1950s): these qualify as individual-level explanations. This means that in relation to varieties of the debate that involve narrow individualist explanations, the dispute may be conceived of as turning on whether to use only narrow individual-level explanations (the position of narrow methodological individualists), or both narrow individual- and social-level explanations (the position of methodological holists). Accordingly, proponents of the level conception may still insist that when it comes to narrow varieties of the debate, that is, varieties that involve narrow conceptions of individualist explanations, the debate should be conceived of in level terms.

There are good reasons, however, to oppose even this view. As amply brought out by the foregoing discussion, explanatory levels are, in the context of the individualism–holism debate, taken to be derivative upon ontological levels. As a result, the notions of individual-level and social-level explanations go hand-in-hand with the view that the social world is organized into an individual and a social level. Or differently put, the notions imply an endorsement of the ontological assumption that the social world is levelled. But why saddle the narrow varieties of the debate with this ontological assumption? What is gained by this? After all, the methodological individualism–holism dispute is one about explanation, not ontology. In the absence of convincing answers to these questions, it is preferable to understand all varieties of the dispute in traditional terms, that is, as relying on a distinction between individualist and holist explanations: these notions carry no commitments to the social world being levelled.

In response, proponents of the level formulation may perhaps argue that conceiving of the narrow varieties of the debate in level terms has the advantage of preempting various objections to explanations in terms of social phenomena. For instance, assume that the supervenience picture is taken to underwrite the level formulation of these narrow varieties. In that case, it may be claimed that because nations, schools, etc. supervene on individuals, explanations containing terms like “nation,” “school,” etc. cannot be accused of positing dubious ontological phenomena. Supervenience functions as a “no mystery constraint on social facts” (List and Spiekermann 2013, p. 633).

While this may be granted, note first that, in the individualism–holism debate, there are numerous arguments that do not presuppose any account of the organization of the social world. For instance, methodological individualists and holists also employ arguments that appeal to: diverse conceptions as to what constitutes a satisfactory explanation, different views of explanatory depth, disparate notions of causation, ideas about individual autonomy, and pragmatic concerns.Footnote 15 This being the case, it is not clear why the attending to ontological worries about social phenomena should be privileged by building ontological assumptions into the very conception of narrow varieties of the debate (that is, by conceiving of narrow individualist explanations as individual-level explanations and holist explanations as social-level explanations). To conceive of narrow varieties of the debate in level terms is to exaggerate the relevance of the issue of how the social world is structured for the individualism–holism debate about explanation. Multiple other considerations are, to repeat, equally central to the dispute. In addition, note that even though the supervenience and emergence pictures are the dominant level accounts, they have each met with serious criticism (see footnote 12). The level conception—even when only applied to varieties of the debate that involve narrow notions of individualist explanations—unnecessarily entangle the individualism–holism debate about explanation in an unresolved dispute about the structure of the social world.

In view of these reflections, the claim that at least narrow varieties of the methodological individualism–holism debate should be understood in level terms may be rejected. It is better to go along with the traditional conception of the dispute as involving a distinction between individualist and holist explanations. This does not mean that specific arguments in favor of methodological individualism or holism may not appeal, say, to the supervenience or emergence picture and that no discussion of levels should take place in the context of the dispute. What should be avoided is only the very characterization of the methodological individualism–holism debate as tied up with a commitment to the social world being levelled.Footnote 16

10 Conclusion

In this paper, I have argued against the widespread view that the general methodological individualism–holism debate is apt to be conceived of in level terms. First, I showed that, since the 1950s, the majority of methodological individualists have relied on inclusive notions of individualist explanations that, relative to the two dominant accounts of the social world as leveled (the supervenience and emergence pictures), fail to qualify as individual-level explanations. For this reason, the level formulation is inadequate as a characterization of the general individualism–holism debate. Next, I rejected one way in which to dispute this conclusion, namely by arguing that inclusive individualist positions should be dismissed from the methodological individualism–holism debate. Moreover, I discarded the objection that the level formulation is defensible if underwritten by some alternative account of the social world as levelled. Lastly, I considered the consequences of my rejection of the level conception of the general individualism–holism debate for the ongoing dispute. Perhaps it should be stressed that my discussion should not be taken to question the general importance of analyzing and mapping out the structure of the social world whether in level or other terms. Rather it shows, more modestly, that ontological accounts of the social world as levelled should not be incorporated into the very conception of a classic debate about explanation, viz. the methodological individualism–holism debate.