For decades, psychologists and psychiatrists have relied on the DSM to bring order to a very complex human landscape. Diagnostic labels have helped with communication, research, and treatment planning. But it has also come with obvious limitations.
Everyone who has worked clinically knows the Diagnostic and Statistical Manual of Mental Disorders (DSM) isn’t an infallible bible.
Two people can both meet the criteria for the same diagnosis and look completely different in the room. There are 79,794 combinations of ways to meet the diagnostic criteria for PTSD (Galatzer-Levy & Bryant, 2013).
One person may be highly anxious, perfectionistic, socially withdrawn, and traumatised. Another may present with the same diagnosis on paper, but with a very different developmental story, personality structure, social context, coping style, and treatment needs.
At the same time, many people meet criteria for several diagnoses at once, which raises a deeper question: are these really distinct conditions, or are we imposing neat labels on something that is far messier and more human?
A recent narrative review article by Eiko I. Fried offers a thoughtful way forward.
Its central idea is both simple and profound: mental disorders may be better understood not as fixed natural categories waiting to be discovered, but as homeostatic property clusters. In plain language, that means mental health problems are made up of overlapping patterns of biopsychosocial features that tend to cluster together, without always having sharp boundaries.
I think this is one of the most important directions in contemporary mental health science.
Psychiatric classification has struggled for a long time with the same core issues.
We’ve long known that diagnostic categories are highly heterogeneous (Forbes et al., 2024). Comorbidity is the norm, not the exception. Reliability is sometimes weaker than we would like.
Clinical utility can be modest. And despite multiple revisions since DSM-III in 1980, the field has not fully resolved these problems.
The DSM can be useful, but it often compresses rich, living, complicated people into administrative shorthand. It can create the impression that a person is their diagnosis, rather than a person who happens to show a particular pattern of difficulties at a particular point in time.
The HiTOP approach has been a leading alternative to the DSM, reclassifying symptoms into real empirically derived clusters, but it still just focuses on symptoms.
People are rich and wonderful. They are shaped by temperament, attachment, trauma, biology, culture, stress, strengths, relationships, values, and history. The DSM and HiTOP, by necessity, reduce that complexity down to symptoms. Sometimes that is practical. But it is never the full story.
Fried’s article argues that mental health classification can learn from the ways we classify things in biology. Have you ever argued with someone about whether a tomato is a fruit or a vegetable?
Rather than imagining diagnoses as fixed “natural kinds” with essential properties, we might think of them as clusters of traits, symptoms, experiences, risks, and protective factors that tend to travel together. These clusters are probabilistic, not deterministic. They overlap. They shift over time. They may look different across cultures and contexts. And the categories we draw on top of them are, to some degree, human constructions designed for particular purposes.
It means that the way we carve up the mental health landscape may never produce one perfect, universal classification system. And perhaps that has been the wrong goal all along.
Fried (2026) proposes a different vision: building a kind of mental health atlas.
Instead of beginning and ending with diagnostic boxes, we could map the many properties that matter in mental health: symptoms, personality traits, cognitive styles, coping patterns, trauma histories, treatment responses, social determinants, biological processes, functioning, beliefs, lifestyle factors, and more.
From there, we can ask better questions.
What properties cluster together most strongly? Which patterns are stable, and which are dynamic? Which factors cut across traditional diagnoses? Which mechanisms appear central? What predicts treatment response? What differs across settings, cultures, and developmental stages?
This is a much more ambitious approach than counting symptoms to see whether someone crosses a threshold. It is also much more faithful to clinical reality.
This is at the core of NovoPsych’s research mission, to help clinicians get to the heart of a client’s issues.
In practice, many psychologists like me already think this way. We notice sleep, emotion regulation, avoidance, shame, attachment insecurity, trauma, social isolation, neuroticism, perfectionism, inflammation, financial stress, and family context. We know these factors do not sit neatly inside one diagnosis. They interact.
NovoPsych already measures many of these factors, and I’m linking here to various measures that hone in on particular constructs in our assessment library
But a new, more unified framework that takes diverse factors into account in a single model would give expansive clinical intuitions a stronger conceptual foundation.
This perspective matters because labels do not explain people.
A person does not sleep poorly because they “have generalised anxiety disorder.” Rather, the label summarises a pattern that includes worry, hyperarousal, tension, attentional bias, avoidance, and often poor sleep. The diagnosis is descriptive shorthand, not the mechanism.
Don’t even get me started about the shortcomings of the borderline personality disorder label.
That distinction between labels being a descriptive shorthand, not an explanation of the underlying mechanism, matters enormously for treatment.
If we focus too much on categories, we risk treating the label. If we focus on the person’s profile of processes and properties, we are more likely to target what is actually maintaining distress. This is why psychologists care so much about psychological formulations.
That is where I think transdiagnostic and dimensional approaches have so much value. Interventions that improve sleep, reduce avoidance, build social connection, strengthen self-efficacy, or target perfectionism often help across multiple diagnostic presentations. That makes sense if shared processes matter more than tidy categorical boundaries.
It also helps explain why idiographic care matters. Two people with the same DSM diagnosis may need very different interventions because the property clusters driving their difficulties are not the same.
At NovoPsych, this is not just an abstract academic discussion. It is a research and development direction we care deeply about.
We believe the future of mental health measurement will move beyond narrow, siloed diagnostic thinking towards richer and more flexible models of human distress and functioning. That means better tools for dimensional assessment, transdiagnostic formulation, and understanding people as more than a checklist of symptoms.
We have already taken a step forward with the inclusion of advanced models of measurement, including the Brief Hierarchical Taxonomy of Psychopathology (B-HiTOP). That matters because frameworks like HiTOP are getting us closer to how people actually work. They are part of a broader effort to organise psychopathology in ways that better reflect empirical patterns and dimensional variation. But even HiTOP doesn’t take into account someone’s context, or fully appreciate a biopsychosocial formulation of a problem.
This is NovoPsych’s direction: away from asking “what box does this person belong in?” and toward asking “what are the major dimensions, properties, and mechanisms shaping this person’s mental health?”
That is a better scientific question, and often a better clinical one too.
Having looked at dozens of attempts at unifying models, my strongest conclusion is that no single classification system will optimally serve all purposes.
That is worth sitting with.
Clinicians need diverse tools that support formulation and treatment planning. Psychologists use models like CBT, ACT, Internal Family Systems because we know treatment isn’t one size fits all. And the same is true with formulation and psychometric assessments.
So perhaps the aim is not to replace the DSM with one new master system. Perhaps the aim is to accept pluralism: different frameworks for different functions, all grounded in better science.
When trying to understand psychopathology, complexity is reality.
And mental health classification isn’t defective because it is messy. It’s human-messy.
People can be anxious and high-functioning. Depressed and socially charming. Traumatised and resilient. Avoidant and deeply longing for connection. Highly conscientious and unraveling inside. Struggling in ways that change over time, across relationships, and across contexts.
A classification system that flattens all of that into a few categories will always miss something important.
The answer is not to abandon classification altogether. We do need structure. We do need shared language. But we also need a framework big and adaptive enough to hold the richness of actual human lives.
There is a reason why the opening question a clinician will ask a client an broad question like “what brings you here today”? The client might respond with their life context, symptoms, or a recent event. We need scientific models that can handle this.
That is why I find Fried’s property-cluster approach so promising. It gives us a way to think scientifically without pretending that people fit into clean boxes. It aligns with biopsychosocial thinking, dimensional models, network approaches, and transdiagnostic treatment. And it opens the door to a more personalised, mechanistic, and humane mental health science.
That is a future worth building toward, and I’m pleased that our research at NovoPsych is a part of that.
View over 150 of the current psychometric tools available on NovoPsych here.
Dr Ben Buchanan
Clinical Psychologist
NovoPsych Co-founder
[email protected]
LinkedIn
References:
Galatzer-Levy, I. R., & Bryant, R. A. (2013). 636,120 Ways to Have Posttraumatic Stress Disorder. Perspectives on psychological science : a journal of the Association for Psychological Science, 8(6), 651–662. https://doi.org/10.1177/1745691613504115
Forbes, M. K., Neo, B., Nezami, O. M., Fried, E. I., Faure, K., Michelsen, B., … Dras, M. (2024). Elemental psychopathology: distilling constituent symptoms and patterns of repetition in the diagnostic criteria of the DSM-5. Psychological Medicine, 54(5), 886–894. doi:10.1017/S0033291723002544
Fried, E. I. (2026). Mental disorders as homeostatic property clusters: A narrative review. JAMA Psychiatry. Advance online publication. https://doi.org/10.1001/jamapsychiatry.2026.0073
Hegarty, D., Forbes, M. K., Buchanan, B., Smyth, C., Baker, S., & Bartholomew, E. (2025). A Review of the Clinical Utility and Psychometric Properties of the Brief Hierarchical Taxonomy of Psychopathology (B-HiTOP): Norms, Percentile Rankings, and Qualitative Descriptors. https://doi.org/10.17605/OSF.IO/U5WD8