m. schraefel, Eric Hekler
The foundation of inbodied interaction is adaptation. Our inbodied, complex, nonvolitional processes are always on, always immediately responding and adapting to our context. Inbodied interaction invites us to explore what kind of adaptation we seek to create, as well as where, when, how, and for how long in an adaptation cycle. We refer to this inbodied intervention process as tuning. The focus of this piece is to explore particularly what we mean by tuning for health, and to consider what that approach opens up for HCI research and design.
The concept of tuning relates to aligning one component with another in order to work together. Something is always in tune with something else. This tuning is necessary for these related components to work together. In music, tuning relates to achieving certain harmonic resonances within an instrument and across instruments. In other physical systems, like engines, tuning relates to the timing or resonance of various components in order for them to operate in sync with each other. Often there are references to tune against, such as a timing gun for an engine set to a particular reference pulse, or an orchestra tuning all instruments to 440hz. Tuning in physical systems is also a necessary and regular process, as contexts can and do affect the state of a system, and thus affect that system's timing.
Tuning, in the context of inbodied interaction, is much like becoming resonant and harmonic with oneself and one's context, and simultaneously can involve engaging in actions that tune up the body to enable positive adaptation in context. In terms of the inbodied interaction framing of the body as the site of adaptation, adaptation is the automatic response of the nonvolitional inbodied processes overall; tuning is the set of volitional acts one engages in to effect adaptation. A key takeaway is that, as with an instrument or an engine, tuning is active and dynamic: It is regularly required since both use of the system and the environment in which that system sits impact the degree to which tuning is maintained.
As with an instrument or an engine, tuning is active and dynamic: It is regularly required since both use of the system and the environment in which that system sits impact the degree to which tuning is maintained.
What, then, are we tuning to? In music, an orchestra tunes to a common tone (e.g., concert A, 440hz). For a car, tuning occurs in relation to desired performance, such as achieving certain targets for acceleration, speed, longevity, and so on. To understand what we tune to with regard to health, the WHO definition is instructive: "a state of complete physical, mental, and social well-being, and not merely the absence of disease or infirmity." An important, often underrecognized word within this definition is state. Health can be understood as the condition someone is in within a particular context and at a particular time—a state. Inbodied interaction frames this resulting health state as always the output of the inbodied state.
A health state, like the complexity of the inbodied interactions that inform it, of course, is always complex, as it involves balancing over time physical, mental, and social well-being, which is influenced by the inbodied state at any moment. The inbodied state, in turn, is influenced by the constant dynamic impact of different behaviors: how context might facilitate or compromise one's health now, in the past, and over time; the impact of other organisms and other people on one's health; the many different life rhythms (e.g., seasons, changing life roles); and so on . A goal of tuning is therefore to improve a person's adaptive resilience across contexts. Inbodied interaction emphasizes that our mental, physical, and social interaction capacities are all fundamentally enabled and affected by the state of our always-on, always-adapting, nonvolitional inbodied systems. As one approach to tune these systems, inbodied interaction proposes the in5 model as five essential volitional processes—move, eat, engage, cogitate, and sleep—that one can tune interactively to support healthful adaptivity within and across contexts (see sidebar).
Tuning offers a broader conceptualization of health than what is normally used in health sciences. The dominant health paradigm espoused by medicine and epidemiology is built around diagnosing, preventing, and treating disease. In this paradigm, everything seeks to avoid mortality. Disease and related markers are created and ordered in importance relative to their impact on mortality. Successful interventions reduce deaths and, by extension, the incidence of diseases.
There is obvious value in this disease-oriented framing: If a person has an infection, is exposed to second-hand smoke, has developed colon cancer, has type II diabetes—each of these leads to a person adapting to the need to counteract these major issues. In acute circumstances, the healthcare system plays a pivotal role, offering one approach to diagnosing and treating disease: It supports health as an urgent, critical, single-factor response to adaptivity in context—inject this drug; undergo this operation.
While the WHO definition explicitly states that health is "not merely a lack of disease or infirmity," most positive health-promotion-oriented interventions (e.g., behavioral interventions, policies, contextual changes) are framed as prevention. Even for scientific studies with behavioral, policy, or contextual factors, the common metric defining whether it was valuable for health ultimately links back to reductions in disease and deaths.
Research on prevention, just like diagnosis and treatment, is vitally important. That said, just like treating disease, preventing disease is also just a small part of the broader possibility of adaptivity in context. In inbodied interaction, we recognize and foreground the complexity of the inbodied body. Health is likewise complex. Tuning for healthful adaptivity in context is of greater inherent complexity than the dominant paradigm for studying and advancing health currently supports.
To help illustrate how even the logic of prevention is limiting, it is instructive to think through how health behavior research is done. In particular, health behaviors are often defined according to national guidelines, such as engaging in 150 minutes of moderate-intensity physical activity per week. This is an extremely valuable general benchmark for understanding activity. That said, its emphasis, particularly in isolation from context and other behaviors, sets up a bias toward a more reductive, single-factor approach. Associated interventions are designed in terms of a single, or only a few, factor goals, such as eating vegetables, steps per day, or weight loss, reinforcing the implication that these could be advanced separately, rather than being intrinsically linked. A great deal of work from behavioral science and inbodied interaction, as well as common sense suggest that these behaviors are all highly interactive. As framed in the in5 model, for example, what someone eats influences how much physical activity they may engage in; how active someone is may influence how much sleep someone gets. Further, as discussed in , these behaviors interact in a highly dynamic, idiosyncratic, and multicausal way. When looked at through this lens, single-factor health guidance is not dynamic, not context-dependent, and not helpful for enabling a resilient approach to health.
Our tuning approach seeks to advance health dynamically by building and supporting a person's adaptive resilience across contexts. Central to this approach, tuning seeks to make such practices readily accessible without compromising the inherent complexity, in three ways. First, a tuning approach explicitly seeks to work in context and for the specific individual. Second, determining whether progress is being made occurs using the referent of a person's past, present, and desired future, instead of comparing them to other people, which is the more common referent in health sciences. Finally, a tuning approach reduces complexity by actively engaging with the person for whom health is being advanced to enable some options (e.g., certain behaviors to try, resources to use to advance care) to be quickly ruled out. We speak of this as designing and assessing interventions as they support a person gaining knowledge, skills, and practice of how to tune their health across contexts.
Of course, this logic of tuning is not unknown. A tuning approach fits well with more Eastern forms of care, such as yoga and Ayurvedic medicine. More recently within Western medicine, integrative health is being used to acknowledge the kind of tuning targets we seek. It is also common in sports science, where one could think of training as tuning to support peak performance at a particular event. As such, a great deal can be learned about a tuning approach from these other perspectives. Thinking about this more from the perspective of HCI, the work on self-experimentation could be considered a starting point toward a tuning approach.
If the classic health science model is biased toward very slow timescale effects (i.e., death) and works backward, the tuning approach is biased toward the current moment to move forward from there.
With each of these, a more healthy state is co-defined by individuals (e.g., the desire to sleep more , the desire for reduced irritable bowel syndrome symptoms ), and then a process is offered to them to tune their actions, to shift them toward these more desired new states. Further, the movement toward just-in-time adaptive interventions also fits with this general logic, as does the use of control-systems engineering methods for supporting health behavior change . In both of these intervention domains, support is provided in context, and progress is defined dynamically and in relation to each individual's personal prior and present actions and desired future state. For example, the MyBehavior study  could be thought of as a tuning intervention. It used a multi-arm bandit that had a simple, multiple-behavior, dynamic definition of health used to define success for each person within context—namely, walk more and eat fewer calories. To support this, it offered adaptive, context-relevant suggestions, such as alternative walking routes to planned destinations that were slightly longer than the normal path. These suggestions were timely and relevant to the person in context, offering them small, meaningful adjustments to make in context (i.e., tune) that, over time, would increase their adaptivity in context. Within inbodied interaction, the Experiment in a Box approach offers participants multiple paths for exploring health heuristics for each of the in5 . The approach enables participants to test out the heuristics in terms of a simple tuning evaluation, how do you feel? (see article on insourcing), to evaluate if and how a particular heuristic may work for them in their own contexts, before adopting any approach longer term.
Within HCI and the broader digital health research communities, there is already activity in advancing interventions that could be thought of as helping to tune one's health. We wish to stress that a particular framing from inbodied interaction is the need to communicate and emphasize that tuning itself is multifactored and interconnected. We need to challenge ourselves in our designs to incorporate this multicomponent interaction, in line with the complexities outlined in inbodied interaction and as implied with "small data" methods.
If the classic health science model is biased toward very slow timescale effects (i.e., death) and works backward, the tuning approach is biased toward the current moment, the current experience of how do you feel? to move forward from there: to improve one's current and ongoing normal—to make normal better. One area of work is to advance ways for understanding how these approaches may work synergistically together to advance health.
These different vectors also emphasize the foundational importance of thinking far more deeply about time, including the various timescales with which factors influence our health. What do we want our health adaptations to look like not only today, right now, but also in terms of how today enables five years, 20 years, from now? How do we design today for a lifetime of adaptations for ongoing resilience, while recognizing that our context and past experiences influence what is possible today? Such design thinking needs to include how we support social structures that foster longitudinal dynamic adaptation from the built environment to changing social contexts.
From a tuning perspective, these could be thought of as contextual factors that provide boundaries on what are plausible adaptations for a person or community to act on. They also imply the importance of understanding how past factors, such as a person's learning history, prior developed habits and routines, or social interaction styles influence what is possible, and, thus, tunable, in a given moment toward improved adaptivity in context. As work increasingly shows, zip code, for example, often situates ethnicity, socioeconomic status, and education, each affecting differences in opportunities to tune the quality of those volitional to nonvolitional interactions. The implication is that it is important for a person not only to learn how to tune but also to find ways to change context to enable more healthy actions to tune to.
Tuning offers HCI novel pathways to explore health design. For example, tuning is regular but periodic, not constant. Interactive technology to support building the knowledge, skills, and practice to be able to tune oneself—potentially independent of that interactive technology once the skill is achieved—may be different from tools (like trackers) designed to measure practice or tools (like software audio instruments) to support performing the same activity over time, devoid of context. We may wish to consider, from these contexts, what the minimal data dose may be rather than how to acquire constant data for health management. This approach to tuning also plays to HCI's strengths, based on the bias toward defining success in the present moment. If more multifactorial definitions of healthy in a given context can be defined, then HCI may more meaningfully contribute novel solutions around proximal outcomes as indicators of success .
We have proposed 1) adaptivity in context as a way to conceptualize health, 2) tuning as a volitional, multifactor approach to influence that adaptivity, and that 3) tuning can be supported at the individual, social, and infrastructure levels. We suggest that tuning advances health in a way that is distinct from prior strategies. First, tuning, within an inbodied interaction frame, respects the complexity of health by foregrounding the multiple pathways that influence it. Second, inbodied interaction embraces this complexity. Its design ethos, with accessible models like in5 that highlight the interaction of shared essential volitional experiences from move to think, aims to empower individuals and groups to be able to explore such key volitional factors to be able to better tune themselves and remain in tune.
Tuning as analog for inbodied interaction broadly, and health particularly, also guides designs to embrace the complex in a gentle way. Like assessing related instruments or components, we discern and test how practices may be appropriate and effective for the adaptation we seek to support, using context to guide how we can simplify without compromising complexity. We can design approaches to help us understand which components may create dissonance today, weakening our inbodied systems, and which ones may produce harmony today, enhancing and strengthening our inbodied selves and communities.
1. schraefel, m.c. in5: a model for inbodied interaction. Extended Abstracts of the 2019 CHI Conference on Human Factors in Computing Systems. ACM, New York, 2019, Paper LBW1818; https://doi.org/10.1145/3290607.3312977
2. Hekler, E.B., Klasnja, P., Chevance, G., Golaszewski, N.M., Lewis, D., and Sim, I. Why we need a small data paradigm. BMC Med 17, 133 (2019). DOI:10.1186/s12916-019-1366-x
3. Lee, J., Walker, E., Burleson, W., Kay, M., Buman, M., and Hekler, E.B. Self-experimentation for behavior change: Design and formative evaluation of two approaches. Proc. of the 2017 CHI Conference on Human Factors in Computing Systems. ACM, New York, 2017, 6837–6849. DOI:https://doi.org/10.1145/3025453.3026038
4. Nahum-Shani, I., Hekler, E,B,, Spruijt-Metz, D. Building health behavior models to guide the development of just-in-time adaptive interventions: A pragmatic framework. Health Psychol. 34, 0 (2015), 1209-19. DOI: 10.1037/hea0000306
5. Hekler, E.B., Rivera, D.E., Martin, C.A., Phatak, S.S., Freigoun, M.T., Korinek, E., Klasnja, P., Adams, M.A., and Buman, M.P. Tutorial for using control systems engineering to optimize adaptive mobile health interventions. J Med Internet Res 20, 6 (2018), e214; https://www.jmir.org/2018/6/e214
6. Rabbi, M., Aung, M.H., Zhang, M., and Choudhury, T. MyBehavior: Automatic personalized health feedback from user behaviors and preferences using smartphones. Proc. of the 2015 ACM International Joint Conference on Pervasive and Ubiquitous Computing. ACM, New York, 2015, 707–718. DOI:https://doi.org/10.1145/2750858.2805840
7. Klasnja, P., Consolvo, S., and Pratt, W. How to evaluate technologies for health behavior change in HCI research. Proc. of the SIGCHI Conference on Human Factors in Computing Systems. ACM, New York, 2011, 3063–3072. DOI:https://doi.org/10.1145/1978942.1979396
m.c. schraefel directs the International Wellthlab (wellthlab.soton.ac.uk). Its mission is to explore where and how interactive technology can help make normal better, for all, at scale. firstname.lastname@example.org
Eric Hekler is director of CWPHS, a center of excellence for digital health within the Qualcomm Institute, and associate professor in public health at the Design Lab at the University of California, San Diego (UCSD). He works on methods for optimizing behavioral interventions and helping people help themselves. email@example.com
Inbodied systems are always on, always adapting, and, in particular, nonvolitional. Tuning is framed as a way to deliberately affect the inbodied system. In5  (the inbodied five) is a model featuring core volitional and essential processes that directly affect the state of the inbodied. These are: move, eat, engage, cogitate, and sleep (MEECS). These five are not necessarily exclusive, but to the best of our knowledge they are each processes that 1) we all share and practice regularly to varying degrees of quality, 2) are essential to (quality of) life (their deliberate extended removal is used as torture), and 3) are processes for which we are physiologically wired. For example, in terms of engagement, we have specific hormones like oxytocin that are triggered by physical human contact and have been shown specifically to enhance trust; we are evolutionarily adapted to move in synchrony with each other; and loneliness—a lack of positive, in-person engagement with others—may be more of a killer than obesity (https://www.sciencedaily.com/releases/2017/08/170805165319.htm), underscoring its importance for a healthy and not just long life.
These five fundamental processes interact with each of the 11 nonvolitional systems all the time. For example, to move demands nutrients from digestion to be translated into energy for muscles; the amount of energy required is gated by the ability of the heart and lungs to pump and refresh our blood. The nervous system builds up these movement patterns throughout multiple areas of the brain to ensure efficiency. The load from movement places demands on our skin and bone to refresh and rebuild, while our lymph system manages the healing of inflammation caused by these adaptive stresses. The urinary system clears waste byproducts produced from these actions.
Design Opportunity: These interactions and their overall effects on our health and performance are not necessarily tangible, so a design challenge is to help people better understand and appreciate the connections between their actions and their nonvolitional state (see design example: Fat Tapestry).
Likewise, the in5 themselves interact constantly and dynamically (Figure 1). This interaction presents a huge design space opportunity for HCI: One can start anywhere, with any of the in5, as everything affects everything. For example, if someone has trouble sleeping, rather than focusing on sleep, designs can help participants explore any of the other in5 that may be easier for the participant to manage at that time, such as practicing social interaction skills. We can also stack MEECS: Going for a walk with someone while talking about ideas for a project brings together movement, social engagement, and cogitation, itself affecting types of cogitation (exploratory versus focused) stimulated by types of movement. Together, these may likewise help fatigue a person sufficiently, on multiple levels, to enable better sleep.
|Figure 1. The inbodied 5 (in5), core volitional and essential processes that directly affect the state of the inbodied.|
Each of the in5 MEECS can be tuned within continua (Figure 2) to support optimizing the inbodied for the current aspiration. We can design tools to help people find their sweet spots for each of these when tuning for a particular adaptation. For instance, students preparing for exams may wish to tune their in5 to support relaxed, effective recall. Imagine tools designed to help self-tune one's movement, social engagement, food choice, sleep, and cogitation/study to support current in5 states, and to further help optimize these qualities based on work to do, time to prepare, and other physical, cognitive, and social interactions.
|Figure 2. Tuning the in5 along their continua.|
Body as Starting Point: 3rd Annual International Workshop on Inbodied Interaction, taking place at CHI 2020. We are inviting work that deliberately leverages the concepts explored in this special topic section, including the inbodied interaction lenses, in5, tuning, discomfort, and insourcing, toward resilient adaptation.
Inbodied Interaction 101: A Human Physiology to Neurology Informed Approach to HCI Research and Design, taking place at CHI 2020. This course is designed to offer an overview of the the nonvolitional systems toward using the in5 for exploring discomfort and enhancing tuning for resilient adaptivity across contexts. Students will leave with a map of the inbodied to help guide their own further physiological explorations.
Inbodied Interaction 102: Understanding the Selection and Application of Noninvasive Neurophysio Measurements for Inbodied Interaction Design, taking place at CHI 2020. This course, building on Inbodied Interaction 101, is focused on insourcing for tuning, and interoception in particular. It presents how to blend formal, noninvasive measures (like EEG and HRV) to support insourcing. It also introduces new assessments of inbodied processes (drawing from neurology) to explore and develop interoceptive awareness for insourcing/tuning. In particular, the course will look at how to bring the central nervous system, limbic system, insular cortex, cerebellum, and midbrain into design to support tuning/insourcing.
Inbodied Interaction Summer School 2020 in Southampton, U.K. Building on the success of the first International Inbodied Interaction Summer School 2019, with participants from Mexico to Russia, and Japan to Australia, and co-sponsored with support from the ACM Development Fund. Get ready for a five-day deep dive into the physiology and psychology of inbodied interaction with leading researchers teaching on topics from epigenetics to social contagion, and the inbodied science of each. The school blends seminars and design studio work.
Inbodied Interaction, the manual, the book, presenting a deeper dive into each of the topics in this special topic, will be coming out in Spring 2021.
Copyright held by authors. Publication rights licensed to ACM.
The Digital Library is published by the Association for Computing Machinery. Copyright © 2020 ACM, Inc.