The science
Built on evidence, not willpower.
Rewarding Habits is designed around decades of behavioral research. No fluff, no pop-science promises — just the mechanisms the app is built on, and the work behind them.
How to read this
Designed around research — not a replacement for care.
We describe the mechanisms the app is built on and cite the work behind them. Rewarding Habits is a skill-building app informed by psychological science. It isn't therapy, and it isn't a crisis service. If you're struggling with your mental health, please reach out to a professional. In a crisis in the US, call or text the 988 Suicide & Crisis Lifeline.
The foundations
Five ideas that shape everything.
Most of what the app does traces back to these. Each is decades deep and widely studied.
1 · Variable-ratio reinforcement
Rewards delivered on an unpredictable schedule produce some of the most durable, motivated behavior known to behavioral science. It's why the app gives random rewards, and why it refuses to count streaks.
Skinner; Schultz (1997); Milkman et al. (2021), N=61,293.
2 · Acceptance & Commitment Therapy (ACT)
Six processes — acceptance, defusion, present-moment awareness, self-as-context, values, and committed action — that help you act on what matters even when it's uncomfortable. The backbone of Surf the Urge and the coach.
Hayes; Harris; A-Tjak et al. (2015), 39 RCTs.
3 · Compassion-Focused Therapy (CFT)
A three-system model of emotion (threat, drive, soothing). After a setback, the app activates soothing first — which is why you'll never find guilt or shame here.
Gilbert; Kirby et al. (2017), meta-analysis.
4 · Self-Determination Theory (SDT)
Lasting motivation comes from autonomy, competence, and relatedness. The app's values & identity system anchors every habit to what you actually care about.
Deci & Ryan (50+ years of research).
5 · Identity-based change
People sustain behavior that matches their self-image. Every check-in is a small vote for the person you're becoming — and difficulty is reframed as importance, not impossibility.
Oyserman et al., N=2,746 validation.
The thread that ties them together
Build people up, never tear them down. Reward effort unpredictably, anchor it to identity and values, and meet every stumble with compassion instead of a broken streak.
No magic. Just many small things that work.
Behavior change is modest and gradual. The app's edge isn't one big trick — it's stacking dozens of well-supported nudges until following through gets easier than not.
The library · building habits
How habits actually form.
It takes ~66 days, not 21
New behaviors automate over weeks to months, on average about 66 days, with a wide range. The "21-day" rule is a myth.
Lally et al. (2010).
Implementation intentions
"When X happens, I will do Y" links a habit to a concrete cue and measurably improves follow-through.
Gollwitzer; meta-analysis of 94 tests, d≈0.65.
Coping plans for the obstacle
"If the obstacle comes, I'll cope by Z." Action plans help you start; coping plans help you keep going.
Sniehotta et al.
Start tiny
Shrinking a habit until it's almost too small to fail lowers the activation energy and builds momentum.
Fogg, Behavior Design (B=MAP).
Environment over willpower
Context cues predict behavior better than intention. Design the cue, and the habit gets easier.
Wood; Duhigg (cue, routine, reward).
Temptation bundling
Pairing a habit with something you enjoy ("only listen to this podcast while walking") boosts follow-through.
Milkman et al. (2014).
The library · motivation
Why random rewards, and never streaks.
The dopamine of "maybe"
The brain responds more to unexpected rewards than predictable ones, the prediction-error signal. Unpredictability is the point.
Schultz (1997), dopamine prediction error.
Streaks invite quitting
"Progress" framing quietly licenses you to slip ("I've earned a day off"). Commitment framing — this is who I am — keeps you consistent.
Khan & Dhar; Blanken et al., meta d≈0.31.
Variety keeps rewards fresh
Good feelings fade through adaptation. Varying the kind of reward, not just the timing, keeps motivation alive.
Sheldon & Lyubomirsky, HAP model.
Intrinsic beats extrinsic
Autonomy, mastery, and purpose sustain effort long after external rewards stop working, so the app pairs surprises with meaning.
Deci & Ryan; Pink.
The library · quitting & urges
Riding out the urge.
Urge surfing
Cravings crest and pass like waves, usually within minutes. You can notice an urge and let it move through without acting on it.
Marlatt; Bowen (2009), MBRP.
Acceptance beats suppression
Fighting a craving can make it stronger (the rebound effect). Allowing it, with curiosity, works better. The goal isn't to erase urges; it's to change how you respond.
Forman et al. (2007).
A lapse is not a relapse
One slip doesn't undo your progress. Treating it as total failure is what turns a lapse into a relapse, so the app uses "lapse," never "failure."
Marlatt & Gordon; Bouton (2000).
Replace, don't just resist
Quitting works better when paired with a competing replacement behavior than with sheer avoidance.
Habit Reversal Training; Bate et al. (2011), d≈0.80.
Picture the future you're protecting
Vividly imagining a meaningful future event reduces the pull of the immediate craving.
Episodic Future Thinking; Ye (2022), Colton (2024).
Name what you feel, precisely
Putting a feeling into specific words engages the prefrontal cortex and calms the threat response: "disappointed," not just "bad."
Lieberman et al. (2007), affect labeling.
Move toward, not just away
"Quit scrolling" is an avoidance goal; pairing it with an approach goal ("read for ten minutes") measurably improves success a year out.
Oscarsson et al. (2020), N=1,066, 1-year follow-up.
The library · self-compassion
What happens after a miss.
Self-compassion, three ways
Self-kindness, common humanity ("others struggle too"), and mindfulness — together they support motivation after a setback better than self-criticism.
Neff; Kirby et al. (2017).
Soothe first, then rebuild
State self-compassion in the moment predicts getting back to it, by restoring your belief that you can handle the next obstacle.
Zhang et al. (2025), barrier self-efficacy.
Unhook from the thought (defusion)
"I'm having the thought that I always fail" creates distance from a thought so it stops running the show.
ACT cognitive defusion; Hayes.
Praise the effort, not the trait
The coach celebrates what you did ("you showed up"), never a fixed label ("you're disciplined"), which protects motivation when things get hard.
Dweck, process praise.
Advise yourself like a friend
We reason more wisely about our own problems when we step back and view them from a distance, Solomon's Paradox. The app's reflections nudge that shift.
Grossmann & Kross; self-distancing studies.
Soothe, drive, and threat
Compassion-Focused Therapy describes three emotion systems. Self-criticism keeps the threat system firing; self-kindness activates soothing, so you can re-engage instead of shutting down.
Gilbert, three-system model.
The coach
An AI coach that stays on your phone.
The coaching is built from the science above, and it runs entirely on your device. No coaching conversation is ever sent to a server.
On-device intelligence
Personalized coaching runs locally with on-device AI, falling back to carefully written, reviewed guidance when needed, always private.
A guilt-and-shame filter
Every message is screened before you see it. Anything resembling guilt, shame, or streak-loss language is blocked, by design.
The full ACT toolkit
Defusion, acceptance, grounding (SOBER), values work, and committed action — matched to what you're feeling in the moment.
Care for hard moments
If a conversation signals crisis, the coach points to real help, in the US, the 988 Suicide & Crisis Lifeline. It's support, never a substitute for it.
How we keep it honest
Science you can trust.
No overclaiming
We describe mechanisms grounded in research. We don't promise to cure anything or guarantee a result.
We cite the work
Every claim here traces to published research. Where the popular version overstates the effect, we use the honest number.
Effects are real, not magic
Behavior change is modest and gradual. The app stacks many small, well-supported nudges rather than betting on one big trick.
Private by design
All of this happens on your device. Your habits, reflections, and coaching never leave your phone.
Sources & further reading
The work behind the words.
A selection of the research the app draws on. These are starting points, not an exhaustive list, the app is informed by dozens more.
- Schultz, W. (1997). A neural substrate of prediction and reward. Science. Dopamine and unpredictable rewards.
- Milkman, K. et al. (2021). Megastudy of habit-formation interventions (N=61,293). Nature.
- Blanken, I. et al. Meta-analysis of moral licensing: commitment vs. progress framing.
- A-Tjak, J. et al. (2015). Meta-analysis of Acceptance & Commitment Therapy (39 RCTs).
- Gilbert, P. Compassion-Focused Therapy and the three-system model of affect.
- Neff, K.; Kirby, J. et al. (2017). Self-compassion and its three components; meta-analysis.
- Deci, E. & Ryan, R. Self-Determination Theory: autonomy, competence, relatedness.
- Oyserman, D. et al. Identity-Based Motivation (validation N=2,746).
- Gollwitzer, P.; Sniehotta, F. et al. Implementation & coping planning (94 tests, d≈0.65).
- Lally, P. et al. (2010). How habits form: a median of ~66 days. Eur. J. Soc. Psych.
- Wood, W.; Duhigg, C. Context cues and the habit loop.
- Milkman, K. et al. (2014). Temptation bundling.
- Marlatt, G. & Gordon, J.; Bouton, M. (2000). Relapse prevention; lapse vs. relapse.
- Bowen, S. et al. (2009). Mindfulness-Based Relapse Prevention; urge surfing.
- Forman, E. et al. (2007). Acceptance vs. suppression of cravings.
- Bate, K. et al. (2011). Habit Reversal Training meta-analysis (d≈0.80).
- Ye, J. et al. (2022); Colton, K. et al. (2024). Episodic Future Thinking.
- Oscarsson, M. et al. (2020). Approach vs. avoidance goals (N=1,066, 1 year).
- Lieberman, M. et al. (2007). Affect labeling and the prefrontal cortex.
- Grossmann, I. & Kross, E. Solomon's Paradox and self-distancing.
- Zhang, J. et al. (2025). State self-compassion and barrier self-efficacy.
- Dweck, C. Mindset and process praise.
Rewarding Habits is a skill-building app informed by this research. It is not therapy or a crisis service, and nothing here is a clinical claim or medical advice.
Put the science to work.
Build the good, quit the bad, and let unpredictable rewards keep you going.