"Chasing Shadows": Understanding Personal Data Externalization and Self-Tracking for Neurodivergent Individuals
Tanya Rudberg Selin, Danielle Uneus, Soren Knudsen · 2026 · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI '26) · doi:10.1145/3772318.3794641
Summary
This CHI 2026 paper examines a mismatch between the promises of self-tracking and personal informatics — rich self-awareness, reflection, insight — and the lived experience of neurodivergent individuals, for whom stable routines, generalised Likert items, and internal consistency checks commonly encoded in tracking instruments impose substantial interpretive and emotional demands. The authors, who position themselves as neurodivergent researchers with ADHD and autism, focus on masking (modifying or suppressing behaviours to fit neuronormative contexts) and unmasking, drawing on Walker, Halsall, and crip-HCI traditions to frame these as ambiguous, context-dependent, emotionally complex phenomena poorly served by conventional tracking designs. The study is a two-phase reflexive-thematic-analysis study with six Swedish-speaking neurodivergent adults (P3-P8, ages 28-31, mixed genders; diagnoses include ADHD, autism, or both). Phase I was a three-hour externalisation workshop in which participants completed the Camouflaging Autistic Traits Questionnaire (CAT-Q), then externalised masking experiences through drawing and writing prompts and discussed them in breakout groups. Phase II extended for three participants (P3, P5, P6) into 1-2 weeks of personalised self-tracking of masking in a context they chose (social situations, school, work); each designed their own logging approach — event-based Likert logs (P3), an hourly 64-item diagram sheet (P5), and minimal X-marks on a work calendar plus reflective notes (P6). Data included workshop and consultation transcripts, externalisations, self-tracking logs, and 40-60 minute exit interviews, analysed via reflexive thematic analysis (Braun and Clarke) conducted by the neurodivergent author team.
Key findings
Four synthesising themes emerged. (1) Defining and identifying masking was itself difficult — P6: "It feels like I've been chasing shadows"; P4: "It is still difficult today … especially when you relate it to yourself." (2) Context-dependency defeated generalisation: P5 summarised it as "it depends," and participants consistently struggled with CAT-Q items that aggregated over "social interactions" in the abstract. (3) Overthinking responses was pervasive; internal consistency checks (repeated or rephrased items, commonly added to validate responses) provoked anxiety, self-doubt, and time blow-out rather than the intended validation effect. (4) Shared, peer-led reflection left a lasting positive impact — P6: "It feels like you have strength in numbers"; four of six explicitly said the workshop made them "not feel alone." From these findings the authors synthesise a working model of three emotional dimensions in personal informatics that most existing frameworks overlook: emotional weight (immediate stress from overthinking during data entry), emotional self-reflection (discomfort from insights the data surfaces, e.g., P8 realising how much he masked), and emotional burden (the accumulated pressure of commitment to ongoing tracking). They argue that self-tracking guidance should treat reflection as a sufficient goal (not just a step toward behaviour change), embrace simple methods and short-term goals, and prioritise adaptability. Short completion-time estimates on tools like CAT-Q (a few minutes) are critiqued as a form of epistemic injustice because neurodivergent engagement requires significantly more time. The authors explicitly propose peer-supported reflection — facilitated group sharing, accountability partnering, body-doubling — as a design direction that offsets emotional weight, deepens self-reflection, and normalises the burden.
Relevance
For practitioners designing self-tracking, wellness, mental-health, or assessment apps, the paper flips a default assumption: that longer deployments and more structured categorisation produce more valid insight. For neurodivergent users, the opposite may hold — brief, flexible, context-rich logging plus peer-supported reflection produces more honest data and better wellbeing outcomes. Actionable implications include: expose reflection as a legitimate endpoint rather than a step toward behaviour change; remove or flag internal consistency checks when targeting neurodivergent populations; allow free-text alongside any quantitative item; publish realistic, not aspirational, completion-time estimates; and integrate facilitated peer-sharing primitives into tracking apps (shared reflections, partnered accountability, body-doubling sessions). The emotional-dimensions framework (weight, self-reflection, burden) transfers to other invisible-disability contexts where authenticity and identity concealment are everyday challenges, such as mental-health apps, LGBTQ+ wellness tools, and chronic-illness tracking. Limitations to flag: convenience sample of six Swedish-speaking, verbally and visually articulate young adults drawn from the authors' own networks, only three completed Phase II, all self-describe as ADHD and/or autistic; the model is proposed as a working hypothesis, not validated. The authors' own neurodivergent positionality is a stated analytical resource but also a potential interpretive bias.
Tags: self-tracking · personal informatics · neurodiversity · autism · ADHD · masking · unmasking · qualitative research · data visualization