A lot hinges on what we think explains social phenomena. Consider the question: Why do so many more women than men quit their jobs to care for their young children? Attribute it to individual free choice alone, and there is nothing more to say, and no injustice to confront with changes in our policies or practices. Attribute it ‘society’ and you risk not only being painfully imprecise, but also failing to respect the autonomy of the women who make these decisions.

Haslanger’s paper undertakes extraordinarily important groundbreaking work in aiming to give an account of structural explanations and of the social structures that figure in them. The account she provides has the crucial advantage of threading around these traditional alternatives, giving an account of structural explanations in a way that makes clear that they are not in conflict with individualistic explanations and are compatible with acknowledging the role of actors as autonomous agents making rational decisions. But while structural explanations are compatible with individualistic explanations, she argues convincingly that structural explanations appealing to social structures are “in many ways preferable to individualistic explanations”.

I begin by reviewing what structural explanations are, how they are compatible with acknowledging agent autonomy, and the ways in which they may be preferable to individualistic explanations. In all of this I think Haslanger is entirely successful. My only real worry concerns the breadth of the account. For while the main account Haslanger offers works well in explaining how a social structure may figure in the best explanation, even where rational choices are involved, it may not yet do enough to capture all the deep and sometimes insidious ways in which social structures can and should figure in explanations. I’ll suggest that attention to the normativity inherent in social structures may enable us to go further in developing structural explanations of other kinds, including some that may not meet the ‘agent autonomy’ requirement.

1 Structural explanations and their virtues

Structural explanations in general are, roughly, explanations of something in terms of its being part of a larger system that constrains it—as we explain the trajectory of the dog treat in terms of the ball that it is packed in. By contrast, individualistic explanations appeal primarily to the behavior of the individual—in the social cases at hand, more specifically, to the psychological states of the individual. We can also distinguish (with Dretske) between structuring causes (appeal to which can explain why this outcome rather than another one happened) and triggering causes (what causes the event to happen now). An explanation that appeals to a structuring cause is not in conflict with one that appeals to a triggering cause such as an agent’s decision.

In what ways might social-structural explanations be superior to individualistic explanations of social phenomena? Haslanger suggests several ways:

  1. 1.

    Structural explanations in general better enable us to intervene (as the ball-based explanation would better enable us to intervene to save the treat).

  2. 2.

    They enable us to offer explanations that are better in the sense of enabling us to ask, and answer, better questions: properly addressing the ‘foil’ cases (as it is only by taking into account the structure of possibilities laid out by the ‘curve’ grading system that we can properly explain why Mary got the (only) A).

  3. 3.

    They enable us to offer more ‘stable’ explanations, in the sense of explanations that apply to a broader ‘type’ and would continue to hold good even under various changes.

  4. 4.

    Social-structural explanations also enable us to better identify and critique social injustice than individualistic explanations can (as in the case of Lisa and Larry seeking childcare).

2 Structural explanation and the autonomy constraint

The appeal to structural explanation in social science is controversial, since some hold that explanations of behavior must be in terms of psychological states (prominently, belief and desire) and that “no social explanation ought to be acceptable, if it assumes that people behave otherwise than in an agent-autonomous way” (Jackson and Pettit, as quoted p. 16). Some of Haslanger’s initial examples might make one worry on that score: we might resist a model of human behavior within a social structure that made us look ‘constrained’ in a way that the behavior of a dog treat in a thrown ball is. Nonetheless, I think it is clear once one looks at her view as a whole that structural explanations, in her sense, are perfectly compatible with this autonomy constraint.

This comes out once we see how social structures are related to social explanations. Social structures ‘impose constraints on our action’ not in the way that the movement of a ball constrains the movements of the treat enclosed in it, but rather (for example) in that they set up a ‘choice architecture’ for individuals who occupy various ‘nodes’ in the social structure:

Even if individual psychology provides triggering causes of our choices and actions, not only what we do, but what we can do, depends on social structure. The structures socially constrain our behavior by providing the options available…

It might be helpful to spell this out in detail using the central example: Lisa and Larry (the parents with parity but without childcare) occupy different nodes in our social structure. As a result, the choices available to them differ: Larry can choose both to keep his job, retain more than half of the family income, and to have his child well cared for; Lisa cannot. In more detail (but of course still massively simplifying: most importantly away from the social approval/disapproval incurred by various choices) Lisa can choose (a) to keep her job, keep less than half of the family income and have her child well cared for; (b) or to quit her job, keep more than half of the family income, and have her child well cared for; or (c) to keep her job, keep the full family income, and leave her child without decent care. The social structure determines the combinations of options available to her—and they are different from the combinations of options available to Larry. This is what enables us to identify this form of social injustice—an injustice that persists even if one says ‘but Lisa chose to quit her job’. The immediate ‘triggering cause’ (figuring in the individualistic explanation) is still Lisa’s beliefs and desires. And indeed we can say that, given her preferences and the options available, she chose rationally: her choice is rationally explicable and we can treat her as an autonomous agent. So we can meet the ‘autonomy constraint’.

But crucially, while we meet the autonomy constraint, the structural explanation enables us to identify an injustice in the situation that would be completely papered over if we limited our explanations to the ‘triggering’ causes in individual choices. The crucial and insightful feature we get out of the structural explanation is, first, a more stable explanation (that will be generalizable to explain and predict the choices of other individuals who occupy the same ‘nodes’ in the structure), and secondly and more importantly, a way of identifying a social injustice: that “women as a group are structurally situated so that it is rational for them to choose options that keep them subordinate”.

I think this is right, crucially important and insightful. In her aim to show why structural explanations are important and often superior to individualistic explanations, and that they are compatible with acknowledging agent autonomy, I think Haslanger is entirely successful.

3 Broadening social-structural explanations

But while Haslanger makes it clear that social structural explanations are compatible with acknowledging agent autonomy, there are many other cases in which the model of explanation that appeals to autonomous choices within a socially-structured choice architecture doesn’t seem so apt:

  • Why don’t more women go into engineering (even when they have the skills and aren’t physically or institutionally prevented)?

  • Why aren’t more working class British kids applying to Oxford and Cambridge, when they aren’t prevented, and many have the capability and could get the funding?

  • Why don’t more women speak up in class? Why are women more often interrupted, not as often given meetings with professors, etc.?

  • Why are resumes of minority candidates evaluated as worse than those of comparable white candidates?

I don’t mean to suggest that no choice-architecture explanations could be forthcoming for any of these cases—perhaps by appealing to expectations of praise and blame or other social incentives, rather than just economic incentives—but for many cases like these they don’t seem so promising. Haslanger concurs, saying that the model that says merely that “social constraints… have their effects by creating social expectations and anticipation of blame and praise” is ‘inadequate’. As is generally acknowledged, problems like those above aren’t mainly the product of the people in question considering the option but rejecting it as disadvantageous given (in part) social pressures. Instead, there is a deeper problem: that so many girls never think of pursuing engineering, so few working class kids even consider pursuing an Oxbridge education. Similarly, people not only don’t consciously and explicitly choose to interrupt women more or downgrade minority resumes, but don’t even realize they do so. So we might look for a broader account of social-structural explanation that can also help explain social facts like these.

Haslanger seems to agree that social-structural explanations are not limited to the rational choice within a choice architecture model. Indeed as she elucidates the account of social structures in the second half of the paper, she writes:

…structural constraints need not be a matter of causal processes that trigger action; social constraints set limits, organize thought and communication, create a choice architecture; in short, they structure the possibility space for agency… what we do, and what we can do, depends on [the] social structure within which we act.

Here, creating a choice architecture is listed as only one of the ways in which social constraints are relevant to what we do (and to explaining what we do). Social constraints also set limits—one form this takes, which Haslanger identifies, is setting material limits in terms of what resources are available (childcare centers, idli steamers). They also ‘organize thought and communication’.

In what sense do social structures ‘organize thought and communication’? Haslanger says less about this here, but does say that schemas ‘provide templates of interaction’ and public meanings that ‘are not up to the individual agent but depend on collective understandings and the resources that have been organized by those understandings’. I want to say a bit more about what these public meanings might be, and how they might figure in social structures and in giving other forms of social-structural explanation—explanations that don’t appeal to rational choice within a choice architecture.

I’ve argued elsewhere (2014), developing some ideas from Heidegger (1927/1962) and Ingarden (1961/1989), that norms of treatment—of how we are to regard and behave towards the relevant things—are constitutive of our pubic artifact kind concepts.Footnote 1 Part of what it is to be a fork is to be an object that is to be used in certain ways, to do certain things, by certain (sorts of) individuals. Part of what it is to be a church (rather than a movie theatre or night club) is to be a structure that is to be behaved in in certain ways—indeed in different ways by individuals occupying different nodes in a social structure (by priests versus choir members versus working electricians, sometimes by men versus women, by the confirmed and unconfirmed…). Part of what it is to be a flag is to be a piece of cloth that is to be treated and used in some ways but not others (not in the ways a dishcloth or tablecloth is to be treated) (Ingarden 1961/1989, 260).

Generalizing this, one might hold that at least many of the meaningful categories by means of which we navigate our social world (not merely ‘think’ about it) are intrinsically normative, building in ways in which we are to regard, to treat, to behave towards individuals and objects that occupy various nodes in the structure (as well as ways the occupants of various nodes are themselves to behave).Footnote 2 So similarly, I would suggest, social roles that form nodes in social structures are in part constituted by norms: part of what it is to be a police officer, a professor, a priest, or a waitress is to be someone to be regarded and behaved towards in specific ways (quite different from each other)—as well as to be someone who is to behave in some ways rather than others.Footnote 3 Social structural explanations, as Haslanger presents them, involve understanding ‘the individual as the instance of a type’. What I am suggesting is that, for many social types, part of understanding what it is to be an instance of the type is understanding the norms governing members of the type: what they are to pursue, to avoid, how they are to behave, how they are to regard people and objects that belong to other social categories…, and how others are to regard and behave towards them.

Part of the insidious nature of social nodes (of gender, race, class…) we identify by sex, appearance, national or historical origin, income, and the like is that they, too, come with norms for behavior towards and regard of the relevant individuals.Footnote 4 The norms for behavior towards women differ from those for behavior towards men: even in tiny, seldom noticed (but well documented) ways such as in what ways they are to be attended to in conversation (when or whether it is ok to interrupt or talk over them, when close attention must be paid, etc.), when one is to defer or free to contradict, how they are to be compensated for their work, as well as what the individuals themselves are to do in various contexts, towards various individuals (parallel points could no doubt be made about norms for treatment of people of different racial groups in particular societies).

The idea that our social categories are linked to behavioral norms is one Haslanger herself has suggested elsewhere:

Typically, the act of classifying someone as a member of a social group invokes a set of ‘appropriate’ (contextually specific) norms and expectations. It positions her in a social framework and makes available certain kinds of evaluation; in short, it carries prescriptive force. (2012, 241)

What I aim to do is to suggest how this idea may play a key role in broadening the role social structures can play in our explanations beyond the methods Haslanger identifies in the present paper, where the focus is on the way in which social structures create an architecture within which explicit rational conscious choices are made.

I have been suggesting that social structures, and the nodes (and resources) within them are at least in part structured by norms. Social practices, too, are normative in ways that matter. Haslanger describes social practices (in the central cases) as ‘collective solutions to coordination or access problems with respect to a resource’. But she herself marks various reservations about this idea in a footnote (p.16n 10); I think we should have even more. Social practices certainly need not involve solutions that are ‘rational or mutually advantageous’ (p.16n 10), and they do have an explicitly normative dimension not captured in the standard definition: they involve ways things are to be done around here.

Haslanger suggests that schemas are ‘social meanings conventionally associated with things in our social world… but are also internalized and guide our behavior’. If we accept the idea that these social meanings and practices are normative, we can more easily see how these meanings (which ‘are not up to the individual agent’) guide our behavior. If we think of the very categories we employ to grasp the social world, which are constitutive of what it is to be a fork, a teacher, or a wife, as normative, involving ways one is to use, regard, treat… members of the category, it is easy to see how they may guide our behavior by making certain forms of activity show up as to-be-done, or to-be-avoided, or not show up at all. What we are to do is written all over the social world—in simply grasping social meanings and categories, we see what we are to-do (this doesn’t mean of course that one can’t rebel, reject the tugs of such signs and norms—but to do so requires resistance, thought, effort, not just following the well-greased channels).

A nice feature of this phenomenological approach is that it enables us to explain patterns of behavior without either treating everything we do as the product of explicitly considered rational choice (even within a socially structured choice architecture), or as mere mechanistic behavior or the product of blind dispositions. We do not consider all of the various things that could be done with a fork and then decide to eat with it; nor do we (except in strange circumstances) consider all of the various ways we could eat our food and decide to use a fork to eat on grounds of rationally considered preferences (we continue to use it even when no one else is around to scorn us, and when efficiency would call for other methods). In seeing it as a fork, we simply see it as something to be used in eating, and (unless we make a conscientious decision to rebel) proceed accordingly. The little girl who rushes to the pink aisles of the toy store and comes to examine the baby dolls and vacuums rather than the trucks and construction toys does not make an explicit rational choice that these things are preferable; nor is she simply ‘programmed’ by nature or society: instead, given her initiation into local social practices, the toys of the pink aisles simply show up to her as ‘for her’, as ‘to be played with’.

Do such explanations that go beyond the choice-within-a-choice-architecture model interfere with the idea that agents ‘behave in an agent-autonomous way’? It depends what is meant by that. If the autonomy constraint means that all social explanations must just go via seeing the action as the product of a conscious rational choice among all the (physical) possibilities, then that is a model we have good empirical (and phenomenological) reason to do away with (as I am sure Haslanger agrees). On the other hand, we certainly can accept that social structures make some options show up as ‘live options’, still others as to-be-pursued or avoided, while others are never noticed—without denying that agents choose or are self-moving actors. And we can accept that norms shape how we see ourselves, what behaviors show up to us as appropriate or inappropriate, coarse or polite or off-limits or simply weird, without mechanistically treating people as cause–effect cogs in purely mechanistic explanations. This analysis, after all, comes from phenomenology, from understanding people’s experience from the first-person point of view. But it is a first-person perspective in which some thing show up as to-be-done and others as to-be-avoided, or objects or people show up as to-be-treated-in-this-way-or-that. Understood in this way, phenomenology can play a key role in giving social structural explanations, pointing to ways in which social structures not only create choice-architectures, but shape our very perception of what choices there are, of what the proper ways of treating objects and individuals are, of what we are to do. These norms guide our behavior in ways that we often don’t explicitly and consciously recognize, and that don’t even require or call for considering various possibilities according to their costs and benefits and rationally choosing among them within a choice architecture.Footnote 5

Acknowledging these built-in norms might figure into social explanations in a variety of ways:

  • In determining which (among the various possible choices in a choice architecture) even show up to us as ‘live options’ for pursuing.

  • In channeling the ways we navigate our social world, even when we are not making explicit and conscious choices among options that we weigh in a choice architecture based on our beliefs and desires. (e.g. Norms shape the way children navigate a toy store, and the way adults sit, move, and speak in a job interview, in a classroom, etc., in ways that we generally don’t consciously choose, and that do affect our potential for success).

  • In channeling the ways others regard and interact with us (interrupting, deferring, attending to, criticizing, shunning, fearing)—again, often in ways that aren’t consciously chosen or even acknowledged.

The first two ways might be seen as ways in which social structures (the normatively-constituted nodes we occupy) may figure in social explanations that don’t go via the belief/desire/choice-within-a-choice-architecture model. For they shape our behavior in ways in which we typically don’t notice at all—and that don’t figure into our conscious choice among items in a choice architecture. The third might be seen as another factor that structures the choice architecture: e.g. given the norms of the social node associated with a professional woman, a female manager can’t choose to both act as aggressively as male managers and be liked (a male manager can). A young African–American man can’t choose to both walk through certain white neighborhoods in a hoodie and be regarded without suspicion; a young white man can.

Calling attention to the role these norms play also opens room for different kinds of critique and intervention, complementary to those we can get by looking at the choice architecture. For, once brought to light, we can critique the forms of injustice and practical problems that these subtle built-in norms can lead to (not only leading to a situation when fewer girls end up in well-paying jobs, but where there simply aren’t enough people going into engineering). Moreover, while these norms make certain ways of engaging with those who occupy various nodes to-be-pursued or to-be-avoided, that of course is not to say that we have good or moral reasons for acting in these ways. Our structures of norms may be critiqued on various grounds: of health or safety or efficiency or public good, or on more purely moral grounds: that these (morally) are not the norms we ought to have. So acknowledging the role of norms in social structures enables us to do still more to identify sources of injustice, beyond the fact that social structures create ‘architectures of choice’ that lead some choice combinations to be accessible to some, but not other, members of a population. Norms for treating women, people of color, immigrants, etc. may be subjected to critique if, in our reflective state, we don’t feel that members of these groups should be subjected to these different norms (in job interviews, by police or fellow citizens…).

All of this is frankly sketchy and programmatic, and none of it undermines the important achievements of Haslanger’s paper. What I have suggested is simply a broadening project, based on seeing how understanding the norms embedded in social meanings and the nodes and practices of social structures may shape our ways of navigating the social world. I think the focus on norms is quite compatible with work Haslanger has done elsewhere, as well as with some other threads of her present paper. By bringing the role of norms more upfront, we may make available social explanations that both impact explanations in the choice-within-a-choice-architecture model, and reach beyond it—opening the way for other forms of social-structural explanation and other forms of critique and intervention.