Focus Mode for Digital Wellbeing

 A cognitively informed design for a smartphone launcher


MS Thesis

Supervisor: Prof. Harish C Karnick |  CGS, IITK

ABSTRACT

It is possible to design digital spaces that exploit users while the users are unaware of how they are being used to benefit someone else. There are abundant digital means for almost every aspect of human life. To stay in the market, businesses are in the game of seeking continued user attention. In order to take control and leverage the capabilities of technology, it is important to design technology as per the needs of the human mind and resist any exploitation of its limitations. Digital Wellbeing envisions an optimal balance of technology usage to promote work productivity among other goals to achieving wellbeing. 


A number of design choices make up today’s ubiquitous UI on smartphones. Traditional smartphone interfaces propagate interruption-driven usage there by hindering task-driven usage during device launch. We hypothesise that Interruptions and Distractions both lead to Goal Slippage. Our study focuses on the pervasive menu-grid interaction for smartphone app selection and challenges the heuristic design principle ‘Recognition rather than Recall’ that recommends it against plain search. We investigate the phenomenon of Goal Slippage in digital devices through an experiment that captures the extent of Goal Slippage and Loss of Content with the altering presence of external interruptions and distractions. We find that external interruptions have a significant effect on response times suggesting Goal Slippage but distractions do not have any effect. However, a wider spread of response time is established during distractions suggesting disruption if not complete goal slippage. 


Based on the results, we attempt to design a smartphone launch interface ‘Focus Mode’ for digital wellbeing backed by a set of design principles. 

Survey

We have conducted a user survey(n=35) on smartphone usage for the age group 20 to 38 years. From the survey results, many concerns about smartphone usage and the need for digital wellbeing become evident. Users are aware of their dependency and a lack of control over their usage. We also see that smartphones are being considered one-for-all solutions, where they are the go-to devices for both Work and Leisure.

Problem Statement

A digital gadget user should be able to use their gadget for their needs and solely for it. In today’s world, everyone uses their device for more than just that. We daily spend 3 hours and 15 minutes on our phones, a number that can easily shoot up considering the post-pandemic times. We experience a lack of control over our usage. A part of the problem is deliberate design. Interfaces are primarily interrupt-driven by design. Can we pursue a task-driven design approach and undo the disruptive nature of gadgets? We think that the current best practices and principles of Interface design will not lead there.

To establish empirical evidence, we set out to capture the extent of Goal Slippage and Loss of Content with the altering presence of external interruptions and distractions. For the level of distractions, our study challenges the pervasive menu-grid interaction for smartphone app selection and thereby challenges the heuristic design principle ‘Recognition rather than Recall’ that recommends it. We present plain search as an alternative. For the level of interruptions, a mandatory interruption task is employed to represent the interruption-driven usage against task-driven usage. If results support our line of prediction, we will propose a set of design principles and attempt to design a smartphone launch interface that mitigates goal slippage.

Hypothesis

While interacting with a smartphone for accomplishing a specific goal/task, loss of content and goal slippage are Significant when the primary task is externally interrupted.

While interacting with a smartphone for accomplishing a specific goal/task, loss of content and goal slippage are Significant when distractions are present.

Experiment

A within-subjects design was employed where two factors acted as independent variables with two levels each. The factors were ‘Interruption’ and ‘Distraction’ with levels being presence and absence. In the ‘Hold’ stage a Blank Screen meant the absence of Interruption and the Word Game indicated the presence of Interruption. The Search component during the ‘Navigate’ stage meant the absence of Distraction, and the Menu Grid meant the presence of Distraction. For each participant, 25% of the trials belonged to each condition of the 2 × 2 factorial design.

The four conditions are (Blank Screen, Search), (Blank Screen, Menu Grid), (Word Game, Search), and (Word Game, Menu Grid).

Links to Experiment

Instructions and Practice Trials: link
Main Experiment: link
Survey on Task Variability: link



Encode / Framing of the Primary Task (above)Distraction-present(left) and Distraction-absent(right) variants

Result

The trials where participants could not recall the goal (i.e. the target app) are assigned under ‘Goal Slippage’. The trials where the goal is recalled but content keywords were either not recalled or recalled incorrectly are assigned under ‘Goal Retention but Loss of Content’. Participants’ recall is presented below out of the 420 trials in each condition.

Condition-wise Goal Slippage and Loss of Content
Probability Density Functions of condition-wise RT GOAL
Cumulative Density Functions of condition-wise RT GOAL

Two Way Repeated Measures ANOVA of RT Goal

RT Goal was analysed using a 2 × 2 repeated measures ANOVA with the within-subject factors Interruption (Blank Screen, Word Game) and Distraction (Search, Menu Grid) to evaluate whether they modulated RT Goal at a significant level.

Medium Effect (0.26 Cohen’s f) of Interruption was observed, F(1,1569)= 110.9, p < 0.001. Distraction and Interactions did not have any significant effect on RT Goal.


Right-End Tails between Distraction Conditions

Since the focus of interest are the tails of the distributions we compared the occurrence of large RT Goal values between distraction conditions. We analysed the differences for the right extreme 10% of all the values belonging to Complete Retention trials for each condition.

To evaluate the differences in RT Goal means under Blank Screen condition, Welch’s t-test was performed, and a significant difference was found.

As variances are similar for distraction conditions under Word Game, an independent two-sample t-test was carried out to evaluate the difference in means, and a significant difference was found.


Welch's t-test (Blank Screen Condition)
Two-sample t-test (Word Game Condition)

Conclusion

With our experiment, we found that Interruption has a significant effect on RT Goal, and Distraction has no effect, both in terms of differences in means and number of assigned Goal Slippage and Loss of Content instances between conditions. Compared with an assumed baseline for immediate recall, every condition had a significantly different trial count suggesting that disruption is associated with delayed task execution. Under the Blank Screen condition, distraction conditions were found to have significantly different variances. While selectively analysing the Right tail(10%), RT Goal’s differences between distraction conditions were found to be significant under both the Interruption conditions. This indicates clear disruption from a MenuGrid if not complete goal slippage. We looked at the practical significance of this difference in variance and recommend that interface designs need to eliminate distractions whenever possible. With a stricter recall duration and/or a stronger interruption task, we predict that there will be a clear preference for Recall over Recognition based on our main result. Task related factors, ‘Familiarity’ and ‘Pleasantness’ were found to differ significantly between groups suggesting that familiar and inherently pleasant tasks are easier to recollect in general.

Design for 'Focus Mode'

We have proposed a Design System, Focus Mode, that re-imagines the smartphone launch experience, guided by the design principles. The launcher consists of minimal interaction elements while supporting touch and voice input. The home screen is set up to receive utterances of the user’s primary task. A specific screen of a target app can be accessed directly from the launcher home screen using semantic categorization of the user’s utterance.

Design Principles

During the analysis, certain unifying elements were found to be guiding our design solution. Here we establish them as design principles.

Generelized Flowchart for Task-driven Interaction

THESIS REPORT

19128410_Sruzan_MSR_Thesis.pdf