Causal powers and social ontology

Abstract

Over the last few decades, philosophers and social scientists have applied the so-called powers ontology to the social domain. I argue that this application is highly problematic: many of the alleged powers in the social realm violate the intrinsicality condition, and those that can be coherently taken to be intrinsic to their bearers are arguably causally redundant. I end the paper by offering a diagnosis of why philosophers and social scientists have been tempted to think that there are powers in social realm.

1 Introduction

Since the early 1970s, non-Humean causal powers—often referred to simply as powers—have made a prominent return to philosophy.¹ Initially invoked primarily in the philosophy of physics and chemistry (e.g. Harré 1970, 1997; Harré and Madden 1975; Bhaskar 1975/2008; Cartwright 1983), powers were subsequently put to use in metaphysics (e.g. Shoemaker 1980/1997; Johansson 1989/2004; Ellis 2001; Molnar 2003; Bird 2007; Mumford and Anjum 2011), the philosophy of biology (e.g. Dupré 1993), and, in recent years, the philosophy of social science and social ontology (e.g. Archer 1995; Cartwright 1999; Lawson 1997, 2013; Porpora 2008; Elder-Vass 2010; Groff 2011; Kaidesoja 2013; but see already Bhaskar 1979/1998). Indeed, there is now an interdisciplinary movement, called critical realism, which among other things applies powers to the social realm. In this paper, I question the deployment of powers within the social domain.² I also offer a diagnosis of how and where defenders of that deployment go astray. I do so by distinguishing between true ability predications and powers, between constitutive (or “performative”) abilities and powers, and between deontic powers and causal powers. These distinctions tend to be jumbled up in the literature.

The plan is as follows: I begin by identifying the core features of powers, as they are generally characterized in the literature (Sect. 2). I then look at powers in the social domain, and I argue that they are either contradictory or causally redundant entities (Sect. 3). I end the paper with my diagnosis (or diagnoses) of why philosophers and social scientists have been tempted to think that there are powers in the social realm (Sect. 4).

2 What are powers? A brief recap of their core features

Powers are generally taken to be real—i.e. ontic, non-nominalistic (sometimes called “sparse”, as opposed to “abundant”³)—properties with the following characteristics:

• (a) they are intrinsic to their bearers;⁴

• (b) they have their causal abilities essentially.⁵

Characteristic (a) entails that powers are not extrinsic properties, i.e. properties that are had by objects in virtue of their standing (or not standing) in certain relations to external entities.⁶ Characteristic (b) makes powers distinct from categorical properties. Categorical properties are properties that have their causal abilities (if any) only contingently (e.g. via contingent laws of nature). They are so-called quiddities: properties whose identity is not tied to what they can do. By contrast, a power’s identity is tied to what it can do.

Powers thus understood are usually invoked for two purposes: to defend strong production, as opposed to regularity, views of causation (e.g. Harré and Madden 1975; Bhaskar 1975/2008; Shoemaker 1980/1997; Cartwright 1983; Johansson 1989/2004; Ingthorsson 2002; Molnar 2003; Mumford and Anjum 2011; Marmodoro 2017), and to eliminate or deflate the role of laws of nature in explanations of events and cross-time regularities (Cartwright 1983, 1999; Ellis 2001; Mumford 2004; Chakravartty 2007; Bird 2007; Cartwright and Pemberton 2013). Powers in the social domain are typically postulated to serve both of these functions (see e.g. Elder-Vass 2010 for an instructive overview and discussion).

3 Putative powers in the social domain

Concrete examples of powers at work in the social domain are surprisingly rare in the literature. Defenders of powers in the social realm tend to focus more on abstract theorizing, and on programmatic formulations and defences. Nevertheless, examples are occasionally supplied. The Cambridge economist Tony Lawson, for example, maintains in a recent volume dedicated to Roy Bhaskar (who was a student of Rom Harré’s and co-founder of the modern conception of powers):

My contention is not only that there are indeed causal properties, powers and entities reasonably so identified as social, but also that such features are just as real or objective as those of any other domain, bearing their own irreducible causal powers, justifying and indeed warranting their separate, specialised and relatively autonomous form of scientific study. […] Among the global powers of social systems [e.g. tribes, families, universities, firms, trade unions, political parties, research groups, sports clubs, armies, etc. (p. 297)] are the abilities to stage Olympic Games, wage wars, raise taxes, hold elections, establish international treaties, conduct strikes, form monetary unions, and so on. (Lawson 2013, pp. 286, 298)

And Lawson says about individuals in, or members of, social systems that:

…individuals positioned as members of a particular community become the bearers of emergent positional powers. Thus a police officer can arrest suspects, a judge can pass a sentence, and doctors prescribe drugs. (Lawson 2013, p. 298)

Lawson is not alone in presenting these kinds of example: similar claims can be found scattered in the work of other supporters of a powers-based social ontology.⁷ Lawson’s formulations are particularly succinct, however, and I will therefore attend primarily to them in what follows, referring to additional examples and claims when needed.

Now, the question that should be asked—but which is not asked by proponents of powers in the social realm—is how the postulated powers fare with respect to benchmarks (a) and (b). I shall argue that they fare badly on (a). However, in focusing on (a) I am not suggesting that their compatibility with (b) is unproblematic. Far from it—I think there are problems here, but that these become largely redundant given what I argue below.⁸

So: Are the putative powers mentioned by Lawson intrinsic to their bearers? I think it is evident that a large proportion are not.

Consider to begin with individuals bearing so-called positional powers: the purported powers of policemen, judges and doctors. Clearly—as no doubt Lawson would unhesitatingly agree—it is only relative to a societal context that a certain person, with a certain history, is a policeman, judge or doctor, and has the ability (in the society in question) to arrest suspects, pass sentences, or prescribe drugs. To dramatize: an intrinsic duplicate of the person in question (as the person is at a certain time) existing alone on a remote planet would not have these powers.⁹ It is precisely because these abilities are relative to a societal context (i.e. to the attitudes of external agents, to regulatory structures, and so on) that they fall under the province of social ontology (cf. Searle 1995) and can be called social abilities. If they were not relative to social context, they would be mere individual abilities and not features of social reality. Again, I think Lawson would readily agree with this verdict of extrinsicness: indeed, he is explicitly developing a social ontology (e.g. Lawson 2013, pp. 285, 288 ff), and the powers in question are supposed to be positional powers, i.e. powers attaching to a position within society (ibid, pp. 292–293, 298). But then the obvious objection is that positional powers cannot really be Aristotelian/Harréan/Bhaskarian powers, not as long as they attach to a position the features of which are socially grounded. That is, it must be false, strictly speaking, that “individuals positioned as members of a particular community become the bearers of emergent positional powers” (Lawson 2013, p. 298, my emphasis).

Consider next the putative “global” (or higher-level) powers of social systems. In principle, abilities of groups and other complex social entities can be said to be social in the trivial sense that they are abilities of, or ascribed to, complex entities consisting of many people. Several of the examples here, however, concern abilities that are relative to a larger societal context in the same way that the social abilities of individuals are social, i.e. in the sense of “social” or “positional” referred to in the paragraph above: let us call this extrinsic mode of being social, social^E. (Lawson’s own brief characterization of “social features” does not strictly distinguish between these two modes of sociality: “Primary candidates for features to be classified as social, I take it, are those […] that arise out of, and depend necessarily upon, human interactions (clearly constituting a unique mode of being); those, if any, that could not exist in the absence of human beings and their doings” (Lawson 2013, p. 285). This kind of ambiguity is endemic in current social ontology literature.) For example, it is only relative to a wider societal context that a certain group of people constitutes a government, and has the ability to raise taxes, declare war, or ratify international treatises; and it is only relative to a larger societal context that a firm exists, and has the ability to issue shares, sell products, and hire and dismiss employees; and so on. Thus, the objection mentioned in the previous paragraph generalizes: typically, even the alleged powers of social systems are extrinsic, and hence not powers in the technical sense.¹⁰

One way of rescuing these alleged powers, and showing that they are after all intrinsic properties, is to hold that they are really powers of a larger social system of which the individual or the relevant social system is a proper part. Following this, the powers would be social, not social^E. This adjusted view (or something similar) appears occasionally to be advocated by defenders of powers in the social domain, although not explicitly with the purpose of defending intrinsicality. For example, Lawson—I think incoherently—says of individuals and their alleged positional powers (I suppose he would want to extend the reasoning to positional powers of social sub-systems as well):

…the positional powers are always system properties and individuals remain the agents of these powers only when appropriately positioned and relationally organised as components of the system. Thus when a police officer arrests a suspect, or a judge passes sentence, the powers of arrest or sentencing are those of the community. (Lawson 2013, p. 298, my emphases; see also Elder-Vass 2010, pp. 27–28)

It is important to note, though, that if these powers of individuals and sub-systems really are instantiated at the level of whole communities, then strictly speaking there cannot be any positional powers. For, if the powers are really instantiated by the relevant system as a whole, it cannot truly be said that “individuals positioned as members of a particular community become the bearers of emergent positional powers” and that “individuals remain the agents of these powers” when appropriately positioned and relationally organised (Lawson 2013, p. 298, my emphasis)—unless, of course, people like Lawson want to hold that the relevant powers are instantiated both by communities as wholes and by their relevant sub-components.¹¹ If they do want to defend this last elaboration, the objection from extrinsicality discussed above simply reappears: sub-components can only be bearers or agents of these powers on pain of rendering the powers extrinsic. I take it, then, that the community move can only save the relevant powers as bone fide powers (in all their manifestations) if the powers in question are regarded as strictly emergent properties—that is, as ontic, irreducible properties of complex wholes which are not also instantiated by proper parts of the wholes in question.¹² (Compare with Lawson, who calls the would-be positional powers of individuals within societies “emergent positional powers” (2013, p. 298). It is not crystal clear what concept(s) of emergence Lawson is operating with. One possibility is that he is alluding to Mario Bunge’s unorthodox definition of emergence: “P is an emergent property of a thing b if and only if either b is a complex thing (system) no component of which possesses P, or b is an individual that possesses P by virtue of being a component of a system (i.e. b would not possess P if it were independent or isolated)” (Bunge 1996, p. 20). However, the problem here is that the latter disjunct has application to extrincis properties only; it cannot capture emergence of powers. Only the first disjunct can be of relevance here, strictly speaking).

Postulating strictly emergent properties at the level of communities is deeply problematic, however—even if we disregard the troubling fact that it is unclear what the boundaries of communities are, and hence what the real bearers of these powers are. If a community has strictly emergent causal powers to hold elections, raise taxes, arrest suspects, and pass sentences, then it has, as a whole, causal powers to affect (among other things) its own proper parts and constituents. This type of causation, from wholes to proper parts, is usually referred to as downward causation. Infamously, the notion faces various difficulties. First, it is not clear how downward causation is supposed to work—some have even questioned the very coherence of the notion (see e.g. Kim 1999; Hulswit 2006 for detailed discussions). Secondly, and relatedly, downward causation (assuming it exists) seems in many cases to be a redundant phenomenon given the causal abilities and activities of the whole’s interrelated parts: unless the emergent causal powers of the whole can somehow intervene, and alter the constituents’ interactions, there seems to be nothing left for the higher-level powers to do (see e.g. Merricks 2001, 2005; Bird 2008—this is often referred to as the exclusion argument). Now, I am not aware of a general proof to the effect that there can be no non-redundant downward causation in the social realm (assuming the coherence of the concept as such), but elsewhere (Hansson Wahlberg 2014b) I have argued in some detail, considering various examples from different “levels” of social reality, that there is little reason to believe in such a phenomenon in the social sphere.

Importantly, Lawson himself denies that there is any downward causation from wholes (exerted by their “global powers”) to proper parts in society:¹³

Downward causation is usually defined in terms of an emergent (higher-level) entity having a causal effect by way of causally impacting its own (lower-level) components. But of course, a whole cannot causally impact its own parts; it can only act through them. […] Clearly, if a social totality exhibits powers of a sort not possessed by any of its components [sic], such causal powers nevertheless emerge only through the relational organisation (involving the empowerment [sic]) of its components, and are exercised, as mechanism or process, only through the actions of its relationally organised (human) individual components. An Olympic Games, concert, war or industrial strike cannot be staged other than through the activities of various participants. (Lawson 2013, pp. 287, 298; my emphasis)

Thus, on Lawson’s own view, to account for the unfolding of various processes and events within communities it is in principle enough to postulate suitably interrelated and interacting components (individuals). Emergent causal powers at the level of communities or social systems are not needed: these “global” powers would not do any additional causal work, over and above the causal work done by the suitably interrelated components (see also Searle 2016, pp. 406–407 for a similar complaint).¹⁴

Note that these considerations also apply to mere “humdrum” social powers of social systems (i.e. not only to internalized, prima facie, social^E powers, which are the focus of Lawson’s discussion): any putative, intrinsic social power at the level of a social whole is arguably redundant given the abilities and interactions of the interrelated components. For example, the alleged social power of so-called norm circles (e.g. families and religious communities) to influence, as wholes, the normative beliefs of individual members of the circle (Elder-Vass 2010, Ch. 6) can arguably be accounted for fully by appealing to interactions among the members of the norm circle (see Hansson Wahlberg (2014b) for discussion). There is simply no need to postulate an ontic power at the level of norm circles here. And the same kind of reasoning would seem to apply to alleged social powers that involve affecting entities that are external to the social system in question: for example, the putative power of a group of people to lift a sofa or smash a shop window (cf. Petersson 2007). There is little reason to think that a social system must have an ontic, higher-level causal power if the interrelated members are to be able, collectively, to lift a sofa or smash a shop window.¹⁵

To sum up: Neither individuals nor social systems can have any intrinsic social^E causal powers, because that would be a contradiction in terms. And mere social causal powers of social systems—i.e. ontic causal properties that are instantiated at the level of a social whole but which are not social^E and consequently can coherently be taken to be intrinsic—are arguably causally redundant. Thus, there seems to be no use for Aristotelian/Harréan/Bhaskarian powers in the social realm: they are either contradictory or causally redundant entities.

Before concluding this section I should comment on Lawson’s brief remark that the organisational structure of a social system can exert downward causation. If structures can exert downward causation it might be hoped (as Lawson seems to suggest) that this kind of causation will involve or stem from non-redundant intrinsic social powers, and hence that there is genuine work for powers in the social realm to do after all.

Lawson says:

Along with the emergence of an entity or whole and its [redundant] “global” powers of efficient causation, emerges the entity’s organising structure. […] Confusion arise, because, or where, emergence is used as the criterion of higher or top, in a situation in which (by this criterion) there are at least two higher-level features that might be categorised as higher- or top-level: the emergent whole and the emergent organisational structure. Top-down or downward causation is intelligible in the case of reference to the latter but not the former. (Lawson 2013, pp. 286, 287; see also Lawson 2016b, p. 431)

Introducing “structural” downward causation will not help to save social powers, however. (Unfortunately, in discussing this, I will have to rely to a large extent on my own understanding of what structural causation could involve. Lawson never quite explains how structures could exert any form of downward causation, certainly not of the powers variety. Occasionally, Lawson refers to the causation exerted by organisational structures as “formal causation” (Lawson 2013, p. 287, 2016a, p. 363), apparently alluding to the second of Aristotle’s four causes (Physics, book II, Ch. 3). But “formal causation” is rarely regarded as a genuinely causal concept by modern commentators on Aristotle; it is often described as picking out features of objects that help answering questions of the type “What is it?” (see e.g. Bostock 1999, p. xxvi). Arguably, it is the third of Aristotle’s causes—roughly, efficient causation (“…a producer causes a product and a changer causes a change”, Physics: 194^b29)—which has most in common with modern conceptions of cause (Bostock 1999, p. xxv), certainly of the powers variety (see also Aristotle’s own explicit characterization of power-causation in Metaphysics, Theta, Ch. 5).)

First, causation involving organisational structure—i.e. objects that are interrelated in certain ways—does not as such necessarily involve downward causation. Clearly, causation among objects or individuals must involve “organisational structure” in some form or another. Consider Newton’s law of gravity, according to which the distance between massive objects cannot be ignored. Even if we were to agree that mass has an intrinsic power to attract (and be attracted to) other masses in a mathematically specifiable way (a power that would partly account for the truth of the law-statement, were it true), we would have to concede that the spatial distance between interacting masses affects the magnitude of the gravitational forces created between the bodies involved (cf. Ellis 2001, pp. 137–138; Molnar 2003, pp. 164–165). Likewise, in a society, individuals are manifestly interrelated in various ways (e.g. spatially and deontically), and these relations arguably make a difference to the way people behave (I will return to this below, Sect. 4). But the difference-making abilities of relations within a structure do not necessarily have anything to do with downward causation. The relevant relata (massive objects, individuals, and so on) may very well belong to the same level of reality (assuming here that the notion of reality being stratified into “levels” makes sense); and even if they do not, they do not, as a result, have to be related as parts and wholes to each other.¹⁶ Thus, to suggest that causation involving organisational structure is, as such, a case of downward causation is to invite confusion.

Secondly, and importantly, the relations that partly make up the structure can in any case not themselves be powers in the traditional sense—because, being relations among components, they are not monadic, intrinsic properties. Thus, being a difference maker does not entail being a power. Lawson seems to assume—erroneously—that being a difference maker does entail being a power. He writes, for example: “The organisation or arrangement of the bricks and other components [of a house] makes a difference. And on this criterion of causality, i.e. of possessing the power or ability to make a difference, the relational organisation is causal.” (Lawson 2013, p. 287, my emphasis) Being a difference maker may indeed involve being causal or causally relevant, as discussed above; but it does not need to involve being a power, as Lawson is apparently suggesting. Again, non-powers, such as relations, can “make a difference” by being entities that powers respond to in certain ways.

Thirdly, if it is held that the organisational structure as a whole has emergent causal powers that can be exerted downward, the exclusion argument simply reapplies: all the relevant causation (production, difference making, etc.) appears to be going on within the structure—adding powers that act, so to speak, “holistically” or from the “top” of the structure seems therefore to be a straightforward violation of Occam’s Razor.

Let me end this section by conceding that there is a regress argument looming here: if we reiterate the exclusion argument over and over again it seems that all causal powers drain away, except for those that exist at the level of elementary particles (if there is such a bottom level, see Schaffer 2003), which prima facie is an absurd consequence (see Block 2003; Elder-Vass 2012, 2014). However, on the assumption that there are powers at all, I do not find it evident that such a highly sparse version of the powers ontology is necessarily untenable (see e.g. Heil 2012; Bird 2016; Hansson Wahlberg 2014c for discussion). For example, I think the exclusion argument could be applied with equal force to the putative powers of many of Cartwright and Pemberton’s so-called “nomological machines”, allegedly operating at the physical meso/macro levels: their example of a toilet cistern (Cartwright and Pemberton 2013, pp. 99, 103) is an obvious case in point (see also Elder-Vass’s 2012, p. 84 discussion of a torch). On the other hand, it might very well be that the regress stops before the fundamental level is reached, perhaps at the level of biological organisms or minds (assuming that downward causation as such is a coherent notion). The case has been made for the thesis that organisms/minds have emergent causal powers that defy the exclusion argument (see e.g. O’Connor 1994; O’Connor and Jacobs 2003; Dupré 1993; O’Connor and Wong 2005; Merricks 2001; Bird 2008; Lowe 2013; Harré 2013). Others have argued that the exclusion argument does not apply to chemical compounds (e.g. Needham 2009; Hendry 2017). And see Schaffer (2007, pp. 184–187) for a relevant discussion of emergence in the quantum realm (see also Calosi and Morganti 2016). However, I do not wish to evaluate these highly controversial claims about those specific domains here. In this paper, I am content to restrict my attention to the social domain; and I argue that powers in the social realm (be they social^E or social) are highly problematic entities we have little reason to believe in and should not postulate.

4 Diagnosis

If there is precious little room for powers in the social realm, why have philosophers and social scientists been inclined to postulate them? Here, I shall suggest three such reasons. (I should acknowledge that this section is to some extent speculative: it is hard to ascertain for sure what thought processes lie behind the problematic endorsement of social/social^E powers. Thanks to an anonymous reviewer here.)

First, talk about “abilities”, “capabilities”, “capacities” and “dispositions” may mistakenly be taken to entail, or analytically involve, ascriptions of ontic powers. Recall that Lawson expressed himself this way:

Among the global powers of social systems are the abilities to stage Olympic Games, wage wars, raise taxes, hold elections, establish international treaties, conduct strikes, form monetary unions, and so on. […] a police officer can arrest suspects, a judge can pass a sentence, and doctors prescribe drugs. (Lawson, 2013, p. 298, my emphases)

But statements ostensibly ascribing modal properties to objects or persons need not have truth conditions that involve the existence of ontic powers—at least, not if the relevant modally loaded expressions are used in an ordinary language, rather than technical, sense (and this goes for the very word “power” as well, I think) (cf. footnote 3, above).

Arguably, the truth conditions of day-to-day statements ascribing abilities and dispositions to objects are generally, or often, merely of the following conditional form: if such and such were to happen, then such and such would (in indetermistic contexts: might) happen (cf. Ryle 1949/2002, pp. 43, 123; Mackie 1973, pp. 126–128; Lewis 1973, p. 38; Harré 2013, pp. 131, 138). The truth-makers for such loose statements may be a variety of things, including events in possible worlds (Lewis 1973), actual states of affairs together with laws of nature (Armstrong 1997), macro powers of interacting objects (Mumford and Anjum 2011), and powers of the relevant entities’ suitably interrelated constituents (Heil 2012; Harré 2013; Bird 2016). Consequently, loose ordinary language statements about abilities and dispositions are not, as such, ontologically committal, although they may seem to be, due to their grammatical form. In particular, they do not express a specific commitment to intrinsic powers inhering in the grammatical subjects of the ascriptions; and they certainly do not express a commitment to there being intrinsic powers instantiated at the same ontological level as the grammatical subjects.¹⁷ Accordingly, such statements are compatible with the relevant truth-makers consisting of lower-level objects and features, or objects and features that are external to the prima facie subject of the ability ascription: as just indicated, the truth-makers may involve laws of nature, external states of affairs and objects (possibly with intrinsic powers), and so on. Importantly—and typically this is not highlighted in the literature on disposition/ability ascriptions—the external factors may also include legal regulations, social–historical circumstances, and external individuals with certain attitudes towards the subject of the ascription. Thus, in this relaxed sense there can clearly be “extrinsic dispositions”, “extrinsic abilities” or even “extrinsic powers”: i.e. there can be true ascriptions of disposition/ability/power predicates which are made true by factors external to the grammatical subject. But the idea that there are extrinsic dispositions/abilities/powers in this relaxed or “abundant” sense does not entail that there are extrinsic powers in the technical, ontic, sparse sense. [Cf. Bird’s (2007, pp. 29–31, 125) discussion of McKitrick’s (2003).]

It is true that this simple conditional analysis of disposition/ability ascriptions has been called into question in connection with scenarios involving masks, finks and antidotes (e.g. Martin 1994), and that it has been concluded, in the light of such scenarios, that such ascriptions should after all be taken to involve the postulation of intrinsic powers (ibid., Molnar 1999). However, I think it is plausible to assume that the ascriptions are implicitly associated with some qualifying “ceteris paribus”, “in ideal conditions” or “absent intervening factors” clause (cf. Hansson Wahlberg 2009). That would arguably save the conditional analysis from the alleged counter-examples. Besides, as I have highlighted elsewhere (Hansson 2006), the ontic powers interpretation of disposition ascriptions faces a serious problem of its own which the conditional analysis escapes: the powers account cannot explain certain entailment relations which intuitively connect some ability and disposition ascriptions, and these relations can be straightforwardly explained within the conditional analysis. For example: if “sugar is water-soluble” (or the proposition expressed by the sentence, or the statement made when the sentence is tokened) is true, intuitively it follows that “water is a solvent for sugar” is true, and vice versa. On the conditional analysis, such entailments can be explained easily: the relevant ability ascriptions have the same truth condition, namely: if some sugar were put in water, it would dissolve. (Or, to use an example inspired by Lawson’s putative positional abilities: If doctors can (or have the ability to) prescribe drugs to patients, then intuitively, patients can (or have the ability to) be prescribed drugs by doctors. An ordinary language statement expressing either sort of ability seems to have a truth condition with roughly the following conditional structure: if a doctor were to fill in and sign a document of certain kind, then the patient referred to in the document would be prescribed a drug (of the kind specified in the document in question).) But notice that on the powers semantics, the entailment does not hold, because on that understanding “water is a solvent for sugar” can be true without “sugar is water-soluble” being true, and vice versa. On the powers semantics, the ascriptions have distinct truth conditions involving distinct powers inhering in distinct objects, and one of the subjects may very well lack the relevant ontic power.¹⁸

Thus, all in all, I think the standard, ontologically non-committal, conditional analysis of ordinary ability and disposition statements has a lot going for it. An obvious benefit is that this analysis allows ability ascriptions to the social realm to be true—ascriptions that would be false, or at least highly problematic, on the intrinsic powers account. This notwithstanding, many social scientists and philosophers are, it seems, implicitly adopting the powers reading of such ascriptions. (A notable exception is Lindahl and Reidhav 2017, p. 162 who adopt a conditional analysis). It may be that at least some of them are misled by surface grammar, and perhaps, in some cases, by certain philosophical (and in my view erroneous) doctrines relating to the interpretation of ability/dispositional statements.¹⁹

Secondly, some defenders of powers in the social realm seem to have conflated socially^E grounded “constitutive” abilities (an alternative name would be “performative” abilities, cf. Austin 1962) with causal powers. For example, recall that Lawson mentioned, as examples of powers, abilities of social systems to raise taxes, establish international treaties, and monetary unions, but also abilities of individual persons (police officers, judges and doctors) to arrest suspects, pass sentences, and prescribe drugs (for similar examples, see Harré 1970, p. 92; Harré and Madden 1975, p. 95; Elder-Vass 2010, pp. 73, 173, 199, 2012, pp. 87, 89). Many of these abilities are, I think, first and foremost constitutive in nature (cf. Searle 1969, 1995; Goldman 1970; Smith 2014). Consider, for example, a national governmental body with the ability (the putative power) to establish international treaties. By signing a certain document, the governmental body does not cause the establishment of the relevant international treaty. Rather, the act, in the relevant context (along with other acts such as the signing of the document by other appropriately authorised parties), is or constitutes the establishment of the treaty. Of course, the actual writing process is (at some level) a physical, causal process; but this process (or rather the end stage of it), which more or less anyone who can write can reproduce, non-casually constitutes the establishment of the treaty in the context in question. As John Searle would put it, in the societal context in question, the act counts—in line with his general formula for constitutive rules: X counts as Y in context C (Searle 1969, pp. 34–35; 1995, pp. 43–44)²⁰—as the establishment of the treaty (Alvin Goldman, 1970, pp. 25–26, would say that the act involves “conventional generation”).²¹

Likewise, if a doctor fills in a form to prescribe a drug, she does not thereby cause the prescription of the drug. The act in question is or constitutes the prescription of the drug. And when a policeman grabs a suspect while declaring “You’re under arrest!”, the act does not cause the arrest—it is or constitutes the arrest. The same holds, I would argue, for many of the other putative powers, such as the ability of a governmental body to raise taxes, the ability of a university to assign grades, the ability of a judge to pass sentences, the ability of a manager to fire an employee, and so on.

That we are not dealing with causal powers here is indicated—apart from the obvious extrinsic, conventional grounding of the abilities—by the fact that the relevant “acts” and their putative “effects”, were they genuinely causal, would in many cases violate the actual speed-limit to causal propagation, i.e. the speed of light (see e.g. Salmon 1984, p. 141).²² Consider, for example, the act of grading. When a student is assigned a grade, the student acquires the grade (i.e. the “property” of having such and such a grade, the putative “effect”) as soon as the official grading (the putative “cause”, fundamentally, the inscribing of some marks on a piece of paper, or something similar) has occurred, irrespective of the student’s whereabouts. The student may have travelled to the other side of the earth, and still “acquire” the grade at the relevent time. Similar situations obtain when a criminal is sentenced (although in such a case the distance will typically be shorter), an employee is fired, an academic is promoted to professor, a citizen’s taxes are raised, and so on. Better, I think, to conceive of such putative “effects” as instances of non-causal, so-called “mere Cambridge changes”: a predicate (expressing an abundant extrinsic “property”) begins to apply to a certain person (or a group of people) at a certain time simply in virtue of changes or events that take place elsewhere at that time (see e.g. Kim 1974; Shoemaker 1980/1997; Mellor 1998, pp. 87–88). In the case of grading: a physical act that counts as an official grading of person x occurs at place p and time t, whereby a predicate of the form “x has acquired grade Φ” becomes true of x (i.e., it becomes an institutional truth or fact that x has acquired grade Φ at t), although x exists at distinct place p′ at time t.

Thus, ascriptions of such abilities (to establish international treaties, raise taxes, assign grades, arrest suspects, pass sentences, and so on) to social entities and individuals ought not, as such, to be taken as attributions of causal powers (beyond the required, underlying physical powers). Again, it seems more appropriate to understand these ability ascriptions in line with the conditional analysis. For example, to say that a judge x has an ability to pass a sentence, is to say, roughly, that if x were to do y, a sentence would be passed. No social^E causal power is hereby ascribed to the judge (which would be a contradiction in terms), only a “constitutive” ability, which is held relative to a certain societal context in which a judge’s doing y counts, by convention, as a sentence being passed.

Thirdly, and relatedly, some defenders of powers in the social realm seem occasionally to illicitly equate causal powers with so-called deontic powers (sometimes labelled positive and negative conventional powers, or, in legal literature, legal powers and liabilities)—that is, with rights, privileges, entitlements, obligations, responsibilities, duties, etc. (see e.g. Searle 1995, p. 100, 2010, pp. 8–9; Hohfeld 1913). Consider Lawson again, who says:

…an individual allocated to the position university professor, acquires the social/positional identity of (is accepted within a community as possessing the status of) university professor; and so on. […] a position is essentially a locus of a set of specific rights and obligations, where the accepted position occupants are agents or bearers of these rights and obligations and typically possess a status or identity associated with them. But any given position is always constituted in relation to others. And the rights of individuals in one group over individuals in another are matched by obligations of the latter group members with respect to the former. […] If positional rights and obligations ultimately relate to ways in which certain positioned individuals can influence the behaviour of others, it follows that rights and obligations are in effect positional powers, respectively positive and negative powers. For the agents of rights (positive powers) have the causal capacity intentionally to get others, the subjects of those rights (those with relevant obligations, or negative powers), to do something whether the latter want to do that something or not. (Lawson 2013, pp. 292–293, emphases original; see also Elder-Vass 2010, pp. 153, 159)

However, as Searle stresses, deontic powers are extrinsic (1995, pp. 9–13, 100–101, 104–112), while causal powers are, as we have seen, intrinsic. Deontic powers are generally assigned to persons (or groups or other entities) via laws or other regulations, or simply through people’s acceptance of the assignment. Causal powers, on the other hand, are had by objects in virtue of their intrinsic nature (e.g. Harré 1970, p. 85; Bhaskar 1975/2008, p. 231; see fn. 4 for further references). For this reason, Searlean deontic powers and Aristotelian/Harréan/Bhaskarian causal powers cannot be equated.²³

If Lawson and likeminded people nevertheless want to insist that deontic powers are a species of intrinsic causal power, they will have to adopt the community move involving strictly emergent properties discussed in Sect. 3. However, that would make whole communities the bearers of the relevant “negative” and “positive” powers, and not the individuals/sub-systems therein. I do not think much sense can be made of such a proposal. For one thing, as Lawson himself stresses, “the rights of individuals in one group over individuals in another are matched by obligations of the latter group members with respect to the former” (2013, pp. 292–293, see also Hohfeld 1913, p. 30). This characteristic and crucial feature of deontic powers will be lost if only whole communities are agents of deontic powers. That is, it will no longer be strictly true to say that:

If university teachers have the right to set exams, students have the obligation to sit them, just as students have the right to expect the exams to be marked, and fairly, and teachers have an obligation to undertake this. (Lawson 2013, p. 293)

The community move for equating deontic and causal powers will also have to involve downward causation, and I have argued that causation of this kind is redundant unless the emergent powers of communities can somehow causally intervene and produce new behaviours of the interrelated individuals in the society in question—an idea which seems highly questionable. Finally, the community move in any case faces the obvious objection that deontic and causal powers clearly seem to be distinct inasmuch as deontic powers are normative while causal powers appear to be utterly devoid of normativity. All in all, then, I think deontic and causal powers should be kept distinct.

My denial that deontic powers are causal powers does not entail that the former—apart from their enabling constitutive abilities²⁴—have no causal relevance in a society, i.e. make no causal difference at all. My argument is simply that being causally relevant does not necessarily mean being a power, not even on a powers metaphysics. Recall the discussion of Newtonian gravitation in Sect. 3: we saw there that, even if massive objects have a power to attract one another,²⁵ the magnitude of the ensuing accelerations (absent disturbing factors) also depends on how far apart the powerful objects are. Hence, their distance is, in some sense, causally relevant: it makes a difference to how strongly the mass-powers interact. But again, spatial separation need not itself be a power to make such a difference. (Indeed, distance is a relation, and is typically not regarded as a power in the literature; see e.g. Ellis 2001, pp.137–138.) Plausibly, this kind of causal relevance or difference making can be expressed by a counterfactual: had the distance differed, the resulting gravitational force on the objects would have been different.²⁶ However, the truth-maker for this counterfactual could simply involve the nature of the relevant mass-powers: how they react to increasing/diminishing distance (cf. Chakravartty 2007, pp. 144–147). The situation may be similar, I suggest, with rights and obligations (within obvious limits!). That is, people may be inclined—perhaps in virtue of their intrinsic ontic causal powers, if they instantiate such properties—to react to rights and obligations in certain ways in certain contexts without these rights and obligations themselves being powers. (In a contrast with spatial relations, rights and obligations appear to be even less ontologically robust and objective, being mere institutional or conventional “entities” which are existentially dependent on human actions and attitudes.) Thus, without being causal powers, deontic powers may nevertheless be “causally relevant” in the sense that they “make a difference” to how people behave, plausibly indirectly by being represented and taken into account in people’s decision making: were the deontic powers distributed differently, people would (generally) make other decisions and act differently (cf. Searle 2001). But again, the truth of this counterfactual does not entail that deontic powers are causal powers.²⁷

5 Conclusion

Philosophers and social scientists have become increasingly eager to apply the powers ontology to the social domain. I have argued that this application is misguided—powers in the social realm are either contradictory or redundant entities—and is based on a number of mistakes and conflations. Alleged social causation had better be conceptualized in other terms.