The function of solidarity and its normative implications

ABSTRACT Many lament that solidarity is declining, implying there is something good about it; but what is solidarity and why should we want it? Here, we defend an original functionalist re-interpretation of solidarity. Political solidarity plays a key functional role in a polity’s persistence through time. Thus, we should want institutions that foster solidarity. This paper is divided into three parts. In the first, we draw on the philosophy of biology to pinpoint what counts as a proper function, in a way that is naturalistic, objective, and selective. On this aetiological account, a sharp distinction between functional needs (e.g. the pumping of blood) and functional mechanisms (e.g. the valve that pumps blood) is drawn. In the second part of the paper, we propose that solidarity should be understood as an aetiological function of society. This new conception sheds light on the widely acknowledged, yet seldom clarified connection between two common readings of solidarity: solidarity as a set of feelings of mutual kinship (its functional need), and solidarity as a set of redistributive institutions (a key functional mechanism). The third part concludes that this new functional conception of solidarity provides normative reasons to foster solidarity.


Introduction
To 1 the crises affecting contemporary societies, many respond by invoking increased 'solidarity'. During the 2020 pandemic crisis, political actors called for unity and sincere efforts to cooperate in fighting COVID-19 (Prainsack 2020). Likewise, in the 2007 economic crisis, several appeals for a conjoint response echoed throughout the public sphere (Ferrera 2014). 2 In these cases, lack of solidarity at the political level is often mourned as a deplorable situation to be averted. Political solidarity seems to be a normative value, something that we should want even if we actually do not. Yet, both the meaning and the value of solidarity remain a matter of considerable academic debate (Brunkhorst 2005;Kapeller and Wolkenstein 2013;Stjernø 2009). This article intends to tackle precisely these two questions, thereby bridging the lamented gap between solidarity's nature and its normative value (Voland 1999, 170). In this article, we propose that solidarity ought to be understood in functionalist terms, and that it is valuable because it discharges the crucial function of societal cohesion. So long as we care about the survival of our society, we should want institutions in place that maintain and do not undermine solidarity feelings among its citizens.

The paper is divided into three sections
In the first part, we qualify what counts as a function and whether such a notion can be usefully applied to the social world. To do this, we draw on the aetiological account, which views functions not merely as features of things, but as special properties that explain the persistence of the function bearer by reference to its history (Wright 1976, 81). Hearts both pump blood and emit thumping sounds, but only blood pumping is a 'proper function' (Millikan 1989) because it explains hearts' persistence through natural selection; the sound is merely a by-product.
In the second part, we argue that solidarity can be conceived as a crucial aetiological function of any political community. We start by summarizing current scholarly disagreement on the notion of solidarity, and go on to revive Durkheim's functionalism as a promising point. By innovatively recasting Durkheim's intuition in aetiological terms, we provide a persuasive answer to the usual critiques against functional explanations.
In the final part of the paper, we show that solidarity is a normative concept: it is a desirable feature for societies and one they may in fact lack. Here, we first argue that functions have normative implications (Hardcastle 2002): pumping blood is not something all hearts do (defective hearts do not), but rather something that all hearts should do. A heart that fails to pump blood is a bad heart in this functional sense, and we have reasons to prefer good hearts (Neander 1999). We then conclude that similar logic can be applied to solidarity, if conceived as an aetiological function. Solidarity's functional benefit provides important reasons to lament its absence and justify its presence in modern societies.
Before we proceed, a small clarification is required. This paper is concerned with political solidarity, understood as the willingness to accept the costs that social cooperation entails (Banting and Kymlicka 2017, 1) at the political level (Habermas 2013). Other scholars are instead interested in how solidarity operates as a constituent of social groups (Blum 2007;Salmela 2014;Shelby 2002;Tuomela 2013). While we think our functional argument may apply to social groups as well, here we confine ourselves to the political level. On our account, solidarity is in fact partial: that is, encompassing the needs and channelling the forces of a particular group or social entity. The opposition being intrinsic to solidarity of a 'we' to a 'they', which Kolers (2016) has qualified as the 'agonistic' trait, is widely represented in literature (Heyd 2015). In this paper, we prioritize the discussion of the activity of solidarity at the macro-level, where it is a more precarious and onerous systemic resource to maintain, and the ultimate stake, i.e., the social order, is higher. However, we allow the possibility -that cannot here be explored in more detail -that small-scale social groups can also be dependent on solidarity not dissimilarly from societies. For example, a small protest group must foster solidarity to ensure group cohesion and successfully promote its cause, although its activities might, in so doing, endanger the solidarity mechanisms operating at a wider level. It might then be the case that small-group solidarity is detrimental to a largegroup or societal solidarity.

What functions are
While functionalism was a historically influential paradigm in the social (Parsons 1991) and political sciences (Easton 1979), it has been famously criticized on epistemological grounds (Elster 1982). Recently, however, interest has grown in functional explanations, both in political science (Genschel and Jachtenfuchs 2021;Hooghe and Marks 2009) and political theory (Ceva and Ferretti 2021;Erman 2020). Nonetheless, few in these recent revivals specify exactly what they mean by 'function'.
We often think the function of a pen is writing, the function of the heart is pumping blood, and the function of the parliament is voting laws. Not everything one may wish to do with an object counts as its function. For example, I may want to use a pen as a blowgun for small pieces of paper. In addition, not any property that is regularly exhibited by a certain system counts as a function. For example, hearts are usually of a reddish colour, and tend to emit thumping sounds.
We therefore need an account of functions that is selective. Within philosophy of biology, the aetiological view 3 attempted to identify some criteria to check whether some feature is a 'proper function' (Millikan 1989) or merely an accidental property, like the thumping in hearts. As its name suggests, the aetiological account assigns functions on the basis of the previous causal history. Two conditions are jointly needed: a tendency (a), and a feedback (b). Consider as an example a human artefact: Heating food is the function of microwave ovens if and only if: (a) Microwaves tend to heat food (b) Heating food contributed causally to the persistence of microwaves.
Obviously, in order for something to be a function, it needs to be generally displayed (a). Yet this condition is only necessary, not sufficient. The original contribution of this approach lies in condition (b). Microwaves exist in this world because they were designed and built with the goal of heating food. Persistence describes the reason they spread and persist despite potential adversities (Queloz 2020, 211) (more on this later).
A key advantage of the aetiological view of function is that it is naturalistic: even when applied to natural functions it does not contradict our scientific view of the world: Pumping blood is a function of hearts if and only if: (a) Hearts tend to pump blood (b) Pumping blood contributed causally to the persistence of hearts.
Condition (b) can be accounted for by the causal mechanism of natural selection. There is then nothing mysterious to attributing to the pumping blood of the heart the special status of a function, and then concluding it is significantly more important than other properties of this natural object, such as emitting thumping sounds or being reddish. The pumping of the blood played a causal role that allowed the organism to survive, reproduce, and thrive. The other features did not. Again, it is worth noting that persistence does not refer to the appearance of hearts: the very first mutation. The first appearance is random and might be functional or not. What is crucial for functional attributions is how a new trait is selected: why it spreads and persists through competitive pressure. A second important advantage of the aetiological account is that it grounds objective functions. Some believe functions are arbitrarily assigned by observers, depending on the goal they happen to have (e.g. Searle 1995). Let us call this the instrumental view of functions: The (instrumental) function of pens is writing if and only if: (a) I want to write (b) Pens are a means to write Under this alternative interpretation, functions are clearly epistemically dubious, because they lack the minimal intersubjective reliability of scientific explanations. Different people may have different uses for a pen; thus, the very concept of function becomes almost useless. Instrumental views of functions are notoriously vulnerable to such criticism. Let us return to the pen: (a) I want to shoot bits of papers at my classmates (b) Pens are a means to shoot bits of papers (c) Therefore, the function of pens is shooting bits of papers.
Note that even if both (a) and (b) are true, many would still want to resist (c). This is because the instrumental view does not adequately track our intuitions about functions. That functions cannot be reduced to goals is quite obvious in that unconscious functions operate without anyone in a society knowing. The possible discrepancy between proper social functions and individual intentions was famously accounted for by Robert Merton (1968, 114-123) in terms of the distinction between manifest and latent functions. In Merton's terms, this distinction is intended to 'preclude the inadvertent confusion, often found in the sociological literature, between conscious motivations for social behaviour and its objective [i.e., unintended and/or unrecognized] consequences' (Merton 1968, 114). For instance, regarding the question 'Why do the Hopi dance for rain?' Merton replies that despite the manifest function of the Hopi's rain dance being to produce abundant precipitation, its latent function is to promote group solidarity (Merton 1968, 118). Most importantly for our purposes, social functions are not dependent on individual intentions to persist: the former can operate even in absence of the latter being concertedly oriented to the proper activity ensured by that function.
The aetiological account thus provides more robust criteria that allow the notion of function to yield objective functional attributions.
(a) Pens tend to write (b) Writing contributed causally to the persistence of pens Therefore, the function of pens is writing Such a scheme allows us to say that the function of a pen is not being used as a blowgun: that is not the activity for which it was created (b). This allows for much more reliable explanations: for example, I can expect that pens are widely used to write, while only very rarely used as blowguns. I can even expect people to refill a pen's ink or jettison the whole thing when it fails to write. However, I cannot expect people generally to finetune a pen's accuracy as a blowgun, e.g., by removing the ink cartridge.
In sum, this aetiological view of function is selective because not every property is a function, and not everything we wish something to do is its function. It is naturalistic because it extends to natural functions thanks to natural selection. Moreover, it is also objective, insofar as one can be mistaken in identifying a function and external references are provided to establish the truth.

Social functions and their critics
The aetiological account of functions is applicable to social functions as well: Exerting institutional coercion is a function of armies because: (a) Armies tend to exert institutional coercion (b) Exerting institutional coercion contributed causally to the persistence of armies As is the case for artificial and natural functions, (a) seems unproblematically true. However, (b) is more difficult to ascertain for armies than for hearts or microwaves. Human manufacts have simple causal stories, referring to human intention. None would contest that heating food is why we have microwaves. Natural functions can draw on natural selection. Few would object that the reason hearts are widespread is that they evolved to pump blood. The causal story required by social functions, in comparison, is difficult to reconstruct with the same degree of epistemic confidence. However, the aetiological account still provides two conditions as external references that one may look for. Even if it does not often happen in practice, giving an objective answer is possible. The aetiological account of function prevents two widely used objections: (1) ex-post causation, and (2) missing causal mechanism.
A classic critique of functional explanation is that they rely on an epistemologically dubious ex-post causation (1), i.e., the cause seems to come after the effect takes place. If there are no hearts before the pumping of blood is performed, how can the latter explain the former? Similarly, how can the pen's ability to write explain the pen's persistence, if the ability to write comes after the pen?
This objection is not particularly serious for artificial functions. For artefacts, some prior human intention acts as causal trigger. In other words, the pen was fashioned by someone who wanted to create a writing tool. Biological functions rely instead on natural evolution to do the explaining at the microlevel. The heart was selected based on its ability to pump blood. This is not ex-post causation because, as previously discussed, it is not meant to explain the first appearance of the phenomenon, but rather its spread and persistence over time. What explains the appearance of the heart is random mutations, but what explains why hearts stick around and multiply is their function. The aetiological account of function is powerful precisely because it does not need an a priori intent to qualify something as a function. Instead, it is agnostic about how functions emerge for the first time. What qualifies something as a function is not how it came to be, but how it sticks around. This is what functional explanations are thought to describe: the persistence of a phenomenon through time, its 'resilience' (Pettit 2007) rather than its first appearance.
For example, I cannot use functional considerations to explain the first appearance of money. However, once I know -for whatever reason -money exists, I can use its functional role in the system to predict its ubiquity and robust persistence through time and existential threats. Consider Max Weber's functional explanation of capitalism (Weber 1950). 4 Capitalism is born by chance, as a by-product of the protestant ethic. However, more important than its birth is why it persisted and spread: capitalism is highly efficient, thus likely to prosper and offset alternative models.
A second concern against functional explanations is that they rely on obscure or missing causal mechanisms (2) (Elster 1994;. In other words, they lack a proper micro-foundation, failing the golden standard of methodological individualism (Elster 1982). However, the aetiological account refers to general causal mechanisms in the case of positive selection and filter effects. In the case of natural objects, like the heart, the environment operates as a filter that culls maladaptive mutations and favours adaptive mutations through enhanced survivability and reproduction. Instead, for artefacts and purposeful objects, positive selection is carried out by individual intentions. For most political and social facts it is some combination of these factors. Individual intentions see the functional need and intervene to restore an institution to well-functioning. For example, if new weapons make an army functionally deficient (say, the invention of gunpowder), politicians can take it upon themselves to reform. However, there is also what is known as 'of cultural multilevel selection' (Turchin 2010). If a social system remains dysfunctional, it is likely to die out against more competitive alternatives. Many Japanese feudal lords rejected gunpowder weaponry because it was dishonourable and undermined their political control over peasants, yet their reigns were short compared to those samurai who embraced the new military technology.

Functional mechanisms and processes
One interesting feature of functions is that they join two different yet related aspects: (1) the process that is performed, e.g., the pumping of blood; and (2) the mechanism that performs it, e.g., the actual valves of the heart (Mahner and Bunge 2001).
First, functions involve processes. In this sense, functions are specific behaviours performed within a certain class of systems, be they natural, artificial, or social. Examples of functional processes include the beating of the heart in mammals, the thrust that controls a vehicle's speed, and the revenue gathering in a political institution. Functions in this sense are immaterial: they exemplify a certain behaviour that plays a crucial role within complex systems. In this sense, functions represent systemic needs, and at least in the aetiological account always play a self-selective role in the organism. In other words, if the process in question is absent, or performed poorly, the system will not stick around for long.
Second, functions are taken to refer to the specific internal mechanism that performs a process in question. Functions can denote the valve of the heart that pumps the blood, the pedal that controls how much fuel is released in a vehicle, or a certain form of institutionalized tax collection. In this sense, functions are material and involve specific causal mechanisms.
Interestingly, different processes can be performed by the very same mechanism. For example, the heart pumps blood, thereby energizing the body through oxygenation, but it also regulates blood pressure, by adapting its rhythm to the necessities of the specific situation. Hearts tend to do both, and both play a role in hearts persisting. In the social world, we can say an army protects the social system from external interference, but also insulates political power from internal revolutions. Armies have historically played both roles, and both processes have contributed causally to armies being around.
The reverse is also true: a certain process can be performed by different physical mechanisms. For example, blood pumping can be achieved through the natural heart, or via an artificial pump. Alternative mechanisms may be better or worse at performing this process (see section 3). In the social world, 'pattern maintenance' (Parsons 1991), i.e., socialization in the same system of values, may be performed by different institutions, such as the family, religious rituals, or public schools. The different mechanisms may coexist and reinforce each other's processes, or can be mutually exclusive functional alternatives.
We will now turn to solidarity to investigate what benefits (if any) come from conceiving it as a social function.

Early functionalist accounts of solidarity
There is a strong precedent for conceiving solidarity in functional terms. Émile Durkheim provides a reliable initial ground for such an inquiry: he is both an early advocate of functional explanations (Halmwood 2008;Jones 2007;Pope 1975) and a ground-breaking theorist of solidarity (Bayertz 1999;Brunkhorst 2005;Stjernø 2009).
In The Division of Labour in Society, Durkheim makes a case for the importance of functional explanations in the social sciences. He carefully distinguishes a function from a mere 'aim', 'purpose' or 'goal'. According to him, those terms suggest something 'exists for the sake of results that we shall determine' (Durkheim 1984, 11). In contrast, the term 'function', for him, has no need to assume whether something 'arises from some intended and preconceived adaptation or from some adjustment after the event' (Ibid.). For Durkheim, again, 'to ask what is the function [of something] is to investigate the need to which it corresponds' (Ibid.).
In that book, Durkheim is interested in treating the division of labour as a function; however, in so doing he also exposes a functionalist view of solidarity. While solidarity for Durkheim acquires moral worth in itself, as valuable to communal life, it is also of chief importance to keep social cooperation running. Drain the key solidarity sources, and any society withers away. In this sense, the investigation of solidarity is pivotal to the intellectual mission of social theory, whose 'presuppositional problem' is what holds society together (Alexander 2014). This question is particularly crucial under modern conditions, where individuals are simultaneously more autonomous and more dependent on society (Durkheim 2013, 7). That Durkheim understands solidarity as a function is easily discernible from his presentation of it as a systemic need that can be delivered through multiple different mechanisms. Commenting on his Division of Labour, Steven Lukes notes: 'the functions once performed by "common ideas and sentiments" were now, in industrial societies, largely performed by new social institutions and relations' (Lukes 1971, 139). Accordingly, Durkheim famously distinguishes between 'mechanical solidarity', the agreement in conscience among the members of a simple community, and 'organic solidarity' (Schiermer 2014). Organic solidarity arises in complex societies with an established division of labour because individuals acknowledge they need others to provide what they cannot produce themselves. This structure of economic exchanges creates space for a sense of organic belonging and should prompt a genuine, yet partly self-interested, concern about other people's wellbeing. In other words, Durkheim's functionalist account of solidarity maintains an explanatory power that can apply to several societies in different ways. As body-heating is maintained by diverse mechanisms in cold-blooded and warm-blooded animals, so solidarity is performed by distinct institutions in different kind of societies. On the other hand, modern societies' constituents are thoroughly differentiated and interdependent, and are governed by regulatory processes that fulfil their social need for 'laying down in advance the functioning of each organ' (Durkheim 2013, 237).
To be fair, Durkheim's view of solidarity remains highly controversial. Some interpreters have contended that Durkheim collected insufficient historical references to substantiate the dichotomy of mechanical and organic solidarity: indeed, Poggi (2003,83) remarks that the set of societies in mechanical solidarity Durkheim has reported is almost empty, whereas the set of societies in organic solidarity is overly heterogeneous. Furthermore, textual evidence suggests Durkheim completely got rid of the concept of organic solidarity in The Elementary Forms of Religious Life (Schiermer 2014). Along this line of argumentation, Thijssen (2012) emphasizes that 'Durkheim's plea for the installation of some kind of neo-corporations, notably in his preface to the second edition of Division (1902), can clearly be interpreted as a mechanical rescue operation for a moribund organic solidarity'. While it needs further refinement, Durkheim's theory makes two important points: solidarity is best understood as a function, and solidarity can be discharged through different mechanisms. It is neither our intention here to advocate the claim that Durkheim was a full-fledged functionalist all the way down, nor to recommend a functionalist reading as most appropriate to appreciate the entire span of his works. More cautiously, we argue that his framework exhibits traits which a functionalist perspective can appropriately valorize, if duly updated in the light of the aetiological account.

Etiological solidarity between mechanism and process
Durkheim's functionalist account can be rescued by an aetiological conception. On the descriptive level, such innovative reconceptualization reconnects two common uses of the term in contemporary literature: solidarity as a 'set of feelings' and solidarity as a 'set of transfers' (Parijs 2004, 375). Moreover, this combinatory interpretation of solidarity could be helpful to restoring the analytical utility of the concept, whose obscurity -typically affecting related social concepts, such as community (Gemeinschaft) -is often lamented.
Solidarity's prime meaning is psychological and relatively straightforward: it connotes the set of feelings of belonging together, which supports 'attitudes of mutual acceptance, cooperation and mutual support' (Banting and Kymlicka 2017, 3). Under this respect, it stands in opposition to a 'lack of stable social relationship or bonds or connections, an absence of community or fellow-feeling' (Lukes 1999, 273). Solidarity in this sense is partially grounded in 'an interest in the integrity of a shared form of life that includes one's own well-being' (Habermas 2015, 23), and thus involves a complex combination of both self-interested and collective-oriented motivations.
Another common meaning is institutional, and frames solidarity as a set of redistributive transfers or, to put it in more emphatically evaluative terms, as a 'virtue of institutions' (Laitinen and Pessi 2014, 7). While the concept of solidarity has a millennial history, the term was first employed by Pierre Leroux (1840) following the social turmoil of the French revolution, and later formalized in article 21 of the Declaration of Human Rights of 1739 as the 'holy duty to support the unfortunate members of society'. Modern solidarity thus becomes associated to the idea of solidarity as a 'set of transfer[s]' (Parijs 2004, 375) involving 'redistribution of resources in favour of those in need' (Bayertz 1999) and the national welfare state, built on the intuition that such policies strengthen the community feeling of a plural society (Flora and Heidenheimer 2009).
These two meanings of solidarity, as a set of feelings and as a set of redistributive transfers, seem related yet quite distinct. Viewing solidarity in functional terms reveals these meanings are intimately connected. In order to understand why, we first must update Durkheim's view in the more epistemologically robust terms of the aetiological paradigm.
As we have seen, for Durkheim solidarity is a need because it holds societies together. Without solidarity, any social group simply withers and dissolves. This intuitively implies that any society exhibiting solidarity, exists partly because it does so. This is an implicit version of the feedback condition (b) of an aetiological functional account.
We propose that Durkheim's functionalist account of solidarity might thus be unpacked in aetiological terms as follows: Solidarity is a function of societies because: (a) societies tend to maintain solidarity (b) maintaining solidarity contributed causally to the persistence of societies The scope of such an aetiological scheme is sufficiently broad so as to apply to most societies historically known. In other words, the need for solidarity is not a distinctively modern property, but rather a precondition of any society as such 5 ; after all, 'the phenomenon of group loyalty and sharing resources existed long before the idea of solidarity developed' (Stjernø 2009, 25). Although solidarity can be performed by several mechanisms, some of which have been disclosed only under modernity, its fellow-feeling-producing process remains the same trans-historically and crossculturally. In other words, our account leaves room for a variety of mechanisms performing solidarity, and puts more emphasis on the social need rather than on each mechanism's specificities. Durkheim's theory of social evolution depicts not only a transition from one solidarity mechanism to another but also a progressive loosening of the tie between a given function and a given mechanism. In other words, the function of solidarity can be performed by more mechanisms in modern societies than in traditional ones. Thus, our account leaves room for a plurality of solidarity mechanisms in modern societies and concedes several could be operating both at the micro and macro level of social unity. Yet, at least in modern western societies, many authors point at the welfare state as the primary mechanism through which solidarity is performed (Ferrera 2005).
A further specification of the same aetiological scheme can thus be applied to contemporary western societies, and regards the extent to which solidarity is connected with the welfare state: The welfare state is a mechanism of solidarity because: (a) the welfare state tends to foster solidarity (b) fostering solidarity contributed causally to the persistence of the welfare state The intuition that the welfare state is pivotal to solidarity is present in the philosophical literature that equates such schemes with 'civic' solidarity (Scholz 2015, 730-732) or 'redistributive' solidarity (Banting and Kymlicka 2017, 4). This is also the standard occurrence of the term in political science (e.g., Ferrera and Burelli 2019;Vandenbroucke 2020), which roughly identifies solidarity as 'the sum of political practices that increase equality in people's life chances (that is, the welfare state broadly understood)' (Rothstein 2017). The core claim shared by most proponents of this line of thought is that every political community needs some sort of collective protection for those citizens who are more likely to be affected by social vulnerability and social exclusion (Ferrera 2017a;Scholz 2015, 730-732), and that welfare state schemes fit the bill for this demand. However, it is quite common to find this argument relying on moral grounds; indeed, civic or redistributive solidarity is often framed as a duty governing bodies have to their citizens (Bayertz 1999, 21). This literature is centred on the links between solidarity, the welfare state, and the maintenance of the political order, in a way that matches quite well the idea that aetiological solidarity performed by the welfare state discharges one key societal need, without which societies fail to persist through time. These authors usually argue that the welfare state has played a key role in the historical process of modern state-building (Flora and Heidenheimer 2009), through instilling loyalty towards the political centre (Ferrera 2005), and providing a more diffuse motivational stake in polity maintenance (Burelli 2018).
The previously mentioned distinction in a functional account between the process (e.g., the pumping of blood) and the mechanism that discharges it (e.g., the human heart or a mechanical pump) turns out to be particularly effective in this case. The welfare state provides a specifically modern mechanism to perform solidarity, whose successful outcome results in increased social cohesion within the political community. The link between solidarity as social cohesion-related feelings (its process) and solidarity as welfare transfers (its modern mechanism) goes both ways: on the one hand, redistributive policies contribute to reinforcing societal cohesion-related feelings (Beer de and Koster 2009); and on the other hand, such feelings support redistributive policies as they 'tend to cause people to seek out situations in which there are strong feeling of cooperation, mutual identification, and similarity of status and position' and inequalities result in a 'loss of mutual identification' (Crocker 1977, 263).
Others may question which of the two triggers -feelings or policies -came first. Some claim that 'a direct sense of community membership based on loyalty to a civilization that is a common possession' (Marshall 1949, 96) is a precondition for establishment of redistributive policies. In contrast, alternative approaches such as the 'power source theory' claim that the historical development of the welfare state can be accounted for in purely strategic terms, without postulating any antecedent national solidarity holding the population together (Banting and Kymlicka 2017, 7-8). According to the latter position, it is not solidarity which causally contributed to the persistence of the welfare state, but other social factors such as self-interest and conflict among competing political actors.
Aetiological accounts can remain agnostic on this question. As we have discussed previously, these accounts do not explain the first appearance of a phenomenon, which can just as well be the result of random mutation (in hearts) or human intention (in microwaves). Instead, aetiological accounts examine how a certain structure spreads and persists through time, and goes on to populate the world. The same can be said for solidarity: it is a function of societies independently from how it contingently arises. Indeed, as Kymlicka and Banting (2017: 8) point out: Inclusive welfare states or expanded enfranchisement may have initially arisen as a result of strategic behaviour by actors motivated by partisan or particularistic interests, but these reforms set in motion an evolutionary process which over time contributed to a more comprehensive sense of solidarity. Some authors push this argument further, suggesting that the relationship between solidarity as a set of feelings and solidarity as a set of transfers constitutes a virtuous circle that, once set in motion, accelerates its momentum: solidarity as a set of transfers requires the substratum of fraternal feelings, yet this is in turn reinforced by the transfers (Mueller and Keil 2013, 128-129). The latter disposes people to be more likely to value cooperation and resource-sharing embedded by the notion of social cohesion, as well as to reject inequalities as a determinant of social disintegration ̶ and, on Durkheim's terms, anomie. The aetiological-functionalist account of solidarity we propose can make sense of this virtuous circle, claiming that whatever reason ̶ whether merely instrumental or straightforwardly moral ̶ brought about the establishment of a welfare state scheme, the latter's persistence can be accounted for in virtue of its functional work.
In conclusion, solidarity is best understood as an aetiological function. On one side, this has the advantage of insulating Durkheim's intuition from the usual criticisms to social functions. On the other side, this conceptualization makes sense of the interconnectedness of two different senses of solidarity employed in the contemporary literature: as set of feelings, which we identify with societal cohesion in this context, and as a set of transfers, by which we refer to redistributive policies.

Functions as norms
Viewing solidarity as an aetiological function provides a strong ground to clarify in what sense it is normative: that is, why we should bring it about when we actually lack it.
In artefacts, it is quite common to discuss some 'anticipated good (or apparent good) that the function bearer serves (or is thought to serve) that helps to explain why it is there' (McLaughlin 2001, 57). Consider the case of a pen. Pens are designed to write; therefore, it is safe to say their function is writing. In this very basic sense, a (functionally) good pen is a pen that is good at writing (Thomson 2015, 69). Many other qualities exist, based on which one might positively evaluate a pen: being long lasting, elegant, or smooth to the touch. Nevertheless, a pen that satisfies all these other standards, but cannot write at all, cannot be considered a good pen.
However, can this normative implication be cashed out for natural or social functions, where there is no intended 'anticipated good' to build upon? Many scholars have deemed the aetiological account of functions able to carry normative implications even in the natural and social world (Hardcastle 2002). Perhaps the most explicit is Karean Neander: 'To attribute a natural function [. . .] to something is to attribute a certain kind normative property to the thing. That is, to attribute an evaluative standard to it that it could fail to meet, even chronically (i.e. systematically and consistently and even under ideal circumstances)'. (Neander 1999, 14) One starting intuition to draw normative conclusions is that things can have an aetiological function, and yet be completely incapable of carrying it out. The function of a heart is pumping blood because (a) hearts tend to pump blood. However, this clause expresses a probabilistic tendency, not a universal statement. A single heart may be completely incapable of pumping blood, yet this does not falsify (a). Indeed, if I were shot in the heart, my heart would lose the ability to pump blood, but would we say it had also lost its function? According to the aetiological account, this is not the case. It would remain true that hearts (a) tend to pump blood and (b) exist insofar as they do so. A broken heart still has the same function; it is merely incapable of carrying it out.
Once we detach the attribution of function from the actual ability to discharge it, we can make a further claim. Not only can hearts fail to pump blood at all, but various hearts can carry out this function with various degrees of efficiency. This means that hearts can be good or bad at pumping blood, and can be ranked from best to worst on the basis of their functionality. If our hearts are at the lowest end of functional performance, we have reasons to take medications to fix it. If our hearts are functionally broken, we should resort to a different mechanism to perform the activity: perhaps artificial pumps, or transplanted hearts. As such, functions do operate as natural normative standards: standards of good and bad performance, independent of the observer's intention or linguistic description. If the heart's function is pumping blood around an organism, a good heart is one that performs this task well.
If the previous defence of social functions is sound, functional normativity can also be found in the social world. Consider the case of armies. Suppose the function of armies is exerting organized violence. A good army, therefore, in a functional sense is one proficient at exerting organized violence. We can acknowledge, for example, that the Wehrmacht in the Second World War was a functionally good army, even if we deem Nazi Germany morally repugnant (Burelli 2020). 6 The point here is not that it is impossible to evaluate an army from a moral standpoint. It is obviously plausible to claim that a morally good army is one that fights only in just wars (ius ad bello), and one that fights reasonably justly (ius in bello). However, if an army fully respects the moral requirements, but is completely incapable of exerting organized violence, it can hardly be considered a good army.
It is important to note that functional normativity is not a version of instrumental reasoning, whereby one ought to do what is necessary to a goal one possesses (Schroeder 2009). On the contrary, functional normativity does not depend on any aim a person might want to pursue, moral or otherwise. As we have argued when we assessed the instrumental view of functions, this would make functions arbitrary and ultimately subjective. Etiological functions, in contrast, are attributed objectively and do not depend on the observer's intention. Suppose an eccentric autocrat only cares that the army looks stylish when paraded down the streets, but not at all whether it can fight. Instrumental reasoning would imply the autocrat has a reason to rank armies by the elegance of their uniforms. Functional normativity suggests instead that the most important quality of armies is how well they fight, because that is their essential social function. To the extent that eccentric autocrats ignore the army's function, they risk being eventually deposed. In this sense, proper functions have clear advantages over institutional teloi, which always suffer the criticism of who decides what the telos of an actual institution really is.
The reason for this objectivity is that a function is distinguished from an accidental property by the role it plays in the persistence of the system. Armies have been selected 6 For a critical discussion of this 'realist' account, see Erman and Möller (forthcoming). in the past because the ability to exert organized coercion is necessary to thrive in a conflictual world, not because of the aesthetic qualities of their uniforms. So long as one cares about the system's persistence through time, functions are unavoidable: doing without something that pumps blood or that exerts institutional violence is not a real option for anyone.

Solidarity as a functional norm
If solidarity can be aptly reconceived as a function, and if functions carry normative implications, then solidarity does also.
Many philosophers have sought to argue that solidarity should be seen as a normative value, whose absence we lament (Taylor 2015). Many think that when we discuss solidarity, we have in mind a sort of social bond that encapsulates mutual obligations (Bayertz 1999;Scholz 2015). If not moral in itself, solidarity has been characterized as a necessary companion to justice (Habermas 2013;cfr.;Tava 2021).
To an extent, the moral dimension is also present in Durkheim's framework, which emphasizes that solidarity enables humans to enjoy a fuller moral life. As he put it: When individuals discover they have interests in common and come together, it is not only to defend those interests, but also so as to associate with one another and not feel isolated in the midst of their adversaries, so as to enjoy the pleasure of communicating with one another, to feel at one with several others, which in the end means to lead the same moral life together. (Durkheim 2013, 18) Under this respect, solidarity is not only a kind of association but also a genuine moral need. However, Durkheim also underlined the set of normative expectations any kind of solidarity involves: indeed, along the just-quoted passage, he adds that 'a life lived in common is attractive, yet at the same time coercive'. It is no surprise that such normative dimension of solidarity caught the attention of contemporary political philosophers, who mostly engage in what Kolers names 'the question of moral justification' (Kolers 2016, 28), which scrutinizes when, if ever, solidarity is desirable or even morally mandatory. In this paper we offer a different ground for why we should have solidarity in our society. Solidarity plays a crucial functional role, without which no social group will be able to persist through time.
This functional ground can nevertheless do critical work, because it implies that we should want solidarity, even if we actually do not. Solidarity is an important feature enabling societies to survive through time. Some philosophers have argued instead that we do not need solidarity (Levy 2017). According to the view defended here, even if we do not want solidarity, we need it. Suppose a rich individual like Jeff Bezos does not care much for the welfare system, because he has sufficient money to provide for himself everything he might need. While he may have no instrumental reason to support the welfare state, there remain functional reasons he ought to do so: without an effective welfare system, social cooperation risks breaking down. Or, suppose a rich European state has no direct interest in helping other member states hit by the pandemic crisis; yet, without some system of mutual help, it is highly unlikely that the European Union as a political system would survive recurring asymmetric shocks (Ferrera 2017b). In this sense, solidarity is a condition of possibility for any cooperative venture (Sangiovanni 2015), as it is crucial to overcoming various 'strains of commitment' . 7 Indeed, the idea that society could be stable in the long term without solidaristic redistributions is a delusion of some economic elites that is slowly fading away. To claim the erosion of welfare systems that made our societies less solidaristic also made them less stable is not implausible. Indeed, the entire 'economic insecurity' (Zürn 2022) explanation of the populist wave is predicated on this. Sheri Berman, for example, persuasively argues that marketization in 20th century Europe created a demand for protection from marketization, leading on one side to social democracy and on the other side to fascism (Berman 2006) -the same might be true of populists today (Berman 2021).
Importantly, a functional view of solidarity clarifies why we should want it: on what grounds solidarity is desirable. Solidarity need not be understood as a moral obligation. Although Durkheim also emphasizes solidarity's moral dimension, it is not merely a moral value, whose absence we deplore from an idealistic standpoint of what a good society should be like, or of what we owe to each other. It is not only the case that our lives would be more fulfilling if we were more connected. Lack of solidarity is a much more dire situation, urgently in need of reparation: it implies there is something wrong with the basic functioning of our social system, which if protracted might imperil the system's persistence through time.
Solidarity's functional benefits provide important reasons to foster its process through activating its mechanisms. Many calls for solidarity, while appealing to philosophers (Van Parijs 2017), have little purchase among citizens (Ferrera 2014). Showing that solidarity is required by the very functioning of political institutions, may lead to a more persuasive justification.

Conclusion
This article has investigated what solidarity means and on what grounds it is desirable. In so doing, it has made three original contributions to the literature on solidarity in philosophy and the social sciences.
First, it reconceives solidarity in functionalist terms. It does so by reviving Durkheim's classical understanding of solidarity as a functional need of societies, and re-grounding it in the more refined theoretical framework of contemporary aetiological theories of functions. The advantage of doing so is that functional solidarity is not vulnerable to the usual criticism of functional explanations in the social sciences.
Second, the new functionalist view of solidarity defended in this article enlightens the connection between two widely used occurrences of the terms in different bodies of literature: solidarity as a welfare transfer in political science, and solidarity as fellow feeling in philosophy. Both are distinct mechanisms to perform the same solidarity process, much as both a heart and a mechanical pump can diffuse blood throughout the human body.
Third, a functionalist view of solidarity explains why solidarity is often framed not as a mere descriptive fact, that either is present or absent in our society, but as a normative value whose absence we lament. Lack of solidarity, in fact, is bound to undermine the cohesion of any social group, ultimately unravelling it and dooming it to perish. In this respect, our aetiological-functionalist framework provides a valuable contribution in bridging the gap between the understanding of the nature of solidarity and its normative value.

Disclosure statement
No potential conflict of interest was reported by the author(s).