Replies to Critics

Abstract

There are three prevalent themes among the critics. First, a call to advance the concept of social power by expanding on structural power and demonstrating how telic power can be operationalized. Second, the critics want a stronger objection to the collective intentionality claim (assumed by the standard model). Third, there is a need to clarify the concept of opaque kinds of social facts, especially in relation to economic class. I integrate Khalidi’s clarification of opaqueness, elaborate on the concept of economic class, and introduce a new development of the power view by operationalizing telic power through an empirical case study.

Reply to Ásta

I would like to thank Ásta for her thoughtful and constructive comments. ¹ We agree on key points, such as the use of nonideal theory in social ontology, the importance of the type 1 social phenomena in general, and the need to go beyond deontic powers to account for certain central social phenomena. I also found Ásta’s discussion of what is problematic about idealization in the descriptive case illuminating. The same holds for the discussion of what conferralism implies in the event of a conflict between institutional and communal properties, such as marriage and the “eligible” status example. ²

Ásta’s central point is that I need to show how the power view can accommodate and explain statistical facts, such as those pertaining to the research funding and the Covid cases. ³ With respect to the former, the question is how to operationalize the category of telic power. In other words, can I further elaborate how telic power can be useful in analyzing this and other cases? With respect to the Covid example, Ásta asks for further development of the account of structural power and, I take it, is not entirely convinced that we need categories such as economic class and class systems. The third theme is about the central difference between nonideal and emancipatory social ontology and more generally the role of value, if any, in social ontology. ⁴

I begin with one correction of my view. Ásta notes that I argue “that ideal social ontology cares almost exclusively about the second kind of case and cannot account for the first type.” I do not claim that using the standard model of ideal social ontology means that one cannot accommodate the first kind of social kind, but rather that using it makes these phenomena drop out of sight. I am thus open to the possibility that it can be done, but it remains to be shown how (and it is also worth noting that it has not yet been done). I also argue that “the reflexivity claim and the performativity claim exclude the first kind of social kind from being among the primary and paradigmatic social phenomena because neither reflexivity nor performativity hold for these phenomena” (Burman 2023, 106).

I was intrigued by the two suggestions of how to operationalize telic power. This led me to do new work on that issue with Nemi Pelgrom. ⁵ Ásta writes about the category of telic power: “I would like to be able to operationalize the notion so we can give a clear analysis of the cases.” The case referred to is the research funding case: “A female applicant had to be 2.5 times more productive than the average male applicant to receive the same competence score as he” (Wennerås and Wold 1997, 342). Ásta suggests that it can either be operationalized in a strong way, by building gender into the stereotype of a good researcher, so that there is a necessary connection between gender and being a good or ideal researcher. Alternatively, it can be operationalized in a weaker sense, as a conflict between different norms, so that there is an association between one’s gender and being a good or ideal researcher.

I am sympathetic to both suggestions. First, the definition of telic power is consistent with both. My view is that whether the stronger or weaker connection—or another possible connection—is accurate depends on the case in question. Below, I further develop the research funding case to show this.

The oldest university in Sweden, Uppsala University, was founded in 1477. It took almost 400 years before women were formally allowed to study at Swedish universities (SOU 2011, 79). Before 1870, being a woman necessarily excluded one from being a university student. ⁶ Here we have the stronger connection, with a twist: women are not even being measured against the standard of a good or ideal university student since they could not be university students. ⁷ This point is based on new work by Anna Moltchanova where she also argues that not being measured can be a distinct kind of bias. ⁸

However, in the parliamentary debates about this issue, women were in one sense measured against the standard of an ideal university student—those who favored admitting women into universities often argued that women had intellectual capacities equal to men and thus would be good or even excellent university students. Their opponents denied that women would, or even could, be good or excellent university students because women were presumed not to have the adequate intellectual capacities. Furthermore, the possibility of women becoming university students hinged on perceptions of women and their presumed intellectual capacities. If women were perceived as having equal intellectual capacities as men, they could be good or even excellent university students. If not, they could either not be university students at all or, if allowed to enroll, they would be substandard university students. This case, then, is best understood in terms of the stronger connection (necessary connection) between the ideal and gender that Ásta suggests.

To continue this example from history, a formal requirement to be admitted to a university was to have certain merits in the form of a studentexamen, which basically meant obtaining a degree at the level below university studies. Women were only allowed to get this degree a few years prior to 1870. However, the economic conditions were different for women and men who studied. Men could take part in the state schools, which were free of charge but women could not. Instead, a woman could be a “privatist” who studied independently and then took the exams. This required a certain level of economic means, and few women studied at universities during that period. ⁹

Another argument used by the opponents of allowing women into university was that it could be dangerous for women to study and that doing so would be an obstacle to their fulfilling their role in the private sphere. After 1870, we might then have a conflict between different kinds of norms—participating in the public sphere as a student and fulfilling one’s duties in the private sphere. This and similar cases are best understood in terms of the weaker connection (association) between the ideal and gender: that is, as the conflict between different norms that Ásta suggests.

To turn once again to our example, a new law was passed in 1925 that formally allowed women to hold higher-level government positions and thus to become university professors. That made it possible for women to be evaluated against the standard of an ideal researcher and professor and brings me to a third way of operationalizing telic power, beyond a necessary connection and association.

In a new empirical study “Telic Power and Academic Excellence,” Nemi Pelgrom and I have used the category of telic power to uncover the standard of an ideal researcher in a specific context (Burman and Pelgrom forthcoming). Telic power was used to uncover the actual standard of the ideal researcher used by the reviewers rather than what was stated in job advertisements. For instance, we noted that international experience was emphasized much more in the actual standard than in the formal job description.

This data set consisted of 1,341 external reviewer statements for tenured professorship positions in mathematics, philosophy, and political science at three main universities in Sweden from 2014 to 2024. The reviewers are tasked with evaluating candidates against specific formal criteria that constitute a standard of excellence for research, teaching, and outreach (among other things): that is, a formal standard, or ideal, of academic excellence. Recall that the definition of telic power considers shared ideals, other agents, and perceptions as central components. In the paper, Pelgrom and I analyze whether a shared ideal was used by the external reviewers (the other agents) in practice. So, one way of operationalizing telic power is to uncover actual academic standards and potential gaps between the formal and the actual. We also analyzed how female and male applicants were evaluated in relation to the actual standard: that is, how the external reviewers perceived the different candidates according to gender. Our results show that both genders received a proportional amount of text written about them by external reviewers. Furthermore, only a third of all applicants were women, while women were overrepresented on shortlists. Thus, the women who actually applied appeared to have more positive telic power than the men who applied in the three fields we studied. ¹⁰ In this case, there was no necessary connection or association between gender and academic excellence, at least not the conflict between being an excellent researcher and being a woman as in the original research funding case. In short, a third way to operationalize telic power is to use it in an empirical study that, for example, uncovers the actual standard used in practice and how different genders are evaluated against it.

As to statistical facts and the original research funding case, I agree with Ásta that it is not to be interpreted in terms of a necessary connection.

I take it that Ásta also objects that it is material constraints and enablements, rather than economic class, that explains the Covid case. Consequently, economic class is superfluous.

Is economic class merely a matter of material constraints and enablements? It is helpful to turn to recent sociological work on common understandings of class. Erik Olin Wright offers a typology of different concepts of class and three different approaches to class (Wright 2015). ¹¹

“The first identifies class with the attributes and material conditions of the lives of individuals. The second focuses on the ways in which social positions give some people control over economic resources of various sorts while excluding others from access to those resources. And the third identifies class, above all, with the ways in which economic positions give some people control over the lives and activities of others” (Wright 2015, 3).

Wright refers to the first as “the individual-attributes approach” to class, and it is part of the stratification tradition in sociology. The second is referred to as “the opportunity-hoarding approach,” and it is part of the Weberian tradition. The third is referred to as “the domination and exploitation approach,” and it is part of the Marxist tradition.

None of the three approaches view economic class merely in terms of material constraints and enablements (Wright 2015, 3–10). The individual attributes approach amounts to the idea that class “…is a way of talking about the connection between individual attributes and these material life conditions: class identifies those economically important attributes of people that shape their opportunities and choices in a market economy and thus their material conditions of life” (Wright 2015, 4).

In the book (and the Covid case), I employ the understanding of class as one’s relation to the means of production, but any account of class that does not directly depend on people’s attitudes could serve as an example. All three approaches seem consistent with the power view, but the latter two are more in line with its spirit since they give more weight to power and unequal power relationships in the analysis. I return to the Covid case and the category of structural power in my reply to Muhammad Ali Khalidi’s comments.

Let us turn to Ásta’s comment about descriptive and emancipatory projects. In Nonideal Social Ontology, I write that emancipatory social ontology is a subcategory of nonideal social ontology. Ásta writes that “it is not clear to me why Burman wants to characterize the differences between Brännmark’s and my view on the one hand and Haslanger’s and Jenkins’ on the other as emancipatory or not. A descriptive theory can have emancipatory aims.” This raises the question of identifying the central difference between the two. One core aim of the book is to develop the central differences between ideal and nonideal social ontology, and Ásta is right to point out that there is less detail on how to draw the line between nonideal and emancipatory social ontology. Given the recent developments in the field of social ontology, this is an increasingly important question that requires much more discussion. Here is my sketch of an answer: I agree with the point that a descriptive theory can have emancipatory aims. But the question is what happens in the case of a conflict between the descriptive and the emancipatory. ¹² We can imagine cases where a descriptively correct theory can have non-emancipatory consequences (and the other way around). One way of viewing the difference between descriptive and emancipatory versions of nonideal social ontology is to say that, in case of conflict, the former would prioritize the descriptive aim over the emancipatory, while the latter would elevate the emancipatory over the descriptive.

Consider, for instance, this statement by Brännmark as an example of the descriptive having priority: “We should want a social ontology that is relevant to political theory, not one that has been turned into political theory” (Brännmark 2019, 1062–1063). By contrast, this remark by Jenkins prioritizes the emancipatory: “I write this book primarily because I want to understand race and gender oppression, and I want to understand them because I want to end them. Understanding the ontological aspects of race and gender strikes me as useful given this aim” (Jenkins 2023, Introduction). This interpretation is in line with my power view and Brännmark’s “institutions as distributions” (and I take it conferralism) as descriptive and Jenkins’s as emancipatory.

I take on board Ásta’s interpretation of Haslanger’s ameliorative account of gender and race as descriptive with emancipatory aims. Here, we might distinguish between Haslanger’s broader critical project and the specific ameliorative analyses of gender and race, and it might be an open question what the broader critical project would say in the case of a conflict between the descriptive and the emancipatory: “At the most general level, the task is to develop accounts of gender and race that will be effective tools in the fight against injustice” (Haslanger 2000, 36).

Reply to Arto Laitinen

I want to thank Arto Laitinen for his insightful comments and objections. In writing the book, I was inspired by Laitinen’s talk on power at the Social Ontology 2019 conference in Tampere, and I welcome this opportunity to continue the conversation. The first theme in his comments is about the distinction between ideal and nonideal social ontology and the second is about the concept of power, both core parts of my book.

I begin with how I use the distinction between ideal and nonideal social ontology and how it evolves in the book. I develop it in a three-stage process. I start out with the two-worlds metaphor in which the first world of ideal social ontology is characterized by consensus and the second world of nonideal social ontology is characterized by conflict and contestation. This metaphor captures some key dividing points in the field.

The second step is to use Charles Mills’s distinction between ideal and nonideal theory and show how work in what I call ideal social ontology fulfills most of Mills’s criteria. Note that I use those criteria, such as an idealized social ontology, silence on oppression, and ideal social institutions in the sense of how some institutions (e.g., the family) might systematically disadvantage some of their members. I do not take a stand on the other claims in his article. I also found his criteria helpful as a characterization of the state of the field of social ontology. Interestingly, then, Mills’s distinction has greater applicability than might have previously been noted.

The third and most important step is to develop my own version of the standard model of ideal social ontology. I show that this model is more fundamental than Mills’s distinction since some of the standard model’s features are the same as Mills’s criteria, while other criteria actually fall out by using the standard model. On my view, then, one could not separate ideal and nonideal on the one hand and the standard model and non-standard model on the other. ¹³ Laitinen is using ideal and nonideal social ontology in a different sense in separating between the two dimensions in his matrix. Recall that I develop the standard model by showing how some works in traditional social ontology all share these features, so the real work in my book about the distinction between ideal and nonideal social ontology is my standard model of ideal social ontology. Fulfilling all its features means that one is at one end of the spectrum and that there is a continuum between ideal and nonideal social ontology. This is important in relation to Laitinen’s comments: Given this continuum, the picture (and scoresheet) of Brännmark and me as sharing some features of the standard model is perfectly in order. For example, my power view is an example of descriptive metaphysics.

It is also important to clarify that I do not claim that we should not do ideal social ontology at all; rather, if we want to fulfill the aims of the standard model by providing a general theory of the social world and offering theories that can be the foundation of the social sciences, then we would do better to shift to nonideal social ontology.

However, it is a misrepresentation of the standard model of ideal social ontology to exclude some of the features, as Laitinen does, since all of them are crucial. The category “objects of analysis” in the model refers to what social ontologists take the central, or paradigmatic, social phenomena to be, and it is consistent with these theorists also analyzing other social phenomena that play a less vital role. And what we take those paradigmatic social phenomena to be influences our theorizing to a great extent. So, the category “objects of analysis” with the following three features should be included in the scoresheet:

• Collective intentionality, such as walking together. ¹⁴

• Institutions, such as money and private property.

• Social and institutional facts, such as the fact that Tom is a US citizen.

Laitinen poses a central challenge to my view by claiming that conferralism shares most or even nearly all of the standard model’s features. In response, I use my original standard model of ideal social ontology, including the category of “objects of analysis.” Furthermore, I disagree with some of Laitinen’s interpretations of conferralism, which means that I would give that theory fewer points on the scoresheet and still consider it as an example of nonideal social ontology.

With respect to the objects of analysis, conferralism does not accord with any of the three features. A core contribution of Categories We Live By is to argue for the importance of communal properties of individuals and to analyze those properties. These communal properties are contrasted with institutions and institutional properties; for instance, they differ with respect to the kinds of power they exhibit. Conferralism, then, is not only an account of institutional properties of individuals but broadens the view of social properties to include communal. Furthermore, traditional discussions and examples of collective intentionality are not emphasized in Categories We Live By. Rather, there are many examples about oppression and conflict, such as the central bully example, in the discussion about how communal properties differ from institutional with respect to power.

Additionally, “the conferralist schema” for institutional and communal properties does not rely on an irreducible sense of collective intentionality. I find no other textual evidence that conferralism relies on an irreducible sense of collective intentionality, so I also disagree with the claim that conferralism accepts the collective intentionality claim. Here, the debate between Laitinen and me seems to be largely verbal since we have different interpretations of the collective intentionality claim in mind (with the claim in the standard model of ideal social ontology referring to an irreducible sense).

With respect to the deonticity claim, conferralism states that communal properties of individuals are central and are not deontic powers, and I thus agree with putting parentheses around the deonticity claim.

With respect to the use of generic stylized facts, Ásta uses a mix of generic stylized facts and messier examples from the real world. However, the central example throughout the book—indeed, from its opening pages—is about Rebecca Solnit and Guillermo Gómez-Peña traveling through San Francisco (Ásta 2018a, 1). And the latter kind of examples are most often used for the communal properties of individuals, which is at the core of the book and one of its central innovations. Generic stylized facts thus play a less central role than the messier real-world examples or even the less abstract made-up examples, such as the bully case.

I also disagree with Laitinen’s interpretation of conferralism and the power claim, which states that the enabling and restricting aspects of social and institutional power such as rights and obligations are emphasized as key phenomena. This contrasts with the productive aspects of power: that is, how norms and institutions shape our very preferences. I take it that the applications of conferralism to sex, gender, and other social categories in the latter part of the book are mainly about the productive aspects of power.

In sum, using the original standard model of ideal social ontology, conferralism contains fewer than half of the features of the standard model of ideal social ontology and can plausibly be categorized as an example of nonideal social ontology.

The other core theme is power. There are a number of interesting questions, objections, and new developments. Laitinen raises the metaphysical question about so-called negative powers; there is something metaphysically suspect about talking about powers that do not exist.

When it comes to negative powers, I think our disagreement might be merely terminological. Laitinen discusses my definition of telic power, which refers to positive and negative telic power. ¹⁵ But the definition refers to phenomena that exist; for example, a shared social standard used by other agents to judge a particular person as falling short of the ideal. It does presuppose some baseline; what would the agent have been able to do if she had not been perceived as being substandard? So, it does seem right to me to presuppose a baseline of what would have been the case if this particular social power had not been in place. Thus, I would stick here to referring to social phenomena that can both constrain and enable our actions. One alternative would be to use the terminology of high and low, rather than positive or negative with respect to telic power. I discuss this suggestion in my response to Muhammad Ali Khalidid’s comments. In fact, I find it intriguing that the different forms of power can interact with each other in the ways I suggest in the book (cf. chapter 6).

However, I do not think there is a parallel between my view on power and that metaphysical debate. I do not think I end up mired in those metaphysical troubles. With respect to negative (or low) telic power, there is a negative evaluation that restricts an agent’s abilities, but there are no non-existent entities.

Another central question raised is whether power is simply an ability. If so, are the different forms of social power four different abilities? Or is it rather the case that it is one ability that has four different bases?

Social power is a central concept, and there are a number of approaches one could adopt when analyzing social power. I took some of Barry Barnes’s work as a starting point; he refers to the bases of power and the idea that if we understand those bases better, we can better grasp the nature of power (Barnes 1988). Social ontologists have greatly expanded our knowledge of the bases of power, so if we use institutions, social ideals, and structures, for instance, as some of those bases, we could better understand the nature of power.

I draw a sharp dividing line between physical power, or force, and social power in the sense that I think the power of an engine or tow truck or being able to fly is a fundamentally different concept than social power. ¹⁶ Thus, thinking about social power in terms of physical power can lead us astray. When it comes to social power, I view all social power as an ability and thus talk about agents having power to clearly show that social power can exist even if it is not exercised. In line with the dividing line between physical and social power, it might be helpful to draw a line between abilities that one has in virtue of physical properties on one hand and social properties on the other, or even between physical abilities and social abilities. Not all social powers are abilities since abilities include physical abilities, but all social powers might be social abilities. On the power view, the four forms of social powers are social abilities to effect a specific outcome, and each form of social power has a different basis.

On a related point, it is indeed also grammatically odd to say that I have a negative physical power in the sense that I cannot jump 5 m in the air, while it is not so grammatically odd so to say that I have a negative social power in the sense that I have an institutional obligation to pay taxes.

One reason I refer to power as an ability is that I find it important to make a difference between resources for power (say, money), power as an ability, and the exercise of power. It seems to me that saying that power only exists when it is exercised would be too narrow since one could not then capture domination as a form of power, while power would be too broad if one included resources for power, such as money. After all, one needs to have the ability to use money for it to be transformed into a power. And according to the power view, there are at least four different forms of social power. I would like to think some more about whether these four forms are four different forms of abilities, or one ability with four different bases. I still lean toward the former view.

I would like to thank Laitinen for raising this point and for his other insightful comments and objections that made me look at the book in a new light and to think about future developments of the power view. I want to end on a point of agreement with respect to deontic power. I first clear up a bit of a misunderstanding. I start out with the Searlean notion of deontic power but then shift to Brännmark’s development of deontic power as Hohfeldian incidents as my preferred understanding of deontic power, and this is also the notion I use in analyzing cases (Burman 2023, 210–212). I am thus in agreement with Laitinen here. It is helpful to invoke the Hohfeldian incidents in providing a clearer view of deontic power, and my view on deontic power does that; I apply this more fine-grained of deontic power to some cases as well.

Reply to Muhammad Ali Khalidi

I would like to thank Muhammad Ali Khalidi for the careful treatment of my book, for the suggested improvements of my view, and for the interesting new developments for the research field in general—his discussion of the benefits and drawbacks of the various conceptions of sociality and the discussion of the opaque, semi-transparent, and transparent social kinds.

I take Khalidi’s comments to show agreement with the main themes of my book: that there is and ought to be a paradigm shift from ideal to nonideal social ontology; that power is a central social concept; that it is important to capture class and more generally the first kind of social kind. In other words, class is on a par with what Khalidi refers to as “the big two” of social ontology thus far, race and gender. ¹⁷ His comments also offer a new argument for emphasizing the first kind of social kind: namely, that opaque phenomena might be ontologically primary.

I am deeply indebted to Khalidi’s tripartite distinction between three kinds of social kinds for the argument in my book. In his comments, they are called “opaque,” “semi-transparent,” and “transparent.” Khalidi points to a problem with how I use opaque and transparent kinds of social facts and asks for clarification. He claims that I have not clearly separated three distinctions: opaque vs. transparent, represented vs. not-represented, and overt vs. covert. This is a key point that bears on how I understand the first kind of social kind, which I argue has been overlooked by social ontologists. I agree with Khalidi’s point and want to clarify that I intend the first kind of social facts, called opaque by Khalidi, to be understood in terms of types or tokens that are capable of existing without being represented as existing. Khalidi summarizes the first kind of social kind as “the social kind and its instances can both exist without being conceived or represented as such (opaque).” So, here we have the transparent-opaque distinction at work. Khalidi writes: “opacity means that it is possible for it to exist without being represented.” It is also important to differentiate between an opaque class structure not being known to people in a society and its becoming known to these people, perhaps due to discoveries by social scientists. Applying this to class, one could then say that an opaque class structure has become represented over time.

Khalidi suggests that there are actually two different distinctions at work here; represented vs. not represented and covert vs. overt. Covert kinds are “social kinds that are misrepresented in some way. Thus, covert kinds are ones that are represented, but the concept or representation of the social kind is different from the reality (this is especially evident for kinds like race and gender).”

I agree with Khalidi that these distinctions can cut across each other. To have those distinctions to work with is an improvement that I take on board. The overt-covert distinction is also helpful in accommodating the idea that class structures are often covert in the sense that one might think that a person’s place in society is based on individual merit, whereas parental income levels, for instance, might be more useful for explaining where a person ends up. In general, I think the overt-covert distinction is helpful for debunking projects and what Khalidi has referred to as the big three of social ontology (race, gender, and class). I also found the example and the differences between economic class and the caste system illuminating. A similar example exists between economic and social class.

Clearly separating the three distinctions, understanding the first kind of social kind as opaque, and putting the represented-not represented as well as the overt-covert distinctions to use represent an improvement of my view. These developments are consistent with the central tenets of the book and do not change any of its conclusions.

The second theme is Khalidi’s comments is about the mark of the social and the role of collective intentionality: “But Burman takes issue with this formula, since she does not think that social reality requires collective intentionality at all, based on what she calls the ‘conceivability argument.’” Here, I want to clarify that the conceivability argument is directed at a specific subset of social reality: namely, institutions and institutional facts. Khalidi continues as follows: “Therefore, the place of collective intentionality, at least broadly understood, in social ontology does not seem undermined by Burman’s challenge.”

This could be about at least three separate questions. First, is it about the collective intentionality claim of the standard model? Second, how are we to distinguish between the natural and the social? Third, what categories do we need to accommodate central social phenomena such as class structures? I was engaged in the first and third projects. Here, it is important to distinguish between collective intentionality as it is used in the standard model and the wider debate on collective intentionality. I use collective intentionality in an irreducible sense due to Gilbert, Searle, and Tuomela using it in that sense, with my purpose of building the standard model of ideal social ontology. I argue that the collective intentionality claim, as stated in the standard model, is false. With respect to the third question, to accommodate central social phenomena like a class structure or the illuminating example of caste used by Khalidi, it seems to me that different forms of power rather than collective intentionality (understood in either an irreducible or reducible manner) are more apt to do the job. With respect to economic class, it is also noteworthy that collectivities united by something other than collective intentionality are needed.

With respect to the mark of the social, it is helpful to start with the central unresolved debates at the very foundations of our research field: Which are the social phenomena? What method(s) should we use in answering our research questions? And what conditions of adequacy should a theory fulfill? (Burman, 2023, 6–7).

The unresolved debate of which the social phenomena are, raised early in Ásta’s work (Ásta 2018b) and Khalidi’s comments on my book in this issue, also regards identifying the mark of the social. The objection is that my characterization of social facts is too broad (the bench example). This is a mistake I made in writing the book and is correctly pointed out by both Khalidi and others, such as Kirk Ludwig. What I had in mind was as weak a characterization of the social as possible but one that would still exclude counterexamples of the sort Khalidi raises. This weak characterization would then be used as a basis on which more complex social phenomena depend. One such account is Max Weber’s idea of social, or communal action: “We shall speak of ‘communal action’ when human action, in terms of its subjective meaning, is related to the behaviour of other human beings” (Weber 1913 [2012], 281; italics in original). So, the characterization of a social fact and the social needs to be changed in the book. I offer some initial thoughts on this issue.

Ultimately, this issue comes back to the central unresolved debates. In answering the question of sociality, or what the mark of the social is, what method should we use? What social scientists take the social phenomena to be (with the problems pointed out by Khalidi), conceptual analyses (but it is unclear whether our linguistic intuitions would be helpful), or relying on empirical findings? One might also question whether we have one kind of sociality that comes in degrees or whether there are different kinds of sociality (for humans and other animals)? The third unresolved debate is also applicable to this question; what conditions of adequacy should a theory of sociality fulfill? Thomasson’s conceptual engineering proposal has different conditions of adequacy than Tomasello’s suggestion, for example. I would choose conditions of adequacy stemming from descriptive metaphysics. With respect to the content of the present proposals, there is a clear difference between social ontologists relying on collective intentionality (whether in an irreducible or a reducible sense) as a mark of the social and those who do not. I have not yet taken a stand on this interesting and important debate.

The third theme is about social power. Khalidi poses the following questions:

(1) Spillover power is an interesting form of power, but why is it illustrated solely with cases of unjust uses of power? Can it also be put to positive use, as when a person in a position of power acts as a role model?

(2) Structural power can be either opaque or transparent, but once it is made overt or visible (i.e., correctly represented), does it not then become deontic or telic power? See for example the research results about gender discrimination (Burman 2023, 220–221)—does publishing the research results make it telic power, as in the housewife example?

(3) The characterization of social structures seems to leave their nature somewhat mysterious; it does not really tell us what a social structure is, but rather what it does. Maybe that is as it should be, since it is a purely functional characterization, but can we say anything more about the nature of social structures?

Khalidi correctly points out that I was using only darker examples of spill-over power and suggests brighter ones as well, including positive role models. The definition of spill-over power also works for these kinds of examples, and I welcome this development.

I use social facts as the ontological category in my power view, ¹⁸ so here I would separate the social facts involved and the social powers involved and ask what particular fact or facts we are interested in and what category of power is involved. For example, if one fact about the newly suggested research funding case is that applicants have a new right to appeal decisions, that would be a fact about deontic power. Or we can imagine that the Medical Research Committee put new practices in place, such as evaluating one dimension at a time rather than making holistic evaluations of the candidates, and that this results in the fact that the evaluation of the candidates’ research abilities was in line with actual publications. In that case, women’s positive telic power would have increased. In short, we would need to state what facts we are interested in and then what category those facts belong to. At that point, we can use the taxonomy of social facts to explain the interaction between the different categories (cf. chapter 6).

Regarding the second question, I believe that it depends on what is doing the work. Take the Covid case as an example. If patterns like income level affect survival rates in Covid become overt, explaining why certain persons survived it and others did not, that would still be a case of structural power, while if knowledge about this patterns resulted in political change so that the deontic powers of certain persons increased—say that you were forbidden to drive buses and you had the right to separate living spaces—then deontic power would explain this.

Regarding the third question, I do agree with all commentators that it is important to make further progress on social structures and structural power. One might view the definition of a social structure as a functional definition or as giving the conditions of existence. I am inclined to think of it in terms of the latter, and one could then fill out its nature using different essential definitions of a social structure.

An important part of my definition is that it refers to being systematically restricted or enhanced. Another important aspect is collectivities, so that we can distinguish between random collections of individuals and social groups united by collective intentionality. In addition, my definition does not include an evaluative component and is thus descriptive or analytical. It means that we can have both unjust and just social structures. While I was writing the book, I was inspired by Kate Ritchie’s work on social structures, as well as Sally Haslanger’s with Amie Thomasson’s comments. Some other accounts of the nature of social structures could also fit this definition.

Khalidi also questions the presumed parallel between positive and negative deontic power and positive and negative telic power: “The extent to which a person meets the ideal introduces another dimension to the analysis, which is not found in the concept of deontic power, and it seems as though that aspect of telic power should be theorized not in terms of positive and negative power, but in some other way.” I agree with the point that telic power introduces another dimension to the analysis and thus that the use of positive and negative with respect to telic power might be misleading. One option I was considering in writing the book was using “high” and “low” telic power rather than positive and negative. That would have captured the idea that telic power comes in degrees and show the lack of analogy between deontic power and telic power in a clear way. So, this terminological shift seems like a good option, but it does not change anything substantial in the theory.

I want to thank Muhammad Ali Khalidi once again for these helpful comments. They share some themes with Ayana Samuel’s remarks, which also question how I view the role of collective intentionality and what is ontologically primary. Samuel also convincingly argues that there is a new negative space in social ontology: children as agents.

Reply to Ayana Samuel

I would like to thank Ayana Samuel for the clear and succinct summary of my book. I would especially like to thank her for identifying a new negative space in nearly all of social ontology—the absence of children in theories of the social world. Samuel and I agree on key issues such as the paradigm shift and the importance of forms of power other than the deontic. I focus my reply on her two main critical points—the conceivability argument for the collective intentionality claim being false and the dangers of individualism—and her convincing suggestion to incorporate children as agents into our theorizing about the social world.

There are two clarifications I want to make regarding the summary of my view. The first is that the power view consists partly in a taxonomy of social facts in virtue of social power. It is thus a taxonomy of social facts and not a taxonomy of social power. Note, for instance, the category of “social facts not about social power,” which shows that it is more extensive than just a taxonomy of social power. This category also seems important in accommodating some of the examples of interdependent (presumed) social phenomena to which Samuel draws our attention, such as feeding an infant. ¹⁹ The second clarification is that I argue that a specific formulation of the collective intentionality claim is false. That formulation stems from what the three theories that shaped the field have in common. The collective intentionality claim from the standard model runs as follows: “Collective intentionality is the basic building block of social reality and a necessary condition for the existence of either all institutions or standard institutions.” This will be of relevance for the critical discussion of the conceivability argument to which I now turn.

Samuel does not find my argument for the collective intentionality claim fully convincing. The first reason is that the private property example is too restricted. In other words, even if the more limited statement about legal ownership rests on individual beliefs and common knowledge, it does not show that the general institution of private property does not require collective intentionality. This is a nice point, and I agree with Samuel that I need to offer other examples that are less limited and more convincing. In the book I also use Ylikoski and Mäkelä’s examples of the institutions of money and authority (Burman 2023, 93–94). Both these examples are more general and target the elementary institutions of money and leadership, respectively; thus, they are not vulnerable to the same objection.

Samuel uses Aristotle’s notion of a mixed action in an interesting way in discussing the example of strategic marriage. The objection is that since this it is really an example of a mixed action rather than an involuntary action, collective intentionality still plays a role in it since mixed actions require collective intentionality. Consequently, it does not show that collective intentionality is not required for the institutional facts at play in this case. I think that one would probably need many more details about this case to know what type of action it is, mixed or involuntary, or something else. Still, we can invent an example or look to our actual world to find examples of marriages that are involuntary and thus resting on individual intentions and common knowledge but not collective intentionality (in the irreducible sense).

This discussion leads to the more general question that Samuel raises: “If individual intention and common knowledge are sufficient for institutional facts like marriage, in what sense are such facts social at all?” Here, I would like to emphasize that I do not claim that individual intentions and common knowledge are sufficient for institutional facts like the one that the king and the queen are married. Rather, they are necessary but not sufficient for institutional facts, which can only exist given that there are institutions. If we take institutions to be systems of constitutive rules, then constitutive rules would be another necessary feature (Searle 1995).

This question also points to the problem of my having offered a far too broad characterization of social facts (see my response to Khalidi on this matter). It also relates to the unresolved debate on which are the social phenomena. Samuel draws our attention to a largely neglected area in our theorizing about the social world: children as agents and examples involving children. This discussion offers another set of examples of the social.

I was intrigued by Samuel’s discussion of the place of children in theories of social ontology, or rather the absence of children in theories of social ontology. I take this to be an absolutely central point on another negative space in contemporary social ontology. Samuel draws on another body of philosophical literature to show that although children are engaged in the social world, they are still not able to fulfill the more complex requirements for collective action on which the standard model of ideal social ontology relies. Thus, collective intentionality is not a necessary building block of social reality. Recall that the collective intentionality claim consists of two parts: first, that collective intentionality is the basic building block of social reality; second, that it is a necessary condition for the existence of either all institutions or standard institutions.

I argue against the second part by the conceivability argument and against the first part by stating that if collective intentionality is not necessary for either all or standard institutions, given that these institutions are such central social phenomena, then collective intentionality is not the basic building block of social reality. My interpretation of Samuel’s argument is that it addresses the first part; it offers a new and convincing argument for the claim that collective intentionality is not the basic building block of social reality. However, it does not seem to address the second part of the collective intentionality claim; that is, her argument does not show that collective intentionality is not necessary for either all or standard institutions. The reason is that it seems plausible to say that children below a certain age do not participate in institutions from the first-person point of view (though they are definitely part of many institutions!), but they still participate in less complex social phenomena and thus in social reality.

This discussion comes back to one of the unresolved debates in contemporary social ontology: which are the social phenomena? And which social phenomena do we take to be paradigmatic? What we consider paradigmatic phenomena will greatly influence our theorizing. Annette C. Baier argues, for instance, that if we were to start from the symbiosis between mother and child, “The question ‘How is individuality and individual action possible?’ is just as good a question as the more usual one, ‘How is collective action possible?’” (Baier 1997, 43).

In relation to children’s missing place in theories in social ontology, it is worth mentioning that Amie Thomasson was an early exception to this omission. Her article, “Foundations for a Social Ontology,” is mostly referred to for the two objections she directs at Searle’s early theory—the problems of abstract social objects and of opaque kinds of social facts (Thomasson 2003). Her positive proposal (which is less frequently cited) is to invoke existential rules in addition to constitutive rules to solve these two problems. It is an interesting proposal in its own right and also lets children as agents play a role in theories of social ontology. Thomasson uses the example of children’s play (make-believe) as a parallel to social reality and in introducing existential rules. ²⁰

The central omission that Samuel convincingly points out also relates to Charles Mills’s characterization of ideal theory; one of his criteria is “idealized capacities that amount to individuals having ‘completely unrealistic capacities attributed to them’” (Mills 2005, 168). Children would be still further away from having these idealized capacities than adults. Mills also criticizes the tendency of ideal theorists in political philosophy and ethics to start with the ideal, with the non-ideal phenomena to be dealt with later or not dealt with at all. ²¹ Excluding children as agents and examples incorporating children from our theorizing about the social world risks falling into the same trap. Samuel reminds us about the importance of including this area from the start in our theorizing.

One consequence of giving children their proper place in theories about the social world, according to Samuel, is that holism rather than individualism is the most plausible position. Samuel does not explicitly state whether that she means ontological or methodological individualism or both. Baier’s original article includes both senses. The individualism-holism debate is thorny, extensive, and beyond the scope of the book. I thus restrict my response here to offering some thoughts on the debate over ontological individualism and holism to its relation to the power view. I distinguish between what follows from the theory (the power view) and my own take on the matter. The taxonomy of social facts, including its four different forms of power, is consistent with both views. It is agents having power, and the view is consistent with only individuals having power or groups also having power (interpreted individualistically or holistically). The category of structural power refers to a social structure. One could give an individualistic account of social structures if one were to reduce macro-level phenomena to micro-level phenomena, such as a class structure or gender structure. My own view is that such a reduction is not possible due to the problem of multiple realizability (Burman 2023, 105–106), and I therefore propose an ontologically holistic reading of social structure. In short, the power view gives a framework and then, depending on how one fills out the central components, the specific theory that results is either ontologically holist or individualist.