What is character? According to most contemporary accounts, a personās character consists in a collection of psychological dispositions, known as character traits, which manifest themselves in how she thinks, feels, and acts.1 She can change her fashion, her musical tastes, her favourite food, or her preferred pastime and remain, in all interesting senses, who she was. She may move from one school to another, one state to another, one country to another and still be the same person. She may go so far as to change her political party, her name, and her marital status, and yet remain fundamentally unchanged. But change her character and you change her identity. A personās character goes to the heart of who she is. Even within this core, some aspects are more central than others. A person can act like someone other than she is, i.e. her behaviour can be out of character, without that changing who she really is. Thus, within this core, the most internal aspectsānamely, how she thinks and feelsāare what really matter for the constitution of her character. Actions, on such an account, are important only to the extent that they manifest these inner qualities, revealing what one thinks and feels, disclosing who one is.
However, it has not always been thus. Earlier accounts of character did not share a number of assumptions upon which 21st century accounts depend. During the 17th and 18th centuries, character was thought of in quite different terms. Character, as Hume and his contemporaries conceived it, has a social, rather than an individual, ontology (Pettigrove 2015). What others think of you partially determines your character. One place in which the difference between earlier and later perspectives is reflected is in the way a personās social roles are thought to relate to her character. At present, writers take a personās character and her roles to be independent. Earlier writers, by contrast, took social roles to be ingredients of character.
What should we make of this difference? I suggest it gives us reason to reconsider our account of character. This chapter will not settle the debate between these competing conceptual schemes. It aims, at a minimum, to keep contemporary theorists from dismissing the earlier view out of hand. More ambitiously, it identifies a number of respects in which the earlier account of character surpasses the contemporary alternative. I begin by discussing the nature of social roles. Section 2 looks at the work that our conception of character does for us and argues that it can do its job better if we include roles within our account of character. Section 3 lays out the most pressing objections to taking roles to be ingredients of character. In the final section I respond to those objections and show how we need a conception like the one with which Hume and company were working in order to fill a sizeable gap in contemporary accounts of the moral life.
I. Roles
Social theorists employ the language of ārolesā to designate a number of different things. More often than not, when George Herbert Mead speaks of roles he is picking out individual points of view within the context of a relationship between two or more people (Mead 1934, 141n., 254, 364ā376). So, for example, he speaks of a particular infant taking the role of her parent. At the outset, this merely involves mimicking some of the parentās characteristic actions, but as time goes on and the child develops a more robust awareness of other minds, it can also involve imagining how things seem from the otherās vantage and thinking or feeling some of what they think and feel (Mead 1934, 364ff).
However, sometimes Mead speaks of more generic roles: āA child plays at being a mother, at being a teacher, at being a policeman; that is, it is taking different roles, as we sayā (Mead 1934, 150). Here āroleā no longer means a particular point of view, but picks out a social position that any number of people might take up. And this is the more usual meaning given to the term in social theory (Turner 1968, 552; Hardimon 1994, 335). However, describing it in this way might give one the impression that the agentās point of view had been lost. This would be misleading, for when the child plays at being a teacher, she not only performs actions that are associated with the role, she also notices the kinds of things teachers notice and plays at caring about the sorts of things teachers care about. Thus, the social position she takes up is one that carries with it a stereotypical point of view, as well as distinct ways of acting and interacting.
Peter Berger and Thomas Luckmann highlight how these two uses of āroleā might be related to one another. When two agents interact over a significant period of time, patterns emerge (Berger and Luckmann 1966, 56; similarly, Goffman 1961, 96). One party repeats the same type of action from what appear to be the same kinds of motives and the other party begins to expect her to perform actions of that type and motivation under relevantly similar circumstances, adjusting his behaviour accordingly. As they become aware of these patterns, the parties begin to repeat them self-consciously. Each begins to play her own role, which is to say, each continues to act and feel in the patterned way, but now with an awareness that the other is expecting her so to act and feel. This makes their actions and motivations more predictable to the other and reduces the effort required to determine how to go on together (Berger and Luck-mann 1966, 56ff).
Berger and Luckmann discuss two factors, in particular, that facilitate the shift from roles of this local sort to more generic social roles. The first is the addition of third parties. The more people there are who observe the pattern and begin to form their expectations in light of it, the less it is up to the two actors who started it to change the pattern whenever they wish. The second factor is the process of socialisation. When one comes into a social situation where there are already well-established patterns, they appear less dependent on the preferences and habits of the individuals in question and more a matter of āhow these things are doneā (1966, 59). This is especially true for children who are socialised into an environment in which these patterns are on display. As children, we are prone to take the patterns we have observed to be normative. But it is not exclusive to young children. Similar things happen for new employees who are socialised into the way a company does things. Berger and Luckmann suggest that both observation and socialisation āthickenā or āhardenā social roles. They go from having a merely subjective reality for the original actors to having an āobjectiveā reality as they become fixed in a larger social or institutional context (1966, 59).2
A third factor ā one not discussed by Berger and Luckmann ā that facilitates the shift to generic social roles is the opportunity to notice similar patterns emerging between other individual actors.ā If the pattern we recognise is not just observed in the interactions of A and B, but also in those of C and D as well as E and F, then we are more likely to find it useful to speak of it in terms of a general role that many people might inhabit, rather than a point of view that is indexed to one or two particular individuals. And once language has been introduced or employed to label the pattern, the role becomes even more firmly established.
The kinds of roles that figured in 17th- and 18th-century treatments of character were roles of this more general sort. They included the role of the patriot, the soldier, the sailor, the gentleman, the physician, the philosopher, the magistrate, the elder brother, the wife, the chambermaid, the widow, the judge, and the jailor, to name just a few. Varied though this list might be, one could group these roles together under two primary headings. Generally, the roles that were thought relevant to oneās character were defined in terms of oneās status or function.3 However, as gender-based roles illustrate, the divide between these two categories was not rigid. Gender pertained to both status and function. The elder brotherās status was above sisters in the familyāeven older sistersāand he was expected to perform a number of traditional functions (including taking responsibility for the welfare of the rest of the family when the father died). Similarly, the role of a judge was defined around its function, but it also was associated with an elevated social standing (as compared to the role of the jailor or the physician). So the distinction between status-based and function-based roles should be thought of in terms of two circles in a Venn diagram that partially overlap. Although they overlap, there remain important differences between status- and function-based roles. The most important of these differences pertains to how the role might be lost. One could be a gentleman landlord even if one performed none of the associated functions. One would be a bad landlord and a lacklustre gentleman, but a gentleman landlord all the same. However, someone who performed none of the functions associated with carpentry would not be a carpenter at all. Or if he started off as one, because at one point he worked with wood, he could cease to be a carpenter if he stopped building for long enough.
Some of the roles discussed by 17th- and 18th-century writers, like some of the roles mentioned by contemporary social theorists, are normatively thin. They are general groupings that describe typical behaviours and may aid an astute observer who wishes to predict what those who fit the description will do. Other roles are normatively thicker, picking out not only what role-occupants characteristically do, but also what they should do. The role of the ācollege graduateā would be an example of the former.4 There are skills we may expect a college graduate to have acquired, but there are not moral ideals we expect college graduates to live up to simply in virtue of being graduates. Similarly, the role of āfelonā5 is normatively thin. This is not to say that the concept felon is normatively thin. On the contrary, it is a normatively thick concept; when we apply it to another person we are thereby evaluating her. Nevertheless, the role with which it is associated is normatively thin. The role of felon does not involve a normative script that includes guidelines for how the role-occupant ought to behave, such that the role-occupant need worry about living up to the standards intrinsic to the role. It would be odd for someone to think, āIām a felon, but I havenāt committed a crime in months. I need to lift my game!ā By contrast, normatively thick rolesālike spouse, high court judge, and spiritual leaderādo involve such a script, and if one fails to live up to the roleās standards then one is susceptible to shame and subject to censure.6 There are a number of normative (as opposed to merely predictive) expectations that define the latter. Both thick and thin roles can each be used in utterances like, āOf course you should [X]. Youāre her [Y]. Thatās what youāre supposed to doā (where X picks out an action and Y a role). But the thicker the role, the more (and the more binding) Xās there will be.7
Closely related to the normative thickness of a role are the enablements and constraints attached to it. Some social theorists define social ontology in terms of enablements and constraints, such that X only has a social existence as a certain kind of thing to the extent that being an X enables one to perform certain actions or fulfil certain functions and constrains one from doing others.8 Whether or not that is the best way to characterise social ontology, attending to what they enable one to do or constrain one from doing is an important part of understanding social roles. There are certain actions, such as sentencing a criminal to life in prison, that can only be performed by a judge. Being a judge enables one to pass a legally binding sentence. Occupying the role (of judge) is the condition for the possibility of performing the action (of sentencing). In other cases, occupying the role is not required to perform the action; it is just required if one is to have permission to perform the action. Anyone with resources, land, a basic knowledge of physics, and rudimentary woodworking, plumbing, and wiring skills could build a house. If they have more than rudimentary skills, they might even build it well. However, in many social contexts, if they are not recognised as a carpenter (i.e., as occupying not just the functional role but also the social role of the carpenter), then they will not have permission to build a house. As a result, they may be obliged to take down the house they have built or pay a fine for having built it without the required permissions.
Similarly with constraints, occupying a role may constrain one from performing certain actions. If we ignore social roles for a moment, most cultures are fairly permissive about the giving and receiving of gifts. Anyone can give almost anything of which they are the sole owner to anyone else. However, in many social contexts if one is a candidate for political office, oneās role restricts the kinds of gifts one may receive and the range of sources from whom one may receive them. Likewise, if one is a plaintiff in a criminal case, oneās social role constrains oneās gift giving activities; one may not give a gift to the judge, the district attorney, or members of the jury.
The norms that enable or constrain occupants of social roles will be an important part of the script of normatively thick roles. For example, it is only because her role enables her to marry couples that a minister of the church has to ask whether she should marry this couple. And it is because people in her civic environment or religious denomination are only permitted to have one spouse at a time that her role constrains her from marrying a couple if she knows that one of them is currently married to someone else. However, role-related enablements and constraints need not account for every feature of the script. A minister of the church might also be expected to pray at public events, not because she is the only one who can pray in public (since these words would still count as a prayer if spoken by someone other than a minister of the church) or who may pray in public (since in her social and religious context laypeople are also permitted to pray at public event...