Investigating Interventions in Fine-grained Contexts for Habit Formation

Abstract

Behavior change is important, yet hard to sustain. Habits are automatic responses to specific contextual cues, and can help sustain behavior change. Fine-grained specific contexts are commonly used in habit formation, but interventions in automatically-detected fine-grained contexts have rarely been explored for habit formation.

We investigate habit-formation using interventions in fine-grained mobile, physical-world and digital, computer-based contexts, making three key contributions for each: a survey to identify behavior change needs, a prototype system designed to deliver fine-grained context-specific interventions, and a study to investigate habit-formation using interventions in fine-grained contexts, compared to interventions in less fine-grained contexts. We use the Self-report Habit Index (SRHI) and Self-Report Behavioral Automaticity Index (SRBAI) to measure habit formation and habit automaticity, respectively.

For mobile, physical-world behavior change, the survey of needs (N=53 participants) indicated that participants want diverse and personalized behavior change support in diverse and specific contexts. We created a wearable device with on-device deep learning for interventions in personalized and privacy-preserving egocentric visual contexts. In a 4-week pilot study (N=10), interventions in egocentric visual contexts led to more percentage increase in average habit formation (SRHI) and automaticity (SRBAI) than interventions in coarse-grained contexts based on time, geolocation, and physical activity. The percentage increase in median habit formation was also more for the fine-grained egocentric context group, whereas the percentage increase in median habit automaticity was similar between the two groups. For both groups, the habits persisted in the post-study evaluations 1 and 10 weeks later, without interventions.

For computer-usage behavior change, the survey of needs (N=68) indicated that participants want to reduce excessive/unnecessary use, e.g., social media, and found off-the-screen breaks helpful. We created a Chrome extension to deliver interventions based on specific web activities, and conducted a 6+2-week study (N=31, 6 weeks of interventions and 2 weeks of post-study without interventions). After 6 weeks, interventions in fine-grained website-entry-based contexts led to more percentage increase in mean and median habit formation and automaticity than interventions in coarse-grained interval-based or random contexts. After the additional two-week post-study, without interventions, the website-entry group had the largest percentage increase in mean SRHI/SRBAI, whereas the interval-based group had the largest percentage increase in median SRHI/SRBAI.

Qualitative results from both studies indicated that interventions in fine-grained contexts were delivered at more opportune moments and were less disruptive. We discuss the limitations of our research and present a first step towards investigating interventions in fine-grained contexts for habit formation, potentially for sustainable behavior change, without long-term dependence on technology.