Beyond empathy: Role-Taking as a structural approach to participatory design

Human-computer interaction (HCI) and psychology are longtime disciplinary affiliates, and for good reason. Psychological theories predict and explain mental, emotional, and interpersonal dimensions of the human experience. HCI researchers draw on those theories to illuminate human-machine relations, the social implications of hardware, software, and interface design, and the interpersonal dynamics between designers and (real or imagined) user-publics. It is from this disciplinary pairing between psychology and HCI that concepts like “empathy” and “perspective taking” have risen to the fore, serving as both objects of study and targets of intervention. Empathy, in particular, weaves throughout HCI, taking a variety of forms. This includes machines’ expressions of empathy (e.g., [1, 2]), humans’ experiences of empathy towards machines (e.g., [3, 4]), HCI applications designed to increase out-group empathy (e.g., [5-7]), and methods for improving designers’ empathic capacities and sensitivities towards user-publics (e.g., [8-11])—the latter underscored by a Design Thinking process that positions “empathy” as a first and fundamental step [12, 13].

Although enduring and prolific, the relationship between HCI and psychology is necessarily limited by the individualized focus of psychological approaches vis-à-vis HCI's socio-structural foundations. Human-computer interaction is at its core, a social process, suffused by structural factors that inform both technical design and interpersonal interaction. Consider here the structural-relational turn in affordance theory [14, 15]. Affordance is a central concept in HCI that describes the opportunities and constraints of given objects based on their design features. While early affordance theory positioned designers as psychologists charged with communicating to users, through design, how an object ought to be used [16], the structural-relational turn clarified that “users” are non-homogonous, technologies are never universal nor neutral, and that opportunities and constraints are conditioned by social context [15]. With this turn, affordances were recast in a sociological light, modeling any human-computer interface as an assemblage of social factors and structural forces [17]. Continuing this interplay of sociology and HCI, I propose a shift from the psychological construct of “empathy” to the sociological construct of “role-taking” as a foundation for HCI research and methods, scoped here to implications for participatory co-design.

Participatory co-design is premised on a commitment to community stakeholders and recognition of lived experience as essential expertise [9]. This commitment is part of a broader ethic underpinning co-design through community-based participatory research (CBPR), which seeks to reconfigure power relations between participants and practitioners [18, 19]. This principle of community empowerment drives best practices in co-design such that end-products will reflect community stakeholder interests. However, such community empowerment does not always materialize. Co-design contexts can, and often do, inadvertently re-entrench normative social hierarchies and mechanisms of privilege [20]. I suggest that role-taking has the potential to mitigate this fallibility, enhancing precision in co-design protocols while focusing status and power within co-design processes.

In what follows, I present role-taking as a conceptual scaffold for participatory co-design. Role-taking structurally orients the design process, adds precise targets and indicators that account for both affect and cognition, and draws from and builds upon empirical research about how status and power affect and are affected by, interactional arrangements. I begin by reviewing role-taking as used in sociology, addressing theoretical developments and empirical trends in role-taking research. I then discuss how role-taking can productively figure into HCI co-design processes, with particular attention to status and power. To ground the argument, I draw on examples of HCI co-design protocols geared towards enhancing empathy, pointing to these methods’ weaknesses when addressing social hierarchies in practice. Finally, I suggest modest adaptations to the aforementioned protocols, laying groundwork for testable hypotheses and related interventions that stem from a role-taking research program.

Introduced in the early 20^th century, role-taking was first described by George Herbert Mead as the process and practice of putting the self in the proverbial shoes of another or others [21]. Crucially, role-taking is a structural process, premised on mutual recognition of interrelated role-positions and their situatedness within socially defined communities and contexts. Role-positions are the structural categories people occupy, such as parent, student, doctor, inmate, or patient. Roles are supra-individual, operating as categorical constructs that transcend specific role-occupants. For instance, the parent and student roles remain part of the social order, regardless of who fills those roles at any given time. Each social role comes with shared sets of norms and expectations about the meaning of that role, how occupants of the role will behave, and how people of various roles will interact with each other in different contexts. For example, there are norms and rituals that define interactions between parents and students, which differ from the expectations about interaction norms between parents and doctors, each variously shaped by situational circumstances (e.g., a party versus an appointment). All people occupy multiple roles simultaneously, with various combinations of those roles emerging as more or less relevant in any given situation. Within role-taking theory, each interaction and each person in that interaction are defined by the combination of situationally relevant roles. In this way, social encounters are not remade anew nor are individuals treated as unique entities. Rather, social interactions are structured by shared templates for self, other, and situation.

Stemming from Mead (1934), role-taking integrated into the sociological canon and now underpins myriad theories of self and society. However, role-taking itself remained relatively understudied for many years, serving as an axiom rather than object of inquiry. This changed over the past two decades, as researchers have advanced role-taking's theoretical formulation, tested its parameters and effects, and applied role-taking as an explanatory variable in a variety of fields, including HCI and human-AI interaction [22, 23]—a project for which continued development will benefit both the HCI domain and the role-taking research program.

HCI has a repository of strategies, tools and toolkits meant to elicit greater empathy and perspective taking from design teams, with the goal of building products and services that meet the needs and priorities of diverse user-publics [8, 9]. Many of these tools operate as proxies for users who are not directly involved in the design process, facilitating various forms of role-play among designers. An illustrative sampling of these include stakeholder tokens, which prompt designers to identify a range of direct and indirect stakeholders who will be affected by the design of a system, chart the relationships between those stakeholders, and visually depict the feelings and experiences of stakeholders in relation to the system in question [32]; empathy maps, through which designers imagine the thoughts, feelings, and behaviors of target subjects as they engage with or encounter a technology under consideration [33]; and physical tactics, like IDEO's use of rubber gloves and blindfolds to test products from the perspective of people with mobility and vision impairments, respectively [34, 35]. Although proxy-based methods are not the primary focus of the present work, these examples highlight empathy's entrenchment as an orienting concept in HCI practice.

HCI's focus on empathy extends and amplifies within co-design, where design teams engage stakeholder communities to collectively ideate, build, evaluate, and/or adapt HCI applications. The very premise of co-design is based on enhancing awareness and understanding of various publics, building their perspectives into a product or service. Within HCI, empathy enhancing co-design tools aim to increase empathic sensitivities among designers and design teams, while crafting prototypes and end-products that represent a meaningful collaboration [36]. This imperative drives CBPR in which community stakeholders are the focal point of design processes.

The examples of co-design strategies and toolkits aimed at enhancing empathy are bountiful. The Value Sensitive Design corpus, for instance, offers envisioning cards as one way to imagine the long term and multi-faceted effects of a product or service [37]. Researchers have applied this tool to address mobile-phone safety among youth experiencing homelessness, collaborating with homeless young people, police officers, and service providers to understand how design elements interplay with these stakeholders’ needs, ambitions, concerns, and vulnerabilities [38]. Envisioning cards are but one of many card-based tools that prompt empathic imaginings [39]. Microsoft's AI Ethics Team (now defunct) also emphasized empathy in their Responsible Innovation Toolkit, which I observed firsthand while participating in a pre-release demo session with representatives from the company, who began the presentation with a segment on empathy and closed with a promise of empathic enrichment. The toolkit consists of three ostensibly empathy-enhancing tools: judgement call, harms modelling, and community jury [40]. Community jury is the co-design tool within this kit, and it operates by bringing together diverse stakeholders who “learn from experts” (i.e., design teams) while providing insight and feedback about product design, allowing researchers to better “identify stakeholder values and understand the perceptions and concerns of impacted stakeholders”. Governments and public sector entities, too, evoke empathy as both inspiration and aspiration for co-design practice. For instance, representatives from the USAGov product team boast of their empathic approach from “ideation to launch” for a newly released benefit finder tool, described in a blog post titled “Designing with Empathy” [41]. Here, they describe a combination of expert research and “transcreative interviews” with English and Spanish speaking Americans who recently withstood difficult life events for which the benefit finder would be of use. The blog's authors open with a cautionary note that design without empathy may thwart opportunities for the public sector to effectively support the people and communities they serve.

The strategies, tools, and toolkits just described are by no means exhaustive. Rather, they are exemplar selections spanning academic, industry, and public sectors, showcasing a shared and crosscutting interest in empathic approaches to HCI. I set them out here in preparation for Section 4, where I will make the case that these strategies, tools, and toolkits are best understood, anchored in, and/or reconfigured through the conceptual lens of role-taking, informed by role-taking's attention to structures of status and power. In making this case, I suggest not only different vocabulary (i.e., role-taking instead of empathy), but also subtle adjustments to co-design protocols in service of amplifying and effectively integrating the needs and experiences of diverse user-publics, in line with the motivating ethic of CBPR as a practical and political project.

There are two main reasons to shift from empathy to role-taking as the underlying framework for co-design processes. First, is accuracy and precision. Second, is the evidence-based practical adjustments that role-taking facilitates, attending to asymmetries of status and power within the co-design encounter. Shifting to role-taking is thus a generative move for HCI as a field committed to meaningful incorporation of users’ voices, and in particular, inclusivity for those who have been historically marginalized and/or excluded. In these ways, role-taking dovetails with best practices in community- based research, scaffolding those practices with a multidimensional sociological model.

As a professionalized field, HCI ought to be accurate and precise in its conceptual language. Role-taking, rather than empathy, more accurately describes the goals and practices of co-design¹, while role-taking's multifaceted definition, which includes both affect and cognition, specifies where, how, and to what extent empathy and perspective taking are at play. Recall role-taking's structural underpinnings, in which attention to others’ thoughts, feelings and behaviors are predicated on shared understanding of interrelated social roles rather than the unique inner lives of specific individuals [21]. Co-design is about listening to and understanding the participants involved in co-design activities, but with the larger purpose of understanding and accommodating stakeholders like those specific participants—i.e., those who occupy similar social roles. Unless a technological system is one-off and bespoke, which is unusual, the co-design process is about optimizing role-taking as conceived sociologically, rather than empathy as conceived by psychology. This relates to another matter—how empathy is conceptualized. Within psychology, empathy is a saturated concept, besieged by so many meanings it is difficult to pin down what exactly one is striving for and/or measuring when empathy is set as a target. Role-taking resolves this issue by tightly scoping empathy as an affective component within a broader structural process, defining in clear terms its distinction from the cognitive process of perspective taking [24]. Role-taking is thus a more accurate and specific approach to HCI than is empathy, with the former benefitting co-design protocols and sharpening measures of effectiveness.

Not only does role-taking offer definitional accuracy and clarity to co-design practice, its structural approach and related empirical evidence illuminate the fallibilities of co-design methods and provide a theoretical overlay for best practices. Co-design is inflected by status and power imbalances, whereby designers possess both esteem and authority in ways that community participants do not [9]. Role-taking research alerts us to the effects of these status and power asymmetries [25, 42]. In particular, we can expect status and power to diminish role-taking capacities, such that design teams and UX researchers will face sociological barriers to understanding, integrating, and implementing input from community participants. We can expect this effect to amplify when community participants are from marginalized and thus low status groups, like people with disabilities, elderly populations, stakeholders of color, people experiencing economic precarity, and those at the intersections of these and other axes of inequity—a problem that has plagued HCI practitioners whose efforts at co-design too often reproduce normative social hierarchies [20].

Evidence from the role-taking research program can sensitize HCI practitioners to issues of status (esteem) and power (control) in co-design encounters, while at the same time, suggesting modes of correction. Role-taking studies show that adjustments to status and power relations within an interaction situation can undo asymmetries in role-taking performance. For example, when women are placed in leadership positions, the men with whom they interact role-take with those women more actively and accurately than women in non-leadership positions [42], with similar results in studies of role-taking between black women and white women, where race and leadership role served as intersecting status markers [25]. This means that changing the definition of a situation can change how well people role-take. Applied to HCI co-design methods, these findings indicate that elevating community participants’ status and granting them meaningful forms of power within co-design sessions, will enhance the role-taking environment. While these dynamics are already understood by HCI researchers and embedded within CBPR standards, they lack a cogent theoretical underpinning and are thus at risk of inadequate implementation.

This paper proposes role-taking as a generative framework for HCI co-design, moving beyond empathy as an individualized and psychologically grounded approach. Role-taking is a sociological process of understanding, connecting to, and acting upon others’ thoughts, feelings, and behaviors. It entails comprehensive attention to interrelated and contextually situated social roles, rather than the inner lives of specific individual subjects. Moreso than empathy, role-taking comports with best practices in co-design, especially CBPR, where the purpose is to serve a diversity of user-publics, attending to the needs, aspirations, and vulnerabilities of social roles those user-publics occupy. Within role-taking's formulation, empathy is not lost, but conceptually and operationally scoped as the affective element of a structural whole, interrelated with cognitive processes of perspective taking. Role-taking is thus an accurate reflection of HCI co-design goals while offering theoretical precision and specificity. More than this, role-taking's structural focus and the empirical evidence from role-taking research elucidates the ways status and power figure into co-design practices, fostering adaptations that alleviate asymmetries and enhance meaningful inclusion of community perspectives. This move stands to enrich co-design overall, with particular benefit to co-design initiatives targeting underserved and minoritized groups (i.e., those who are not and have not traditionally been, privileged or default subjects).

Beyond making a case for role-taking in HCI co-design, this paper lays groundwork for testable hypotheses. For example, we can hypothesize that elevating community co-designers into expert roles, remunerating them for those roles, and empowering them with feedback mechanisms and decision-making capacities will enhance role-taking performance among designers and design teams, and result in prototypes and products that are more satisfying/less harmful to the communities involved. Such hypotheses are imminently testable, with modest adaptations to existing protocols, like those suggested in Section 4.1, paired with assessment metrics gleaned from extant role-taking and HCI research. Moreover, if we take seriously the consistent finding from role-taking research that people with lower status and power within the social structure are better at role-taking, then we have evidence-based reasons to diversify design teams, with perhaps over-representation of historically underrepresented groups. This is not just a matter of fairness or inclusion, but of building design teams who are good at their job. This too, can be empirically tested. In sum, role-taking provides a conceptual scaffold, evidence-based justification, and methodological tools to test the efficacy of structurally situated projects and programs, compared against empathy-oriented approaches. Outcomes of such tests will be vital for incorporating role-taking specifically, and sociology in general, into the fabric of HCI.