Psychiatry speaks in nouns. Depression. Schizophrenia. ADHD. Panic disorder. Borderline personality disorder. Substance-use disorder.
Patients experience verbs, adjectives, and degrees.
They sleep less. Fear more. Want less. Think faster. Lose control. Hear a voice. Avoid a room. Become more suspicious. Feel empty. Seek reward despite consequences. Their experiences vary in intensity, duration, combination, and context before clinicians draw a categorical line through them.
Categories are not useless. Medicine needs reliable communication. Services need eligibility criteria. Clinical trials need inclusion rules. A diagnosis can summarize a recognizable syndrome and guide treatment. The mistake is to assume that because a category is useful, the underlying biology must also be categorical.
Many psychiatric phenomena look continuous with ordinary function.
Fear is universal. Panic is not. Confidence is ordinary. Grandiosity is pathological in some contexts. People differ normally in suspiciousness, impulsivity, social detachment, compulsivity, emotional reactivity, and unusual perceptual experiences. The existence of severe endpoints does not tell us whether the underlying dimensions contain a natural boundary.
This is the intuition behind dimensional frameworks such as RDoC and HiTOP, although they approach the problem differently. RDoC emphasizes functional systems that cut across conventional diagnoses. HiTOP organizes psychopathology into empirically derived dimensions and spectra. Personality pharmacology belongs near this family because it asks whether treatment should target control dimensions rather than labels alone.
The strong version of dimensionalism, however, is too easy to overstate.
Not every disorder is simply normal personality turned up. A structural brain lesion can create abrupt psychiatric change. Infection, seizure disorders, endocrine disease, intoxication, neurodegeneration, chromosomal differences, or rare variants can create causal discontinuities. Some neurodevelopmental conditions may contain subgroups with genuinely distinct structures. A population can contain both continuous variation and taxonic components.
The important distinction is between cause and phenotype.
A categorical cause can produce changes along familiar psychological dimensions. A tumor can alter inhibition, motivation, affect, or perception. Conversely, a continuously varying parameter can generate a sudden clinical transition when a threshold is crossed. Gradual sleep loss can culminate in mania. Slowly increasing autonomic arousal can trigger an abrupt panic attack.
The geometry of cause does not dictate the geometry of symptom.
This is why taxometric questions matter. If a syndrome is best modeled as a distinct latent class, that has different implications from a pure continuum. If dimensional and categorical models both fit, hybrid models may be necessary. The book should resist the temptation to declare victory because a symptom score has no obvious gap in a histogram.
Clinical sampling adds another trap. If researchers compare diagnosed patients with healthy controls, they can manufacture a bimodal distribution by design. If they recruit only treatment-seeking cases, the middle of the population disappears. Conversely, population surveys can blur biologically distinct subgroups when measurement is crude.
The right test is predictive. Which representation better forecasts course, impairment, biology, and treatment response?
This leads directly to psychopharmacology. Suppose two people meet identical criteria for depression. One has profound anhedonia and low effort with little anxiety. Another has severe threat sensitivity, rumination, and insomnia. A category tells us both are depressed. A dimensional model tells us they occupy different regions of a psychological space.
If those dimensions predict different treatment responses, the dimensional representation is clinically superior even if the diagnosis remains useful.
The same argument applies to comorbidity. A person with “depression”, “generalized anxiety disorder”, and “alcohol-use disorder” may not have three independent diseases. They may have a shared configuration involving high negative affect, chronic arousal, poor sleep, and state-altering behavior used to suppress distress. Another person with the same three diagnoses may have a different mechanism.
Categories can therefore hide both shared dimensions and hidden heterogeneity.
There is an ethical advantage to dimensional thinking as well. It softens the false boundary between “people with mental illness” and “normal people” without denying severe impairment. A person can recognize that anxiety, unusual beliefs, impulsivity, or low mood exist on dimensions shared by everyone, while still accepting that at sufficient intensity those dimensions can devastate function.
But dimensional language can become dismissive if used carelessly. Telling someone with psychosis that their experience is merely “a point on a continuum” may minimize the qualitative rupture they are living through. Thresholds, feedback, and altered network states can create experiences that are not captured by saying “more of the same”.
The more mature claim is this: many of the functions disrupted in psychiatric illness vary continuously in the population, while illness emerges through combinations of extremity, context, thresholds, instability, causal pathology, and loss of control.
That formulation is less elegant than a universal continuum. It is also more likely to survive.
Once average is no longer synonymous with healthy, psychiatry needs another way to define dysfunction. The next chapter proposes one: distance not from the population mean, but from a context-specific adaptive region.