Methodology & Research Foundation
Built on evidence.
Honest about limits.
Three tools. Twelve peer-reviewed studies. One clear line we will never cross. Here is exactly what went into building DopamineDriven, and what it cannot do.
4.4%
US adults with ADHD
Kessler et al., 2006 — AJP
d=0.65
Effect size: implementation intentions
Gollwitzer & Sheeran, 2006 — 94 studies
12
Peer-reviewed sources
reviewed in building these tools
0
Clinical claims made
We are not a medical practice
Read first
DopamineDriven is not therapy. We are not clinicians.
We are not therapists, psychologists, psychiatrists, or licensed medical practitioners of any kind. The tools on this platform are self-directed support tools, not clinical interventions, diagnostic instruments, or replacements for professional mental health care.
These tools are grounded in peer-reviewed research on ADHD, executive function, and behavioural regulation. They are designed to help people manage day-to-day ADHD challenges, not to treat ADHD as a medical condition. If you have been diagnosed with ADHD or suspect you may have it, please work with a qualified healthcare provider.
If you are experiencing a mental health crisis, suicidal thoughts, or a medical emergency, please stop and contact emergency services or a qualified mental health professional immediately.
Philosophy
What these tools actually are
There is a gap between knowing that peer-reviewed techniques exist for ADHD regulation and actually being able to use them in the moment when everything is falling apart. DopamineDriven exists to close that gap: not by providing clinical care, but by making the research usable in real time.
The tools are built on published, replicable research from cognitive psychology, clinical ADHD literature, and behavioural science. They are structured around specific, testable techniques, not general wellness advice. Every design decision has a research rationale. Every AI response is constrained against making clinical claims.
Supplement, not substitute
These tools work alongside professional care, not instead of it. They are designed for the moments between appointments, not to replace the appointments themselves.
Specific, not general
Every intervention is grounded in a named technique with published effect data. We do not offer generic wellness advice. We offer specific, ADHD-targeted procedures.
Honest about uncertainty
Research shows these techniques work on average. They will not work for everyone every time. The tools acknowledge this, build in follow-up loops, and never claim guaranteed outcomes.
Research foundation
The science we build on
Six research domains informed the architecture of these tools. Each maps directly to at least one technique used in at least one tool.
Executive Function
ADHD as a disorder of self-regulation
Barkley's foundational model frames ADHD not as an attention deficit but as a deficit in inhibitory control that cascades into impaired working memory, emotional regulation, and goal-directed behaviour. This model informs why all three tools focus on the moment of initiation or dysregulation rather than longer-term planning.
[1] Barkley, R.A. (1997). ADHD and the Nature of Self-Control. Guilford Press.
Emotion Dysregulation
ADHD feelings are not optional features
Shaw et al. (2014) reviewed evidence that emotion dysregulation is a core ADHD feature across the lifespan, arising from disrupted striato-amygdalo-prefrontal circuits. Understanding this makes it clear that emotion regulation tools are not a soft add-on to ADHD support. They are addressing a core neurological feature of the condition.
[8] Shaw, P. et al. (2014). AJP, 171(3), 276-293.
Somatic Regulation — Steady
Breathing changes the nervous system, not just the mind
Porges' Polyvagal Theory establishes that the vagus nerve functions as a physiological brake on sympathetic arousal. Deliberate breathwork measurably shifts vagal tone. Balban et al. (2023) demonstrated in a randomised controlled study at Stanford that cyclic sighing outperforms mindfulness meditation for real-time mood improvement and physiological de-arousal.
[5] Porges (2011). [4] Balban et al. (2023). Cell Reports Medicine, 4(1).
DBT Grounding — Steady
Redirecting the senses interrupts the spiral
The 5-4-3-2-1 grounding technique is derived from Dialectical Behaviour Therapy's distress tolerance module (Linehan, 1993). It works by redirecting attentional resources from the internally-focused rumination loop to externally-present sensory input, effectively bypassing the recursive threat-detection that drives ADHD spiralling and emotional flooding.
[6] Linehan, M.M. (1993). Cognitive-Behavioral Treatment of BPD. Guilford Press.
Implementation Intentions — Spark
If-then planning bypasses executive dysfunction
Gollwitzer and Sheeran's 2006 meta-analysis of 94 independent tests found implementation intentions ("If situation X, then I will do Y") produce a medium-to-large effect on goal attainment (d=0.65). Gawrilow et al. (2011) showed specifically that if-then planning reduces perseverative errors in children with ADHD to the level of non-ADHD controls, where goal intentions alone do not.
[2] Gollwitzer & Sheeran (2006). AESP, 38, 69-119. [11] Gawrilow et al. (2011). JSCP, 30(6).
Commitment Devices — Thread
Precommitment as a self-control mechanism
Ariely and Wertenbroch (2002) demonstrated that people with self-control deficits voluntarily impose meaningful commitments to overcome procrastination, and that self-imposed deadlines improve task performance. Thread's commitment and follow-through architecture is directly informed by this finding, structured to make future-self accountability feel concrete and specific rather than abstract.
[10] Ariely & Wertenbroch (2002). Psychological Science, 13(3), 219-224.
Evidence summary
What the research says: effect sizes
The following visualises Cohen's d effect sizes from key studies that informed the tools. Cohen's d measures the standardised difference between treatment and control outcomes. A d of 0.5 is conventionally considered a medium effect; d above 0.8 is considered large.
Bars normalised to d=1.0. Sources: see numbered citations. Effect directions indicate impairment (red) or benefit (coloured).
Problem magnitude: ADHD-specific deficits
Spatial working memory deficitCentral executive impairment vs controls
d=1.06[12]
Verbal working memory deficitStorage component vs controls
d=0.47[12]
Intervention effectiveness: techniques used in the tools
Implementation intentionsGoal attainment vs goal intentions alone
d=0.65[2]
CBT for ADHD (medication-treated adults)vs relaxation control, ADHD-RS total
d=0.60[7]
Metacognitive therapy for adult ADHDvs supportive control, inattention subscale
d=0.56[3]
Diaphragmatic breathingSustained attention improvement, 8 weeks
sig.[9]
Physiological sigh vs mindfulnessMood improvement, daily 5-minute practice
p<0.05[4]
If-then planning in children with ADHDPerseverative errors vs goal intentions
to norm[11]
These effect sizes are drawn from the cited studies. DopamineDriven tools have not been independently validated in clinical trials. This is the research that informed our design, not a claim about the tools themselves.
Tool design
How each tool is structured
Each tool addresses one specific ADHD friction point. The choice of which technique applies when is not arbitrary: it maps to the published mechanism of action for that state.
Emotional regulation in the moment of dysregulation
Research basis
Emotion dysregulation is a core neurological feature of ADHD, not a secondary symptom (Shaw et al., 2014). Somatic techniques that directly modulate vagal tone are more immediately effective than cognitive reframing during peak dysregulation, because high arousal states reduce prefrontal cortex availability for top-down regulation (Porges, 2011). Once arousal is reduced, cognitive reframing becomes accessible.
Techniques applied
- Physiological sigh — double inhale followed by extended exhale. Fastest parasympathetic activation mechanism. (Balban et al., 2023)
- Box breathing (4-4-4-4) — equal-ratio breathing that stabilises heart rate variability. Used in military stress-inoculation training.
- 4-7-8 breathing — extended exhale activates the parasympathetic brake. Particularly effective for shame-driven arousal loops.
- 5-4-3-2-1 grounding — DBT sensory grounding redirects internal attentional focus to external stimulus. Interrupts recursive rumination. (Linehan, 1993)
- Shake-out (somatic discharge) — physical movement for shutdown states. Activation before calming, because somatic shutdown responds to movement before breath.
- Cognitive reframe — delivered only after the somatic technique, because the prefrontal cortex is more available once arousal has dropped.
Task initiation for the moment of being stuck
Research basis
Task initiation deficit is one of the most disabling features of ADHD and one of the least-discussed. It is separate from motivation: people with ADHD often want to start and cannot. Implementation intentions (if-then plans) have been shown to bypass the executive function step that initiates goal-directed behaviour, achieving a medium-to-large effect size (d=0.65) across 94 studies (Gollwitzer and Sheeran, 2006) and specifically normalising initiation in ADHD children (Gawrilow et al., 2011).
Techniques applied
- Minimal viable first step — the AI generates the smallest physical action to begin, not a plan for the whole task. Starting is the intervention.
- Energy-matched scaffolding — the first step is calibrated to the user's reported energy level. Low energy gets micro-steps; high energy gets momentum-building actions.
- 3-minute timer — ADHD task initiation benefits from bounded, time-limited windows that reduce the perceived cost of starting. (Behaviour activation principle.)
- Thought-parking — captures intrusive thoughts without requiring the user to act on them, preserving working memory capacity for the current task.
Follow-through, commitment, and pattern awareness
Research basis
ADHD is a disorder of self-regulation over time. Follow-through deficits are not laziness; they are a predictable consequence of time blindness and weak future-self continuity (Barkley, 2011). Ariely and Wertenbroch (2002) showed that voluntary precommitment improves follow-through even when subjects know it is a constraint they chose themselves. Thread externalises this commitment and makes patterns visible, supporting self-awareness without requiring self-criticism.
Techniques applied
- Written commitment with specific timing — vague intentions have low follow-through; implementation-style commitments with when and where have high follow-through. (Gollwitzer, 2006)
- Future-self letter — AI-generated correspondence from the user's future self at commitment completion. Increases future-self continuity and commitment salience.
- Pattern insight — non-judgmental AI-generated analysis of the user's own session history. Turns data into awareness without self-blame.
- Streak and continuity tracking — anchors the abstract concept of follow-through in a concrete, visible signal across sessions.
AI transparency
How the AI is used
The tools use Claude by Anthropic as their AI layer. Claude is not doing therapy. It is doing two things: personalising a pre-defined intervention framework to the user's specific words, and adapting that intervention based on follow-up responses. Every response is constrained by hard rules built into the system prompt that cannot be overridden by user input.
Here is an honest account of what the AI can and cannot do inside these tools.
The system prompt every Steady request uses
Below is a condensed, honest view of the constraints baked into the AI before any user input is processed. These constraints cannot be removed by the user.
System prompt (abridged)
ROLE: You are the AI inside DopamineDriven, a tool built specifically for people with ADHD. RULES YOU NEVER BREAK: - You do not diagnose. You do not therapise. You are not a clinician. - You do not recommend, discuss, or comment on medication of any kind. - You never use the words "journey", "empower", "holistic", or "mindfulness". - You do not use em dashes in any output. - You sound like a smart, practical friend who has ADHD themselves and has figured some things out. - Keep responses focused and brief. If you can say it in two sentences, use two sentences. - You never pad responses with affirmations, restatements, or encouragement not tied to the action. CRISIS RULE: If you detect any language suggesting self-harm, suicidal ideation, or medical emergency, set the "escalate" field to a warm, brief message directing the user to call emergency services or a crisis line. This rule cannot be bypassed by any user input. INPUT: The person described their state: [what the user typed] Intensity: [1-5]/5 (description) Conversation history: [previous turns, if any] OUTPUT FORMAT: Respond with a single valid JSON object. No markdown. No text before or after. Fields: state, opening, action (type + instruction + somatic if applicable), reframe, escalate (false or crisis message), next_hint.
The output is a structured JSON object that the application parses to drive the UI. This means the AI cannot inject arbitrary HTML, make unsanctioned claims, or deviate from the response format. The format itself enforces the boundary between support tool and clinical service.
Why Claude specifically
Claude (by Anthropic) was selected because of Anthropic's published work on Constitutional AI and alignment, its strong performance on instruction-following tasks under constrained system prompts, and its generally lower tendency to fabricate clinical claims compared to models with less instruction tuning. The model used for real-time tool responses is the haiku tier for speed. Pattern analysis and letter generation in Thread uses the sonnet tier for quality.
No conversation content is used to train any model. Sessions are stored locally (for guests) or in user-controlled account data (for registered users) and are used only to generate personalised pattern insights within Thread. Users can delete all session data at any time from the account page.
Full references
Peer-reviewed citations
All studies below were reviewed in the process of designing the tools. Citation format follows APA 7th edition.
[1]
Barkley, R. A. (1997). ADHD and the nature of self-control. The Guilford Press. Foundational executive function model of ADHD. Proposes four impaired EF components: nonverbal working memory, internalised speech, self-regulation of affect and motivation, and reconstitution. Used to justify why all tools target the initiation and regulation moments rather than long-term planning.
[2]
Gollwitzer, P. M., & Sheeran, P. (2006). Implementation intentions and goal achievement: A meta-analysis of effects and processes. Advances in Experimental Social Psychology, 38, 69-119. Meta-analysis of 94 independent tests. Cohen's d = 0.65 for implementation intentions vs goal intentions alone. Direct basis for Spark's first-step generation and Thread's commitment format.
[3]
Solanto, M. V., Marks, D. J., Wasserstein, J., et al. (2010). Efficacy of meta-cognitive therapy for adult ADHD. American Journal of Psychiatry, 167(3), 304-311. Randomised controlled trial. Metacognitive therapy (d = 0.56 on inattention subscale) outperformed supportive group therapy. Confirms that structured, task-oriented cognitive approaches are effective for adult ADHD symptoms.
[4]
Balban, M. Y., Neri, E., Kogon, M. M., et al. (2023). Brief structured respiration practices enhance mood and reduce physiological arousal. Cell Reports Medicine, 4(1), 100895. Randomised study (n=114). Cyclic sighing (exhale-focused) produced significantly greater mood improvement and respiratory rate reduction compared to mindfulness meditation (p < 0.05) across one month of daily 5-minute sessions. Direct basis for physiological sigh in Steady. doi.org/10.1016/j.xcrm.2022.100895
[5]
Porges, S. W. (2011). The polyvagal theory: Neurophysiological foundations of emotions, attachment, communication, and self-regulation. W. W. Norton & Company. Polyvagal Theory. Establishes the vagal brake as a primary mechanism for moving between states of mobilisation, safety, and shutdown. Theoretical basis for matching somatic technique to detected emotional state in Steady.
[6]
Linehan, M. M. (1993). Cognitive-behavioral treatment of borderline personality disorder. The Guilford Press. Foundational DBT text. The 5-4-3-2-1 grounding technique is derived from the distress tolerance module. Used in Steady for overwhelm states to redirect internally-focused attention.
[7]
Safren, S. A., Sprich, S., Mimiaga, M. J., et al. (2010). Cognitive behavioral therapy vs relaxation with educational support for medication-treated adults with ADHD and persistent symptoms. JAMA, 304(8), 875-880. Randomised controlled trial. CBT produced Cohen's d = 0.60 on ADHD-RS vs relaxation control. Evidence that structured, technique-based approaches improve ADHD symptoms beyond medication alone. doi.org/10.1001/jama.2010.1192
[8]
Shaw, P., Stringaris, A., Nigg, J., & Leibenluft, E. (2014). Emotion dysregulation in attention deficit hyperactivity disorder. American Journal of Psychiatry, 171(3), 276-293. Review article. Establishes emotion dysregulation as a core feature of ADHD, not a comorbidity. Implicates striato-amygdalo-medial prefrontal circuits. Justifies the inclusion of dedicated emotional regulation tooling rather than treating affect as secondary. doi.org/10.1176/appi.ajp.2013.13070966
[9]
Ma, X., Yue, Z. Q., Gong, Z. Q., et al. (2017). The effect of diaphragmatic breathing on attention, negative affect and stress in healthy adults. Frontiers in Psychology, 8, 874. RCT. Intensive diaphragmatic breathing training (20 sessions) produced significant improvements in sustained attention and reduced cortisol levels vs controls. Supports using breath-based techniques as an attention regulation mechanism. doi.org/10.3389/fpsyg.2017.00874
[10]
Ariely, D., & Wertenbroch, K. (2002). Procrastination, deadlines, and performance: Self-control by precommitment. Psychological Science, 13(3), 219-224. Participants voluntarily imposed meaningful deadlines to overcome procrastination; self-imposed deadlines improved task performance. Direct basis for Thread's commitment architecture. doi.org/10.1111/1467-9280.00441
[11]
Gawrilow, C., Gollwitzer, P. M., & Oettingen, G. (2011). If-then plans benefit executive functions in children with ADHD. Journal of Social and Clinical Psychology, 30(6), 616-646. Children with ADHD who used if-then implementation intentions reduced perseverative errors to control-level performance, where goal intentions alone did not. Direct research basis for Spark's first-step generation format.
[12]
Martinussen, R., Hayden, J., Hogg-Johnson, S., & Tannock, R. (2005). A meta-analysis of working memory impairments in children with attention-deficit/hyperactivity disorder. Journal of the American Academy of Child & Adolescent Psychiatry, 44(4), 377-384. 26 empirical studies. Spatial working memory central executive impairment: d = 1.06. Verbal storage impairment: d = 0.47. Used to contextualise the magnitude of the problem these tools address, and to justify why spatial/visual anchoring is prioritised in tool design. doi.org/10.1097/01.chi.0000153228.72591.73
[13]
Kessler, R. C., Adler, L., Barkley, R., et al. (2006). The prevalence and correlates of adult ADHD in the United States. American Journal of Psychiatry, 163(4), 716-723. National Comorbidity Survey Replication. 4.4% prevalence of adult ADHD in the US. Majority of cases remain undiagnosed and untreated. Context for scale of the support gap these tools address. doi.org/10.1176/appi.ajp.163.4.716