5.1 An Internalistic Perspective on Behavior Change
Using interviews and IPA allowed us to focus on aspects of the behavior change experience that are rarely investigated in previous research. The first contribution of this study precisely lies in an in-depth account of behavior change dynamics that are still underexplored, which may provide important insights on how people appraise and manage the process of behavior change, which in turn may lead to the design of more engaging or effective technologies.
Our first research question was: How do individuals experience and account for the changes that they want to produce in their own behavior? The participants in our study highlighted that behavior change tackled matters that went beyond the mere modification of a target behavior and made each process of change quite unique. From the participants’ accounts it emerged that: (1) both the behavior and the process of change are heavily meaning-laden, whereby the meanings constructed by the participants are rooted in their internal dynamics and intimate life, often revolving around important existential issues; (2) behavior change is embedded in a variety of life circumstances, which represent not ancillary details but central aspects of the process of change, deeply affecting how it develops: in this sense, the behavior change attempts are only a part of the life matters that the participants have to manage in their daily life; (3) behavior change unfolds in very long periods of time, having roots in the past of the participants, changing across their different “phases of life,” and being possibly projected into an imagined future.
These findings parallel those reported by Rapp et al. [
2019] with reference to the characteristics of “general change,” showing that behavior change makes no exception in being a fundamentally internal process. In particular, the role of internality clearly emerges in the importance that sense-making has in behavior change. The phenomenological perspective that we adopted allowed us to investigate how meanings were constructed by the participants, also depending on the different life circumstances in which they were situated and the diverse “phases of life” that they were living. This perspective at first glance may resemble those technological approaches that consider the cognitive aspects of behavior change (e.g., Spruijt-Metz et al.
2008; Macvean and Robertson
2013], like the role of self-efficacy [Bandura
1991] and that of motivation [Ryan and Deci
2000], which are allegedly seen as “internal” constructs. However, in such approaches, the “internal” factors of behavior change are mostly objectively and quantitatively appraised, which means that they can be effectively manipulated and studied without taking into account the subjective experience of the person. For instance, motivation may be tackled in its different degrees and types, like intrinsic and extrinsic motivations [Ryan and Deci
2000], but the subjective meanings that motivate people, that is the reasons why they are motivated to change, may be ignored: in other words, what is important is that the person is motivated by a specific motivation type (e.g., extrinsic), not the meanings that she ascribes to the motivation (e.g., “I want to change because I am afraid to die”).
Our study suggests, instead, that researchers start qualitatively considering how individuals subjectively appraise and make sense of the behavior that they want to change and the internal processes that may lead to the desired modification. This is in line with the approach of the third wave/paradigm of HCI, which focuses on the individual's lived experience [Bødker
2006] and her ways of perceiving and understanding the world [Harrison et al.
2007], and with work on health coaching noticing the importance of meaning in behavior change process [Rutjes et al.
2019; Ryan et al.
2022]. For instance, one individual might want to do physical activity because she believes that it is a way to improve herself. Another individual might want to exercise more because it is a means to exert control over her life. Another one might wish to be less sedentary because she wants to be recognized by others. Our participants showed that these meanings do matter because they entail different modalities to start, enact, and maintain (or not) the process of behavior change. This may require the design of different strategies to support the process of change, depending on the particular universe of sense within which the behavior is framed. The novelty of our findings lies mainly in showing that behavior change may be tied to meanings that ultimately connect with important existential concerns, like the need to have control over life and the fear of suffering. In this sense, behavior change may become an existential concern itself, because it may be lived as a central part of the individual's existence and have relevant implications for her whole life. Moreover, we point out that building “knowledge” may help people make sense of their attempts to change. This mirrors Bhattacharya et al.’s [
2018] research findings, which highlight that individuals may proactively seek external information that may increase their understanding about their behavior. We specify that the construction of knowledge about the behavior change process may help people better understand how behavior change can fit in between their overall personal goals, beliefs, and values and how technology can become a support, potentially leading to better outcomes.
Our study also emphasizes that it is not possible to isolate the attempts of changing a behavior from the life circumstances in which it unfolds. Behavior change is not an isolated practice that is enacted in the void, but a situated endeavor that interacts with other daily routines and is affected by a variety of life conditions that may determine the success or otherwise of the process. These circumstances are not superficial details, which can be ignored when designing a “surgical intervention” that targets only the behavior, but central part of the process. These findings connect with studies exploring people's behavior and activities as “social work,” namely complex experiences entangled with social dynamics [Shin et al.
2022], as well as with previous work on the impact of context factors on behavior change [Bhattacharya et al.
2017; Murnane et al.
2018; Rutjes et al.
2019]. For instance, Bhattacharya et al. [
2018] noticed that external conjunctures, be either “negative” like an unexpected illness, or “positive” like starting a family, may lead people to increased resolve toward making the change. Likewise, Agapie et al. [
2016b,
2018] emphasized that people feel that it is important that behavior change plans accommodate their routines and everyday constraints. Moreover, our study echoes behavior change research that relies on social practice theory, which highlights the role that collective routines, broadly accepted social norms, physical environment, and near materiality have in shaping behavior [Blue et al.
2016; Entwistle et al.
2015; Twine
2015]. However, in social practice theory, the subject plays a minor role. The person is only a carrier of collective practices, which are independent from her [Schatzki
2002]: the individual may perform such practices, reproducing them over time [Reckwitz
2002], even with variations and resistance [Shove and Pantzar
2005]. Nevertheless, the real agents of change are the practices themselves.
With respect to this previous work, our study better points out the active role of the individual in managing her behavior change attempts, as well as in leveraging the circumstances of her everyday life. Moreover, we point out the meaning-laden nature of life circumstances: in this sense, it is not possible to identify absolute “positive” or “negative” conjunctures, as they are informed by subjective meanings that may have an idiosyncratic value to the individual. In fact, despite being constrained by circumstances that go beyond their control, our participants are not at their mercy. Rather, they are often able to exploit the right situation, what we have called the Kairos, to begin or reaffirm their endeavor to change. Here, the impact of “life” on behavior change is neither deterministic nor irrelevant: rather, it appears that the individual interacts with the conditions in which she is cast, giving meaning to and reinterpreting them. In so doing, she adapts her behavioral attempts to the varying life constraints and opportunities that she encounters, trying to find a balance among the different matters that make up her everydayness.
Furthermore, the study findings pinpoint the role that temporality has on the process of behavior change. Obviously, behavior change is intrinsically temporal because it occurs in time: but its temporality is not limited to the moment when the change is undertaken. Instead, behavior change may have roots in the distant past of the individual, take place over many years returning in different forms, and be projected into an imagined future. This temporality differs from that implied by technologies based on the “behavioral model” of change, which sees behavior change as an isolated event and focuses on the well-circumscribed time of the behavioral intervention [Rapp et al.
2019]. The TTM, for example, sees individuals progressing through six stages when attempting to change a behavior (pre-contemplation, contemplation, preparation, action, maintenance, and termination) [Prochaska and Velicer
1997]. This temporality, however, refers to the proximal time in which the behavior change occurs: contemplation, for instance, is meant as the moment in which people are intending to change in the next 6 months and is commonly the first stage tackled by technology, being pre-contemplation referred to a moment when the individual does not intend to change, thus falling outside the scope of the technological intervention [e.g., Li et al.
2010]. Likewise, albeit most self-tracking systems are past-centric in nature, addressing the retrospective aspects of data collection and analysis [Lee et al.
2020], the past that they consider is a “near past,” strictly related to the behavior to be changed, like past performances and successes [Kersten-van Dijk et al.
2017]. This proximity is shared by systems that look at more future-oriented interactions [Rho et al.
2017; Lee et al.
2020], considering a future that is close to the site of the intervention. In this narrow temporality, therefore, there is no room for the impact that the experiences rooted in the individual's distant past may have on the present modification of the behavior. By contrast, our study points out that behavior change may have roots in a distant and meaningful past, possibly embracing the individual's whole life course. Moreover, we highlight that people may imagine future life trajectories that evolve far away from the present and relate to imagined future existential concerns. In this sense, the behavior change temporality emerging from our findings is far more subjective, meaning-laden and “stretched” than that tackled by previous research.
In sum, our study depicts behavior change as a fundamentally internal process. This means that behavior change concerns more the internal, subjective aspects of the individual's life, than the external manifestation of the behavior. Being behavior change heavily meaning-laden, and being meanings subjectively constructed by the person, each process of behavioral modification is rather unique. Despite the uniqueness of each process of behavior change, however, some tentative patterns can be identified: the importance of the existential concerns that are connected with the behavior to be changed, the relevance of the life circumstances in affecting the process of change, and the prominence of a dilated time.
5.2 A life Perspective on Behavior Change Technologies
Our second research question was:
How do technologies used to address behavioral matters connect to the wider aspects of people's life? Our study confirms and deepens the scattered findings reported by those HCI studies that attempted to explore behavior change outside the “behavioral model,” which highlighted that technology puts too much emphasis on behavior to the detriment of the person's lived experience [e.g., Rutjes et al.
2019]. Moreover, the study resonates with HCI research on mental health and difficult life moments, which emphasized the role of sense-making and reflective experiences when people use technology to find support [Iacovides and Mekler
2019; Loke et al.
2021; Boldi and Rapp
2022; Boldi et al.
2022]. Building on top of these previous studies, we analyzed more in depth and systematically how technology for behavior change is used by people “in everyday life,” finding that the technologies used by our participants were often unable to help them develop the meanings required to effectively enact the process of change.
This lack of support for sense-making aligns with the findings of research examining the features of popular self-tracking tools [Cho et al.
2022], as well as of work exploring tracking practices in domains like sports and chronic illness management. For instance, amateur athletes may find no guidance in popular sports self-tracking instruments to interpret their own body data, so they may “read” such data as intrinsically valuable numbers: this approach, however, fetishizes the data and might jeopardize the achievement of long-term goals [Rapp and Tirabeni
2018,
2020]. Likewise, Mamykina et al. [
2006,
2008] noticed that popular glucose monitoring devices for diabetes management may not be sufficiently helpful in driving people's sense-making processes, and simple presentation of data may reinforce individuals’ preconceived notions instead of facilitating genuine discoveries. From our study participants’ recounts, however, it appears that sense-making is always enacted when individuals use a certain technology, so that, if such technology is not intentionally designed to support the meaning construction, its outcomes may be unpredictable.
Many of our participants also stressed that as the behavior change systems they used do not consider and tackle the wider life context in which behavior change is situated, they either may fail when such circumstances are modified, or may simply appear more sterile in comparison with other meaningful activities that are wired into a lifeworld. Even though this may not be a bad thing per se, if the person is able to find a more engaging way to change her own behavior, it may signal the inability of some technologies to offer a valid alternative to richer but more “complex” activities (e.g., sports), which clearly require stronger effort to be enacted and may not be suitable for everyone. The poorness of the lived experience reported by several participants when using behavior change systems may further explain the reasons why people abandon them [e.g., Lazar et al.
2015], thus requiring that designers focus more on designing supplementary activities “surrounding” the behavioral intervention, so to create a richer lifeworld and elicit more engaging and meaningful experiences.
Systems used by the participants showed also to ignore the temporal evolution of behavior change, whereby different support may be needed in diverse phases of life. By providing the same modalities and features regardless of the “phase” the person is in, technology risks offering outdated help, thus leading to a rapid abandonment. These “phases” of life, however, are different from the “stages of change” tackled by many personalized persuasive technologies that use models like the TTM [Prochaska and Velicer
1997; Oyebode et al.
2021], where the focus is on the stages’ “objective” characteristics (e.g., “people at the pre-contemplation stage tend to be strongly motivated by self-monitoring”, Oyebode et al.
2021]: in such technologies, the aim is to identify patterns of framing and enacting the change at a particular point of a “standardized” path, in order to deliver an intervention tailored to that point. Instead, the phases highlighted in this study point to the particular universe of sense that a person connects to the behavior at a particular moment in her life, being thus intrinsically subjective. For an individual, different phases of life may entail different ways of framing and managing the process of change. Likewise, different individuals may progress to completely different phases and universes of sense. This perspective aligns with previous research that explains how behavior change goals evolve over time, stressing that technology should account for such evolution [Niess and Woźniak
2018]. With respect to this previous work, however, we point out that the flowing of time may entail completely different life phases, whereby not only the goals, but also the meanings ascribed to the process of change may be modified. This would mean to provide more individualized programs and interaction modalities that take into account the evolution of the person's meanings and design for the temporal evolution of technology use, considering changes of modalities of usage, modifications of the role of technology, and transformations of the meanings attributed to the behavior.
Finally, several participants underlined that the successfulness of the intervention can be achieved by following alternate paths that are not inscribed in the technology designs and may point to the existential concerns that are tied to the target behavior. In doing so, they showed that they were able to appropriate the technology. Appropriation [Dourish
2003; Lally
2002] refers to the person's ability to adapt the use of technologies to her needs, in ways that were not foreseen in their original designs [Quinones et al.
2013]. Appropriation may take several forms, being semantic, behavioral, or technological [Muller et al.
2016]. In this study, we found that several participants were able to ascribe to the technology meanings that were useful to partially address their existential issues. When the participants did not show sufficient competence to drive the process of change through technology, however, several unexpected side effects could also appear. For example, the person may experience a loss of agency and become dependent on the technology, an effect that has been observed even with reference to self-tracking in the sports domain [Rapp and Tirabeni
2018]. Moreover, she may become obsessed by data, as prior research on fertility tracking [Figueiredo et al.
2018], health coaching [Rutjes et al. 2018] and weight monitoring in the context of eating disorders [Eikey and Reddy
2017] has also pinpointed. The novelty of our study, therefore, lies in suggesting that these side-effects might be due to a common root, that is the lack of consideration of the user's personal meanings, life circumstances and time in technology designs.
HCI researchers should then always acknowledge and account for the potential dark sides of technology-based behavioral interventions. More transparency in communicating their potential double-edged effects to individuals could mitigate the risk, especially with reference to people that do not hold sufficient knowledge about the target behavior, thus possibly becoming excessively dependent on technology. The findings of this study could thus inspire the HCI community to focus more on the double-sided impacts that behavior change systems may yield.
5.3 An Existential Model of Behavior Change
On the basis of the study findings, we will now develop a preliminary theoretical model of behavior change that puts in the foreground the internal and existential aspects of the process and embeds the modification of behavior in the wider context of life. At this stage, this is a tentative model, which will require further empirical testing to prove its validity.
The model is thought of as an alternative to the dominant behavioral model of change. The behavioral model primarily focuses on behavior, which is conceived as a “quantum” that can be isolated from the wider life into which is situated and from the life course in which it happens, and is studied “objectively,” from a third-person point of view that considers almost exclusively its external features [Rapp et al.
2019]. Instead, the model that we propose gives value to the internal meanings that people develop throughout their lives and accounts for the existential matters that are intertwined with their behavior change attempts, shifting in this way the focus from the behavior to be changed to the life in which it is situated. It is worth pointing out that this “existential model of behavior change” is not meant to criticize or substitute the behavioral model. Rather, we suggest that it may offer an alternate perspective that may integrate, develop, and amplify the impact of what has been previously done under the behavioral account.
The model highlights that the behavior change process is influenced by:
(i) the meaning that a person ascribes to the target behavior and the process of change. The term “meaning” refers to the sense-making activity that people enact during the process of change: individuals link to behavior change the additional sense that is rooted in their internal dynamics and intimate life. This meaning, in fact, is subjectively constructed by the individual and often points to important existential concerns, framing the interpretation of behavior change: one person may interpret quitting smoking as a way to protect her health and to stave off the fear of death, while another as a means to have more control over her life and be more authentic.
In this sense, the constructed meanings heavily impact on the process of change (Figure
1, Arrow
Meaning→Behavior change), affecting e.g., how people see the behavior and the effort they put in the endeavor to change, their sense of having succeeded or failed, and their decisions to maintain the change or relapse into a previous state (e.g
., I am afraid of dying and quitting smoking may help to prolong my life, I must persevere in my attempts at all costs, otherwise it is a failure vs.
quitting smoking is a way to regain control over myself; however, in certain difficult moments losing control can be positive for me, so it is not bad to momentarily resume smoking). In other words, the meaning puts in the foreground certain aspects of behavior change process, while leaving in the shadow others, modifying how people understand and thus enact their effort to change. This finds confirmation in studies grounded in the phenomenological tradition, which highlights the subjective nature of meanings [Rapp and Tirassa
2017], as well as in research that points out the importance of sense-making, personal motives, and goals in the process of change [Rutjes et al.
2019; Bhattacharya et al.
2018; Rapp et al.
2019].
In parallel, people can also derive meaning from making a behavior change, modifying for example how they interpret their own identity (Figure
1, Arrow
Behavior change→Meaning). For instance, a person who starts exercising may eventually come to identify herself as a sportsperson, and find meaning in that identity. These identity shifts may then further impact behavior change. This aligns with work emphasizing that when a behavioral attempt is successful, an identity shift begins: the increased self-awareness and self-confidence that follow may then fuel continued change [Kearney and O'Sullivan
2003].
(ii) the life circumstances in which the behavior change attempts take place. Life circumstances are all those everyday conjunctures that the individuals consider to be intertwined with the process of behavior change, which do not necessarily pertain to its immediate “surroundings” (e.g., the environment where the behavior change happens), but may involve the practices, routines, and relationships that are seen as connected with the behavior to be changed.
On the one hand, life circumstances directly impact on the process of change, as they constitute a nexus of everyday activities in which behavior change needs to fit in between (Figure
1, Arrow
Life circumstances→Behavior change). For example, a person may not be able to accomplish her everyday exercises because she has to deal with family and work matters, like accompanying children to school, overwork, and so on. This resonates with those studies that emphasize the influence of routines and constraints on behavior change [Rutje et al.
2019; Murnane et al.
2018; Bhattacharya et al.
2017,
2018; Agapie et al.
2016b,
2018].
On the other hand, life circumstances may change the meanings that are ascribed to the target behavior and the process of change (Figure
1, Arrow
Life circumstances→Meaning). Such meanings, then, may modify how the person manages the change. For instance, when a person is with her partner, she may see food as a way to share experiences; instead, when she eats alone, food becomes nutrients that need to be controlled: in the former case, persevering in the effort of changing behavior may become far more difficult.
In parallel, life circumstances are meaning-laden, so that they have a personal and often existential value for the individual. For instance, a person may ascribe to a situation of forced isolation (like a lockdown) an opportunity for being more authentic, which may fuel her willingness to control behavior to align with her “true self.” Another person, instead, may see the same situation as a threat to her personal freedom, so losing control may become a means to regain liberty: this may stop her behavior change attempts. It follows that modifying the meanings attributed to life circumstances (Figure
1, Arrow
Meaning→Life circumstances) may provoke a modification of how the behavior change is managed by the person (e.g., making her see her current situation in a different light may increase her willingness to change).
(iii) the
life time in which the behavior change unfolds. Life time is the time that is experienced by the person in her everyday life and may point to different ages of the individual and even to her entire life course [Rapp
2022]. In this sense, this time aligns with the time of the phenomenological tradition [Zahavi
2012; Husserl 1962], which is fundamentally subjective, embracing not only the (even distant) past of the person as she has experienced it, but also the possible (distant) futures as she imagines them to be.
On the one hand, life time may directly affect the behavior change process, as the flowing of time may transform the life circumstances, the personality, or the body of the individual, and consequently, her willingness to put effort in the change (Figure
1, Arrow
Life time→Behavior change). For instance, a person may lose interest in doing exercise simply because she becomes lazier or less energetic as she gets older. This is in line with research highlighting the temporal evolution of behavior change [Niess and Woźniak
2018].
On the other hand, life time may transform the meanings that are ascribed to the behavior and to the process of change (Figure
1, Arrow
Life time→Meaning), which, in turn, may impact on behavior change. An individual, for example, may pass through different “phases of life,” in which food is first seen as a source of pleasure, then as a means to exert control over the world, and finally as an experience to be shared with others: as meanings change with the passage of time, the way she manages the change will also be transformed.
In parallel, life time is not “neutral” for the individual, but imbued with meanings potentially pointing to existential concerns. In this sense, the person's past may be seen as a time when certain existential matters have been addressed or not, and her future as a field of possible alternatives where novel existential issues could arise. For example, a person may start applying the meanings that pervaded her past, when she perceived her body as frail, to the present time. Another individual, instead, may imagine that in the future she would not be able to persevere in any endeavor. In both cases, the meanings associated with the person's life time may undermine her present willingness to exercise. It follows that modifying the meanings attributed to life time (Figure
1, Arrow
Meaning→Life time) may yield a transformation of how the behavior change is handled by the person (e.g., making her see the future in different terms may encourage her to accomplish the change).
In sum, the existential model of behavior change is intrinsically contextual and temporal. The context and the time that the model accounts, however, are not those that are commonly tackled by behavior change technologies. On the one hand, the notion of context in the behavior change technology field has been traditionally taken by early context-aware research, where it has been conceived as the sum of the physical features of the environment in which the user's action takes place [Dey et al.
2001; Grudin
2001]. Although the debate on context within HCI and Computer-Supported Cooperative Work studies evolved over the years, highlighting the situatedness of our everyday experience that depends on material, social, and cultural circumstances [Suchman
1987; Dourish
2004; Räsänen and Nyce
2006], and widening the notion of context to routines and practical constraints [Rutje et al.
2019; Murnane et al.
2018; Bhattacharya et al.
2017,
2018; Agapie et al.
2016b,
2018], in the behavior change technology domain context is still mostly understood as any information that may characterize the physical (or external) situation of the user [e.g., Prost et al.
2013]. Instead, in the model we propose the context of behavior change becomes internal and “existential,” pointing to all the life circumstances that the person has to deal with in her daily life: these are not only the everyday practices in which the person is routinely involved, but also the meaning-laden conditions that may be connected with the existential issues that she has to face. In other words, the model encompasses a view “from the inside” of the context, whereby more than its material, and even social and cultural aspects, it comes to matter its internal and existential ones.
On the other hand, the time commonly addressed by behavior change technologies is the “time of the machine,” which is also the time traditionally tackled by HCI [Rapp et al.
2022]. In fact, time in HCI has been commonly framed within the “clock perspective” [Rapp et al.
2022], a mechanical instrument that allows the recording of the exact quantity of time, as a measurable, objective, and uniform entity [Starkey
1989]. Even though this view on time has been counteracted by strands of research seeing time as design material [e.g., Odom et al.
2014; Harrison and Cecchinato
2015; Odom et al.
2018], or focusing on the social and cultural organization of time [Lindley
2015; Taylor et al.
2017; Pschetz and Bastian
2018], time as an internal, existential matter, has received far less attention [Rapp et al.
2022]. Likewise, in behavior change technology research, relevant time remains that of the clock, which may refer to the timeliness of the intervention [Lee et al.
2017], the management of eventual relapses [Agapie et al.
2016a], or the duration of the user's adherence to the behavioral program [Kovacs et al.
2021]. Moreover, whether referring to the past or the future, behavior change time is usually narrow, being close to the site of the intervention [Kersten-van Dijk et al.
2017; Rho et al.
2017; Lee et al.
2020]. The model we propose, instead, points to a time that is similar to the time tackled in HCI studies on death [Massimi et al.
2011; Gulotta et al.
2016], legacy [Gulotta et al.
2017; Gulotta et al.
2014], and rituals [Petrelli and Light
2014]. This is an “internal” and “existential” time that is fundamentally tied to the meanings that stem from the individuals’ sense-making of their own existential issues and is affected by how their entire course of life unfolds.
5.4 Using the Model
The existential model of behavior change is still a tentative model, as it needs to be empirically tested in the future. However, in line with other HCI behavior change and data tracking models, developed from single empirical studies and not immediately tested on the field [e.g., Li et al.
2010; Epstein et al.
2015; Murnane et al.
2018; Niess and Woźniak
2018], we believe that it could offer a valuable contribution to HCI research from the theoretical point of view. Recent HCI studies confirming the meaning-laden, contextual, and longitudinal nature of behavior change [e.g., Niess and Woźniak
2018; Bhattacharya et al.
2018; Rapp et al.
2019; Rutjes et al.
2019] further strengthen its validity.
The usage of the model also has practical implications that could aid the design of future systems.
First, the model emphasizes the pivotal role of sense-making in behavior change, thus implying the need to use more prominently inquiry techniques that allow researchers and designers to capture what kinds of meanings people attribute more likely to a specific behavior, using them to inform the whole design process. In this sense, the model can provide guidance in identifying those meaningful aspects that potentially affect a particular behavior change experience. Each aspect of the model, then, could be inspected at best through specific research methods. Interviews as unstructured as possible, projective techniques for the elicitation of meaning [Porr et al.
2011], and IPA for uncovering individual idiosyncrasies [Smith and Shinebourne
2012] may be the optimal solution to study how people make sense of a behavior change. The individual's past time may be best inspected by using life story interview, which is the story a person chooses to tell about the life she has lived, told as completely and honestly as possible, and what is remembered of it [Atkinson
1988]: this technique has been used in HCI to explore how individuals use technologies during the entire life course [Pena et al.
2021]. As for the individual's future, motivational interviews make available a set of tools for supporting participants in envisioning alternate futures [Miller and Rollnick
2013]. Life circumstances, instead, may be better studied using ethnography, observation, and contextual interviews, with a particular focus on the meaning that certain environments, relationships, and practices may have for the person.
Second, the model is intrinsically dynamic as it highlights that how the individuals appraise behavior change changes over time and depends on the life circumstances in which they are situated. This implies that designs that are not “adaptable” and “malleable” might not fully work in the behavior change domain. “Adaptable” systems may tailor their own behavior to the characteristics of the user [Brusilovsky
2001; Frias-Martinez et al.
2006] or other relevant factors, exploiting either user's self-reporting or automated means, like machine learning techniques to extract high-level information from sensors [e.g., Banaee et al.
2013; Perera et al.
2014], lifelong user models, which model user goals and preferences in the long term [Kay and Kummerfield
2009], and data-mining techniques applied to time series to detect anomalies [Izakian and Pedrycz
2014] and rare motifs [Begum and Keogh
2014], which may signal “turning points” in life time. In this sense, the existential model may help designers in identifying those factors that, when they vary, are more susceptible to impact on behavior change. The model also suggests that designers think about their designs not in terms of “complete” and “fixed” products, but as malleable and mutable. This may entail a “modular” approach to design, where multiple design features and behavioral strategies may be made available to the user, who may activate or deactivate them depending on the situation and life phase; or where systems are left open to “add-ons” that could be delivered in response to specific concerns that users may have at a certain time point. This is in line with research recommending that designs should be configurable to support evolving personal goals [Cordeiro et al.
2015].
Third, by emphasizing the subjective nature of behavior change, the model encourages researchers to use more prominent evaluation techniques that consider the users’ subjective criteria for determining whether a system is successful or not, as “objective” assessments may not fully capture the impact that a certain technology may have on the user: for instance, the behavior is modified by the system, but the user perceives that she became more obsessed by caloric intakes. In this perspective, the model may give guidance to researchers on what are the important metrics for evaluating behavior change, like what kinds of meanings the user connects to behavior change before and after the use of the system; whether the system helps her in addressing the existential concerns that are linked to the behavior change; what kind of sense she makes of the intervention and the technology, and whether this sense changes over time or within specific life circumstances. Moreover, the model may help researchers recognize side effects arising from the usage of the technology identifying the reasons for their emergence (e.g., the technology changed the meaning associated with the target behavior in a negative way). This entails the need to explore evaluation methods focusing on internal processes [Baumer et al.
2014]. Diaries, for example, where the user can keep track of her perceptions, feelings, and interpretations over time, may help researchers understand the meanings that users developed during the intervention, tackling their longitudinal nature [Carter and Mankoff
2005; Sohn et al.
2008]. Likewise, experience sampling allows researchers to collect data about the internal aspects of human life (e.g., thoughts, sensations) through self-reports provided by participants, who are proactively triggered at various points throughout the day [Larson and Csikszentmihalyi
2014; van Berkel et al.
2017].
Finally, it is worth noticing that the existential model of behavior change in some cases could lead users to engage in unexpected or undesirable behavioral goals, as well as designers to implement systems that may produce detrimental side-effects. In fact, a behavioral intervention that only relies on the user's meanings might be based on information that does not “reveal the truth” about the user and her behavior, as she may self-deceive [von Hippel and Trivers
2011]. For instance, a person may think that her incapacity to increase her physical activity is due to a fundamental laziness that characterizes her personality, whereby it is actually due to an almost unconscious fear of being harmed while exercising. Trying to change the image of her personality as a lazy person into a more active one, therefore, would not make her exercise more. Likewise, a person may design, with the help of technology, self-experiments that increase the biases about herself (e.g., through a self-confirmation bias effect): for example, she may keep trying to decrease the daily caloric intake to see if her body improves and actually think that this is happening, when in fact it is damaging her health. These potential side-effects suggest that in certain cases, the existential model should be paired with other perspectives that either look more at the “objective” and “external” aspects of the process of change or delve deeper into the often unconscious “truth” of the individual.
Further tensions may arise when implementing the model with reference to the focus on life circumstances. Applications that pay particular attention to the activities surrounding the behavioral intervention, adapting their features to the varying conditions of the user's life and goals, might undermine her capability of appropriating the technology and adapting it to her own ends and needs. As we have seen in the study findings, users may become dependent on technology. Moreover, by delegating tasks to the instrument, people may lose the abilities underlying the execution of those tasks. Besides the well-known effects of automation bias in the Artificial Intelligence field [Goddard et al.
2012], it has been shown that an overreliance in self-tracking instruments, for example in the sports domain, may undermine the opportunities for sense-making [Rapp and Tirabeni
2018]. Designing systems that adapt to the user's life may thus similarly reduce her opportunities for constructing meaning and appropriating technology for her situated goals. This necessitates an approach that values the person's autonomy, developing her sense-making and adaption skills rather than replacing them.
Likewise, applications that pay attention to the user's life time, accounting for her entire history, may reinforce meanings that the user would prefer to dismiss. For example, focusing on the user's past might reduce adaptive forgetting by interfering with adaptive biases, since re-presenting past events may act like rumination [Sas and Whittaker
2013], triggering perseveration on events that might be better forgotten [Konrad et al.
2016]. Moreover, systems offering a life-long, life-wide, perspective may exert more power over the individual. In principle, behavior change technologies inscribe in themselves instances of power that define why and how we should behave in a certain way [Baumer et al.
2012; Rapp
2019]. In this sense, using the existential model of behavior change may encourage designers to address the intervention not only to a specific, circumscribed, behavior but potentially to the individual's whole life. This clearly opens possibilities for designers to control, surveil, or coerce the user [Purpura et al.
2011], subtly shifting her priorities, beliefs, goals, and meanings, without her being fully aware of it. This implies that designers ethically reflect on the (even long-term) consequences of their designs and provide the user with means to rebalance the power relation: for instance, transparency, explainability, and scrutability [e.g., Miller
2019; Kay and Kummerfeld
2013] may allow users to inspect, understand, and change the image of their self on which the system relies, thus empowering them.
To summarize, the existential model of behavior change suggests that researchers and designers start considering the internal and existential aspects of behavior change because such aspects are essential in determining how the behavior change process is accounted for and managed. On the basis of the study findings and the model we surfaced, we will now point out a series of barriers that users of current behavior change technologies may encounter and design suggestions that could help people overcome them.