The Awareness-Agency Gap
Digital behavior change interventions have proliferated for a decade, but most still operate on a behavioral prompting model: segment the user, deliver a cue, track the response. This paper presents the empirical foundation for adaptive identity mirroring theory (AIM-T), which argues that durable change in digital environments operates at the level of identity rather than behavior, and that the mechanism is the reflection of a person’s own values, language, and self-articulations back to them at contextually enriched moments.
The study combined a nine-month institutional SMS pilot with first-year university students, 76 qualitative interviews with students and practitioners, sixteen longitudinal coaching transcripts from a primary beta participant, and a user profile library from earlier Journey cohorts. Heart-emoji reactions on the SMS channel gave a behavioral record of acted-on nudges alongside self-report. Analysis followed grounded theory (Glaser & Strauss, 1967) and a hypothesis-testing interview framework adapted from Fitzpatrick (2013). Full sample sizes are in the Methods section.
What students cared most about was not the surface behavior — how many minutes they spent on a phone — but the texture of the moment underneath it. They described an awareness-agency gap: reaching for the phone before noticing the reach, using the device as a pacifier for stress or boredom, looking up an hour later with a sense of having been somewhere else. AIM-T does not treat this gap as a deficit to be corrected but as a developmental challenge: what kind of support helps a person stay in contact with their own intentions inside an environment engineered to pull them away.
Move the cursor to widen the space — watch impulses bounce inside it and leave changed
Between stimulus and response there is a space. In that space is our power to choose our response. In our response lies our growth and our freedom. — Viktor Frankl
AIM-T proposes that sustainable digital behavior change operates at the level of identity, not behavior. The stronger claim, which emerged repeatedly across the data, is that reflecting a person’s own language, values, and articulations back to them in contextually enriched moments is mechanistically distinct from conventional personalization. It is not a better version of the same thing. It is a different thing. The four processes that make it work — values alignment, contextual awareness, reflective connection, and speaking agency — are developed as continuums later in the paper.
An advisor on this project observed that our current relationship with these devices is the very tip of the iceberg compared to what is coming as the systems become more capable, more persuasive, and more woven into daily life. That is the larger reason this research matters: not because today’s apps are uniquely harmful, but because the gradient is steep and the need for identity-level scaffolding will grow. SMS matters here not because it is sophisticated but because it is low-friction, familiar, and capable of meeting students inside their actual behavioral rhythms rather than asking them to open a dedicated platform at exactly the moment they are least likely to do so.
The Behavioral Prompting Paradigm
The field of digital behavior change has invested heavily in a particular model of intervention. That model, rooted in nudge theory (Thaler & Sunstein, 2008), the Behavior Model (Fogg, 2009), and the dual-process framework of behavioral economics (Kahneman, 2011), proposes that well-designed environmental cues, delivered at moments of high motivation and low friction, can redirect behavior without restricting choice. Applied to digital contexts, this model has produced a generation of mobile applications, notification systems, and habit-tracking platforms that segment users by demographics, deliver personalized content from prompt libraries, time delivery to moments of opportunity, track response rates, and optimize for engagement.
This paradigm has also been applied to the growing problem of problematic digital behavior among emerging adults. It is worth pausing on what makes that problem distinctive. Tristan Harris and others at the Center for Humane Technology have argued that contemporary platforms apply many of the same behavioral design tools that the wellness field uses, but in service of an opposite goal: maximizing time-on-screen rather than supporting the user’s stated intentions (Harris, Center for Humane Technology). The result is an attention environment in which the most sophisticated behavioral engineering ever deployed is pointed away from the user’s own values. Students describe the consequence in their own vocabulary, “brain rot,” scrolling sessions in which thirty minutes vanish without a single recoverable thought, and the descriptions track what the design literature would predict from systems tuned to exploit primitive reward circuitry during exactly the moments when willpower and self-direction are lowest. Unlike alcohol or tobacco, there is not yet a shared cultural script for when, where, and how device use is appropriate. We have inherited norms about not drinking in bed or smoking around children; we have not yet developed comparable norms about phones, and the absence of those norms is part of what keeps the behavior continuous. University students now report unprecedented rates of compulsive scrolling, device dependence, and the associated emotional consequences of guilt, anxiety, numbness, and disconnection (Haidt, 2024; Twenge, 2017; Primack et al., 2017). Institutional responses have ranged from screen-time reduction campaigns to wellness applications to digital detox programs. Most share a common logic: frame the problem as excess, reduce the behavior, measure the outcome in minutes or sessions.
The behavioral prompting paradigm is not wrong — it accurately describes the conditions under which a single behavioral instance occurs — but it has not solved the problem it claims to address. Digital nudge platforms built on these principles achieve engagement rates of 30–40% (Lentferink et al., 2017), retention decays within weeks in a pattern called the novelty curve (Hamari, Koivisto, & Sarsa, 2014), and most behavior change applications are abandoned within thirty days of download (Consolvo, McDonald, & Landay, 2009). Three decades of optimization within this model have not meaningfully improved those figures.
The present study was developed in response to a hypothesis that emerged from preliminary intervention work with university students and coaching practitioners. The hypothesis was that problematic digital behavior among emerging adults is often not best understood as a failure of discipline, willpower, or information. It may be better understood as a failure of awareness, agency, and values alignment in an environment structurally optimized to capture attention (Harris, Center for Humane Technology). On this view, the central challenge is not to prompt a different behavior. It is to support the development of a self that can notice, choose, and increasingly direct its own action.
That hypothesis draws on a theoretical tradition distinct from the behavioral prompting paradigm. Motivational interviewing, the most empirically validated methodology in clinical behavior change, achieves its effects not through prompting action but through reflective listening: mirroring the client’s own language, values, and stated intentions until the discrepancy between current behavior and self-concept becomes untenable (Miller & Rollnick, 2013). Carl Rogers demonstrated that values clarification, reflecting a person’s own articulated commitments back to them in a nonjudgmental context, produces more durable change than directive advice (Rogers, 1961). Self-determination theory demonstrates that interventions supporting autonomy, competence, and relatedness produce intrinsic motivation, while external prompts, regardless of design quality, produce compliance that extinguishes when the prompt is removed (Deci & Ryan, 2000). Acceptance and commitment therapy positions values clarification as the mechanism through which psychological flexibility develops, with values functioning not as goals to be achieved but as directions to be lived (Hayes, Strosahl, & Wilson, 2012). James Clear, synthesizing these traditions for a contemporary audience, captures the distinction with precision: “The goal is not to read a book. The goal is to become a reader” (Clear, 2018, p. 31).
The clinical literature solved this problem decades ago. The digital intervention field has acknowledged the findings in reviews but has not built them into systems — the dominant technical architecture still segments users by demographics, delivers content from a library, tracks response, and optimizes for engagement. That architecture cannot produce identity-level change because it treats the user as a target rather than a self whose own values and language are the actual mechanism of transformation.
The present study sought to generate a theory, grounded in data, of how identity-level mechanisms operate in digital behavior change, and of the processes that support emerging adults in developing what we term digital awareness resilience: the capacity to notice compulsive patterns, interrupt them, and act in alignment with self-identified values.
Method and Evidence Base
Aim
The purpose of this study was to generate a theory, grounded in data, that explains: (a) why existing digital behavior change interventions frequently fail to produce sustained change; (b) how emerging adults experience the relationship between identity, awareness, and digital behavior; and (c) the processes and mechanisms through which identity mirroring, as distinct from behavioral prompting, supports durable behavior change.
Design
Grounded theory methodology, as originally developed by Glaser and Strauss (1967) and refined by Glaser (1978), informed the research plan. The grounded theory process consists of five basic components: theoretical sensitivity, theoretical sampling, coding, theoretical memoing, and sorting. These five components were integrated by the constant comparison method of data analysis. The goal of the research was to understand the participants’ main concerns related to digital behavior and digital intervention, and to develop a theory that explains how participants resolve those concerns.
The methodology was supplemented by a hypothesis-testing interview framework adapted from Fitzpatrick’s (2013) approach, which emphasizes reducing social-desirability bias by focusing on specific past behaviors rather than hypothetical preferences or reactions to the product idea itself. This framework proved essential in a domain where enthusiasm for wellness technology often substitutes for genuine behavioral evidence.
Sample
Theoretical sampling was the primary sampling method. In line with grounded theory principles, participant selection was informed by ongoing analysis and coding of interview data (Glaser, 1978). Purposive sampling supplemented theoretical sampling to ensure diversity across institutional roles, student populations, and intervention contexts.
The complete dataset for this study included the following components:
Qualitative interviews (n = 76). Seventy-six end users (termed “travelers”) were interviewed, including university students and recent graduates who had used the Journey SMS system for at least four weeks. Interviews focused on timing, personalization, routine building, self-direction, tone, and the experience of behavioral interruption. Six practitioners (termed “guides”) were interviewed, including coaches, academic advisors, and wellness practitioners who worked directly with students. Guide interviews focused on the between-session gap, scalability pressure, accountability mechanisms, and the structural limitations of episodic support.
Institutional stakeholders. Three university administrators involved in the nine-month institutional pilot were interviewed across repeated meetings during the partnership period. Their perspectives informed institutional context and implementation findings.
Coaching transcripts (n = 16). Sixteen longitudinal coaching-session transcripts from a primary beta participant were analyzed to derive identity mirroring principles. This corpus produced 123 nudges based directly on the participant’s language, metaphors, and identified patterns.
Longitudinal pilot (N = 100). A nine-month pilot at a mid-sized private university in the northeastern United States included 100 first-year students enrolled in an SMS-based intervention from April 2025 through January 2026.
Survey subset (n = 48). A structured survey was administered to 48 active users, generating preliminary self-report findings on engagement, behavior change, perceived personalization, and helpfulness.
User profiles (n = 237). Individual user profiles from 2022–2026 were analyzed to understand how tone, content, and language preference differed across students.
Data Collection and Analysis
Data sources included participant interviews, coaching transcripts, field notes from conversations with content experts and institutional partners, and field notes from research team meetings.
All interviews utilized adjusted conversational interviewing, regarded as the most effective grounded theory approach (Glaser, 1978, 1998). The three primary questions asked during initial interviews were: “How would you describe your relationship with your phone and digital devices?” “What have you tried to change about your digital habits, and what happened?” and “When something actually worked, when you did change a pattern, even briefly, what made that moment different?”
To formalize the analysis of signal strength, responses were coded against a six-level evidence hierarchy:
1. Actions: Observable behaviors, demonstrated commitments, or time and money already spent.
2. Specific past: Detailed accounts of particular incidents with concrete detail.
3. General patterns: Recurring behaviors described in broad terms.
4. Opinions: Stated beliefs or preferences without behavioral evidence.
5. Hypotheticals: Claims about what someone might do under future conditions.
6. Compliments: Positive feedback about the intervention without evidentiary weight.
This hierarchy served as a guardrail against overinterpreting weak data. Compliments and hypothetical enthusiasm were treated as low-value evidence. Specific past episodes and observable actions carried the greatest analytical weight. This approach is methodologically consistent with grounded theory’s emphasis on earning relevance from data rather than importing it from assumptions (Glaser, 1978; Charmaz, 2006).
Raw self-report enters at the bottom — only specific past behavior and observable action survive the climb
The constant comparative method was used to analyze data line by line. Memos were developed to capture emergent concepts and their relationships. The primary focus of the analysis was identifying the participants’ main concerns and the emergence of a core variable. Each interview was analyzed with four guiding questions: What is the participant actually describing? What do they care about? What are they worried about? What explains the different behaviors and outcomes across participants?
As additional interviews occurred, categories were reconceptualized and the properties informing each category were identified. Selective coding began after the initial interviews when a core concept emerged and the data were saturated across categories and their properties. Additional interviews were conducted for verification.
Ethical Considerations
All participant data were collected with informed consent. Participants were assured that notes would not include identifying information. The institutional pilot operated under the university’s student affairs data governance framework. The coaching transcripts were analyzed with the participant’s written permission.
What the Data Surfaced
The theory emerging from the coding process is adaptive identity mirroring theory (AIM-T). What follows is the working definition of the mechanism, the main concern it addresses, and the four continuums through which the mechanism produces digital awareness resilience.
Defining the Mechanism
The definition of identity mirroring that emerged from the research is: the practice of reflecting a person’s own values, language, metaphors, and self-articulations back to them in contextually enriched moments, moments when the gap between who they are becoming and what they are currently doing is most visible.
Participants did not describe effective intervention in the language of behavioral prompting. They did not say the system reminded them to do something. They said the system saw them. The distinction is not rhetorical. “Sees me” describes a specific experiential quality: the felt sense that the intervention was generated from knowledge of one’s own life, their lived experiences and their reflections around it, rather than from a generic filtered library of predetermined nudges. This quality emerged as the single strongest predictor of sustained engagement across both user and guide interviews.
Across the data, participants contrasted two qualitatively different intervention experiences. The first they described as prompting, an external cue arriving from a system that does not know them, telling them to do something they already know they should do. The second they described using language that consistently invoked recognition, presence, and relationship: the intervention felt like an encouraging friend, like someone who knew their life, like hearing their own voice at the right moment. These two experiences produced categorically different outcomes. Prompting produced short-term compliance that decayed. Mirroring produced engagement that, in some cases, strengthened over time.
The Identity Mirroring Construct
AIM-T proposes that identity mirroring is a psycho-social-contextual construct. The psychological component relates to the individual’s self-concept, values, and the ongoing process of identity formation, what Markus and Nurius (1986) termed “possible selves,” the representations of who one could become. The social component relates to the relational context in which mirroring occurs, mirroring is most powerful when it emerges from or leads toward genuine human connection, consistent with the established finding that relatedness is a fundamental psychological need (Deci & Ryan, 2000). The contextual component relates to the temporal, spatial, and situational dimensions that determine whether an intervention arrives at a moment of genuine leverage or is experienced as noise.
Across the interviews, participants described experiences that could not be reduced to any single dimension. A nudge that was psychologically resonant but contextually mistimed was ignored. A nudge that was contextually well-timed but psychologically generic was registered but not acted upon. A nudge that combined identity resonance with contextual precision was described in language that invoked involuntary recognition, what one participant called the behavioral equivalent of seeing colors when you hear music.
In earlier work on the digital extended self (Bischoff, Berezan, & Scardicchio, 2019), Belk’s (1988, 2013) theory was used to show how digital platforms reorganize the relationship between person and object in an extractive direction — possessions of attention, memory, and identity move outward from the person into systems optimized against them. AIM-T takes the same framework and inverts the vector. An identity-mirroring intervention is a cognitive object that enters the interval between question and judgment, but unlike extractive digital objects, it is designed to return the user to their own capacity rather than to deepen dependence on the platform. The measure of a humane tool is whether its user needs it less over time. The self can be reassembled only when the objects carrying its memory are loyal to the self, not to the system.
The Main Concern: The Awareness-Agency Gap
The main concern is composed of three interlocking concepts: automaticity, disconnection, and misalignment. Each is a challenge on its own, but it is their intersection that constitutes the core phenomenon — the Frankl space between impulse and action compressed, by environmental design and by the atrophy of reflective capacity, until the interval is effectively zero.
Hover the bands — particles rise from the primitive reflex layer into connection, awareness, and agency
Automaticity. Participants consistently described reaching for their phones without awareness of having decided to do so. The behavior preceded the intention. This automaticity was not experienced as choice, it was experienced as something happening to them. The distinction between automatic behavior and chosen behavior emerged as foundational. Participants who could not reliably notice the moment of reaching described feeling helpless regardless of their knowledge about digital wellbeing. This finding aligns with research on meta-awareness: the capacity to notice one’s own cognitive processes as they occur is distinct from, and prerequisite to, intentional regulation (Schooler et al., 2011).
Disconnection. Automaticity produced disconnection, from bodily signals, from emotional states, from the present environment, and from other people. Participants described the experience of looking up from a scrolling session and realizing they had been physically present but psychologically absent for 30 minutes, an hour, sometimes more. Clinicians working in digital harm reduction reported that the deepest effects of prolonged device use are not merely attentional or emotional but somatic: people become physiologically disconnected from their own signals of hunger, fatigue, discomfort, and satiation. The device overrides the body’s natural stopping signals.
Misalignment. The convergence of automaticity and disconnection produced the most painful dimension of the participants’ experience: misalignment between what they said mattered and how they actually spent their time. Participants knew what they valued — connection, presence, creative work, physical health, academic engagement. They also knew that hours of their days were absorbed by activities they could not recall choosing and did not enjoy. The resulting emotional experience was not merely frustration. It was closer to an identity threat: the gap between who one is becoming and who one wants to be.
This three-part main concern explains why information-based and compliance-based interventions consistently underperformed in the data. The problem was not that participants lacked knowledge about the harms of compulsive scrolling. Every participant could articulate the problem. The problem was that knowledge did not bridge the gap between intention and action, between knowing and noticing, between what matters in principle and what happens at 11:40 PM after a stressful day.
The Digital Reactivity Web
Participants most often experienced the awareness-agency gap within what AIM-T terms the digital reactivity web: interlocking environmental, social, and psychological triggers that sustain automatic behavior. The web structure is adapted from Brown’s (2006) shame web because participants described the same felt quality — being trapped inside layered, competing expectations from which no single interruption can free them. The analytic move AIM-T borrows from Brown is structural, not thematic: when a main concern is sustained by a web rather than a single cause, the intervention must address the web, not a single strand.
Three nodes — environmental, social, self-reinforcing — hold the user inside a closed loop
The digital reactivity web has three structural properties. First, it is environmentally reinforced. Every application, notification design, and algorithmic feed is optimized to sustain engagement by reducing the interval between impulse and action (Harris, Center for Humane Technology). The architecture of the digital environment is, in functional terms, an architecture of compression, compression of the space in which reflection could occur.
Second, the web is socially embedded. Participants described feeling that opting out of compulsive digital behavior would disconnect them from peers, miss critical social information, or mark them as different. This social dimension meant that individual intention was insufficient. The web did not merely capture individual attention; it captured the social infrastructure within which attention operates.
Third, the web is self-reinforcing. Compulsive scrolling produced guilt. Guilt produced avoidance. Avoidance produced more scrolling. This loop tracks established models of experiential avoidance (Hayes, Strosahl, & Wilson, 2012) and the harm-reduction observation that self-condemnation tends to sustain the behaviors it condemns (Marlatt, 1996).
AIM-T Continuums
AIM-T proposes that digital awareness resilience, the capacity to notice, interrupt, and redirect compulsive patterns, is best understood on a continuum. On one end lies the main concern: automaticity, disconnection, and misalignment. On the opposite end lie the components of resilience: awareness, agency, values alignment, and connection.
AIM-T further proposes that digital awareness resilience, as indicated by location on the resilience continuum, is the sum of: (a) the degree of values alignment between the individual’s articulated priorities and the intervention’s content; (b) the level of contextual awareness governing when and how the intervention arrives; (c) the capacity for reflective connection, both self-connection and interpersonal connection; and (d) the ability to speak agency, to possess the language, concepts, and emotional competence to describe one’s own patterns, name them, and externalize them.
Like the overall construct, each of these four component processes is best conceptualized as its own continuum.
Values alignment · contextual awareness · reflective connection · speaking agency
The values alignment continuum
The values alignment continuum represents the degree to which an intervention reflects the individual’s own articulated values, language, and priorities rather than externally imposed goals.
AIM-T proposes that interventions grounded in the individual’s own words and self-identified goals produce qualitatively different engagement than interventions delivering the same behavioral content without values anchoring. This finding emerged most clearly from the coaching transcript analysis. Sixteen longitudinal sessions were coded for three recurring elements: mirror material, linguistic patterns, and inflection points.
Mirror material referred to moments where the participant’s own language, reflected back during a later moment of decision, produced a visible shift in understanding. When the participant had described his creative process as an airport departure sequence (“security check, VIP lounge, then flight, you can’t skip the lounge”), reflecting that exact metaphor in a later nudge asking whether he was trying to board too soon produced immediate recognition and action. These moments became the highest-performing nudge type across the entire dataset.
Linguistic patterns included recurring forms of self-description and orientation. The most consequential pattern was the distinction between obligation language (“I have to,” “I should,” “I need to”) and agency language (“I want to,” “I’m drawn to,” “I allow myself to”). When nudges were reframed from the obligation register to the agency register, engagement changed measurably. This became a design rule grounded in data: orient toward positive agency, not deficiency correction.
Inflection points were moments where the participant’s understanding changed in real time, often marked by shifts in wording, energy, or embodied response. These occurred when the system surfaced a tension the person had not yet articulated, not by providing an insight but by returning a question that only this person could answer.
At the institutional level, the same mechanism operated through the intake assessment. When a student who had selected “make friends” as a priority later received a nudge before a campus event noting that peers with similar goals would be attending, the nudge carried weight not because it was well-written but because it connected to a self-declared priority. Survey data confirmed the pattern: 71% of respondents said nudges felt personalized, and 57% said they aligned with their actual goals.
The values alignment continuum therefore ranges from externally imposed behavioral prompts (low alignment) to interventions built from the individual’s own language, metaphors, and articulated commitments (high alignment). AIM-T proposes that interventions operating at the high end of this continuum produce engagement driven by recognition rather than compliance, and that recognition-driven engagement does not decay the way compliance-driven engagement does.
Most candidate nudges never reach the user — each gate filters for values, context, timing, personality
The contextual awareness continuum
The contextual awareness continuum represents the degree to which an intervention accounts for the individual’s temporal, spatial, physiological, and situational context at the moment of delivery.
This continuum extends Dey’s (2001) framework for context-aware computing, which identifies identity, location, time, and activity as primary context types, by adding relational context and internal state as dimensions essential for behavioral intervention. These six dimensions were not proposed theoretically and then tested. They emerged from the data as the dimensions the system needed to evaluate before a nudge could be experienced as relevant rather than intrusive.
Time emerged as an emotional modifier, not merely a scheduling parameter. Morning nudges oriented toward intention; evening nudges toward reflection. Post-class nudges showed the highest engagement for physiological recovery content (hydration, posture, movement) but the lowest engagement for cognitive content (task management, career planning). This finding is consistent with research on just-in-time adaptive interventions, which emphasize that “right time” is not merely a moment of need but a moment where need, opportunity, and receptivity converge (Nahum-Shani et al., 2018).
The most striking evidence for the timing mechanism came from within-student comparisons. When the system delivered a recovery nudge after a heavy class block, engagement was comparable to the highest-performing category overall. When it delivered a cognitive nudge in the same window, same student, same afternoon, engagement dropped to the lowest category. The mechanism was not the message but the match between message and moment.
Three streams oscillate independently — the nudge fires only in the rare moment they meet
AIM-T proposes that contextual awareness is not a peripheral design feature but a primary mechanism of action. The same content delivered with high contextual fit is an intervention. The same content delivered without contextual fit is noise.
The reflective connection continuum
The reflective connection continuum represents the individual’s capacity for two forms of connection: connection to one’s own internal states (self-connection) and connection to others who share similar experiences (interpersonal connection).
Participants reported that when they experienced a moment of genuine self-recognition through an identity-mirroring intervention — when they felt seen in their own terms — the experience increased their capacity for reflective self-connection. They became more able to notice their own states, more willing to pause, more practiced at the micro-skill of interrupting automaticity. This is consistent with Rogers’ (1961) observation that accurate empathic understanding is a precondition for therapeutic change.
At the interpersonal level, participants reported that one of the most important consequences of engaging with values-mirroring interventions was the discovery that their experiences were not unique. The recognition that compulsive scrolling, attention fragmentation, and the awareness-agency gap were shared phenomena, not personal failures, reduced the isolating quality of the experience.
Guides independently identified the between-session gap as the structural challenge that most constrained their effectiveness. The period between appointments, coaching sessions, or support encounters was the space where change most often succeeded or failed. Universities, advisors, and wellness professionals knew that support was episodic while behavior change was continuous. The identity-mirroring intervention addressed this gap not by replacing human connection but by extending its continuity, carrying forward the user’s own language and goals between live touchpoints.
The speaking agency continuum
The speaking agency continuum represents the individual’s fluency in the language of awareness, agency, and identity as they relate to digital behavior. It is the capacity to name one’s own patterns, externalize them, distinguish between automatic behavior and chosen behavior, and articulate the felt difference between prompting and mirroring.
AIM-T proposes that fluency in this language is both a product of and a contributor to digital awareness resilience. Participants who possessed conceptual vocabulary for their experience, who could say “that was automatic, I didn’t choose it” or “I noticed the pull but made a different choice”, demonstrated measurably higher capacity to interrupt compulsive patterns. Participants who lacked this vocabulary described the experience as undifferentiated distress: they knew something was wrong but could not name what was happening or why.
This continuum is modeled directly on Brown’s (2006) “speaking shame” continuum — not because shame and digital behavior are the same experience, but because Brown demonstrated that acquiring vocabulary for a previously undifferentiated experience is itself a mechanism of change, not merely a consequence of one. AIM-T claims the same dynamic operates in the digital domain: teaching participants the concepts of automaticity, the awareness-agency gap, and the reactivity web equips them with the framework that makes the experience visible in the first place, which is what makes it interruptible.
Participants who developed this fluency described a characteristic shift in self-relation. Rather than experiencing compulsive scrolling as a personal failure, “I have no willpower,” “I’m addicted,” “something is wrong with me”, they began to experience it as an environmental and developmental challenge: a gap between current capacity and desired agency, situated within a digital environment designed to exploit that gap. This reframing was not merely cognitive. It changed behavior. Participants who understood the mechanism, who could speak agency, were more likely to interrupt compulsive patterns because they had an alternative interpretation available in the moment.
Preliminary Quantitative Findings
The nine-month institutional pilot and survey subset produced findings consistent with the theoretical categories described above.
Engagement. The platform achieved an 85% active nudge response rate, defined as the proportion of enrolled students who performed at least one response action within any 7-day rolling window. This rate sustained without significant decay over the nine-month pilot. Comparable institutional wellness platforms report 30–40% engagement rates with rapid decay (Lentferink et al., 2017). The engagement differential is not explained by superior content. The same behavioral content delivered without personality calibration and values matching produces standard engagement rates. When the same content is delivered in a voice that matches the recipient’s identity and context, engagement increases substantially.
Wellbeing trajectory. The composite wellbeing score improved from 72 to 82 over one semester. Institutional stakeholders noted that comparable improvements typically require 12–18 months through conventional programming. Pilot-level data recorded 14,222 total student actions across the 100-person cohort.
Survey findings (n = 48). Among the survey subset: 89% of students remained engaged during the study period; 57% reported significantly less compulsive scrolling; 44% to 56% reported increased awareness across mental health, study habits, and help-seeking; 71% said the nudges felt personalized; 57% said the nudges aligned with their actual goals; and 71% rated the system as helpful for focus and time management.
Category performance. Tangible, physiologically oriented nudges (hydration, posture, movement) consistently outperformed abstract or aspirational content (task management, career planning) across all personality types and demographic segments. The highest-performing nudges combined physical immediacy with humor or identity recognition. This paradox, that nudges not targeting stated goals outperformed those that did, is explained by AIM-T’s identity mechanism. Hydration and posture nudges function not as behavioral instructions but as invitations to exercise agency. They communicate: you are a person who can act on your own behalf, right now, in this moment. Each successful micro-action deposits a unit of self-efficacy (Bandura, 1977) that compounds over time into an identity consolidation: “I am someone who takes care of themselves.”
Crisis intervention. The most consequential single finding involved a student in acute emotional distress who, reaching for his phone during a crisis moment, encountered a Journey nudge offering a grounding exercise instead of a social media application. The student completed the exercise, experienced a measurable emotional shift, and four months later continued to cite that moment as defining his relationship with the platform. From a behavioral prompting perspective, a single prompt should not produce four months of sustained behavioral change. From an identity mirroring perspective, the finding is expected. The crisis moment crystallized an identity choice: “I am someone who, in moments of distress, reaches for a tool that reflects my values back to me.” The nudge’s content was the vehicle. The mechanism was identity-level.
Where AIM-T Sits in the Literature
The purpose of the grounded theory literature analysis is to demonstrate how the hypotheses and theoretical concepts that emerged from this research support and question existing literature. AIM-T proposes a contextualized and multidisciplinary understanding of digital behavior change that is not easily categorized into any single existing paradigm. The model brings together behavioral economics, clinical psychology, identity theory, context-aware computing, harm reduction, and the emerging field of digital wellbeing.
Relation to Nudge Theory and the Behavior Model
AIM-T does not propose that nudge theory is wrong. Thaler and Sunstein’s (2008) framework accurately describes how environmental design shapes individual choices. Fogg’s (2009) Behavior Model accurately describes the conditions under which a single behavioral instance occurs. AIM-T proposes that these frameworks are insufficient for the outcomes they claim to pursue. Prompting behavior is not the same as changing identity. Cue-routine-reward is not the same as values clarification. A notification, no matter how well-timed, is not the same as a mirror. Clear (2018) reformulates the habit loop to place identity at the foundation, identity leads to process, which leads to outcome, but even Clear’s model, when operationalized in digital systems, typically collapses back into cue-routine-reward because the technological infrastructure does not know who the user is. It knows what the user did.
Relation to Motivational Interviewing and Values-Based Change
AIM-T is probably best supported by, and lends the most support to, the motivational interviewing tradition (Miller & Rollnick, 2013) and values-based behavior change frameworks including acceptance and commitment therapy (Hayes, Strosahl, & Wilson, 2012). The central mechanism of motivational interviewing, evoking the client’s own reasons for change through reflective listening rather than providing external motivation, maps directly onto the identity mirroring mechanism. Rogers’ (1961) core condition of accurate empathic understanding is, in functional terms, what the identity-mirroring system attempts to implement at scale through technology.
The fact that reflective connection and values alignment play central roles in both motivational interviewing and AIM-T might be attributed to the highly inductive methods used to generate both frameworks. Similar to the grounded theory methodology used in this study, motivational interviewing was developed through careful observation of what actually worked in clinical practice rather than through top-down theoretical imposition (Miller & Rollnick, 2013).
Relation to Identity Theory
AIM-T operationalizes Markus and Nurius’s (1986) “possible selves”: the identity-mirroring nudge does not create an aspiration, it returns one the user has already articulated at a moment of decision. Stryker and Burke’s (2000) identity salience offers a likely mechanism of action — the mirror increases the salience of the aspirational identity relative to the automatic pattern. Belk’s (1988, 2013) extended self and our earlier work on the digital extended self (Bischoff, Berezan, & Scardicchio, 2019) are developed in the Findings section’s bridge note.
Relation to Context-Aware Computing
The contextual awareness continuum extends Dey’s (2001) framework from computing into behavioral intervention. Dey and Abowd identified four primary context types: identity, location, time, and activity. They noted these “not only answer the questions of who, what, when, and where, but also act as indices into other sources of contextual information.” AIM-T adds relational context and internal state, dimensions that emerged from the data as essential for the intervention to produce the felt quality of presence rather than mere notification. This extension aligns with the just-in-time adaptive intervention (JITAI) literature, which formalizes the principle that effective intervention requires the convergence of need, opportunity, and receptivity at a specific moment (Nahum-Shani et al., 2018).
Relation to Harm Reduction
AIM-T adopts a digital harm reduction frame rather than a digital abstinence frame. This positioning aligns with Marlatt’s (1996) foundational harm reduction work, which proposed that reducing harm and increasing agency within the conditions people actually inhabit is more productive than demanding abstinence. Applied to digital behavior, harm reduction does not treat all technology use as pathological. It begins from the practical question of how to increase awareness and agency within an attention-extractive environment.
Relation to Shame Resilience Theory
AIM-T is methodologically and structurally indebted to Brown’s (2006) shame resilience theory (SRT), and the debt is worth naming precisely because the two theories address different problems. What AIM-T borrows from SRT is a form, not a theme: a grounded-theory-generated main concern composed of three interlocking concepts, continuums rather than categories, and the claim that acquiring language for the experience is itself a mechanism of change. AIM-T does not treat the digital awareness-agency gap as a shame phenomenon. It treats it as a distinct psycho-social-contextual construct that happens to share SRT’s architecture because, in both cases, the main concern is sustained by a web of forces rather than by a single cause — and theories of that shape require continuums, relational resilience, and vocabulary-as-mechanism to describe how people climb out.
Relation to Metacognition and Awareness Research
The awareness-agency gap is consistent with research on meta-awareness, the capacity to take explicit note of one’s own mental processes as they occur (Schooler et al., 2011). Schooler and colleagues distinguish between experiential engagement (being absorbed in a process) and meta-awareness (noticing that one is absorbed). AIM-T proposes that compulsive digital behavior is characterized by experiential engagement without meta-awareness, and that identity mirroring supports the development of meta-awareness at moments of automatic behavior. This is consistent with Shapiro and colleagues’ (2006) analysis of mindfulness mechanisms, which identifies attention regulation, body awareness, and emotion regulation as trainable capacities. The identity-mirroring intervention is not a mindfulness practice, but it operates through a related mechanism: creating a moment of meta-awareness within the flow of automatic behavior.
What This Means in Practice
For Digital Intervention Design
If identity mirroring is the mechanism, the architecture of digital behavior change has to shift. Four principles follow from the continuums: personalization must begin from the user’s own words, not demographic segments; timing must be treated as a primary mechanism rather than a scheduling layer; the system must be designed to be outgrown, not to maximize dependency; and crisis availability is likely the highest-leverage single investment, because one well-timed mirror during acute distress appears to carry effects forward for months.
For Institutional Practice
Guides named the between-session gap as the single structural limitation of current student support. AIM-T offers a way to bridge that gap without replacing the human relationship: carrying the student’s own language and goals forward between live touchpoints, acting as a mentoring capacity multiplier rather than a replacement.
For Psychoeducation
The speaking agency continuum implies that naming the pattern is part of changing it. Teaching emerging adults the concepts of automaticity, the awareness-agency gap, and the reactivity web gives them a vocabulary for experiences that are otherwise undifferentiated distress.
Limitations
The qualitative sample is modest, the pilot cohort (N = 100) sits at a single institution recruited through a course pathway, and there is no randomized control. The composite wellbeing measure is not psychometrically validated. Self-report is vulnerable to recall and social-desirability bias. The coaching corpus derives from a single longitudinal relationship. The researcher is also the system’s developer.
Future Research
Priorities follow directly from those limits: a comparative design pitting values-mirrored nudges against category- and timing-matched generic nudges; a waitlist RCT for stronger causal inference; validation of the wellbeing composite against PHQ-9, GAD-7, and SWLS; and longitudinal follow-up to test whether identity-level change persists after nudging stops. The planned 1,800–2,500-student expansion provides power for subgroup analysis by personality type, demographic, and risk tier.
Works Cited
Bandura, A. (1977). Self-efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84(2), 191–215.
Belk, R. W. (1988). Possessions and the extended self. Journal of Consumer Research, 15(2), 139–168.
Belk, R. W. (2013). Extended self in a digital world. Journal of Consumer Research, 40(3), 477–500.
Bischoff, J., Berezan, O., & Scardicchio, L. (2019). The digital self and customer loyalty: From theory to virtual reality. Journal of Marketing Analytics, 7(4), 220–233.
Brown, B. (2006). Shame resilience theory: A grounded theory study on women and shame. Families in Society, 87(1), 43–52.
Brown, K. W., & Ryan, R. M. (2003). The benefits of being present: Mindfulness and its role in psychological well-being. Journal of Personality and Social Psychology, 85(4), 822–848.
Castleman, B. L., & Page, L. C. (2015). Summer nudging. Journal of Economic Behavior & Organization, 115, 144–160.
Charmaz, K. (2006). Constructing grounded theory. Sage.
Cialdini, R. B. (2009). Influence: The psychology of persuasion (Rev. ed.). HarperCollins.
Clear, J. (2018). Atomic habits. Avery.
Consolvo, S., McDonald, D. W., & Landay, J. A. (2009). Theory-driven design strategies for technologies that support behavior change in everyday life. CHI ’09, 405–414.
Creswell, J. W. (2013). Qualitative inquiry and research design (3rd ed.). Sage.
Davidson, R. J., & Begley, S. (2012). The emotional life of your brain. Penguin.
Deci, E. L., & Ryan, R. M. (2000). The “what” and “why” of goal pursuits. Psychological Inquiry, 11(4), 227–268.
Dey, A. K. (2001). Understanding and using context. Personal and Ubiquitous Computing, 5(1), 4–7.
Eisenberg, D., Gollust, S. E., Golberstein, E., & Hefner, J. L. (2007). Prevalence and correlates of depression, anxiety, and suicidality among university students. American Journal of Orthopsychiatry, 77(4), 534–542.
Fitzpatrick, R. (2013). The Mom Test. Robfitz Ltd.
Fogg, B. J. (2009). A behavior model for persuasive design. Persuasive ’09, 1–7.
Frankl, V. E. (2006). Man’s search for meaning. Beacon Press.
Glaser, B. G. (1978). Theoretical sensitivity. Sociology Press.
Glaser, B. G., & Strauss, A. L. (1967). The discovery of grounded theory. Aldine.
Haidt, J. (2024). The anxious generation. Penguin Press.
Hamari, J., Koivisto, J., & Sarsa, H. (2014). Does gamification work? HICSS ’14, 3025–3034.
Hayes, S. C., Strosahl, K. D., & Wilson, K. G. (2012). Acceptance and commitment therapy (2nd ed.). Guilford Press.
Hunt, J., & Eisenberg, D. (2010). Mental health problems and help-seeking behavior among college students. Journal of Adolescent Health, 46(1), 3–10.
Kahneman, D. (2011). Thinking, fast and slow. Farrar, Straus and Giroux.
Lentferink, A. J., et al. (2017). Key components in eHealth interventions. JMIR, 19(8), e277.
Lipson, S. K., Lattie, E. G., & Eisenberg, D. (2019). Increased rates of mental health service utilization. Psychiatric Services, 70(1), 60–63.
Markus, H., & Nurius, P. (1986). Possible selves. American Psychologist, 41(9), 954–969.
Marlatt, G. A. (1996). Harm reduction: Come as you are. Addictive Behaviors, 21(6), 779–788.
Miller, W. R., & Rollnick, S. (2013). Motivational interviewing (3rd ed.). Guilford Press.
Nahum-Shani, I., et al. (2018). Just-in-time adaptive interventions in mobile health. Annals of Behavioral Medicine, 52(6), 446–462.
Orben, A., & Przybylski, A. K. (2019). The association between adolescent well-being and digital technology use. Nature Human Behaviour, 3(2), 173–182.
Oreopoulos, P., & Petronijevic, U. (2018). Student coaching: How far can technology go? Journal of Human Resources, 53(2), 299–329.
Primack, B. A., et al. (2017). Social media use and perceived social isolation. AJPM, 53(1), 1–8.
Prochaska, J. O., & DiClemente, C. C. (1983). Stages and processes of self-change of smoking. Journal of Consulting and Clinical Psychology, 51(3), 390–395.
Rogers, C. R. (1961). On becoming a person. Houghton Mifflin.
Schooler, J. W., et al. (2011). Meta-awareness, perceptual decoupling and the wandering mind. Trends in Cognitive Sciences, 15(7), 319–326.
Shapiro, S. L., Carlson, L. E., Astin, J. A., & Freedman, B. (2006). Mechanisms of mindfulness. Journal of Clinical Psychology, 62(3), 373–386.
Strauss, A., & Corbin, J. (1998). Basics of qualitative research (2nd ed.). Sage.
Stryker, S., & Burke, P. J. (2000). The past, present, and future of an identity theory. Social Psychology Quarterly, 63(4), 284–297.
Thaler, R. H., & Sunstein, C. R. (2008). Nudge. Yale University Press.
Twenge, J. M. (2017). iGen. Atria Books.
Weiser, M., & Brown, J. S. (1996). The coming age of calm technology. Xerox PARC.
