Dopamine and Beyond: Toward a pluralistic model for the schizophrenia spectrum

Abstract

Importance.

Schizophrenia and related psychosis spectrum disorders are clinically heterogeneous conditions whose underlying neurochemical mechanisms remain incompletely understood. Although dopaminergic dysregulation has guided pathophysiological models and antipsychotic drug development for decades, this framework does not adequately explain treatment resistance, cognitive and negative symptoms, or the marked biological diversity observed across patients.

Objective.

To critically evaluate dopaminergic and non-dopaminergic neurochemical models of schizophrenia and related psychosis spectrum disorders and to assess whether these disorders are best conceptualized as arising from a final common dopaminergic pathway or from multiple, partially independent pathophysiological mechanisms.

Evidence Review.

This narrative review synthesizes evidence from human neuroimaging, postmortem studies, genetics, pharmacologic challenge paradigms, randomized clinical trials, and translational animal models published primarily between 1980 and 2025. Literature was identified through expert knowledge and targeted searches of PubMed and related databases, with emphasis on systematic reviews, meta-analyses, and high-quality human studies. Evidence was evaluated qualitatively, prioritizing convergent findings across complementary methodologies.

Findings.

Robust evidence supports increased presynaptic dopaminergic activity in the associative striatum as a key mechanism underlying positive psychotic symptoms, particularly in treatment-responsive individuals. However, approximately 30% of patients exhibit treatment resistance and show normal striatal dopamine synthesis capacity,suggesting alternative mechanisms. Converging data implicate glutamatergic dysfunction (notably NMDA receptor hypofunction), GABAergic interneuron abnormalities, cholinergic deficits—especially involving muscarinic M1/M4 receptors—serotonergic modulation, endocannabinoid signaling, and non-neurotransmitter processes including oxidative stress, mitochondrial dysfunction, and neuroinflammation. Recent evidence demonstrating antipsychotic efficacy of the muscarinic agonist combination xanomeline–trospium highlights the feasibility of effective non-dopaminergic treatments, although it is unclear if its efficacy was indirectly mediated by dopaminergic modulation. Across modalities, findings support the existence of neurochemically distinct subgroups within psychosis spectrum disorders that converge on overlapping clinical phenotypes.

Conclusions and Relevance.

Available evidence does not support dopamine dysregulation as a universal final common pathway across symptom domains for schizophrenia and related psychosis spectrum disorders. A pluralistic model that recognizes multiple interacting neurochemical and cellular mechanisms better accounts for clinical heterogeneity and translational failures. There is greater support for dopaminergic hyperactivity in the associative striatum as a final common pathway for core psychotic symptoms across various conditions. Progress in treatment development will require biomarker-informed stratification, mechanism-based clinical trials, and a shift beyond dopamine-centered paradigms toward circuit- and pathway-informed therapeutics.

Schizophrenia and related psychotic disorders (SRD) are complex, heterogeneous psychiatric conditions whose precise neurochemical basis remains elusive (1). The classical dopamine (DA) excess hypothesis, and its later revisions have guided treatment development for decades. However, DA dysfunction may not adequately explain the full range of symptoms, particularly cognitive and negative domains, and fails to account for treatment-resistant cases (around 30% of the schizophrenia spectrum). Alternate models have emerged inspiring novel interventions, but few non- dopaminergic interventions have proven efficacy raising a central question: do all neurochemical abnormalities ultimately converge on dopamine dysregulation, or do they represent distinct pathogenic pathways requiring unique therapeutic interventions? We critically examine current knowledge and knowledge gaps for each pathophysiological model, explore their possible convergence on dopaminergic function, and discuss whether SRDs result from a final common DAergic dysregulation or from biologically distinct processes. We discuss SRD broadly, given the substantive overlaps in clinical manifestations, biology, etiology and treatment response across primary psychotic disorders (1). Addressing this question will be crucial for developing more effective, individualized treatments for SRD

What we know and don’t know about neurochemical pathophysiology of SRD

Dopamine

The widely held DA hypothesis posits increased dopaminergic functioning underlying psychosis (2) based on the clinical efficacy of antipsychotics that block dopamine D2 receptors and the psychotomimetic effects of dopamine-enhancing drugs (e.g., amphetamines). Subsequent positron emission tomography (PET) studies support increased presynaptic striatal DA release (3) A revised version of the DA hypothesis added regional specificity and implicated mesolimbic DA excess in psychotic symptoms and prefrontal hypodopaminergia as underlying cognitive deficits and negative symptoms (4). In support, PET evidence, albeit inconsistent, shows reduced dopamine release capacity and altered dopamine D1 receptor binding in the prefrontal cortex in schizophrenia the latter is related to cognitive impairment and negative symptoms (5,6). The pro-cognitive effects of D1 receptor agonism first identified in non-human primates (7), is currently under investigation (8); results are awaited. Kapur(9) proposed that dopamine mediated aberrant salience, i.e. assignment of importance to unimportant stimuli is related to psychosis. An alternative model (10) suggests that diminished stimulus (reward) induced dopamine release may lead to negative symptoms.

Howes and Onwordi (11) further refined the dopamine model, proposing that multiple genetic and/or environmental factors make synapses vulnerable to excessive glia-mediated elimination triggered by stress during later neurodevelopment. Loss of synapses, especially glutamatergic, may lead to negative and cognitive symptoms; disinhibition of mesostriatal DA may lead to psychosis (10). Grace and Gomes (12) suggested that the increased DA system responsivity in the striatum is due to an increase in the proportion of active DA neurons, consistent with the increased fluorodopa in the associative striatum (3). Accumulating data support the link between excess dopamine signalling, increased prediction error abnormalities, aberrant salience, and psychosis ⁽13).

The DA model does not account for the facts that a) none of the current dopamine receptor antagonists are effective for negative and cognitive symptoms, and b) Many patients with psychosis do not respond to current dopamine D2 antagonists. This highlights the need for broader neurobiological approaches. In summary, data indicate that dopaminergic dysregulation can significantly explain core psychotic symptoms across disorders, but cannot explain other aspects of psychopathology.

Glutamate

The glutamatergic theory of schizophrenia (14), which posits hypofunction of the NMDA (N-methyl-D-aspartate) receptor, is based on observations of psychosis caused by NMDA antagonists such as phencyclidine (PCP) and ketamine (15) as well as NMDA receptor encephalopathy (16), which presents with positive, negative, and cognitive symptoms. The glutamatergic model, supported by excitotoxic glutamatergic damage resulting from early brain adversity, i.e. perinatal hypoxia provides a neurodevelopmental framework for schizophrenia (17).

Glutamatergic compounds have been widely investigated as adjunctive treatments, particularly for negative symptoms. Glutamatergic modulators modestly benefit negative symptoms when added to antipsychotics (18). These include direct co-agonists, such as glycine and D-serine, and partial agonists ( e.g. D-cycloserine), glycine transport inhibitors (e.g., sarcosine, bitopertin) and D-amino acid oxidase inhibitors (e.g., luvadaxistat); Pomaglumetad, a metabotropic glutamate 2/3 receptor modulator may be effective in early course schizophrenia (19), and evenamide, a glutamate modulator, is promising in treatment-resistant schizophrenia (20)); these findings await replication.

The precise mechanisms underlying glutamate dysfunction also remain uncertain. Potential causes include disturbances of presynaptic glutamate release, intrinsic NMDAR dysfunction, abnormal intra- vs. extrasynaptic NMDAR distribution, disturbances in glycine/D-serine metabolism, and abnormalities in post-synaptic signal transduction (14). Few efforts have focused on precision treatments based on molecular subtypes of NMDAR dysfunction. Rare individuals with glycine decarboxylase gene triplication—leading to reduced glycine levels— may respond to glycine or D-cycloserine (21).

While the glutamatergic theory provides a compelling account of schizophrenia’s complex clinical features, therapeutic translation has been difficult. The heterogeneity of glutamatergic dysfunction, phase-specific alterations (prodromal vs. chronic), and challenges of target engagement complicate progress.

Gamma amino butyric acid (GABA)

Postmortem studies demonstrate reduced cortical levels of mRNAs and proteins essential for GABA neurotransmission, most prominent in the parvalbumin and somatostatin neurons (22); this may contribute to cognitive dysfunction in schizophrenia (23). It is unclear whether lower GABA neurotransmission is primary or a secondary response to altered excitation in cortical circuits. In either case, the resulting disturbances could contribute to excessive DA release in the associative striatum, although direct evidence is lacking.

Therapeutic strategies targeting GABAergic dysfunction have yielded inconsistent benefits. MK-0777, a GABA A2 partial agonist was effective in a pilot study in improving cognitive function and prefrontal gamma band power (24) but not replicated in a larger study (25) Benzodiazepines, which are positive allosteric modulators of the GABA-A receptors, are limited by sedative side effects, potential for abuse (26), and lack of demonstrated antipsychotic efficacy [27].

Serotonin

The serotonin hypothesis of schizophrenia originated from the observations that serotonin agonists such as LSD can induce psychosis. Second generation antipsychotics (e.g., clozapine, risperidone) also have high affinity for 5-HT2A receptors, supporting a modulatory role of serotonin in psychosis (28). Serotonin significantly influences perception, mood, and cognition. However, serotonin dysfunction alone does not explain the full spectrum of SRD. Cortical 5-HT(2A) receptor binding differs between first-episode antipsychotic-naive schizophrenia patients and healthy controls (29). Osugo et al (30) recently showed that increased prefrontal serotonin release is associated with negative symptoms. This deserves replication. Direct evidence linking serotonin abnormalities to SRD pathophysiology, however, still remains limited.

Serotonin agonists such as LSD cause hallucinations, perhaps by increasing glutamate release (31). Psychosis caused by serotonin agonism, which mainly involves hallucinations (32) differs somewhat phenomenologically from psychosis induced by dopamine agonists such as methamphetamine (characterized mainly by delusions with minimal thought disorder or negative symptoms (33).

Drugs targeting serotonin receptors are inconsistently effective in treating schizophrenia, though pimavanserin, a serotonin inverse agonist, is effective for psychosis in patients with parkinson’s disease. Selective 5HT3 antagonists such as ondansetron and tropisetron, may improve sensory gating and thus have precognitive effects (34). Future studies should investigate the clinical significance of these findings.

Acetylcholine

The cholinergic hypothesis of schizophrenia suggests that altered muscarinic (M1, M4) and nicotinic (especially α7) signaling may contribute to psychosis and cognitive deficits. The muscarinic hypothesis is supported by several observations: (i) psychosis caused by belladonna (which contains atropine) does not respond well to antipsychotics (35); (ii) betel nut, which has the muscarinic agonist arecoline, has antipsychotic effects (36); (iii) postmortem and imaging studies show reduced muscarinic receptor density (37) (iv) clozapine’s efficacy in treatment-refractory psychosis has been linked to its partial M-1 muscarinic agonist activity, or to clozapine’s pro-muscarinic effect reduces striatal synthesis of dopamine (38, 39). The muscarinic M1/M4 agonist xanomeline is the first antipsychotic drug without direct DA effects to be approved for the treatment of schizophrenia (40) While M4 positive allosteric modulator, emraclidine was not effective, several clinical trials of molecules with related mechanisms are under way (41).

The nicotinic hypothesis stems from observations of high smoking prevalence in schizophrenia, nicotine’s procognitive effect and normalization of P50 gating deficits seen in schizophrenia. A recent meta-analytic review showed a widespread decrease in muscarinic and nicotinic receptor levels in schizophrenia, evident in both neuroimaging and post-mortem studies (37). While partial nicotinergic agonists like encenicline showed cognitive benefits in initial trials (42), subsequent studies have produced inconsistent results (43).

Endocannabinnoids

The cannabinoid hypothesis of schizophrenia focuses on dysregulation of the endocannabinoid system, particularly CB1 receptor overactivity. Cannabis use, especially during adolescence, increases risk for schizophrenia, and higher levels of the endocannabinoid anandamide have been observed in patients. CB1 receptors modulate neurotransmitter release (including dopamine, glutamate, and GABA), suggesting a possible role in psychosis (44). However, the causal relationship remains unclear—endocannabinoid changes may reflect a compensatory response rather than a primary pathology. Therapeutic targeting of this system is still experimental, with inconsistent results. While one study (45) found a beneficial effect of cannabidiol as an adjunct to antipsychotic treatment in schizophrenia, another study (46) did not replicate this finding.

Non-neurotransmitter pathophysiological processes

Impaired mitochondrial function and disrupted energy metabolism have been proposed (47). Early phosphorus magnetic resonance spectroscopy studies showed evidence of decreased ATP turnover suggesting decreased prefrontal metabolism (48). Studies have shown reduced expression of mitochondrial genes, decreased ATP production, altered oxidative phosphorylation, and increased markers of oxidative stress in patients (49, 50). These deficits may impair synaptic plasticity, neurotransmission, and neurodevelopment, contributing to cognitive dysfunction and negative symptoms. However, findings are heterogeneous, and it remains unclear whether mitochondrial changes are primary or secondary. There is a large literature on treating metabolic impairment in diabetes and other conditions but these have not been tried systematically in SRD, with a recent promising exception, the ketogenic diet (51)

The oxidative stress theory of schizophrenia posits an imbalance between reactive oxygen production and antioxidant defenses leading to neuronal damage, particularly in parvalbumin-positive interneurons (49, 50). Elevated markers of oxidative stress and reduced antioxidant molecules (e.g., glutathione) have been found in blood, cerebrospinal fluid, and postmortem brain tissue of patients. This may contribute to synaptic dysfunction, impaired neurodevelopment, and cognitive deficits. Oxidative stress also interacts with mitochondrial dysfunction, neuroinflammation, and NMDA receptor hypofunction (50). However, it’s unclear whether oxidative stress is a cause, consequence, or downstream effect of other pathophysiological processes. Antioxidant-based treatments remain experimental.

Neuroinflammation may also play key roles (52). Elevations in pro-inflammatory cytokines (e.g., IL-6, TNF-α), activated microglia, and altered immune gene expression have been observed in patients. Epidemiological studies link prenatal infections and autoimmune conditions with increased schizophrenia risk. Inflammation may disrupt neurodevelopment, synaptic pruning, and neurotransmitter systems such as dopamine and glutamate. However, the causal role of inflammation remains uncertain, as findings are inconsistent and may reflect secondary effects. Anti-inflammatory treatments show promise in subgroups but have not consistently improved symptoms across the broader population.

Toward a pluralistic model

While the dopamine final common pathway” model has yielded valuable insights, it may not fully capture the biological complexity of SRD. We propose a pluralistic model recognizing multiple, partially independent neurochemical and cellular pathways converging toward overlapping clinical phenotypes (Figure 1). This framework accepts causal diversity and interaction—where distinct pathophysiological processes can each perturb common neural circuits related to salience, cognition, and self-processing. Rather than assuming hierarchical causation, this model suggests that SRDs represent a family of related pathophysiological routes rather than a single neurochemical endpoint. This model has critical implications for development of precision treatments.

Figure 1.

Pathways to schizophrenia and related disorders.

Dopamine dysregulation may be only one of many pathways to SRD

The view that dopamine is a final common pathway for SRD is supported by a) all approved antipsychotics to date, possibly including the xanomeline–trospium combination, directly or indirectly modulate dopamine neurotransmission; b) many implicated neurotransmitter systems—glutamate, GABA, serotonin, and acetylcholine—modulate dopaminergic signaling (Table 1); and c) no non‑dopaminergic interventions has shown efficacy across symptom domains in schizophrenia.

Table 1.

Most pathophysiological mechanisms in schizophrenia impact dopamine

Factor	Effect on Dopamine	Key Reference
Glutamate (NMDA hypofunction)	Disinhibits cortical excitatory output → increases subcortical DA release	Lisman et al., 2008
GABA (Interneuron dysfunction)	Reduces inhibition → disinhibits excitatory drive to DA neurons	Lewis and Moghaddam 2006; Enomoto et al 2011
Serotonin (5-HT2A modulation)	Modulates glutamate → alters DA tone in cortex and striatum	Meltzer & Massey, 2012
Acetylcholine (Nicotinic/Muscarinic)	Modulates cortical circuits → indirectly regulates DA	Tregellas et al., 2014
Oxidative Stress	Damages GABA interneurons & DA neurons → alters DA regulation	Do et al., 2009
Inflammation (Cytokines)	Cytokines alter DA metabolism and glial support of DA signaling	Khandaker et al., 2015
Bioenergetics (Mitochondrial dysfunction)	Disrupts energy supply to DA neurons and top -down cortical control	Shao et al., 2008

However, this dopamine‑centric framework has notable limitations. Psychotomimetic agents such as phencyclidine (an NMDA antagonist), and LSD (a 5‑HT2A agonist) as well as anticholinergic drugs cause psychosis like conditions which are poorly responsive to antipsychotics (15; 60). Moreover, dopaminergic blockade offers little benefit for cognition (61) and only modest effects on negative symptoms (62). Xanomeline shows promise for cognitive enhancement, but these findings require replication (63).

A pluralistic model may better fit neurochemical heterogeneity in SRD

Accumulating evidence indicates neurochemically distinct subtypes within SRD. Patients with elevated presynaptic dopamine synthesis typically display prominent positive symptoms and respond well to D2 antagonists (64). In contrast, individuals with glutamatergic abnormalities—such as NMDA receptor hypofunction or astrocytic glutamate dysregulation—often exhibit cognitive impairments, negative symptoms and treatment resistance (65); psychosis in NMDA encephalitis does not respond to antipsychotics (15). Cholinergic deficits, particularly involving muscarinic receptors, may define another subgroup characterized by attentional and sensory‑gating abnormalities (66). These observations bolster the rationale for mechanism‑targeted treatments.

A multifactorial, multifinality framework, (used categorically, dimensionally or both) may better capture SRD’s heterogeneity. Diverse upstream insults—genetic, inflammatory, mitochondrial, oxidative, or neurodevelopmental—interact with neurotransmitter systems and neural circuits at different developmental stages, producing varied clinical phenotypes through overlapping yet distinct mechanisms (Figure 1). The knowledge graph (Figure 2) illustrates this complexity, showing how multiple neurochemical and cellular pathways can culminate in different outcomes, not all of which center on dopamine. While there may be several biases in generation of such knowledge graphs using AI, it provides a visual illustration of the incomplete nature of our understanding, and the several knowledge gaps.

Figure 2.

A knowledge graph from ChatGPT to represent the current understanding of the etiopathology of schizophrenia. Each construct is a node, and edges represent the causal relationships between these constructs. The weight of the edges represent the strength of the evidence. Search history was turned off to ensure agnostic search. List of source references will be available upon request.

Knowledge gaps and research priorities

Significant translational gaps remain between mechanistic insights and effective treatments. While non-dopaminergic hypotheses have broadened our conceptual framework, these insights have yet to yield robust, widely adopted non-dopaminergic therapies. Several factors may contribute to this translational bottleneck (67).

Heterogeneity.

SRDs are phenotypically, neurochemically and etiologically heterogeneous, and this may point to different treatment approaches (Table 2). Elevated synaptic dopamine is predictive of good antipsychotic treatment response in antipsychotic-naïve schizophrenia patients (64). On the other hand, those with normal DA synthesis capacity have higher glutamate levels and may respond to clozapine; their symptom reductions are correlated with reductions in glutamate levels (65).

Table 2.

Neurochemical heterogeneity of psychosis

SUBTYPE OF PSYCHOSIS	TARGETED INTERVENTION
• Parkinson Psychosis	• Pimavanserin
• Amphetamine psychosis	• Dopamine antagonists
• Anticholinergic psychosis/delirium	• Physostigmine
• Treatment responsive SRD with increased dopamine in striatum	• Dopamine antagonists
• Anti-NMDA psychosis	• IV immunoglobulins, steroids, plasmapheresis

Several genetic disorders are associated with psychosis, including 22q deletion syndrome, porphyria, Wilson’s disease, etc (68). Genetic studies suggest several glutamate related genes linked to schizophrenia ⁽69); a lack of benefit from glutamatergic modulation in many patients may be because their illness is not primarily driven by this pathway. Longitudinal heterogeneity is also a significant challenge for characterizing pathophysiology, as neurochemical signatures may vary over the course of illnesses (70⁾. While a glutamatergic drug may be more effective early in the illness (19), dopamine antagonists are effective both early and later in the illness. Longitudinal investigations to chart neurochemical trajectories from prodrome through chronic stages are therefore critically needed.

Heterogeneity is not simply an excuse for inconsistent, weak, or poorly replicated findings, but offers an opportunity explain observed variance and thereby develop more specific interventions (1).

Need for conceptual clarity and formulation of explicitly testable models

A key barrier to translational success in SRD is our lack of conceptual clarity associated with formulation of weakly testable models; models must explain a certain body of “established” facts but also generate a set of testable predictions (71). Suboptimal studies fail to develop effective treatments or to disconfirm the model. Mechanistic models need to specify exactly what aspect/s or dimension/s of the syndrome and at what stage/s of the illness that mechanism is relevant. That would enable proper subject selection and study design, rather than a post-hoc explanation of why the study’s findings failed to conform to the model.

Lack of actionable biomarkers

There are no actionable biomarkers in psychiatry despite decades of research (72). Greater investment in biomarker discovery is needed to guide personalized interventions in psychosis. Clinical trials should prioritize biologically defined subgroups rather than syndromic diagnoses alone. Pharmacodynamic target engagement biomarkers used early in treatment development, can help identify compounds likely to succeed in phase 3 trials. Examples of potential biomarkers to resolve neurochemical heterogeneity in SRD include MRS glutamate and GABA levels, mismatch negativity, and auditory steady state gamma activity, PET dopamine transporter availability and neuromelanin-senstive MRI, PET cholinergic receptor availability for acetylcholine, and MRS measures of GABA.

Clinical trial limitations

Methodological issues contribute to the lack of demonstrated efficacy, including short trial durations, small sample sizes, high placebo response rates, and the frequent inclusion of chronic, treatment-resistant patients who may be less responsive to any intervention. Interestingly, while pomaglumetad, a glutamate metabotropic allosteric modulator was ineffective in an overall schizophrenia sample, a post-hoc analysis showed evidence of efficacy in early course patients (19). Gill et al (73) have shown that prior antipsychotics, which may cause dopaminergic supersensitivity, prevent response to novel drugs in the methylazoxymethanol acetate (MAM) model of schizophrenia. Thus, testing novel compounds on chronically-treated schizophrenia patients may be ineffective, suggesting that studies of antipsychotic naïve patients, or those treated with drugs unlikely to cause DA supersensitivity such as aripiprazole, may be informative.

Regulators will also need to reassess how drugs are evaluated. Currently, antipsychotics are approved if they result in greater improvement than a comparator – usually placebo – in the symptoms of acute schizophrenia. Psychopathological assessments typically focus on positive symptoms with inadequate coverage of negative symptoms, cognition, functioning and quality of life. Thus, studies comparing drugs with different mechanisms may miss important differences.

Innovative trial designs, including adaptive and enrichment strategies, may improve detection of signal in heterogeneous populations. Finally, translational models incorporating human stem cell-derived neurons, in vivo imaging, and computational simulations can improve the predictive validity of early-phase research.

Clinical trials have tended to prioritize syndromic eligibility over mechanism-based enrichment. The lack of biomarker-driven patient selection likely dilutes potential treatment effects in clinical trials, as only a subset of patients may have the relevant abnormalities for a given pathophysiological target. Finally, the pathophysiology of schizophrenia likely involves complex interactions between glutamatergic, dopaminergic, and GABAergic systems, as well as other non-neurotransmitter mechanisms, and targeting a single pathophysiological process may be insufficient to address the full spectrum of symptoms

Moving on from what we know

While substantial evidence supports dopaminergic hyperactivity in the associative striatum as a final common pathway for core psychotic symptoms across various conditions, it fails to account for several aspects of SRD such as treatment resistant psychosis, negative symptoms and cognitive deficits. This translational gap requires a shift from dopamine-centered paradigms to pluralistic, circuit-based models that acknowledge and resolve heterogeneity, underlying the full spectrum of symptoms in SRD. Such a model will also need to accommodate psychological models; for example, it has suggested that treatment resistance in some cases may not be neurochemical but rather results from accumulation of spurious autobiographical memories formed during previous acute psychotic episodes (74)

Transdiagnostic frameworks such as Research Domain Criteria (75) and the Bipolar Schizophrenia Network for Intermediate Phenotypes (BSNIP) Biotypes (76) complement a pluralistic model by shifting the focus from symptom-based categorical diagnoses to better reflect the distributed, overlapping neurobiological pathways implicated in psychosis. Aligning pluralism with revised nosology can be transformational, and map specific neurochemical or circuit-level disturbances to broader domains of functioning—such as cognitive control, reward processing, or social cognition—thereby bridging molecular diversity with clinical heterogeneity.