Neurobiology and genetics of mental disorders

01Models of causation

Mental disorders rarely have a single cause. Two models organise the many causes that research has found.

The biopsychosocial model, set out by George Engel in 1977, weighs biological, psychological and social factors for each patient, and is the frame of every psychiatric formulation.

The stress–vulnerability model, described by Zubin and Spring for schizophrenia, adds time. Each person carries a vulnerability, set by genes and early development, and a disorder appears when stress exceeds what that vulnerability allows. The same life event tips one person into illness and not another, and protective factors (a confiding relationship, treatment, avoiding cannabis) raise the threshold again.

The three Ps turn the models into a plan: the precipitants explain why now, and the perpetuating factors are what treatment can change.

02Neurotransmitter systems

Most of what is known about neurotransmitters in psychiatry began with drugs that happened to work, and with drugs that cause symptoms.

Dopamine

Dopamine runs in four pathways, and the effects and side effects of antipsychotics follow from them.

The dopamine hypothesis began with pharmacology: the potency of antipsychotics follows their affinity for the D2 receptor, and amphetamine, cocaine and levodopa can cause psychosis. Its current form comes from PET imaging. People with schizophrenia make and release more dopamine in the striatum, most of all in its associative (dorsal) part; the rise is present in many people with a prodrome who go on to psychosis, and receptor numbers barely differ. It is a final common pathway that genes, stress and drugs reach by different routes.

1. 1

   Sensitisation

   In a vulnerable person, stress and stimulant or cannabis use raise presynaptic dopamine synthesis and release in the striatum.
2. 2

   Aberrant salience

   Dopamine marks ordinary events and thoughts as important, so they feel loaded with meaning.
3. 3

   Symptoms

   Delusions arise as the person's explanation of these experiences, and inner speech experienced as alien becomes hallucinations.
4. 4

   Treatment

   D2 blockade dampens the salience over weeks; the fault remains, which is why stopping treatment often brings relapse.

An antipsychotic usually works once it occupies more than about 65 per cent of striatal D2 receptors; prolactin rises above about 72 per cent and parkinsonism becomes likely above about 78. Some patients who do not respond have normal striatal dopamine synthesis, which suggests another mechanism. All drugs of abuse raise dopamine in the nucleus accumbens, and drugs for ADHD such as methylphenidate block its reuptake.

Glutamate

The NMDA receptor needs glutamate, a co-agonist (glycine or D-serine) and depolarisation to open. The evidence that it works too little in schizophrenia is strong:

• Ketamine and phencyclidine, which block it, produce positive, negative and cognitive symptoms in healthy volunteers and worsen them in patients, which amphetamine, acting on dopamine alone, does not.
• Anti-NMDA receptor encephalitis, in which antibodies remove the receptor, often presents with psychosis before seizures, dyskinesia and autonomic instability appear.
• Genetic studies implicate glutamatergic genes, among them GRIN2A, which codes for an NMDA receptor subunit.

The circuit model joins the two theories: NMDA receptors drive inhibitory GABA interneurons, and when they work too little the pyramidal neurons fire excessively and an overactive hippocampus drives the midbrain dopamine neurons. Drugs that boost the glycine site, such as glycine transporter inhibitors, have so far failed in large trials.

Glutamate matters elsewhere too. Esketamine, authorised in the EU since 2019 as a nasal spray for treatment-resistant depression, lifts mood within hours; memantine, an NMDA antagonist, treats Alzheimer's disease; and a surge of glutamate when heavy drinking stops explains withdrawal seizures.

Serotonin and noradrenaline

Serotonin, from the raphe nuclei, acts on more than a dozen receptor subtypes: the 5-HT1A receptor is linked to anxiety, the 5-HT2A receptor is where LSD and psilocybin act and second-generation antipsychotics block, the 5-HT2C receptor affects appetite and the 5-HT3 receptor causes nausea. Noradrenaline from the locus coeruleus sets arousal and is overactive in panic, PTSD and withdrawal from alcohol and opioids.

The monoamine hypothesis of depression came from chance observations in the 1950s: reserpine, which depletes monoamines, was linked to depression, while iproniazid, a monoamine oxidase inhibitor first given for tuberculosis, and imipramine, which blocks reuptake, lifted mood. Tryptophan depletion briefly brings back symptoms in people who have recovered on an SSRI.

GABA

GABA is the main inhibitory transmitter. The GABA-A receptor, a chloride channel, is where benzodiazepines, Z-drugs, barbiturates and alcohol act, which explains their shared sedative and anticonvulsant effects, their cross-tolerance and why withdrawal from any of them can cause seizures. In schizophrenia, post-mortem studies show fewer markers of one type of GABA interneuron in the prefrontal cortex.

Acetylcholine

Cholinergic neurons of the basal forebrain project to the cortex and sustain attention and memory. Their loss in Alzheimer's disease and Lewy body dementia is the reason for cholinesterase inhibitors, and anticholinergic drugs are among the commonest causes of delirium in older people. Stimulating M1 and M4 muscarinic receptors reduces psychosis without blocking dopamine: xanomeline, combined with trospium to block its effects outside the brain, was approved for schizophrenia in the United States in 2024, but is not authorised in the EU and was not yet licensed in the UK in mid-2026.

03Stress and the HPA axis

The hypothalamic–pituitary–adrenal axis is the hormonal arm of the stress response. The hypothalamus releases corticotrophin-releasing hormone (CRH), the pituitary ACTH and the adrenal cortex cortisol, which feeds back through glucocorticoid receptors in the hippocampus, hypothalamus and pituitary to switch the axis off.

1. 1

   Acute stress

   CRH, ACTH and cortisol rise, then fall as feedback restores the baseline.
2. 2

   Chronic stress

   Repeated or uncontrollable stress keeps cortisol high and blunts feedback.
3. 3

   Brain effects

   Sustained cortisol reduces neuroplasticity in the hippocampus and prefrontal cortex and sensitises the amygdala, raising the risk of depression and anxiety in the vulnerable.

In severe depression, especially with melancholic or psychotic features, cortisol is often raised and fails to suppress after dexamethasone, though the test is not diagnostic. Cushing's syndrome and high-dose corticosteroids cause depression, mania and psychosis. Childhood abuse and neglect leave a lasting sensitivity of the stress response, perhaps through epigenetic marks on genes such as FKBP5, a line of work still early. In PTSD, by contrast, cortisol is often low.

04Neurodevelopment and neuroinflammation

The neurodevelopmental hypothesis holds that schizophrenia begins long before its first episode, in a brain that developed differently. The evidence:

• obstetric complications, especially lack of oxygen at birth, maternal infection and famine in pregnancy, and birth in late winter and spring, raise the risk;
• children who later develop schizophrenia have slightly delayed motor milestones and lower school performance;
• brain changes are present at the first episode, without the gliosis that marks damage in adult life;
• the illness usually starts in late adolescence, when synaptic pruning in the prefrontal cortex is at its height.

The strongest genetic signal in schizophrenia lies in the major histocompatibility complex on chromosome 6, explained in part by variants of the complement gene C4 that raise its expression; complement tags synapses for removal by microglia, so the finding links immunity to excessive pruning. Autism, ADHD and intellectual disability are neurodevelopmental disorders by definition, and ICD-11 groups them so.

Inflammation has a role in some patients. About a quarter of people with depression have low-grade inflammation, with a C-reactive protein above 3 mg/L, and interleukin-6 is raised on average; interferon-alpha causes depression in a large minority of those who receive it, and sickness behaviour resembles depression. Anti-inflammatory drugs have not yet proved themselves as treatments.

05Brain imaging

Imaging studies compare groups, and their findings overlap widely with healthy people; in the individual patient a scan is done to exclude a structural cause.

MRI is preferred to CT. Whether every first episode of psychosis is scanned varies between services; scanning is clearly indicated with neurological signs, cognitive decline, atypical features or late onset.

06Genetics

Twin, family and adoption studies

Families share homes as well as genes. Twin studies separate the two by comparing identical (monozygotic) twins, who share all their genes, with non-identical (dizygotic) twins, who share half, and adoption studies compare adopted children with their biological and adoptive families. Together they give the heritability.

Many genes of small effect

The common mental disorders are polygenic: risk comes from thousands of common variants, each raising it by a few per cent. Genome-wide association studies (GWAS) find them by testing millions of variants in very large samples, pooled for psychiatry by the Psychiatric Genomics Consortium (PGC), founded in 2007.

The variants found so far explain much less of the risk than twin studies predict, and a polygenic risk score, the sum of a person's risk variants, explains less than a tenth of the variation in liability to schizophrenia, and less still for depression. Scores are research tools that diagnose nothing in one person, and they work less well outside the European-ancestry samples most studies used.

Rare variants of large effect

Copy-number variants are deletions or duplications of stretches of DNA, usually more than 100,000 base pairs long. A handful raise the risk of schizophrenia several- to many-fold (deletions at 22q11.2, 3q29, 1q21.1, 15q13.3 and in NRXN1, a duplication at 16p11.2), and the same variants raise the risk of intellectual disability, autism and epilepsy. Together they are found in a few per cent of people with schizophrenia; ultra-rare variants that disrupt single genes, such as SETD1A and GRIN2A, have similarly large effects.

The 22q11.2 deletion syndrome (velocardiofacial or DiGeorge syndrome) is the strongest known genetic risk for schizophrenia. It is found in about 1 in 2000 to 4000 births, and about 1 in 4 people with it develop schizophrenia or another psychotic disorder. The clues are a conotruncal heart defect, a palatal anomaly or nasal speech, low calcium and recurrent infections; a chromosomal microarray detects it, and is the first genetic test in intellectual disability or autism, and in psychosis with these features.

Single genes cause a few disorders that present to psychiatrists, among them Huntington's disease, Wilson's disease, fragile X syndrome and the early-onset familial forms of Alzheimer's disease. The APOE ε4 allele raises the risk of late-onset Alzheimer's disease without deciding it.

Genes and environment

Genes and environment act together in two ways.

• Gene–environment interaction: the effect of an exposure depends on the genes, or the reverse. Cannabis raises the risk of psychosis more in people with a family history, and childhood maltreatment raises the risk of depression more in those already vulnerable.
• Gene–environment correlation: genes shape the environments people meet. Parents pass on both genes and a home, and a temperament can draw life events towards a person.

Early studies reported interactions with single candidate genes, such as the serotonin transporter variant 5-HTTLPR with life stress for depression. These did not replicate in large samples, and candidate gene findings in psychiatry are now treated with caution. Epigenetic marks, changes in gene expression without changes in the DNA sequence, may be how adversity is written into biology, but human evidence is still limited.

Shared genetic risk

Genetic risk does not respect diagnostic boundaries. Psychiatric disorders share much of their genetic risk with each other but little with neurological disorders such as Alzheimer's disease or epilepsy. A study of 14 disorders, published at the end of 2025, grouped their genetic risk into five factors that explain on average about two-thirds of each disorder's genetic variance:

The overlap fits the clinical picture, in which comorbidity is the rule, and supports dimensional views of psychopathology alongside the categories of ICD-11 and DSM-5-TR.

07What it means for patients

Patients and families often ask whether an illness is "in the genes" or whose fault it is. A balanced answer explains that several causes act together and that nobody is to blame. Purely biological explanations reduce blame but can increase pessimism about recovery, so biology is presented as one part of the story, and treatment works whatever the cause.