“Phaedrus: And what is the other principle, Socrates?
Socrates: That of dividing things again by classes, where the natural joints are, and not trying to break any part, after the manner of a bad carver.“
Plato, Phaedrus [265e]
What are mental disorders?
Mental suffering has been defined and treated in psychiatry for over 150 years. This suffering is all too real, yet, there remains a persistent lack of consensus regarding what kind of things mental disorders actually are (1). Diagnostic manuals such as the Diagnostic and Statistical Manual of Mental Disorders (DSM) and the International Classification of Diseases (ICD) primarily treat mental disorders as discrete diagnostic categories, although dimensional diagnostic approaches are on the rise in both diagnostic systems. The DSM-5 acknowledges several structural problems with the categorical approach, including symptom heterogeneity within disorders and high levels of comorbidity (2). In clinical reality, the borders between mental disorders are fuzzy, and hallmark symptoms of one disorder are commonly found in other mental disorders, like symptoms of disturbed sleep or anxiety. Prominent suggestions for how to understand the permeable borders of mental disorders include the p-factor theory and the network theory of mental disorders. The p-factor theory suggests that mental disorders and symptoms are best understood as specific manifestations of a transdiagnostic psychiatric liability, called the p-factor, giving rise to different kinds of mental disorders and explaining the high symptom overlap across disorders. The network theory of mental disorders, on the other hand, shifts the explanatory focus to the symptom themselves by suggesting that disorders arise from symptom interactions, with disorders seen as emergent properties of interconnected symptom-clusters, or networks. From this perspective, comorbidity is explained by direct causal connections between the symptom networks of different disorders. These two perspectives provide radically different views on what mental disorders are, with correspondingly different implications for clinical use. Over the years, the difficulty of defining what mental disorders actually are have rendered a plethora of clinical and scientific challenges, in part also impeding the advancement of prevention, treatment, and care.
The utility of diagnostic categories
For humans to be able to think, communicate, and act intentionally, we typically need to categorise naturally continuous phenomena into more or less arbitrary discrete entities. In the clinical context, these categories are ideally both precise and clinically useful. The most recent revisions of DSM and ICD have an explicit aim to increase clinical utility. It has been argued, however, that current diagnostic categories for mental disorders still provide suboptimal clinical guidance for individual-level treatment and prognostics (3). Individuals with different diagnoses often respond similarly to a treatment, e.g., SSRI or CBT, while there may be considerable heterogeneity of treatment-response within diagnostic categories. Prediction models have made some progress in forecasting individual treatment outcomes, highlighting the need to incorporate individual characteristics beyond diagnostic categories to improve prognostic acuity.
The difficulties of defining mental disorders has also been highlighted as a major obstacle in understanding their aetiology. Epidemiological and clinical research have produced an impressive list of potential causes of mental disorders, ranging from biological agents to social influences. Together with the absence of strong specific causes, like single disorder-coding genes, mental disorders are today considered predominantly polyfactorial, meaning that they are caused by a range of disparate factors. Unlike for other multi-causal diseases such as cancer and CVD however, the accumulation of potential causes has done little to advance prevention and treatment of mental disorders. This failure has been argued, at least in part, to stem from the fact that today’s codified diagnostic categories do not map sufficiently well to actual disorder entities (4). In 2009, this led the American National Institute of Mental Health to launch the Research Domain Criteria (RDoC). RDoC advocates for shifting the focus from mental disorder diagnoses to empirically derived basic dimensions of human functioning, such as attention, reinforcement and fear. Many of these narrow dimensions proposed by RDoC constitute unobservable latent constructs, for example fear, and this proposed path forward relies heavily on psychometric methods to capture them properly.
Measuring mental disorders
Unlike instruments that gauge other bodily phenomena, e.g., machines for measuring blood-pressure or white blood cell counts, it is not entirely clear what our mental disorder instruments are trying to capture. The prevailing approach has been to view mental disorders as latent constructs and to develop psychometric scales to try to estimate these constructs at the individual patient level. This has led to a plethora of widely used clinician- and self-rated scales, like the Montgomery-Asberg Depression Rating Scale and the Generalised Anxiety Disorder Scale (GAD-7). Still, there is a built-in tension between psychometric methods and the current conceptualisation of mental disorder categories. Psychometric approaches favour homogenous and one-dimensional instruments including highly correlated items. Mental disorders diagnoses, on the other hand, often contain heterogenous constellations of symptoms. Narrowing the scope of an instrument, for instance by including only the core symptoms of a disorder, may therefore improve its psychometric properties, but reduces its ability to cover the full symptomatology of a disorder. We believe that this is not necessarily a bad thing. Narrowly defined dimensions of human functioning is what is explicitly advocated through the RDoC approach, and instead of trying to cover the full symptomatology of current diagnostic categories in a single measure, a variety of narrower measures will result in improved measurement properties, and may also provide more insight into mechanisms and clinical intervention targets.
Embrace variation and accept uncertainty
We believe that current diagnostic categories are needed to communicate and guide clinical understanding and decision-making. At the same time, clinicians and researchers should acknowledge that these are uncertain and coarse models of much more nuanced and complex real-life phenomena. Uncertainty is an unavoidable part of almost everything in life, yet, the allure of clear-cut answers is hard to resist, both in science and clinical practice. Methodologist Sander Greenland coined the term dicotomania to describe our urge, almost a compulsion, to divide continuous measures of uncertainty into discrete all-or-nothing categories (6). Rather than treating our tentative and uncertain diagnostic categories as dichotomous black-or-white units, we believe that clinicians and researchers would be well served by heeding the words of statistician Andrew Gelman, “embrace variation and accept uncertainty” (7). By supplementing diagnoses with validated tools for assessing a variety of specific dimensions of behavior and emotions, clinical heterogeneity and individual differences could be embraced rather than treated as nuisance. This pluralistic perspective is well-aligned with the current process-based approach to psychotherapy (8) where clinical interventions are guided by assessment of the ideographic processes leading to suffering. We believe this would help avoid simplistic and reductionist explanations of the human mind and its malfunction, with the hope to advance both psychiatric research and clinical practice. At the end of the day, uncertainty and complexity remains as foundational elements for worthwhile scientific inquiry. No matter how well we understand a topic, the unknowns tend to outnumber the knowns.
In the classic Plato's Phaedrus dialogue (9) quoted in the beginning of this text, the concept of "carving nature at its joints" refers to the idea of properly dividing or categorising things in a way that reflects their true nature or essence. With one quick look at mental disorder diagnostics, this seems an endeavour that is a priori bound to fail. We would however argue the opposite. Because our study and treatment targets are difficult-to-define latent constructs, we need to progress with both diligence and care when partitioning mental suffering into actionable concepts, knowing that the contemporary psychiatric taxonomy is by no means complete and the categories we use today may not be useful tomorrow. Because it is particularly difficult to partition nature at its psychiatric joints, we hold that there is also particular potential for improving the diagnostic categories through which we understand, treat, and care for our patients. □
References by request
Kendler KS. The nature of psychiatric disorders. World Psychiatry. 2016 Feb 1;15(1):5–12.
American Psychiatric Association. Diagnostic and statistical manual of mental disorders : DSM-5. Arlington, VA: American Psychiatric Association; 2013.
Maj M. Why the clinical utility of diagnostic categories in psychiatry is intrinsically limited and how we can use new approaches to complement them. World Psychiatry. 2018 Jun 1;17(2):121–2.
Kozak MJ, Cuthbert BN. The NIMH Research Domain Criteria Initiative: Background, Issues, and Pragmatics. Psychophysiology. 2016;53(3):286–97.
Carrozzino D, Patierno C, Fava GA, Guidi J. The Hamilton Rating Scales for Depression: A Critical Review of Clinimetric Properties of Different Versions. Psychother Psychosom. 2020 May 5;89(3):133–50.
Greenland S. Invited Commentary: The Need for Cognitive Science in Methodology. Am J Epidemiol. 2017 Sep 15;186(6):639–45.
Vasishth S, Gelman A. How to embrace variation and accept uncertainty in linguistic and psycholinguistic data analysis. 2021;59(5):1311–42.
Hayes SC, Hofmann SG, Ciarrochi J. A process-based approach to psychological diagnosis and treatment:The conceptual and treatment utility of an extended evolutionary meta model. Clin Psychol Rev. 2020/09/16 ed. 2020 Dec;82:101908.
Plato. Plato in Twelve Volumes, Vol. 9 translated by Harold N. Fowler. Cambridge, MA, Harvard University Press; London, William Heinemann Ltd. 1925.