A research publication · not a clinic · not a campaign
Investigating social media, psychiatric self-diagnosis, identity formation and the changing expression of mental health in the digital age.
Social media has transformed how young people encounter information about mental health. But does increased awareness simply help people recognize experiences that were always there—or can digital environments themselves influence how symptoms are interpreted, expressed and incorporated into identity?
The Question
This site investigates a possible pathway without assuming it. Every arrow is a separate research question. The work resists two simplistic narratives: that social media is creating mentally ill teenagers, and that social media merely allows previously invisible disorders to be recognized. Both are hypotheses. Both require evidence.
A more interesting possibility is that digital environments may simultaneously reveal existing distress, provide valuable support, reshape how psychological experiences are interpreted, create new forms of social reinforcement, and—in particular circumstances—contribute to changes in symptom expression. The task is to determine which of these effects are supported, under what circumstances, for whom, and how strongly.
What we know
Average associations between time on social media and adolescent well-being are small and heterogeneous. Online peer support can help. Some specialist clinics documented a distinct wave of functional tic-like behaviours during 2020–2022. High-reach ADHD TikTok content is often clinically inaccurate.
How we know it
Umbrella reviews and specification-curve studies for time-use; clinical series for FTLB; content analyses for diagnostic videos; help-seeking and stigma reviews for benefits and barriers.
What else could explain it
Selection and reverse causation; destigmatization and awareness; access barriers; pandemic stress; diagnostic-practice change; concept creep. Multiple hypotheses may operate at once.
What we don't know yet
Whether diagnostic-content exposure increases later self-diagnosis; whether recommenders matter more than time; how accurate self-diagnosis is by condition; whether reinforcement changes symptoms or only posts.
What Changed?
Diagnostic language used to arrive mainly through professionals, families, and schools. It now also arrives through recommended video, comments, and identity communities.
The mature well-being literature mostly measured time. The questions here are about ranking, repetition, models, labels, and belonging.
Rising identification can mean better recognition, more distress, looser concepts, social learning, functional symptoms—or several of these together.
The Algorithmic Environment
Engagement-based ranking can loop exposure. That is an engineering fact, not a psychological conclusion. This project asks whether that loop changes perceived relevance to the self. SeeAlgorithms.
Recognition vs Self-Diagnosis
Recognition can be a first step toward care. Self-diagnosis can be a rational response to inaccessible assessment. It can also be a premature conclusion shaped by misleading content. Accuracy is not a single number. It likely differs by condition, age, severity, source, and diagnostic complexity. See Self-diagnosis.
Mental Health & Identity
A diagnostic label can organize self-compassion and community. It can also become a totalizing story. Having an illness-related identity is not, by itself, pathology. The research question is when labels help people live and when they narrow a self still under construction. SeeIdentity.
Social Reinforcement
Likes, views, belonging, and creator incentives can reinforce posting. That is not the same as reinforcing symptoms, and it is not a licence to call people performative. We keep posting behaviour, interpretation, expression, and severity on separate lines. SeeReinforcement.
Psychiatric Contagion
Emotional contagion, behavioural modelling, mass sociogenic illness, and “mass social-media-induced illness” are different objects. Evidence from functional tic-like behaviours does not automatically travel to ADHD, autism, or DID. See Social contagion.
Functional Symptoms
Functional neurological symptoms are experienced as real. They are not malingering. The FTLB literature is where socially mediated presentation is currently best documented—and still not causally settled. See the functional tic case study.
Benefits of Online Communities
These benefits are not a rebuttal of risk. They are part of the same map. A community can destigmatize and still reinforce a narrow illness identity. Evidence must say which, when.
What Evidence Actually Shows
What We Don't Know
The most important claims in public debate—content exposure causes self-diagnosis; algorithms mint identities; communities manufacture disorders—are exactly the claims with the thinnest direct tests. Research gaps lists them as questions, not insinuations.
The core interactive model
Path A is care. Path B is support after self-labelling. Path C is interpretative capture. Path D is a reinforcement loop. Path E is functional presentation. Evidence status is attached to each arrow, not to a mood.
Shared trunk
Mental-health content
Videos, posts, infographics, creator narratives, and peer testimony about symptoms, diagnoses, and recovery.
Shared trunk
Algorithmic recommendation
Ranking systems that predict what will hold attention and therefore what is shown next.
Shared trunk
Repeated exposure
The same themes, labels, and symptom models returning across sessions, not a single encounter.
Path A
Content helps people recognize genuine symptoms, improve literacy, seek assessment, and obtain support.
Mental-health content
Algorithmic recommendation
Repeated exposure
Symptom recognition
Mental-health literacy
Professional assessment
Treatment / support
end of path A
Path B
Recognition leads to self-diagnosis, peer community, reduced stigma, and improved coping—even without, or before, clinical diagnosis.
Symptom recognition
Self-diagnosis
Community support
Reduced stigma
Improved coping
end of path B
Path C
Repeated exposure heightens symptom attention, recodes ambiguous experience, and integrates a label into identity.
Repeated exposure
Heightened symptom attention
Reinterpretation
Self-labelling
Identity integration
end of path C
Path D
Self-labelling leads to community membership, social reinforcement, and possible increases in symptom signalling, which may loop.
Self-labelling
Community membership
Social reinforcement
Symptom signalling / expression
Possible reinforcement loop
end of path D
Path E
Exposure contributes, in susceptible people and particular conditions, to functional or socially mediated symptoms that then present clinically.
Repeated exposure
Functional / socially mediated symptoms
Clinical presentation
end of path E
Solid arrows: enough evidence to discuss as a live empirical pathway. Dashed arrows: hypothesized. Click any arrow.
Conceptual distinctions
A diagnostic conclusion following appropriate professional assessment, using history, context, differential diagnosis, and—where relevant—standardized measures. It is a clinical judgement, not a revelation of a hidden essence.
Not the same as a self-label, a viral symptom list, or a positive online quiz.
A person concluding that they have a particular condition without formal diagnosis. It can be careful or hasty, accurate or inaccurate, provisional or identity-defining.
Not the same as symptom recognition, and not the same as lying.
Recognizing experiences in oneself—restlessness, low mood, tics, sensory overload—without necessarily claiming a disorder.
Not automatically self-diagnosis or illness identity.
The degree to which a condition or perceived condition is integrated into self-concept: “a thing I have” versus “what I am.” Can organize recovery or organize impairment.
Not inherently pathological, and not proof that symptoms are unreal.
Attention, belonging, validation, status, or other social consequences associated with particular behaviours, narratives, or identities.
Not a moral accusation of “attention seeking.” Reinforcement can also support disclosure and care.
A family of claims that behaviours, symptoms, affects, or interpretations can cluster or spread through social networks. The term is often used more loosely than the evidence.
Not a single mechanism. Emotional contagion, behavioural modelling, and mass sociogenic illness are different objects.
Symptoms experienced as genuine that arise through mechanisms different from the neurological or medical disorder they may resemble. Diagnosed by rule-in clinical features, not by proving someone is faking.
Not malingering. Not “all in the mind” as a dismissal. Not the same as self-diagnosis.
Intentional production or exaggeration of symptoms for external incentives. A distinct clinical and forensic judgement that requires its own evidence.
Must not be inferred from self-diagnosis, atypical presentation, functional symptoms, or social-media use.
Research Timeline
1968 · RESEARCH
Erikson’s work on adolescent identity (and later Marcia’s statuses) establishes why labels and belonging are developmentally hot in this age window—long before platforms.
1993 · RESEARCH
Hatfield, Cacioppo and Rapson popularise emotional contagion as interpersonal affect transfer. Later writers sometimes stretch this term to cover psychiatric identification. They are not the same claim.
2002 · RESEARCH
Bartholomew and Wessely review the protean, historically shifting presentations of mass sociogenic illness: symptoms follow available cultural models.
2004 · RESEARCH
Piper and Merskey’s critical review is a reminder: debates about iatrogenesis and role enactment in DID predate TikTok by decades.
2009 · RESEARCH
White and Horvitz document how web search can escalate medical concern. An analogue for later recommender loops—about search, not short-form video.
2010 · RESEARCH
Gulliver, Griffiths and Christensen summarise stigma, literacy, and self-reliance as barriers. The access and destigmatization hypotheses have a pre-platform evidence base.
2012 · RESEARCH
Dalenberg et al. and Lynn et al. publish competing syntheses on dissociation. The scientific split remains live.
2016–2018 · PLATFORM
Musical.ly / TikTok scale rapidly. Recommendation, imitation, and sound-based templates become ordinary adolescent media, not a niche.
2016 · RESEARCH
Haslam describes the expansion of harm and pathology concepts. A non-platform mechanism that can still change what people call “trauma” or “disorder”.
2016 · RESEARCH
Holland and Tiggemann synthesise social-network associations with body image and disordered eating—an earlier, content-specific literature.
2018 · EPIDEMIOLOGY
Twenge and colleagues argue that post-2010 rises in adolescent depressive symptoms track smartphone adoption. The paper becomes a pole in a still-unresolved debate.
2018 · RESEARCH
Nesi, Choukas-Bradley and Prinstein argue that social media transforms peer relations along specific dimensions (quantifiability, publicness, permanence).
2018 · CLINICAL
Espay et al. summarise contemporary FND: genuine symptoms, not a diagnosis of last resort or of faking.
2019 · RESEARCH
Orben and Przybylski show that technology–well-being associations are small and analytic-choice dependent. A methodological reset for crude screen-time claims.
2020 · PANDEMIC
Isolation, school closure, distress, disrupted care, and increased digital communication arrive together. Any later clinical wave has to survive this confounder.
2021 · CLINICAL
Movement-disorder and paediatric centres describe explosive, atypical tic-like presentations, often in adolescent females, during the pandemic.
2022 · CLINICAL
Müller-Vahl and colleagues argue for a sociogenic, socially mediated presentation distinct from Tourette syndrome. The claim is influential and contested in emphasis, not in the existence of atypical cases.
2022 · RESEARCH
Yeung, Ng and Abi-Jaoude find that popular ADHD TikTok videos are frequently misleading relative to guidelines.
2023 · RESEARCH
Foulkes and Andrews argue that awareness efforts may increase reporting and some genuine problems via interpretative loops—not only uncover hidden illness.
2026 · CULTURE
Signal & Self begins as a static, evidence-mapped investigation: competing pathways, condition-specific claims, and explicit unknowns rather than a verdict.
Evidence Explorer
Open the explorer to filter by evidence level, causality status, and condition. The research library holds the papers—never invented, with commentary labelled as commentary.