Case study · Behavioral design · TikTok extension · HCDE 590–591 Capstone · MindOS

Project Rewire

A TikTok-integrated extension that detects passive scrolling and redirects users toward meaningful engagement — without removing user autonomy.

UX Researcher · team of 3 27 participants 2 iteration rounds U-Michigan

What I owned

Research synthesis (10 of 22-paper lit review, affinity + journey mapping) · designed and ran 2 focus groups and 8 of 16 think-aloud sessions · connected Four Drive Theory to the avatar system · authored the non-coercive ethical framework · wrote microcopy across prompts, challenges, and onboarding · originated the Friend-Authored Challenges feature.

22Sources in the literature review

27Participants across interviews, focus groups & 2 testing rounds

2Expert clinical reviews validating the concept

+0.092VADER sentiment · p < .05 · extremely large effect — the study's one statistically significant result

Sentiment improved significantly between iterations; every other measure trended positive but needs a larger N. Honest read: the concept is psychologically grounded and expert-validated — the measurement design is what needs to scale next.

Problem Research Artifacts Insights Solution Iterations Results Experts Final design Reflection

The problem

Brain rot is a design problem.

Short-form platforms are engineered for engagement, not wellbeing — users lose control of their time and intentions. Unlike blockers and screen-time limits, Rewire aligns intention with behavior instead of just reducing usage.

"How might we redesign traditional gamification to foster meaningful engagement — and reduce the cognitive effects of brain rot?"

DESIGN QUESTION

01 — Research

Five methods. One clear picture.

22Lit review

Behavioral science grounding; surfaced JITAI

7Competitors

Mapped the gamification landscape

5Interviews

1:1 depth on scroll habits

2Focus groups

Group dynamics around social media use

16Think-alouds

N=8 per round, 2 iteration rounds

~90 coded observations clustered into 5 themes. The consistent thread: users aren't failing — platforms are exploiting compulsive behavior patterns. Competitor analysis confirmed no product combines intrinsic motivation, moment-of-behavior timing (JITAI), and autonomy-preserving design.

"I know it's bad for me but I literally can't stop. I open it without even thinking."

SARA L. · PRIMARY PERSONA "THE AWARE ADDICT" · ~2.5 HRS/DAY

Research artifacts

The synthesis work behind the concept.

The redesigned case study keeps the narrative tight, but these artifacts show how the research translated into product strategy: a journey map for the scroll loop, an affinity diagram for user needs, and a competitive analysis for the market gap.

3Artifact types

~90Coded observations

5Need clusters

7Competitors mapped

Think-aloud + interviews Passive scroll session Current + future state

What it feels like to lose control.

The journey map contrasts the current unintentional scroll loop with the intervention-supported path Rewire creates. The key design opportunity is the emotional gap: the moment users move from entertained to aware, guilty, and unable to exit cleanly.

Current state With Rewire

The graph reframes the journey as an emotional arc: without Rewire, early reward turns into a late guilt crash; with Rewire, the threshold prompt redirects the session before self-blame becomes the dominant state.

Current-state failure

Awareness arrives after the user has already lost time, which makes the intervention point emotionally loaded and hard to act on.

Rewire opportunity

Move the intervention earlier, keep it user-controlled, and offer a meaningful next action instead of a punishment.

Interviews + focus groups Literature synthesis ~90 observations

Five clusters. One system shaped by compulsive design.

Raw observations were grouped into five underlying requirements. The thread across clusters was consistent: users wanted social media to meet real needs, but the platform mechanics turned those needs into passive, hard-to-exit behavior.

5Underlying requirements

Relief from boredom or stress 7 observations

Users open short-form feeds for decompression, humor, inspiration, or empty waiting moments.

Sense of community and belonging 5 observations

People value shared experiences, identity cues, and the feeling of being less alone.

Balance between engagement and wellbeing 7 observations

Users want engagement, but they also notice fatigue, anxiety, lost time, and focus depletion.

Alignment with personal identity 4 observations

The intervention needs to match user style and self-image or it gets ignored quickly.

Progress and recognition 6 observations

Visible progress helps users sustain effort, especially when the goal is additive rather than restrictive.

The need is real.

Learning, belonging, decompression, and self-expression are legitimate goals. The problem is the delivery mechanism.

Personalization is key.

The design has to meet users' goals and mental models, which led directly to avatar matching and customization.

Visible progress sustains change.

Progress cues and reflective feedback keep users oriented toward intentional behavior without shame.

7 platforms reviewed Gamification landscape Theory + autonomy fit

Seven platforms. One clear gap.

The competitive analysis mapped gamification and engagement platforms against motivation type, behavioral theory depth, personalization, timing, and autonomy. Rewire's white space is the combination of intrinsic motivation, moment-of-behavior timing, and non-coercive design.

Platform	Model	Motivation	Timing	Autonomy	Gap surfaced
Chatter	Contribution scores	Extrinsic	Post-hoc feed	Medium	Collaboration infrastructure, not behavior change.
SharePoint	None native	None	None	Medium	No behavioral feedback or intentional engagement layer.
Packback	Curiosity score	Intrinsic	During writing	High	Strong theory fit, but locked to education contexts.
Bunchball Nitro	Points + badges	Extrinsic	Scheduled missions	Low	Reward loop boosts activity without changing attention habits.
Centrical	Adaptive narratives	Mixed	Workplace coaching	Medium	Theory depth is enterprise-focused, not consumer-facing.
Badgeville	Behavior platform	Extrinsic	Analytics-driven	Low	Gamifies platform behavior but does not preserve intrinsic motivation.
Kudos	Peer recognition	Extrinsic social	Recognition feed	High	Supports culture, not moment-of-use intervention.
Rewire	Adaptive JITAI	Intrinsic	Mid-session	High	Combines behavioral timing, personalization, and user-controlled friction.

Strategic white space

Consumer-facing behavior change grounded in theory, not enterprise productivity or education-only tooling.

Reward loops are thin

Points and badges can spike engagement, but they do not resolve the passive-scroll habit loop.

Timing is the differentiator

Rewire intervenes during active behavior, when a gentle redirect has the most leverage.

02 — Insights

The patterns were structural.

INSIGHT 01

The value disconnect is the opportunity

Users want learning and connection but default to mindless content. The gap between stated values and behavior is where design intervenes.

INSIGHT 02

Intrinsic beats extrinsic

Points and badges spike then fade. Identity-driven engagement — goals that feel personal — is what sustains change.

INSIGHT 03

Intervene at the moment of behavior

Redirect, don't block; give users control. This is JITAI in practice — and no competitor does it.

03 — Solution

A behavioral system. Not just an app.

Rewire detects prolonged scrolling, prompts through a personality-matched avatar, and redirects into one of three challenge types — all dismissible, all on the user's terms. Built on JITAI + Four Drive Theory + Community of Inquiry.

Avatar system

A short quiz maps users to a drive-aligned guide that builds identity and intrinsic motivation.

User-controlled intervention

Users set their own threshold; the prompt is always dismissible. Autonomy is the non-negotiable core.

Challenges homepage from the early design.

Challenge system

Learn (trivia), Connect (message a real person), Reflect (journaling) — mapped to Community of Inquiry.

Rewire Mode

An opt-in feed of high-quality, socially valuable content — an upgrade, not a restriction.

OwlDrive to Learn

ElephantDrive to Bond

LionDrive to Acquire

BeaverDrive to Defend

Non-coercive. Transparent. User-controlled. Nothing is blocked, every prompt is dismissible, users set their own limits — a behavioral system, not a parental control.

Design iterations

What changed. And why.

Every iteration was driven by a specific failure observed in testing, not aesthetic preference.

Microcopy: from excessive to just enough

Tried → Learned → Shipped

Round-1 microcopy that testing showed was too long to read. — Participants wanted to know what Rewire does — but the detail required too much reading.

Round-2 microcopy rewritten to be scannable. — Rewritten to be scannable and fast to digest, including challenge scoring.

Result: round-2 participants reported better understanding and ease with informational text.

Challenges: from gamification to genuine engagement

Tried → Learned → Shipped

Early challenges page with no option to create a challenge. — Users could only complete challenges, not send them — we were blocking engagement they wanted.

Final design adding the Make a Challenge ability. — Added Make a Challenge: user control over AI assumptions, tap over text.

Result: feedback shifted from "tasks" to "things I actually wanted to do" — the intrinsic framing working.

Avatar onboarding: from personality test to personal commitment

Tried → Learned → Shipped

TRIED

Participants skipped their avatars — clicking next without identifying with them.

SHIPPED

Naming the avatar at assignment turns a label into a commitment.

Result: avatar-match scores trended toward significance (p < 0.1, large effect) and users said avatars "matched" them.

"Look, this is fun. I've literally been wanting something like this."

PARTICIPANT 17 · THINK-ALOUD · ROUND 2

04 — Validate

What moved. What didn't.

Two think-aloud rounds (N=8 each) tracked three avatar questions, SUS, and VADER sentiment.

Q1 PERSONAL—not sig.

Q2 ENGAGING+0.75not sig. · med-lg effect

Q3 MATCHES ME+0.63p<0.1 · large effect

SUS+7.19not sig. · medium effect

VADER+0.092SIG ✓ p<.05 · extremely large

Boxplot showing statistically significant VADER sentiment improvement between iterations 1 and 2.

VADER — the significant one. Sentiment improved between rounds with reduced variance.

Q3 — trending toward significance after the avatar-naming iteration (p < 0.1).

The pessimistic read

The intervention isn't strong enough, or the avatar system doesn't move usability perceptions.

The methodological read

N=8 per round is too small for the medium-to-large effects the variance reduction suggests are present — four of five measures tightened round-over-round.

My honest take

The methodological read is right. The next study needs ~30 per round, longer exposure, and behavioral outcomes — not just self-report.

Expert review · Dr. Siefert (Clinical Psych) & Dr. End (Behavioral Psych)

Two experts. One direction.

Both agreed

The problem is real and worth solving
Friction matters more than persuasion — break the loop, don't lecture
The avatar system is the strongest emotional hook
Low psychological risk; harm-reduction framing is appropriate

Cautions raised

Behavior change won't be universal — strongest for users who already want help
Direct messaging needs moderation safeguards
Extrinsic rewards should scaffold, not dominate, the intrinsic core
Framing must stay ethically transparent about what prompts do

Final design

The screens.

Personalization — avatar customization

Threshold prompt — avatar-driven intervention

Challenge selection in the final design.

Challenges — Learn / Connect / Reflect

Learn — trivia + cognitive engagement

Insights page showing behavior statistics.

Insights — behavior at a glance

Reflect — self-awareness journaling

Reflection

What this project taught me.

The hardest tension: changing behavior without policing it. Every added point of friction risked feeling controlling; every reduction risked no impact. The answer was to make user autonomy the non-negotiable core and let everything else be a trade-off.

Next: position Rewire as a friction-based scaffold for users who already want help, scale to N=30+ per condition over 4–6 weeks, and replace single-session self-report with behavioral measures — Stroop, memory recall, actual screen-time data.