Solidarity as a Social Kind

I. DISAGREEMENT, SOLIDARITY, AND CONCEPTUAL ENGINEERING

Suppose two interlocutors propose different characterizations of what solidarity is. To fix ideas, it is useful to have in mind two working examples. (Note that the purpose of these examples is meant to stage a contrast and not to argue in favor of one as opposed to the other; other hypothetical disagreements would work just as well.) According to the first (on behalf of which I will argue at greater length below), solidarity is a form of acting together to overcome significant adversity, where participants identify with one another and are disposed to come to each other’s aid in the pursuit of the group’s ends.⁵ The second interlocutor proposes a broader, more encompassing alternative. Solidarity, on this alternative, is to be disposed to act in a prosocial or altruistic manner with those with whom we identify. On this view, returning a lost wallet can be an act of solidarity, as can helping a fellow cyclist to get their bike onto a train.⁶ Note that, unlike the first view, there need be no joint or otherwise cooperative action, and our dispositions to aid others can be entirely unilateral (i.e., they need not be reciprocated).⁷ Similarly, the account allows for acts of solidarity that are entirely private or ‘silent’. Rosa Luxemburg, for example, is famous for refusing, as a girl, to eat chocolate for several weeks in solidarity with the workers at the front.⁸

Is this a genuine disagreement? On what does it turn? One way to interpret the disagreement as genuine is to assume that the two interlocutors are trying to provide a definition of the English word ‘solidarity’. If this were so, the disagreement could be adjudicated empirically by, for example, conducting well-designed surveys, or by other, more sophisticated, ways of tracking individuals’ disposition to affirm and reject statements containing the term.

This kind of strategy looks unappealing. It seems possible, indeed likely, that the two interlocutors—supposing they are philosophers or at least philosophically minded—don’t really care about what ordinary usage is, or what some empirical study in such usage would reveal. What they are arguing about is what concept we should use now and around here. The two interlocutors are advocating on behalf of their preferred concept of solidarity. Each participant claims that their preferred concept should be associated with the word rather than their opponent’s. The two interlocutors are, on this reading, metalinguistically negotiating or coordinating which word-concept pairing to use in particular contexts.⁹

What counts as an appropriate usage of a word-concept pairing will depend on the features of the context, and on what hangs on using one word-concept pairing than another. It might be that, say, usage of a term associated with one concept of solidarity rather than the other would lead people to act more decisively, or more humanely, or more reliably. Or it might be that using one word-concept pairing rather than another would lead people to act in morally more commendable ways. One usage might be preferred because it is clearer, or more precise, or helps clears up some confusion that is preventing more effective coordination. We might be judges discussing how we should understand the use of ‘solidarity’ in a legal text, so that we can assess whether the term should have some legal effect in a given case (for example, in interpretation of some more specific provision or derogation).¹⁰ Or we might be leaders of a protest wondering how to use the language of solidarity to express our need for support, and so wondering whether one particular usage might ‘put people off’, or, on the contrary, encourage them to join. Finally, we might be social theorists interested in distinguishing different aspects of social reality in order to provide a schema for discussing them from moral and evaluative perspectives. In each of these cases, if there is a genuine disagreement at stake, then it will turn on (a) whether the competing word usages associated with the proposed concept really do have the effects or functional roles we are concerned with, and (b) whether it is a good or bad thing for them to have those effects or functional roles in the first place.

This article is itself a form of metalinguistic negotiation over the term ‘solidarity’ for use by philosophers and social theorists (including sociologists, anthropologists, social psychologists, and so on)—and so fits the last context identified in the previous paragraph.¹¹ I will argue that we ought to pair the word with the particular social phenomenon mentioned in the first example above, where solidarity names a distinctive kind of joint activity.

Doing so has significant epistemic and normative benefits. First, pairing the term with the type of collective activity briefly outlined above and discussed in more detail below will help us to pick out an important, even fundamental, social kind that would otherwise lack a name. Picking it out via the schema I will present helps us to grasp it in thought. It helps us to study and understand it—both empirically and normatively—in ways we would otherwise not be able to do.

Second, the account, as I will demonstrate, will aid us in distinguishing solidarity from other related concepts, including fellow-feeling, camaraderie, social justice, and community; in this way, it will help us address familiar worries with the vagueness and redundancy of the concept.

Third, the schema will provide a framework for asking and answering normative questions that arise in the presence of solidaristic action: When and why should those who identify with one another act together? What reasons do they have for sharing each other’s fate? Is identification as a basis for joint activity illegitimately exclusionary? Under what conditions? What value, if any, does acting in solidarity have?

Fourth, I will suggest that the schema in fact tracks the phenomenon that the term was used to point to in its nineteenth-century efflorescence. While this is not dispositive (we are not interested in the everyday meaning of the word), it does help to put in focus the social, political, and cultural importance of the phenomenon itself. And, if I am right that the term ‘solidarity’ historically referred to the very phenomenon I identify, then this diminishes a common worry with conceptual engineering of the kind I am engaged in, namely that it either just changes the subject¹², or that it engages in an impossible or chimerical project, namely to change the (semantic) meaning of the words we use¹³.

The account therefore has an argumentative burden to bear. It must show that solidarity is in fact a social kind of significant and distinctive social, historical, and political importance rather than an arbitrary collection united by nothing more than the characteristics used to identify it. It is to this task that we now turn. In Section II, I outline a typology of social kinds, and distinguish their main features. In Section III, I then demonstrate how my schema for solidarity picks out a social kind with the characteristics outlined in Section II. I end by returning to the question: Why is it important to call this kind ‘solidarity’ rather than something else?

II. SOCIAL KINDS

In this section, I will outline a typology of social kinds. This is necessary to pinpoint exactly what type of social kind solidarity is, and to set up a framework for identifying the key features of solidarity as a social phenomenon.

All the red things in my house fall into a category, but the category lacks any unity deeper than the fact that each the objects share the property pairing that marks out the category (‘red’, ‘in my house’). By studying one of the red things in my house, not much can be reliably predicted about what properties the other objects in the category will share (other than that they will be red and in my house), and there won’t be much to say, from studying the sample, about what explains why the other objects in the category have the properties they happen to have. The properties of red objects are, that is, only weakly projectible. (The same goes for other arbitrary collections, such as cloudy liquids, sharp objects, soft things, and so on.)

Natural kinds, on the other hand, have instances whose properties are reliably and systematically projectible (I turn to social kinds below). I can study the properties of a sample of (pure) water and use my study to explain and predict the properties of other samples of water, even if I have not observed them. The category has a unity that an arbitrary collection lacks.¹⁴ In the case of chemical compounds—such as water, gold, iron sulfide, and so on—that unity is explained by the chemical characteristics of its combined constituent elements. Species like the Hairy Woodpecker are also natural kinds.¹⁵ By studying a sample of Hairy Woodpeckers, I can learn about the properties of other Hairy Woodpeckers, even if I haven’t observed them. This projectibility of a set of properties co-instantiated by typical members of a species is not as reliable as in the case of chemical compounds (through mutation, for example, a particular individual Hairy Woodpecker might lack some of the typical properties, but still be a Hairy Woodpecker), but it is still much more reliable than the arbitrary collections listed in the previous paragraph.¹⁶ In Boyd’s now classic terminology, the core properties form a cluster—a cluster that tends to be reproduced (sometimes imprecisely) in each of the members of the kind.¹⁷ In standard cladistics, the unity of a species is explained not by micro-structural characteristics, but by the evolutionary history of the species, including details about gene flow and reproductive isolation. In Millikan’s terms, any one member of the species tends to have similar properties as other members because it is, at root, a copy of an ancestor shared by all members subject to broadly similar environmental pressures.¹⁸

Note that, for a set of projectible properties to be representative of a kind, the properties cannot be typically co-instantiated by mere chance. If the red objects in my house all are made of the same type of plastic, the characteristics of that plastic (which are co-instantiated in all the red things in my house) are related by chance to the fact that they are in my house and that they are colored red. While the plastic forms a kind, the red things in my house do not. There must be an underlying mechanism (e.g., Boyd), a causal network among the properties (e.g., Khalidi), or some other, weaker kind of stability that secures the unity and projectibility of the kind (e.g., Slater).¹⁹ Philosophers differ regarding which relations among co-instantiated properties are necessary for something to constitute a kind.²⁰ I will not take a stand here. I will assume that any one of these (or some combination) can be sufficient for a category to constitute a kind.²¹ The important thing, as we will see below, is to identify what this structure is in our central case, namely solidarity, rather than to establish what the best account of a kind in general is. We can afford to be ecumenical. I will argue that a particular causal structure (rather than some weaker form of stability) secures unity in the phenomenon of solidarity.

The sort of unifying causal patterns typical of natural kinds also structure social kinds, such as marriage, gender, government, recession, and income inequality. There are two types of social kinds that we need to distinguish.²² The first type are social kinds that have the form they do because of constituting norms, conventions, or practices that establish the conditions that need to be satisfied for something to count as a member of the kind. Paradigmatic examples include marriage and money. What it is to be a marriage is just for people to believe it is a marriage (that a union, that is, satisfies the defining set of constituting conventions). Both marriage and money must therefore be represented as existing (via constituting norms, conventions, practices, beliefs, and so on) before they come into existence; put another way, there is no money or marriage without beliefs (and other attitudes) about money or marriage. Our conceptual practices of categorization create the kind rather than track its pre-existing borders and character. This is why the label socially constructed (or, as I will sometimes say, recognition-dependent) is apt for this type.²³ Some more examples: banks, hotels, corporations, professional basketball teams, and schools. Not all socially constructed kinds are, however, formal (often legal) institutions. There are also informal but still socially constructed kinds, such as particular literary and musical genres (the novel, jazz, opera) or cocktail parties.

Socially constructed kinds are unified by the constitutive norms and conventions that set out what must be satisfied to count as an instance of the kind.²⁴ To find the structure supporting the unity in the kind, then, we look to individuals’ actions, which, insofar as they conform and comply with the constitutive norms and conventions, enact and realize the kind in society. (Note that the constitutive norms and conventions can be followed without anyone knowing that they are followed, or being able to articulate what they are.²⁵) We can then study, for example, why the norms and conventions are complied with, what further functions the institution or practice plays once it is enacted, what further norms and conventions it gives rise to, what further causal properties and links it brings into being as people adapt to and use the institution or engage in the practice. Once banks come into existence, we can then study, for example, the different forms of security that banks use to protect their deposits; while security systems are not constitutive of banks, they are projectible properties of banks. These causal patterns—in addition to the patterns of copying and learning that initially led to the spread of the conventional norms in the first place—ensure that the typical properties of the kind are projectible across instances.

The second type of social kind is recognition-independent.²⁶ Examples include things like social status, kinship structure, recession, and income inequality. There can be instances of each of these without anyone knowing them as instances of social status, kinship structure, recession, and income inequality. And the individuals involved need not even have the concepts required to make sense of what is happening as a ‘recession’, as an assignment of something called ‘social status’, or as a manifestation of a ‘kinship structure’. The category exists and is unified independently of anyone’s recognizing it as a category and in the absence of any norms and conventions constituting it as a category. Unlike socially constructed kinds, our categorization practices do not create the kind. What counts as an instance of recession, kinship structure, and so on, does not then depend, necessarily, on a set of widely recognized norms and conventions outlining the conditions for belonging to the category. Recessions are recessions even if no one has any attitudes (beliefs, hopes, desires, etc.) about them as recessions.

What makes recognition-independent categories more than arbitrary collections? As before, the categories are unified if there is a causal structure that offers a basis for prediction and explanation of typically co-instantiated properties. For example, what makes the category of social status fruitful for sociological study is that there are causal connections between the typical properties of the kind that are reproduced across different societies and types of social status. There are enduring features of social ranking, that is, across all societies despite their many differences. Sociologists, evolutionary psychologists, and so on, are, among other things, engaged in isolating what underlying causal mechanisms might explain the unity in the phenomenon. What makes us status-seeking and status-sensitive creatures? Why might social status be an enduring feature of all human societies past and present? If there are fruitful answers to these and similar questions—and, in the case of social status, it seems likely that there are—then there is enough structure and unity in the phenomenon to constitute a kind.

This also raises the question whether recognition-independent kinds must depend on the existence of socially constructed kinds. Must all recognition-independent social kinds be a causal or conceptual consequence of socially constructed ones? Recessions obviously are (since recessions depend on the existence of, among other things, money, whose circulation causally gives rise to recessions). Other recognition-independent kinds also seem to fit the bill: racism requires the social representation of race as a category, sexual harassment requires the background of the (socially constructed) sex-gender system, and so on.

It is difficult to imagine any recognition-independent social kind that does not depend on the existence of socially constructed ones. But this may be only because so much of our social life is mediated by language and convention. There is, I believe, nothing in the very idea of a recognition-independent social kind that necessarily requires social construction.²⁷ Take social status again. And think of ‘lower’ primates like lemurs, tarsiers, and orangutans.²⁸ Such primates have structured systems of social dominance and ranking—and hence social status—but do not have concepts that map those systems and their components in any detail, or that represent those systems as existing. It is more likely that normative structures of social dominance and rank (and their corresponding beliefs) emerge from repeated interactions among individual primates and a result of direct conflict, conflict avoidance, and conflict resolution.²⁹ What makes such social interactions different from, say, bees, is that they are mediated by propositional attitudes such as beliefs, desires, and intentions. In nonhuman primates such as apes, gorillas, and chimps, to be sure, they are also mediated by more complex, cognitively sophisticated concepts that represent different individuals as belonging to a lower or higher rank.³⁰ The key point is that whatever beliefs and other attitudes such primates have that enable them to exhibit the phenomenon of social status, they are unlikely to involve socially constructed concepts like gender, nation, occupation, race, class, ethnicity, and so on, not to mention the concept of status itself.³¹ The categories that constitute nonhuman primate social life are therefore recognition-independent but not socially constructed. (They cannot, for example, change the categories through new representations, and they cannot identify and self-identify with each category as a category [as members of this or that nation, ethnic group, etc.])

So far, we have highlighted (a) the importance of causal structures supporting the projectibility of social kinds, (b) the distinction between socially constructed and recognition-independent social kinds, and (c) recognition-independent kinds that depend on underlying socially constructed kinds, and those that do not. Before applying this analysis to solidarity, I want to distinguish between interactive and non-interactive social kinds. As we will see below, solidarity will turn out to be, at root, a recognition- and construction-independent phenomenon. It is therefore closest in structure to phenomena like social status. The analysis, in addition to telling us something about solidarity, highlights a type of kind that is understudied by social ontologists: kinds where categorization and classification practices shape the content and character of the kind, but do not create the category ex nihilo.³² There can be, then, recognition-dependent kinds that have emerged through social modification rather than construction. Solidarity is among them.

Interactive social kinds are kinds whose typical properties shift in response to individuals’ coming to see themselves as participants in the kind. The boundaries of such kinds, that is, are subject to feedback loops in ways that, say, chemical compounds are not.³³ Hacking gives the examples, among others, of Multiple Personality Disorder (MPD). He argues that MPD was not a kind until people began to associate a variety of unrelated or weakly related psychological forms of malaise as falling under a single rubric that was worthy of medical and scientific attention.³⁴ Once it became a generally recognized category of illness, in part promoted by psychologists who benefited by its classification, people unconsciously began to alter their behavior in accordance with the category. As more and more went in for treatment, the category, in turn, received further validation. This process yielded new classifications and new ‘discoveries’ about the illness, which in turn affected those identifying with it. Such feedback loops can go on indefinitely.

But, if I am right that there are recognition-independent social phenomena, such as the ones discussed above, then there must be social kinds that are not interactive in the required sense.³⁵ However, while a recognition-independent kind need not be interactive, it can become interactive through being recognized. Phenomena like sexual harassment provide an example. While we may suppose that sexual harassment, as a social phenomenon, is as old as the sex-gender system, it only recently became recognition-dependent. Only in the late 1970s did women begin to put a name to a practice that long pre-dated the baptism.³⁶ Once the phenomenon acquired a name, entered a social movement, and become theorized, it rose to consciousness and became interactive. Individuals began to identify themselves as victims of sexual harassment, which had transformative effects both on their own awareness but also on society at large. As a result, the causal patterns supporting the phenomenon itself began to shift. As people began to use the category to protest, to change laws, and to identify and self-identify, the causal structures uniting the category changed and consolidated around new nodes. And with those changes also came a change in the normative valence and social meaning of the phenomenon, and so in the content of the kind itself. From a phenomenon that had been treated as just part of how things are between (mainly) men and women (and hence not needing a name), it became something to be criticized, understood, and monitored; that change, in turn, altered the way in which the phenomenon itself manifested (for example, in the workplace). The fact that the causal networks shift in response to a changing normative landscape need not mean, however, that the term loses touch with the kind or goes out of sync with it; rather, what has happened instead is that the reference of the term has evolved along with the kind it names.³⁷ There has been no impingement of the normative on the epistemic; rather, the kind itself has shifted in virtue of the fact that it has become theorized and politicized. To put the point another way: to have an accurate picture, any historian or sociologist interested in understanding the phenomenon (rather than trying to change it) would need to take note of the changing norms and practices associated with its coming to light as a phenomenon. Unlike Hacking-style kinds (such as Multiple Personality Disorder), there can therefore be social kinds that are fundamentally recognition-independent but can become both recognition-dependent and interactive. These kinds will then not be socially constructed but socially modified.³⁸

It is possible, then, for social kinds to enter periods of interactivity and periods of passivity. Periods of interactivity are periods in which the kind becomes theorized, politicized, mobilized for different ends; it enters people’s everyday consciousness and changes their everyday behavior. There can then be periods of passivity, in which the kind becomes less unstable, less subject to revision, debate, and theorization, or periods in which it languishes or disappears (think of dead musical or literary forms for example). Or a kind can be passive insofar as it is recognition-independent, and so not present to consciousness; in such cases, while the phenomenon exists, there are no concepts available to describe it (recall: sexual harassment before the 70s). This interweaving of interactivity and passivity is one way in which a social kind can shift over time.

I have, so far, allowed for two possibilities. In the first, a phenomenon that predates any naming begins to be tracked by new terms. While the intension of the naming term might shift and adapt to the phenomenon as the term begins to latch on, and people become conscious and react to the phenomenon, the phenomenon itself doesn’t change. This is one way to understand what happened with sexual harassment: as a phenomenon, sexual harassment didn’t itself change in response to consciousness-raising and political and legal mobilization. Rather, what happened was that the consciousness-raising and mobilization just became more attuned to the phenomenon itself.

In the second, the naming of the phenomenon (and the associated political, legal, social, cultural engagement with it) initially tracks the phenomenon, but then, with time, changes its nature. (This kind of possibility, as we have seen, is more in line with the arguments pursued by Hacking, with the important caveat that the naming doesn’t create the phenomenon ex nihilo.) The idea that usage and changing norms and practices can change a social phenomenon shouldn’t be controversial. This happens all the time with recognition-dependent, socially constructed categories like, for example, money (which used to require a material substrate but no longer does) or marriage (which, in some places, now includes a union between two men or two women). But it can also happen with recognition-independent categories such as friendship: changing practices and norms can change what friendship is, even if the phenomenon of friendship (we can suppose) predates any naming of the practice within any particular language. And I have suggested (without doing much to establish) that it also happened with sexual harassment: changing norms and practices transformed the normative valence and causal patterns associated with the phenomenon, shifting it from something that was accepted and expected to occur, and hence part of the social background, to something that acquired a name and became a basis for shame, blame, and criticism.³⁹

But how, one might wonder, do we know whether we are tracking the phenomenon more accurately through changes in intension (in line with the first model), or whether the phenomenon itself has changed, among other things, by naming and describing it (in line with the second)? To determine whether it falls in one category or another, we need to engage in historical, sociological, and philosophical interpretation. For example, a much fuller account of sexual harassment that fits the second model would need to show that sexual harassment, as a social phenomenon, is normatively loaded: it would be false to say that there is an underlying phenomenon—sexual harassment—and, separate from it, a set of normatively valenced attitudes towards it. The attitudes shape and structure the phenomenon itself, changing the way it manifests and the way people understand themselves in relation to it, in the same way as analogous attitudes shape the nature of, say, marriage or friendship. Just as with sexual harassment, it may be difficult to determine which model better describes any particular phenomenon. But even if it is difficult to determine the truth of the matter, this does not mean that there aren’t better and worse interpretations: better interpretations will be sensitive to the sociological and historical facts about the phenomenon, and have better theoretical models explaining those facts.

III. SOLIDARITY AS A SOCIAL KIND

So far, I have laid out a series of distinctions and characteristics needed to understand the nature of social kinds. If solidarity is a social kind, it will have properties that make it articulable in terms of these characteristics and distinctions. As I have already mentioned, framing solidarity in this way will have a series of benefits. First, it will allow us to answer the charge that solidarity is an amorphous and vague concept that can mean anything to anyone. On the contrary, solidarity is, as we will see, a social phenomenon in its own right, worth studying for its own sake. The account of solidarity as a social kind will therefore aid in setting the parameters of a research program for further historical, sociological, philosophical and political inquiry. Second, seeing solidarity as a social kind will provide a strategy for resolving genuine disagreements about solidarity. Which view better explains and unifies the phenomenon? Which one better distinguishes it from other, related phenomena? Third, the account aids us in more clearly isolating the normative and evaluative aspects of solidarity, and so in understanding why it is often perceived as a social ideal. And fourth, the account helps us to make sense of its history, and hence brings to light its social and political significance for us today.

I begin with the account (briefly adverted to above). We are in solidarity when, as a result of mutually identifying with one another on the basis of a role, cause, condition, set of experiences, or way of life, (a) we each intend to do our part in overcoming some significant adversity, X, by pursuing, together, some more proximate shared goal, ф⁴⁰; (b) we are each individually committed to X and ф (we have, that is, a settled and reliable disposition to set aside narrow self-interest in our pursuit of X and ф), (c) we are committed to not bypassing each other’s will in the achievement of X and ф⁴¹; (d) we are committed to sharing one another’s fates in ways relevant to X and ф; and (e) we trust each other with respect to (a), (b), (c), and (d) (where trust is reliance plus normative expectation⁴²). On this picture, solidarity is omnilateral, requiring a dense network of mutual attitudes and dispositions. By contrast, a unilateral disposition to provide aid or support to others, if not reciprocated with others with whom one identifies and with whom one seeks to act to overcome adversity, is not solidarity but humanitarianism or altruism. Solidarity, furthermore, is not reducible to social cohesion (where social cohesion is something like a society’s capacity to withstand breakdown and internal conflict), or to an emotion like fellow-feeling or camaraderie—although its presence can promote both. It is also not synonymous with social justice (it describes a sociological category rather than a purely normative one). While it is a form of cooperative joint action, it is not the same as cooperative joint action: making a computer in a factory, painting a house together, or dancing a tango can be instances of cooperative joint action, but they are not, in normal cases, instances of solidarity (in normal cases, the cooperation in each case is not grounded on any mutual identification, there is not enough commitment to other participants or to the goal, and the goal is not to overcome some significant adversity). Similarly, not every socially salient social group exhibits solidarity: the employees at Amazon form a social group but they do not, in normal cases, stand in solidarity (except when, say, joining in a protest or strike); the same goes for teenage children.⁴³

This also helps us to distinguish solidarity from coalition-building.⁴⁴ While solidarity often involves (as part of its characteristic activity) coalition-building, it is also much more demanding with respect to its core set of attitudes and much more open-ended with respect to the ends pursued (on which more below). Solidarity requires identification, a disposition to help others in the pursuit of the struggle, and a disposition to set aside self-interest; coalition-building may be much more narrowly self-interested, not require any disposition to help others, and lack any identification among members of the coalition (who may only be coordinating for strategic ends). Representatives in Congress, for example, build coalitions in order to get legislation through, but they need not be in solidarity when they do so. Solidarity, as I have characterized it, is therefore distinct from other, related notions with which it is often confused.

Several other features are worth highlighting. First, the schema is not intended to pick out a set of necessary and sufficient conditions. The schema picks out a cluster of properties that often, but not always or necessarily, co-occur.⁴⁵ For any given instance of solidarity, the more properties are missing, the more the phenomenon in question should be classified as a borderline rather than a paradigmatic case; in such cases, we should be unsure if the instance is, or is not, an instance of solidarity.⁴⁶ If the account is informative, it should help us to explain why instances that are missing one or more features are borderline rather than paradigmatic. Suppose, for example, that a group of activists is protesting the conditions within a prison. And suppose that, though the prisoners don’t know about the protest and have not taken any action themselves to protest their conditions, they would act on the inside (were they to become aware of the situation) alongside the protesters outside. Is it right to say that the activists are acting in solidarity with the prisoners? In cases like this, there would be willingness to act together but there is no joint activity. Solidarity between protesters and prisoners, we might say, is latent but not actual. Or suppose a more accurate picture reveals that the activists see themselves as acting on behalf of or for the prisoners, but make no attempt (and would make no attempt) at engaging the prisoners’ own agency in their advocacy? In this case, it is better to say that the protesters are not in solidarity with the prisoners (which paradigmatically requires mutual engagement of one another’s agency and commitment) but are in solidarity with each other on behalf of the prisoners. And finally suppose that though the protesters would engage the prisoners’ agency in all the right ways, the prisoners are not willing to reciprocate; they are unwilling, that is, to act alongside the protesters (thinking that their action is futile). In this case, once again, there is no joint activity, and so the solidarity, on my view, is only aspirational, rather than actual or latent. The account helps to explain why these are all borderline cases: solidarity paradigmatically requires a willingness to engage others’ agency, joint activity, and a willingness to share one another’s fate.⁴⁷

Similar things can be said about levels of commitment. In any solidary group there will be more and less committed actors. Above some threshold level of commitment (which can be very low—sufficient to override narrow self-interest in a range of cases) they will count as being in solidarity. But this threshold is blurry and will vary by group, and within any individual across time. And we can also say of someone (Gerry) who joins up just out of narrow self-interest (only to impress a girlfriend, to make some extra cash) that Gerry does not act in solidarity, and that others who act under the assumption that Gerry is in it for the right reasons are acting under an illusion that makes their solidarity in an important sense defective (toward Gerry). So there can be weaker and stronger forms of solidarity that vary depending on level of commitment, mutual willingness to aid, etc., and for many of these cases, it may be unclear whether the instance is in fact an instance of full-blown solidarity or not.

Second, the account is entirely descriptive.⁴⁸ Solidarity does not have to be good or right or pursue the just to be solidarity.⁴⁹ Neo-Nazi groups can be in solidarity with one another. To be sure, participants must take themselves to have good reasons to overcome some significant adversity together, and to take their mutual identification as a good reason to do so. But these reasons may not, in fact, be good reasons. This is not to say that solidarity cannot itself be valuable. Its value may simply be conditional: it is good (even good-in-itself) but only if it is not used to pursue wicked ends. Indeed, characterizing solidarity in the right way should allow us to pick out and explain its value in much clearer ways than if we had used an inaccurate or otherwise inadequate schema to track the phenomenon. An analogy: to understand the value (if any) of, say, free jazz, we need to understand what makes that genre distinct from related genres. If our concepts of jazz genres didn’t cut the jazz world at the joints, it is unlikely that we would be able to give a successful account of the value of any one such genre.⁵⁰

Third, the account also tracks the most important practices associated with the term, especially since the nineteenth century.⁵¹ The characterization, that is, resonates with and makes sense of the political and social uses of solidarity in the main contexts in which it is salient. As a term, ‘solidarity’ enters our vocabulary only in the modern era (which is not to say, as we will see below, that solidarity itself is a solely modern phenomenon). While one can trace the term to its roots in Roman law—where an obligation in solidum was a joint contractual obligation in which each signatory declared himself liable for the debts of all together—its use as a term denoting a type of broadly social (rather than narrowly legal) relation becomes prevalent in Europe—and especially in France—during the early nineteenth century.⁵² Why then? As any cursory glance at the major early texts (e.g., Saint-Simon, Fourier, Renaud, Leroux, Comte) would reveal, the language of solidarity emerges as a response to growing anxiety regarding the expansion of commercial society, large-scale industry, and the perceived collapse of traditional communities.⁵³ From this perspective, it is no surprise that the language of solidarity emerges in France, where the upheavals of the Revolution and its aftermath had first placed the ideal of republican fraternité firmly on the map. If societies were to hold together in the presence of emerging class conflict and the centrifugal, individualistic pull of markets, then something must replace the old ties of rank, guild, family, and traditional religious practice.⁵⁴ That something was thought to require a social bond between strangers, a form of identification strong enough to give individuals the sense of being connected to a larger whole on which they depend, and which in turn disposes each to share in the good and bad fate of all the rest.

This concern runs like a thread through the five main traditions of thought that have shaped solidarity (as a practice) since then, namely Socialism, Solidarism, Nationalism, Christianity (especially Catholicism), and contemporary social movements. Each one grounds solidarity among workers, citizens, nationals, human beings, and disadvantaged groups (women, blacks, disabled, and so on), in a distinctive notion of identification, and advocates a distinctive kind of collective action designed to overcome significant adversity. Within Socialism, solidarity among workers, for example, is grounded in identification based on a shared condition as exploited, and is realized through collective action designed to overthrow capitalism.⁵⁵ Within Christianity, solidarity among human beings is grounded in identification on the basis of interdependent vulnerability, and is realized through forms of organization and cooperative action (e.g., family, community, country, church) designed to alleviate the suffering that interdependence brings in its train.⁵⁶ Within Solidarism, identification with other producer-citizens on the basis of our coordinated role in a wider division of labor necessary for the prosperity of all grounds a sense of shared fate, and a commitment to institutions that serve to secure each citizen against old age, ill health, and unemployment.⁵⁷ A similar pattern unites each of the other cases.

Fourth, while the account allows for a single action to count as an instance of solidarity (e.g., fellow spelunkers working together to break out of a collapsed cave), the more common case is better described not as an action, but as an activity, where the emphasis is on an open-ended process of overcoming adversity requiring many actions over a broader span of time. In the latter kind of case, the solidary group is disposed and committed to acting in significant ways to overcome whatever adversities they face, but, as long as the group remains so disposed and committed in the interim, the solidarity does not only appear during any particular action that forms part of the larger activity.⁵⁸ It is important, however, that in cases like this the mere commitment or disposition (without any accompanying action) is not enough for solidarity (except in very special cases of thwarted action). Even as an activity, solidarity is always purposive and oriented to something we do together in overcoming adversity.⁵⁹

With this account by way of background, we can return to the characteristics and distinctions central to social kinds, and see how they apply. If they apply in a rich and informative way that can sustain explanatory, predictive, and normative work on solidarity, then that is evidence that we have, indeed, isolated a social kind. The first, and perhaps most important, characteristic is causal structure: What causal structure explains the unity in the properties I have associated with solidarity? How rich in inductive, predictive, and explanatory content is it? If there is no such structure, then the category I have highlighted is merely an arbitrary collection, united by nothing more than the properties used to identify it; the category I have associated with solidarity would be analogous to the category composed of all the red things in my house. If this were the case, then solidarity would fail to be a social kind.

There are two main elements to this causal structure. First, at the heart of solidarity is identification. As a psychological mechanism, identification has three features: affective, epistemic, and normative. Affectively, identification involves a disposition to empathize and sympathize with those with whom we share a role, cause, set of experiences, condition, or way of life. When we identify with another on one or more of these grounds, we desire to enter their point of view and see the world with their eyes. Epistemically, identification involves a positively valenced desire to construct and share a perspective with those with whom we share one or more of the bases just mentioned. Normatively, identification involves seeing this shared perspective as significant in how we shape and run our lives. Our identity as, say, Italians, cancer survivors, and so on, makes a difference not simply to who we are, but to what we do and how we do it.

As a psychological-cum-causal mechanism, identification disposes us to band together in the struggle against forms of adversity that threaten the group. It makes it more likely that we will share one another’s fate in ways relevant to our shared struggle, more likely that we will set aside narrow self-interest, less likely that we will bypass one another’s wills, and more likely that we will trust each other.⁶⁰ Identification then serves to strengthen our bonds in ways that mere tit-for-tat reciprocity (which requires repeated interactions with the same group of people) or narrower self-interest could not do alone.⁶¹ And identification, importantly, is scalable in ways that tit-for-tat reciprocity and narrow self-interest are not: when we identify with others, we identify not with each of them de re but with each of them as members of the group picked out by some shared feature (way of life, cause, role, etc.). The normative, epistemic, and affective dimensions of identification extend to all members in virtue of our shared role, way of life, and so on, rather than in virtue of some specific set of repeated interactions with them. Identification, in sum, causally sustains the joint activity, dispositions, beliefs, desires, and other attitudes constitutive of solidarity.

This is also important, one might speculate, for understanding the way in which identification may have evolved among human beings. In the absence of centralized punishment, cooperation tends to break down in groups as the numbers get larger and the interactions thinner.⁶² In such larger groups, the mutually beneficial interactions with specific others that drives reciprocity in smaller groups is sparse. Less hinges on any one pairing of benefit and contribution for the overall production of the cooperative good; the good is produced more diffusely. It becomes easier and more tempting to free ride. The common knowledge that this is the case erodes trust. From the narrower perspective of self-interest, it seems less and less rational to contribute when the benefits are available anyway. Because identification moves us toward engagement with our fellows affectively, epistemically, and normatively, it makes such free-riding less likely, extends our self-interest to encompass the group (by helping you I help myself), makes trust easier to come by, makes us more disposed to pay costs to support our fellows, and allows reciprocity to be more diffuse. When adversity threatens the group, it then provides a motor and backbone for solidaristic action: identification makes the social cooperation at the heart of solidarity more reliable and robust than tit-for-tat reciprocity or narrow self-interest would alone. This may have been one of the ways in which identification as a psychological mechanism evolved, since human groups that act in solidarity are much more likely to survive than groups that don’t.⁶³

So far we have seen how identification is an underlying causal mechanism that holds the properties co-instantiated in solidarity together. In Boyd’s terms, it secures homeostasis. But there is another element in the causal structure that makes solidarity a kind. Identification doesn’t merely support and hold together the other features of solidarity, but is also supported in turn by them. Indeed, each of the elements in solidarity are, given the right conditions, mutually reinforcing.⁶⁴ When solidarity is successful, it will tend to reinforce identification among the members of the group engaging in the joint activity. Similarly, success breeds trust, trust reinforces a willingness to share others’ fate, a willingness to share other’s fate helps to overcome narrow self-interest, and all these things together help to prepare the group for struggles against future forms of adversity. The properties co-instantiated in particular instances of solidarity are therefore causally connected in a network of mutually supporting elements. Of course, as with all social cooperation, solidarity can be fragile, and, because it is demanding, also rare. But this does not make it any less a kind. Francium, after all, is a chemical kind, but it is rare and unstable.

We are now in a position to apply the other features of social kinds to solidarity. Is solidarity recognition-dependent or -independent? At its root, I want to argue that it is recognition-independent but, as we will see, it has also gone through at least one recognition-dependent phase. Solidarity, then, is more like social status, sexual harassment, racism, income inequality, and recession. People engaged in collective action against adversity that meets all the other conditions listed above can, that is, be in solidarity even though they don’t have the concept of solidarity to describe what they are doing. Solidarity describes what they are doing ‘from the outside’ looking in. Indeed, as I hinted at above, it is likely that solidarity, as a fundamental feature of social reality, was an important feature of human societies from the very beginning. While the bases of identification likely had a narrower character (based on the way of life of a particular human group, or on other distinguishing symbolic characteristics), it is hard to doubt that solidarity is a universal human phenomenon with many local variations.

We can go further. As a fundamental social category, solidarity is not construction-dependent in the sense discussed above. Among our recognition-independent examples, solidarity is therefore most like social status. It is possible to imagine human (and non-human) groups that exhibited all the features we have identified, but where those features were not built on or out of any socially constructed categories. While solidarity can be (and most often was) built on such socially constructed categories (usually on the basis of conventional representations of nationality, shared history, symbolic attachments, etc., on which identification was grounded), it need not have been. Identification, for example, can be mediated by psychological pressures and dispositions that do not go via conventional representations, and so can joint action, trust, and mutual aid.

Solidarity has, however, passed through interactive, and hence recognition-dependent, phases. The most important, for our purposes, kicks off in the beginning of the nineteenth century in early socialist circles, and comes to fruition in the late nineteenth century first in the international worker’s movement and then in the French Solidarist movement in France.⁶⁵ During this period, solidarity becomes theorized and politicized as an ideal uniting associations, states, and social movements.⁶⁶ Once it enters the political, sociological, and philosophical lexicon, attempts are made to shape it to modern conditions. Perhaps most significantly, it is reworked (by Durkheim⁶⁷, among others) to apply across modern states, but not on the basis of a shared way of life or cultural identity, but on the basis of the functional interdependence of all individuals on an extensive division of labor. According to this conception of solidarity, identification is based on our shared, differentiated role as producers of a joint social product. In the twentieth century, it becomes a central pillar of social movements such as the Civil Rights Movement and Women’s Liberation (and, more recently, Black Lives Matter and MeToo); in these cases, identification based on a condition as racialized (or feminized) comes to the fore.

The main features of solidarity that have shifted in response to becoming interactive are not so much located in the higher order characterization I have provided; they are not, that is, changes in its content. Rather, the main changes have come in the possibilities for types of identification and shared action that become available once people begin to see themselves as engaged in solidaristic struggle against common forms of adversity. Once people acting together see themselves as in solidarity, the phenomenon becomes recognition-dependent in a way it wasn’t before: people begin to theorize why, for example, sharing a condition as oppressed might provide a basis for identification, or how to think of solidarity among citizens and co-producers in a modern industrial economy. As a result, it opens solidarity up to being formalized in institutions, parties, and organizations, and to be reworked and redeployed in new contexts. This happens, for example, through various forms of copying-qua-learning—think of the influence of the workers’ movements on Civil Rights movements.⁶⁸ Furthermore, in the wake of its politicization, solidarity explicitly becomes an ideal of social unity to strive for. Its value is recognized and transformed into an ideal for group relations. Note that this does not make solidarity essentially moralized: solidarity is still a descriptive phenomenon, but a phenomenon whose value becomes recognized and championed by individuals in their local struggles.

IV. CONCLUSION

In this article I have tried to provide an account of solidarity that is rich in explanatory, historical, predictive, and normative power—an account, that is, that vindicates its status as a social kind. I have also tried to show how it is distinct from related notions, such as humanitarianism, altruism, fellow-feeling, social cohesion, and justice, and therefore how it can be a productive category in its own right for understanding the social world. We are now able to return to some of the issues raised in the first parts of this article with a fresh perspective. Recall the problem of disagreement regarding solidarity: what’s in a name? To put it another way, why is it important to call the social kind I have identified ‘solidarity’ rather than something else? There are three main reasons to pair ‘solidarity’ with the phenomenon I have discussed. First, the social phenomenon, I have argued, was in fact the referent of the term as it was used in its nineteenth-century heyday. In the late nineteenth century, social theorists, sociologists, and political actors were pointing to the phenomenon in trying to ground, build, and motivate the nascent structures of the welfare state and the campaigns of international social movements. Second, using the term to denote the kind allows us to make its use more precise, and so more amenable to empirical and normative study and elaboration in the ways discussed in Section III. Third, using the term to denote the social kind identified brings out what makes solidarity distinct from other social and political concepts, and so aids in our understanding of its social, normative, and political significance.

This is not to say, of course, that the term can’t be used in more stipulative ways, or to denote other (potentially broader) phenomena. I do not claim that someone using the term in these other ways must be speaking falsely. What I do claim is that such uses would be missing a phenomenon of central social, political, and normative importance—one associated with its paradigmatic and historically most significant uses. Of course, such usages might not be trying to isolate such a phenomenon, but in that case they are speaking about something different (recall: lawyers identifying the meaning of ‘solidarity’ in a legislative statute). Furthermore, using the term to denote other phenomena or other sets of loosely connected properties would also be likely to muddy our understanding by confusing or blurring other concepts with it (as I argued was the case with its association with attitudes and actions that would otherwise be called ‘humanitarian’). There is much to be gained in our understanding of the social world by, for example, disambiguating solidarity from humanitarianism, altruism, coalition-building and related concepts.

Along the way, we have also seen how the schema can help us to isolate a powerful, but understudied, type of social kind, namely a social kind that is not socially constructed (and so unlike categories such as race, gender, and marriage) but nonetheless subject to social modification through being politicized and mobilized (analogously to categories like social status, racism, and sexual harassment). Each of these categories exists and is unified independently of being recognized or conceptualized in our linguistic practice. But once it becomes an object of social reflection by being named, and people begin to identify and self-identify as participants in the category, the category’s boundaries (along with the normative valences and broader practices associated with it) can shift. In such cases, the kind becomes subject to social modification rather than social construction.

ACKNOWLEDGEMENTS

This article was written with the generous support of an ERC Consolidator grant (grant number 771635). I am deeply grateful to Jude Browne, Rainer Forst, Marion Godman, Eliot Michaelson, David Papineau, and Philippe Van Parijs for their invaluable feedback on earlier drafts of this article. Finally, I am grateful to two anonymous referees and Bob Goodin, the editor of this journal, for their thoughtful comments and suggestions.

COMPETING INTERESTS

The author declares that he has no competing interests.