A psychiatric diagnosis is supposed to help people make sense of distress and guide individualised care. Yet psychiatry holds an extraordinary power: criticism of its own diagnostic constructs and treatments can be reinterpreted as evidence that the person criticising them is ill or “lacks insight”. It is a Catch-22: accept the interpretation and the diagnosis stands; challenge it, and the disagreement can reinforce it.
In our new essay published in PLOS Mental Health, When psychiatry does not learn, my co-authors and I call this ‘institutional non-learning’. We describe this as a cycle whereby the systems that give diagnoses, control access to care and sometimes coerce or harm people, also impose limits on whose warnings are heard, whose accounts are believed and what accountability involves.[1] Too often, accountability goes little further than expressions of concern, defensive routines, and unrealised promises of improvement.[1]
Diagnostic weaponisation: when a label silences
Diagnosis itself can change who is believed. We describe diagnostic weaponisation as using diagnostic authority beyond what the evidence justifies to discredit, punish, exclude or control.[1]
During the United States civil rights era, schizophrenia diagnoses were disproportionately applied to Black men, pathologising resistance and anger.[2] Racialised inequities persist: Black people have been found less often diagnosed with attention-deficit/hyperactivity disorder and more often with highly stigmatised conduct-related disorders than White people, routing distress and support needs into more punitive frameworks.[1]
Psychiatry has likewise pathologised sexual and gender diversity. Homosexuality was classified as a “sociopathic personality disturbance”, while gender diversity was pathologised through diagnoses such as “gender identity disorder”.[3] Research shows that the similarly contentious diagnosis of “Borderline personality disorder” (BPD) has been given more frequently to lesbian, gay and bisexual patients and trans and gender diverse patients, even after relevant clinical characteristics have been considered.[1,3] Put simply, sexual orientation or gender identity thus become part of why someone receives a BPD diagnosis, rather than underlying psychopathology.
When the treatment can do no wrong
Challenge a diagnosis and you may “lack insight”. Say treatment is not helping, or is harming you, and the objection may become “resistance”, “non-compliance” or “treatment-interfering behaviour”. We call this diagnostic recursivity: the challenge becomes evidence for the interpretation being challenged.[1]
Participants in a study of negative effects of dialectical behaviour therapy (DBT) described feeling “I’m the problem” while “DBT can do no wrong”; crying could be treated as therapy interference and concerns about treatment as signs of psychopathology.[4] Participants in research on inpatient care for anorexia nervosa similarly described care deepening suffering, with questioning care or expressing frustration risking dismissal as “manipulative” or “difficult”.[5]
At its worst, treatment promotion can look less like science and more like dogmatic or cult-like allegiance. “Gold standards” are treated as settled, and when people do not improve or report harm, deflection of responsibility can shift back onto the patient, including claims that they are self-sabotaging and lacking the motivation to recover. No treatment should be protected from criticism, disconfirming evidence or accounts of harm.
One defence is that harm was “unintentional”. But lack of intent does not erase responsibility and the need for accountability.[1] After decades of psychiatric survivors and Mad communities being ignored when describing recurring harms, “unintentional” starts to ring hollow. This is “ignorance culture”: warnings can be minimised, dismissed or reframed instead of changing practice or leading to meaningful reform.[1] At what point does repeatedly failing to listen and act become a choice not to know?
Benefit is not safety
Think about ibuprofen. You take it for a headache and the headache disappears. On that measure, the treatment worked. But if the same medication also causes a serious, potentially fatal bleeding stomach ulcer, nobody would point to the disappearance of the headache and conclude that the medication was therefore safe.[7] Both things can be true at once: the treatment helped and the treatment harmed you.
Psychotherapy should meet the same standard. Yet a review of 132 trials found no indication that harms were monitored in 79%, while only 3% described both adverse events and how they were collected.[1] Someone can improve on a symptom scale while also leaving therapy traumatised or afraid to seek help again. Not measuring those outcomes means research records the benefit and misses the harm.[4]
There is another problem: universalising “evidence-based” claims. Findings from mainly white, cisgender and neurotypical samples cannot simply be stretched to people barely represented in the research. Culture, identity, trauma history, and communication needs can shape treatment experiences. For autistic, trans and gender diverse people, and Indigenous communities especially, the question is: evidence-based for whom?[1,8]
From consultation to responsibility
Defensiveness and denial also block learning. My co-author James Downs describes “a profound resistance” to recognising that treatment might itself have harmed him.[9] People reporting long-term adverse effects after electroconvulsive therapy have likewise described memory loss, inadequate follow-up and dismissal.[10] These accounts should trigger investigation, not trivialisation.
Our alternative is a “responsibility culture”: independent oversight, proactive harm monitoring, paid lived experience decision-making authority, meaningful remedy and redress, correction where warranted and publicly available evidence of improvement.[1] A mental health system capable of learning must acknowledge, apologise and change when it gets things wrong.
Abbreviations
BPD: borderline personality disorder
DBT: dialectical behaviour therapy
References
- Cobbaert, L., Elwyn, R., Downs, J., James, S., & Jackman, M. (2026). When psychiatry does not learn: Epistemic humility, ignorance culture, exclusion culture and diagnostic weaponisation. PLOS Mental Health, 3(10), e0000741. https://doi.org/10.1371/journal.pmen.0000741
- Jackman, M., Sharma, P., Sugianto, A., Van der Schaar, A. M. P., Hajj-Assaf, S., McIntosh, K., et al. (2026). “Revising the DSM-VI”: Global perspectives on power and classification from lived expertise leadership. PLOS Mental Health, 3(5), e0000626. https://doi.org/10.1371/journal.pmen.0000626
- Cobbaert, L., Maloney, E., Harding, K., & James, S. (2026). Dismantling the diagnostic construct of borderline personality disorder: A critical discourse analysis. International Journal of Mental Health Nursing, 35(2), e70241. https://doi.org/10.1111/inm.70241
- Lawson, Z., & Farquharson, L. (2025). First do no harm: Client and staff experiences of negative effects from dialectical behaviour therapy. Psychology and Psychotherapy: Theory, Research and Practice, 98(3), 570–589. https://doi.org/10.1111/papt.12578
- Sandsten, A., Gabrielsson, S., Strömbäck, M., Looi, G. M. E., & Lindgren, B.-M. (2025). “Being kept alive—but not being supported to live”: Experiences of general psychiatric inpatient care among persons with anorexia nervosa. Journal of Eating Disorders, 13, 282. https://doi.org/10.1186/s40337-025-01483-9
- Junqueira, D. R., Phillips, R., Zorzela, L., Golder, S., Loke, Y., Moher, D., et al. (2021). Time to improve the reporting of harms in randomized controlled trials. Journal of Clinical Epidemiology, 136, 216–220. https://doi.org/10.1016/j.jclinepi.2021.04.020
- Health Products Regulatory Authority. (2025, March 10). Nurofen 200 mg coated tablets: Summary of product characteristics. https://assets.hpra.ie/products/Human/23657/Licence_PA0979-032-006_10032025160937.pdf
- Cobbaert, L. (2026). Cognitive behavioural therapy is not universally evidence-based: Implications for eating disorders. Journal of Eating Disorders, 14, 60. https://doi.org/10.1186/s40337-026-01558-1
- Downs, J. (2025). Whose trauma is it anyway? Creating more equitable mental healthcare in a system that harms. BJPsych Bulletin, 49(4), 265–268. https://doi.org/10.1192/bjb.2024.103
- Shipwright, E., & Murphy, D. (2025). Long-term adverse effects after electroconvulsive therapy (ECT): A narrative analysis exploring people’s experiences, meaning-making, and coping. Qualitative Health Research, 35(12), 1365–1377. https://doi.org/10.1177/10497323241303391