Wearable Computing for Health and Fitness: Exploring the Relationship between Data and Human Behaviour

Natalie Wilde, Katrin Hänsel, Hamed Haddadi, Akram AlomainyarXiv 2015 · arXiv

Introduction

The market for health and fitness wearables has exploded, yet the industry faces a significant challenge: an estimated 85% dropout rate. While modern sensors are highly capable of logging raw physiological data, they often fail to present this information in a way that sustains long-term engagement. This "meaningfulness gap" suggests that logging data is not synonymous with changing behavior. This paper explores how wearable technology can move beyond simple step-counting toward systems that support intrinsic reflection and personalized, contextual feedback.

Categorizing Health Technologies

The research synthesizes various projects and products into two primary paradigms:

  1. Persuasive Technologies: These focus on driving behavior through extrinsic motivators like gamification (points, badges, leaderboards) and social influence (competition and social proof).
  2. Supportive Technologies: These prioritize reflection and self-efficacy. This includes "Quantified Self" approaches and biofeedback that encourage users to understand their own patterns rather than chasing a score.

Theoretical Foundations

To evaluate these technologies, the authors utilize established psychological frameworks, most notably Fogg’s Behavior Model.

As shown in the model above, behavior change occurs when motivation, ability, and a trigger converge. If a task is "hard to do," it requires high motivation to cross the activation threshold. Conversely, if a task is "easy to do," lower motivation is sufficient. Triggers only succeed when they occur above this threshold. Wearables often fail because they provide triggers (notifications) when the user either lacks the ability or the motivation to act in that specific moment.

Key Insights: From Extrinsic to Intrinsic Motivation

A critical finding is the "clouding" effect of rewards. While badges and points can jumpstart activity, they often fail to internalize the behavior. If the rewards are removed, the healthy behavior typically ceases because it was never truly integrated into the user's lifestyle.

Ambient vs. Literal Displays

Raw numbers can be overwhelming or discouraging. The paper highlights "ambient" displays—such as the UbiFit system, where a virtual garden blooms as a user walks—as more persuasive. These displays require less cognitive effort to process and provide an emotional connection to progress, making the data more meaningful.

Just-in-Time Interventions

Feedback is most effective when it is "context-aware." A wearable that detects high stress and suggests a breathing exercise in the moment is far more powerful than a weekly summary report. This shift toward "just-in-time" support aligns triggers with the user's current state and ability.

Results and Case Studies

The survey highlights several specific outcomes from the field:

  • Cognitive Dissonance: The Fatbelt, which physically inflates to simulate weight gain when a user overeats, successfully reduced calorie consumption by making future consequences tangible and immediate.
  • Social Proof: The Fish’n’Steps application demonstrated that seeing a peer's progress (represented by a growing fish avatar) boosted activity levels across the entire group, emphasizing the power of social cooperation over pure competition.
  • Personalization: Systems that allowed for "Tiny Habits"—small, achievable goals—showed much higher long-term retention than generic "10,000 step" targets.

Limitations and Future Directions

The field faces a "six-month wall," with a lack of longitudinal data proving long-term efficacy beyond the initial novelty phase. Furthermore, the tension between privacy and utility remains; providing "just-in-time" support requires access to sensitive personal data like location and calendar. Future research should focus on adaptive interfaces that evolve their motivational strategies based on the user's personality and their current stage of behavioral change.

Summary updated

BibTeX
@misc{wilde2015wearable,
  author = {Natalie Wilde and Katrin Hänsel and Hamed Haddadi and Akram Alomainy},
  title = {{Wearable Computing for Health and Fitness: Exploring the Relationship between Data and Human Behaviour}},
  year = {2015},
  eprint = {1509.05238},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/1509.05238},
}
Figure