Can believing mental health can change help someone begin using self-help tools?
Original Chinese title: 「心理狀況可以改變」這一句，會讓人真的打開自助工具嗎？
A randomized trial of 252 employees links a brief mental health growth mindset intervention to repeat self-help engagement among those with poorer baseline well-being. This feature separates initiation, symptoms, functioning, and safe referral, and examines autonomy and workplace data governance.
王莉如
Counseling psychologist supervisor focusing on trauma recovery, couples and family relationships, professional ethics in helping professions, and social and emotional support.
Can encouragement open a door?
Having a mental health tool on a phone does not mean someone can begin using it when life becomes difficult. Some people cannot choose a feature; others worry about who might see their information. Some have come to understand their distress as something that can never change. A carefully expressed belief that mental health can improve may make a small experiment imaginable. An overconfident version can create another pressure: if change is possible, why have you not recovered? That distinction matters when reading the new research.
A randomized trial published in npj Digital Medicine on 13 October 2026 tested whether a brief growth mindset intervention could encourage renewed engagement with self-help features. The published paper and its original preprint point to a specific finding: among participants with poorer baseline well-being, those receiving the intervention were more likely than active controls to engage again. This is evidence about beginning self-help behavior. It cannot be rewritten as recovery from anxiety, depression, or trauma, or as proof that a different attitude overcomes every difficult circumstance.
The study measured repeat engagement
The original preprint describes 252 full-time employees from 83 companies, randomly assigned to the growth mindset intervention or an active control. Baseline measures covered stress, anxiety and depressive symptoms, and positive mental well-being. App behavior was followed for three months. The primary outcome was binary: whether a participant completed at least a second self-help feature. That captures one further step, but cannot incorporate frequency, quality of practice, symptom improvement, and recovery of everyday functioning in a single number.
Results depended on initial psychological circumstances. Repeat engagement was more likely among people with poorer well-being, while different baseline measures related differently to low- or moderate-intensity features. Here, intensity classified the time and effort needed to complete a feature. It was not a scale of clinical illness or a direct measurement of treatment strength. Readers should not translate these findings into a claim that greater distress requires more phone-based self-help, or that more use already demonstrates greater benefit.
The study also discloses interests related to the product company: Oliver Sündermann is Intellect's vice president and clinical director and holds equity; Sean Toh and Feodora Kosasih were clinical research associates at the company during data collection. These facts should be read alongside methods, comparisons, and results. They do not by themselves establish bias, and should not be omitted when presenting the evidence.
Separate four kinds of outcome
Digital mental health involves at least four distinct questions. Does someone begin or return? Do anxiety, depression, or stress symptoms change? Does everyday functioning improve, including sleep, work, relationships, or caring responsibilities? Can the tool recognize its service limits and help someone obtain appropriate professional support? Measurement of the first three and safety design for the fourth differ. A login cannot represent an entire pathway of care.
This distinction also changes how we understand attrition. Someone may stop because a tool is unsuitable, its language is difficult, time is scarce, or data use feels untrustworthy. Someone may also leave because they have obtained other support and no longer need that feature. Treating every departure as failure encourages products to keep people longer without asking whom that serves. Evaluation should begin with users' needs and recognize use, pauses, alternatives, and seeking help as reasonable choices.
Before-and-after change is not a specific treatment effect
A study published in 2025 assigned 731 adolescents from seven secondary schools to four single-session online interventions, including different growth mindset materials and an active supportive control. Its abstract reports declining anxiety symptoms across all four groups. When reading that participants improved after an activity, readers should therefore ask which comparison was used and whether a particular intervention added anything. Natural changes, shared activities, attention, and other support may affect differences over time. They cannot all be credited to encouragement.
A separate randomized study of a single-session digital emotion intervention in 530 adults did not find significantly better primary emotional distress outcomes than an active nature-video control, although some secondary outcomes and results for participants with higher baseline distress differed. This does not declare brief tools worthless; it means value must be measured separately. Recruitment, interventions, and outcomes differed between employees, adolescents, and general adults. The three papers cannot be combined into one success rate or used to endorse any commercial product.
Encouragement should leave room
A more supportive invitation might be: your situation may change with time, support, and suitable methods; you can try a step you are willing to take. This places hope alongside resources and acknowledges different speeds of change. Saying that belief alone will produce recovery removes social conditions, traumatic experiences, physical health, and unequal resources from view. It turns outcomes into a test of personal will. People who have not improved do not need another failure label or an obligation to disclose feelings to prove effort.
A small step might involve reading a privacy explanation, trying part of a short exercise, or taking a question to a trusted qualified professional. Understanding withdrawal, deletion, and costs often supports autonomy better than a streak counter. Pausing or adjusting an uncomfortable exercise is compatible with believing that change is possible. Seeking fuller assessment and care is also compatible with self-help; it is not evidence of having failed at it.
Counseling must consider people and daily life
Professionals integrating a tool into care can clarify its purpose with the person: recording feelings, understanding a technique, practicing an action, or supplementing support while waiting for services. Different purposes require different reviews of outcomes. Practice can supply material for a conversation, but a single questionnaire or usage log cannot replace understanding a person. Someone who never logs in may still be working to care for themselves. Someone who logs in daily may have difficulties the tool has not recognized.
Relevant considerations for professional assessment include persistent or worsening distress, substantial effects on daily functioning, and safety concerns. A qualified person must judge the situation in context. A tool should not demand more courses before allowing access to other help. Providers should explain the availability, qualifications, and responsibilities of human support so automated replies are not mistaken for continuous personal monitoring. These are practical safety recommendations, rather than effects established by the repeat-engagement trial.
Optimize meaningful use
Designers can separate renewed use, understanding, symptoms, functioning, negative experiences, and referral into different measures. If revised encouragement increases second use but heightens fear of failure, increases privacy concerns, or delays care, retention growth alone cannot establish success. Evaluation needs defined comparisons, observation periods, and ways of responding to adverse outcomes. It should identify which groups show differences. Baseline variation is a reason for further investigation, not a ready-made basis for precision marketing.
Users and counseling practice can also generate research questions. Does the language sound blaming? Can people with different literacy or language backgrounds understand it? Can someone safely return after interruption? Does the interface help rather than delay referral? Such questions move evaluation beyond clicks toward the circumstances in which action supports autonomy and need. Artificial intelligence may help organize explanations, but should not independently diagnose from conversations or use emotional vulnerability to select promotions.
Paying does not confer access to private records
The employee setting makes data governance especially important. Workplace tools may lower some access barriers while raising concerns about managers seeing emotions, diagnoses, usage times, or help-seeking records. This article proposes separating payment, service delivery, and personnel management: funding should not grant access to individual psychological content. Aggregate reporting should specify fields, purposes, retention periods, and minimum group sizes, and prevent identification through combinations of small teams, job titles, or time periods.
Consent must also be understood within actual power relationships. If employees fear that refusing will affect appraisal, a polished checkbox cannot establish free choice. Providers can offer independent questions channels, understandable refusal and withdrawal routes, and alternatives that do not require sharing individual records. Implementing these proposals in Taiwan requires attention to actual contracts, services, and legal circumstances. An international trial does not establish that local employers already follow them, or authorize companies to label psychological states as productivity attributes.
Hope needs care alongside it
The study matters because it turns an everyday question into observable behavior: when people believe their circumstances can change, are they more likely to return to a tool? Beyond the entrance, suitable content, trustworthy data handling, human support, and routes to other services remain necessary. For readers in Taiwan, the useful response is to relate claims to the outcomes actually measured and retain authority over judgments about their own lives.
Future research tracking symptoms, functioning, safe referral, and varied experiences could more fully explain connections between belief, action, and care. Until then, a small beginning can be recognized alongside the questions it leaves unanswered. Mental health can change without making a person solely responsible for every change. A tool that can support hope should make support accessible and preserve the option to take another suitable path.
Sources
Role-Driven Inquiry
Yuan Media AI | Guided Questions by Role
Select a role first, then review the complete questions from that perspective; clicking a question will display Yuan Media AI’s response in real time.
Select a role
Tool users
Examine the article's evidence, governance boundaries, data rights, and actionable steps from the perspective of a Tool users.
Choose a question
AI use and content-safety disclosure
AI assisted the preparation of this article and role questions using public research and institutional materials. Research findings and practical recommendations are distinguished. The cover is an AI-generated concept image.