Neural mechanisms and circuit foundations of conscious and unconscious dysfunction in schizophrenia

Abstract

Schizophrenia is associated with marked abnormalities in consciousness, specifically in the interplay between implicit and explicit cognitive processing. However, whether these deficits originate from early sensory degradation or late-stage access failures remains heavily debated. This narrative review synthesizes psychophysical, electrophysiological, and neuroimaging literature to evaluate these contrasting models. Visual masking and mismatch negativity paradigms suggest abnormalities extending from preattentive stages to conscious access. Metacognitive impairments in error and source monitoring further suggest that implicit processing may fail to achieve stable explicit representation. These abnormalities have been linked to predictive coding errors, thalamocortical dysconnectivity, and aberrant neural synchrony. This integrated framework seeks to connect micro-level sensory deficits with macro-level psychotic phenomenology and may help guide future transdiagnostic neuromodulatory interventions.

Key Words: Schizophrenia; Conscious access; Unconscious processing; Neural mechanisms; Circuit basis

Core Tip: Schizophrenia features a stage-dependent dissociation between conscious and unconscious processing, challenging the simple preserved-implicit vs impaired-explicit dichotomy. While basic subliminal extraction often remains intact, patients exhibit profound deficits in preattentive sensory gating, mismatch negativity, late-stage conscious access, and metacognitive self-evaluation. These abnormalities stem from disrupted large-scale neural synchronization, thalamocortical dysconnectivity, and impaired predictive coding. By mapping these specific circuit-level dysfunctions, this review provides a mechanistic framework linking foundational sensory deficits to clinical psychotic phenomenology, highlighting novel targets for transdiagnostic neuromodulation.

INTRODUCTION

Schizophrenia affects millions of people globally, representing a leading cause of psychiatric disability and markedly disrupting human cognition[1-3]. The conceptualization of this severe psychiatric disorder has undergone significant transformations since its early descriptions by Emil Kraepelin and Eugen Bleuler. While early psychiatric taxonomy studies focused heavily on the overt manifestations of psychosis, such as hallucinations, delusions, and severe formal thought disorder, modern cognitive neuroscience has allowed the transition toward understanding the foundational neurocognitive architectures that underlie these clinical phenomena[4-6]. Within this paradigm shift, an important area of inquiry has emerged concerning the structural and functional relationship between implicit and explicit processing in schizophrenia-spectrum disorders[7]. The fundamental scientific question is not merely whether individuals with schizophrenia exhibit generalized cognitive decline but also whether neural information can be processed sufficiently to shape behavior, autonomic responses, and lower-level neural activity while simultaneously failing to become consciously reportable, confidently accessible, or accurately attributed to its proper source[8-10]. To navigate this complex literature, rigorous operational definitions are imperative[11]. For the purposes of this review, implicit processing refers to the cascade of neural and cognitive operations that occur outside the stable explicit report. This processing modulates behavioral responses and neurophysiological states without requiring stable reportable access[12-14]. Paradigmatic examples within the experimental literature include subliminal semantic and visual priming, automatic preattentive deviance detection [e.g., as indexed by event-related potentials (ERPs)], unconscious conflict monitoring, and subthreshold affective biases in social cognition[15-18]. Conversely, explicit processing encompasses the domain of consciously reportable experiences and higher-order reflective cognition[19,20]. This includes the subjective visibility of a sensory target, the conscious detection and evaluation of a behavioral error, the subjective confidence associated with a perceptual decision, and the deliberate monitoring of whether a thought or memory originated internally or was derived from external sources (reality and source monitoring)[21]. In this review, implicit processing refers broadly to behaviorally relevant processing that does not require explicit reporting, unconscious processing refers more narrowly to processing that occurs without subjective awareness in a given task, and preattention processing refers to operations that can occur before or without voluntary attention, such as sensory gating or mismatch responses[17,20]. These terms overlap in some paradigms but are not interchangeable. Similarly, conscious access here refers to the transition by which information becomes stably reportable and available for flexible use rather than to consciousness in the broadest philosophical or clinical sense[8,15,19]. However, despite extensive theoretical models linking these processing abnormalities to psychotic phenomenology, their clinical application remains limited[4,8,22]. Three primary challenges contribute to this gap in the current literature. The primary limitation lies in the inconsistent terminology surrounding “awareness”, which is highly polysemous and often generates significant conceptual confusion. It is vital to distinguish trial-level perceptual or cognitive awareness—the momentary conscious access to a specific stimulus or cognitive error—from clinical insight, which refers to a patient's global awareness of their illness, its symptoms, and the need for treatment[23]. While some recent theoretical frameworks attempt to map momentary access failures onto global insight deficits, relying on blurred definitions fails to reflect the precise underlying neurobiological mechanisms[7,15]. The complex cognitive outcomes associated with schizophrenia-spectrum disorders create confusion regarding the true nature of the deficit[3,5,24]. Early cognitive theories often posited a strict, categorical “preserved-implicit/impaired-explicit” dichotomy, assuming that basic sensory processes were completely intact while higher-order conscious access was universally broken[17,20]. Conversely, other models have proposed a generalized cognitive deficit hypothesis in which all processing stages are uniformly degraded[25]. The current evidence does not fully support either extreme. Instead, the empirical evidence reveals a highly stratified and task-dependent architecture[15]. Certain modalities of implicit processing remain remarkably robust, whereas others are characterized by increased neural noise and inefficiency[8,9]. Concurrently, the mechanisms responsible for explicit access and metacognitive evaluation appear to be systematically destabilized. Finally, the neural mechanisms driving these dissociations across different sensory domains may not universally apply[15]. Variations in experimental paradigms, ranging from visual masking to auditory mismatch negativity (MMN), complicate the integration of findings[1,26-29]. It remains unclear whether the failure to translate implicit signals into explicit awareness is driven by early sensory degradation, late-stage amplification failure, or a disruption in large-scale network connectivity[15,30,31]. Studying the boundary conditions under which implicit processing succeeds but explicit access fails may therefore offer targeted insight into the neurobiology of psychosis[8]. This minireview focuses on human studies of conscious access, subliminal and preattention processing, metacognition, self-monitoring, and large-scale network dysfunction in patients with schizophrenia, with the selective inclusion of computational and circuit-level work that helps interpret these findings mechanistically. This article emphasizes visual masking and related psychophysical paradigms, MMN and related electrophysiological markers, metacognitive and source monitoring tasks, and neuroimaging studies of the frontoparietal, salience, default mode, and thalamocortical systems. It is intended as a focused narrative synthesis rather than an exhaustive systematic review[32-37]. Taken together, the available evidence suggests that schizophrenia may involve difficulty in transforming sensory or internally generated signals into stable explicit representations, but this difficulty is likely to be stage dependent and task dependent rather than uniform. Therefore, the present review considers whether different behavioral and neurophysiological paradigms converge on selective abnormalities in conscious access, preattention prediction, metacognitive evaluation, and circuit-level integration.

THE DISSOCIATION BETWEEN IMPAIRED CONSCIOUS ACCESS AND PRESERVED UNCONSCIOUS PROCESSING IN SCHIZOPHRENIA

Schizophrenia is often characterized by a dissociative pattern in cognitive function: Patients may show marked deficits in conscious reporting and reflective processing of stimuli, whereas their capacity to process subliminal information remains measurable in selected paradigms[8,18,38]. The Global Neuronal Workspace (GNW) theory offers a useful framework for conceptualizing this dissociation[13,19,39]. This theory posits that conscious access depends on a distributed prefrontal-parietal network[27,40]. This framework proposes that conscious access depends on distributed frontoparietal processing and large-scale amplification, sometimes described as global ignition, through which information becomes reportable and available for flexible cognitive use[12,13,15,19,41].

Ubiquitous impairment of conscious access and its core mechanisms

In psychophysical paradigms such as visual masking and attentional blink, individuals with schizophrenia often demonstrate an elevated threshold for conscious perception[10,31,42]. Neurophysiological evidence is consistent with impairment in late access-related amplification[12]. ERP studies have revealed a marked reduction in the amplitude of the late P3b component, which is often treated as an electrophysiological marker associated with conscious access[13]. During inattention, early sensory ERP components in patients are often indistinguishable from those in healthy controls, suggesting that the initial bottom-up accumulation of sensory evidence may be relatively preserved[15,43]. The primary deficit resides in a failure of top-down attentional amplification, which prevents information from crossing the threshold of consciousness[8]. Consequently, sensory representations remain stranded in a preconscious state, failing to ignite the global workspace for widespread broadcasting[10,15,17,44]. Moreover, the literature does not support the simple conclusion that all unconscious or early processing is intact because other preattention and implicit operations, such as mismatch responses, habituation, and self-generated sensory attenuation, are often abnormal[33,45,46].

Preservation and complexity of unconscious processing

Research within the GNW framework indicates that subliminal processing capabilities may remain relatively preserved in patients with schizophrenia[7,10,42]. For instance, under stringent masking conditions, patients exhibit semantic priming effects comparable to those of healthy controls; similarly, in change blindness tasks, patients can generate accelerated saccadic responses toward perceptual changes that they cannot explicitly report[8,18,31]. It is important to emphasize that this preservation of unconscious processing is highly specific and should not be generalized to all forms of unconscious or preattention processing[15,47]. It most plausibly refers to the extraction of semantic or perceptual content from isolated subliminal stimuli under selected conditions. This must be distinguished from more complex, sequential preattentive processing streams—such as sensorimotor gating and the automatic detection of rule violations—which are robustly impaired in patients with schizophrenia[25,48]. Consequently, schizophrenia presents a highly stratified cognitive profile: While basic subliminal processing remains intact, indicating that localized sensory substrates are not globally compromised, the failure of conscious access coupled with deficits in specific preattentive temporal integration collectively defines its pathological landscape[8-10,31].

Stage-dependent selectivity across information streams

The apparent selectivity of schizophrenia cannot be reduced to a simple preserved implicit vs impaired explicit dichotomy. Different tasks place different demands on local synaptic gain, inhibitory precision, temporal integration, and long-range coordination. Visual masking studies suggest that early sensory responses and some forms of subliminal content extraction may remain relatively intact under selected conditions, whereas late amplification associated with conscious reporting is often weaker[10,42]. In contrast, mismatch responses, sensory gating, habituation, and corollary discharge are frequently abnormal, indicating that automatic processing streams that depend on predictive updating or self-generated sensory attenuation are often compromised[33,45,49,50]. This stage dependence becomes more interpretable at the circuit level. The work on MMN and computational modeling is consistent with contributions from N-methyl-D-aspartate (NMDA) receptor-related recurrent signaling and prediction error formation[35,51]. Moreover, oscillatory and circuit studies have implicated GABAergic inhibitory mechanisms and thalamocortical coordination in shaping precision, filtering, and temporal binding[33,52,53]. The selective profile may therefore reflect different circuit requirements across tasks rather than a single preserved or impaired processing class.

Theoretical integration and circuit-level context

The GNW theory attributes this dissociation to a fundamental disruption in global information integration[15]. The disorganization of long-range functional connectivity and neural oscillatory synchrony—particularly gamma-band activity within the prefrontal-parietal network—serves as the direct physiological substrate for this failed integration[12,13,19,27,54]. At the synaptic and circuit levels, the NMDA receptor hypofunction hypothesis offers a plausible mechanistic explanation for this failure in top-down signal amplification and long-range synchronization[8,15], although current evidence favors a more plural and task-dependent account rather than a single settled mechanism[55-57]. In summary, the GNW framework remains useful for organizing evidence on conscious access abnormalities in patients with schizophrenia. The most defensible conclusion is that conscious access is often impaired, some isolated subliminal effects remain measurable, and several preattention processes are abnormal. The main paradigms used to study conscious access and subliminal processing are summarized in Table 1.

Table 1 Core paradigms used to study conscious and unconscious processing in schizophrenia.

Ref.	Paradigm/method	Processing level emphasized	Main use in the review	Main limitation
[10,28,42,43,48]	Visual masking/object substitution	Conscious access threshold; subliminal processing	Distinguishes impaired reportable perception from preserved masked processing in selected paradigms	Sensitive to visual deficits, attention, and task design
[8,10,15]	Attentional amplification paradigms	Top-down contribution to conscious access	Used to discuss reduced amplification and elevated access threshold	Interpretation depends on task demands
[16,42]	Semantic/affective priming	Subliminal or implicit processing	Used to discuss preserved unconscious semantic or affective processing in specific conditions	Does not generalize to all unconscious or preattentive functions
[31]	Change blindness/implicit detection	Explicit report vs covert sensitivity	Used to discuss dissociation of explicit and implicit responses	Often behaviorally indirect
[1,11,21,26,29,38,70]	ERP measures (P50, MMN, P3/P3b, error-related signals)	Preattentive filtering, deviance detection, conscious access, performance monitoring	Used to relate physiological signatures to conscious and unconscious processing	ERP components do not map one to one onto single cognitive operations
[60,76,83,95]	Resting-state fMRI	Intrinsic large-scale organization	Used to discuss baseline network organization and unconscious neural dynamics	Weak specificity for individual cognitive operations
[22,40,83]	Structural connectivity imaging	White matter and network organization	Used to relate structural connectivity to access-related deficits	Relatively indirect for task specific mechanisms
[23,34,54,67,103]	Metacognitive tasks and insight measures	Confidence calibration, self-evaluation, insight	Used to discuss explicit self-monitoring abnormalities	Must control first order performance

ERP: Event-related potential; fMRI: Functional magnetic resonance imaging; MMN: Mismatch negativity; P3/P3b: P3 (P300)/P3b late positive event-related potential component (often linked to conscious access); P50: P50 auditory evoked potential component used in sensory gating paradigms.

CONSCIOUS FUNCTION IN SCHIZOPHRENIA: NEURAL MECHANISMS AND CIRCUIT BASIS

Conscious functions, encompassing explicit memory retrieval, active attentional allocation, and goal-directed executive behavior, are profoundly impaired in patients with schizophrenia[7,8,18,58]. The neurobiological substrates of these deficits are fundamentally tethered to dysfunctional integration within the frontoparietal control network and the concurrent failure to appropriately suppress the default mode network (DMN)[59,60]. The conscious functions discussed in this review and their associated neural systems are summarized in Table 2.

Table 2 Conscious functions in schizophrenia: Domains and neural systems.

Ref.	Functional domain	Main features described in the review	Neural systems discussed
[8,10,15,19,42,43,63]	Conscious access	Elevated threshold for reportable perception, impaired explicit access	Prefrontal-parietal workspace systems, access-related cortical dynamics
[15,61,63]	Explicit memory retrieval	Reduced recollection accuracy, increased reaction time, disproportionate frontal recruitment	DLPFC, VLPFC, hippocampal-prefrontal circuitry
[38,61,62]	Reality monitoring/self-monitoring	Altered explicit monitoring of internally generated information and performance	ACC, medial frontal systems, prefrontal control circuitry
[39,54,61,64]	Conscious social cognition	Altered theory of mind, self-other distinction, explicit social interpretation	mPFC, TPJ, anterior insula, ACC, control/salience systems
[65,75]	Conscious emotion processing	Altered explicit emotional-network engagement and transdiagnostic comparison	Frontal-limbic and large-scale emotion-related systems
[9,12,77,85]	Stream of consciousness/self-related continuity	Fragmented continuity and altered temporal integration	Large-scale temporal organization, self-related network interactions

ACC: Anterior cingulate cortex; DLPFC: Dorsolateral prefrontal cortex; mPFC: Medial prefrontal cortex; TPJ: Temporoparietal junction; VLPFC: Ventrolateral prefrontal cortex.

Prefrontal compensation and pathological activation during explicit memory retrieval

During explicit memory retrieval, patients with schizophrenia frequently exhibit reduced accuracy in recollection-based tasks alongside increased reaction times and a disproportionate recruitment of prefrontal regions, particularly the dorsolateral and ventrolateral prefrontal cortex[61,62]. Rather than reflecting efficient executive support, this hyperactivation may index compensatory effort in response to disrupted hippocampal–prefrontal coordination and weakened contextual reinstatement. For example, source-memory and associative-recognition paradigms commonly show greater frontal activation despite poorer behavioral performance, suggesting that additional prefrontal engagement does not normalize retrieval but may reflect an inefficient and potentially pathological control process[15,63].

Abnormal effective connectivity in conscious social cognition

During conscious social cognition, individuals with schizophrenia often show impaired performance on explicit theory-of-mind, emotion-recognition, and self–other distinction tasks, accompanied by abnormal effective connectivity among the medial prefrontal cortex, temporoparietal junction, anterior insula, and anterior cingulate regions[38,42,61]. Rather than reflecting a focal deficit in any single social-cognitive node, these findings suggest a systems-level disturbance in the dynamic coordination required to consciously interpret intentions, beliefs, and affective cues[9,24,64]. In particular, abnormal top-down regulation from prefrontal control regions and aberrant salience signaling may distort the contextual integration of socially relevant information, leading to misattribution, reduced mental-state inference, and unstable interpersonal judgments[4,12,19,59].

Disease-specific neural deviations in conscious emotion processing

With respect to the conscious processing of emotional information, recent normative modeling studies—anchored by large-scale neuroimaging data from healthy reference populations—have revealed that the neural deviation patterns in patients with schizophrenia differ significantly from those observed in patients with major depressive disorder and bipolar disorder[65]. The macroscopic neural dysfunction characterizing schizophrenia is embedded within highly specific, nonoverlapping emotion-processing networks, and these aberrant topological patterns correlate directly with the severity of affective symptoms. Translating these network-level findings to the cellular level, the dysregulation of excitatory neurons has emerged as a replicable, disease-specific cell-type association underlying these affective deficits in patients with schizophrenia[1,6,54,59]. These transdiagnostic comparisons confirm that while emotional blunting is a transsyndromal clinical feature, the precise neural and cellular architectures governing conscious emotion processing in schizophrenia maintain a distinct degree of neurobiological specificity[65]. The main unconscious and preattentive functions discussed in schizophrenia, together with the related circuit systems, are summarized in Table 3.

Table 3 Unconscious and preattentive functions in schizophrenia.

Ref.	Domain	Main features described in the review	Neural/circuit systems discussed
[7,8,10,18,30,31,42]	Subliminal processing	Relative preservation in selected paradigms	Local sensory and perceptual processing systems
[1,25,26,68]	Sensory gating	Reduced inhibitory filtering of repetitive or redundant input	Auditory cortex, hippocampal-prefrontal circuitry, thalamocortical pathways
[1,25,68,69]	Habituation	Reduced attenuation to repeated sensory stimulation	Sensory cortices, inhibitory networks, thalamocortical systems
[1,26,29,70]	Mismatch negativity	Reduced MMN amplitude and altered automatic deviance detection	Primary auditory cortex and related predictive networks
[4,71,72]	Predictive coding at preattentive level	Altered prediction-error processing and vocalization-related predictive coding	Auditory and action-related predictive systems
[16,64,75]	Implicit social/affective processing	Altered responses to socially or emotionally salient stimuli outside explicit report	Amygdala, insula, frontal regions, fusiform and related systems
[7]	Unconscious motivational processing	Preserved unconscious motivational effects in a specific paradigm	Motivation-related processing without explicit awareness
[60,76,83,95]	Resting-state intrinsic activity	Altered baseline large-scale organization reflecting ongoing unconscious neural processes	Resting-state whole-brain functional architecture

MMN: Mismatch negativity.

Metacognitive dysfunction and explicit self-evaluation

Metacognitive dysfunction warrants separate discussion because it links conscious access to self-evaluation, symptom formation, and functional outcomes. Studies of confidence judgment, source monitoring, introspective accuracy, and insight suggest that schizophrenia is often associated with impaired evaluation of one’s own cognitive processes rather than only impaired first-order performance. At the neural level, this work implicated the anterior cingulate, medial and lateral prefrontal, frontoparietal, and default mode systems[23,34,66]. Importantly, a meta-analysis suggested that metacognitive deficits can be overestimated when first-order task performance is not equated; thus, this literature should be interpreted with methodological caution[67]. Even so, convergent findings support the view that explicit self-monitoring and confidence calibration are often unstable in patients with schizophrenia and may contribute to psychotic misattribution and poor functional outcomes.

UNCONSCIOUS FUNCTION IN SCHIZOPHRENIA: NEURAL MECHANISMS AND CIRCUIT BASIS

Unconscious functioning comprises a heterogeneous array of operations, encompassing implicit learning, automatic perception, and preattentive memory consolidation[9,20,47]. In schizophrenia, this domain is not uniformly preserved or uniformly impaired. Some isolated implicit effects remain measurable in selected paradigms, whereas sensory filtering, habituation, mismatch responses, and self-generated sensory prediction are frequently abnormal[25,26]. The neurobiological substrates governing these automatic operations rely critically on seamless temporal coordination among the thalamocortical circuitry, the limbic system, and primary sensory cortices[12].

Molecular and circuit mechanisms of sensory gating and habituation

Sensory gating and habituation depend on coordinated molecular and circuit mechanisms that suppress redundant input and preserve access to behaviorally relevant signals. In patients with schizophrenia, these processes are frequently disrupted, as suggested by reduced P50 suppression, impaired mismatch responses, and diminished attenuation of repeated auditory or sensory stimuli[25,26,29]. Converging evidence implicates NMDA receptor hypofunction, deficient GABAergic interneuron-mediated inhibition, and abnormalities in cholinergic modulation—particularly involving the alpha-7 nicotinic receptor (α7 receptor) and nicotinic receptors—as key molecular substrates of this deficit. At the circuit level, dysfunction across the hippocampal–prefrontal, thalamocortical, and auditory cortical networks may weaken inhibitory filtering and impair adaptive habituation[1,6,54]. This pattern is consistent with the view that sensory flooding in schizophrenia may arise in part from a failure of inhibitory gain control, whereby repetitive inputs are not properly dampened and are instead granted pathological salience[4,59,68,69].

Preattentive processing deficits and MMN

MMN, a robust ERP component reflecting the automatic detection of rule violations at a strictly preattentive stage, is highly replicable in terms of the reduction in amplitude in patients with schizophrenia[26,29,70]. Neural generation in the MMN has been linked to intact NMDA receptor function within the primary auditory cortex, which is subject to fine-tuning by the dopaminergic system. Pathological hypofunction within these glutamatergic networks directly compromises patients' ability to preattentively detect subtle environmental deviants[1]. Furthermore, recent computational modeling posits that the phase entrainment of chaotic neural oscillations offers a biologically plausible mechanism for automatic deviance detection. In pathological states such as schizophrenia, the disruption of this phase of entrainment precludes the nervous system from implicitly updating its predictive models of the acoustic environment[4,54,71,72]. Recent computational work has further refined this interpretation[73]. Model-based studies suggest that mismatch responses in patients with schizophrenia may reflect changes in effective connectivity and synaptic parameters rather than a unitary deficit, and recent reviews emphasize that connectivity, decoding, and cognitive models capture different aspects of mismatch processing[1,36]. These findings strengthen the value of MMN as a mechanistically informative marker but also argue against overly simple one-to-one interpretation[74].

Aberrant neural responses in implicit social cognition and transdiagnostic commonalities

At the level of implicit social cognition, individuals with schizophrenia exhibit highly anomalous neural responses to negative facial expressions, particularly signals of fear or threat, even when presented below the threshold of conscious awareness. Functional neuroimaging studies have indicated that during implicit facial emotion processing, patients often display paradoxical hyperactivation in the left amygdala, insula, and inferior/middle frontal gyri, reflecting exaggerated, automatic reactivity to subliminal affective salience[20,75]. Conversely, transdiagnostic research comparing schizophrenia with autism spectrum disorder reveals a shared neural deficit during specific implicit negative facial emotion recognition tasks, characterized by blunted activation within the amygdala and the fusiform gyrus. This transdiagnostic neural signature confirms that while these psychiatric conditions present distinct clinical phenotypes, their profound impairments in unconscious social cognition are anchored in shared, phylogenetically older neural substrates[76].

Resting-state brain activity as a window into unconscious neural processes

Resting-state functional magnetic resonance imaging (rs-fMRI) provides an indispensable methodological perspective for investigating the intrinsic, continuous, and unconscious neural dynamics that form the baseline state of the brain[12,15,19]. Advanced analytical approaches, particularly self-supervised deep feature extraction methods, have demonstrated that the resting-state functional architecture in schizophrenia patients contains robust, patient-specific information sufficient for individual identification. These spatiotemporal features capture the highly disrupted intrinsic functional connectivity patterns that define the pathological state[42,60]. Moreover, machine learning classifiers integrating whole-brain functional connectivity matrices, graph theory centrality metrics, and cortical gradient features can effectively distinguish neuropsychiatric disorders such as schizophrenia from a healthy baseline. The success of these topological classifiers demonstrates that despite the staggering heterogeneity of the disease, the pervasive disruption of unconscious, resting-state network organization remains a defining macroscale feature of schizophrenia[54,76].

INTERACTION AND NEURAL CIRCUIT INTEGRATION OF CONSCIOUS AND UNCONSCIOUS FUNCTIONS

Conscious and unconscious functions do not operate in isolation but engage in dynamic interactions through highly integrated neural circuits[15,17,20]. In patients with schizophrenia, the disruption of this interaction may contribute to the fragmentation of conscious experience and the distortion of self-perception[8]. Shared molecular and circuit mechanisms linking the conscious and unconscious domains are shown in Table 4.

Table 4 Neural circuit and molecular mechanisms linking conscious and unconscious functions.

Ref.	Domain	Neural/circuit systems discussed
[8,12,13,15,19,20,41]	Global neuronal workspace/conscious broadcasting	Framework for conscious access and ignition-like large-scale integration
[8,22,54,59]	Dysconnection	Large-scale failure of coordinated integration across neural systems
[4,8,15,25,54,59,71]	NMDA receptor-related dysfunction	Reduced recurrent integration, altered filtering, impaired predictive coding and conscious access
[8,54,77,104]	GABAergic dysfunction/synchrony disturbance	Altered long-range oscillatory coordination and weakened integration
[25,26,68]	Cholinergic modulation	Sensory-gating-related modulation, including nicotinic mechanisms
[1,6,24,76]	Thalamocortical disturbance	Abnormal relay, filtering, and integration of sensory and higher-order signals
[39]	Claustrum	Discussed as a multimodal integration hub
[78-82]	Orbitofrontal latent-state representation	Supports inferred state representation, latent learning, and flexible updating
[13,15,54,77,104]	Neural synchronization/temporal binding	One mechanism linking unconscious processing with conscious integration

NMDA: N-methyl-D-aspartate.

Fragmentation of the stream of consciousness and impaired temporal binding

The neurophenomenological hypothesis proposes that anomalies in the “stream of consciousness” in schizophrenia may be especially evident in the impairment of its “transitional parts”, indicating a failure of the intrinsic temporal binding mechanism that connects disparate conscious contents. This disruption in the continuity of conscious experience may be directly linked to abnormalities in the temporal structure of the resting-state spontaneous neural activity of the brain, reflecting a decoupling between fundamental neurodynamics and the structure of subjective awareness[9,12,77].

The role of the claustrum in multimodal integration and coherent perception

The claustrum, a hub with extensive whole-brain connectivity for multimodal information integration, plays a pivotal role in coordinating conscious and unconscious information flow[39]. Its relatively uniform cytoarchitecture supports a divergent role in integrating incoming information and forming a coherent percept[15]. Studies have indicated that alterations in the shape, structure, and volume of the claustrum observed in individuals with neurodevelopmental disorders such as schizophrenia may directly impact the formation of conscious sensation and motivational processes related to reward behavior.

Orbitofrontal cortex function in latent learning and state recognition

Evidence indicates that the orbitofrontal cortex supports latent learning and state recognition when behavior depends on inferred, rather than directly observable, contingencies[78]. In sensory preconditioning, orbitofrontal cortex function (OFC) disruption impairs behavior and learning when value must be inferred, while it leaves performance relatively intact when cached value is sufficient, which argues against a purely reinforcement-based account[79]. OFC neurons also acquire incidental stimulus–stimulus associations before reward is introduced, which is consistent with encoding associative structure itself[80]. Medial OFC lesions further impair the retrieval of action consequences in partially observable situations, and in humans, hidden task states can be decoded from OFC activity, with stronger decoding predicting better performance[81]. Together, these findings support the hypothesis that the OFC constructs latent-state representations that enable flexible generalization and adaptive updating[78,82].

Neural synchronization as a bridge between conscious and unconscious processes

Neural synchronization is often considered a key physiological mechanism bridging hierarchical levels of conscious and unconscious processing[13]. Theoretical models suggest that the neural correlates of conscious and unconscious mental states may involve parallel but incompletely synchronized activity across neural networks at different levels[12,15,41,77]. In patients with schizophrenia, aberrant neural phase locking between these two mental strata may limit the timely transition of processed information into stable conscious awareness. This possibility remains especially relevant to accounts that link temporal binding and large-scale integration to positive symptoms such as hallucinations and disturbed agency[8,42]. Furthermore, cyclic models of the brain state based on neuromodulator dynamics suggest that the balance between excitatory and inhibitory neurotransmitter concentrations in the cortex regulates the boundary between creative and psychopathological states, with extreme cyclic dynamics potentially leading to a collapse of integrative cognitive function[12].

Structural network segregation and the neural basis of expectancy processes

Research on structural network segregation provides systems-level evidence for understanding these interactions[83]. Graph theory analyses indicate that temporal and spatial expectancy rely on distinct structural connectivity networks. Within these networks, centrality measures of regions such as the left putamen, right caudate nucleus, left frontal operculum, and right inferior parietal lobule are closely related to behavioral performance. This structural segregation likely reflects distinct neural substrates for conscious and unconscious expectancy processes[9,77]. In patients with schizophrenia, the disruption of this fine-grained structure-function relationship may lead to a disruption in the integration of top-down predictive signals with bottom-up sensory evidence, constituting its core predictive processing deficit[59,71,84,85].

Converging and competing frameworks

The GNW framework remains useful for interpreting failures of reportable access and late amplification in masking paradigms, but it is not the only relevant model[19,42,43]. Predictive coding has become especially influential in schizophrenia because it offers a computational language for abnormal precision weighting, prediction error signaling, and hierarchical inference[4,59,84,86-88]. Recurrent processing views place similar weight on feedback and re-entrant signaling but do not require a discrete global ignition mechanism. Integrated information theory shifts the emphasis toward integration and differentiation at the systems level rather than access or reportability[15,19]. At present, no single framework fully explains all the findings in schizophrenia[55,71,76,89]. A balanced view is that several models converge on impaired recurrent integration and large-scale coordination but differ in how they explain the transition from local processing to explicit reporting and belief formation[9,19,41,69,77,90].

A TRANSDIAGNOSTIC PERSPECTIVE: FROM NEURAL MECHANISMS TO INTERVENTION STRATEGIES

The neural mechanisms underlying schizophrenia exhibit significant transdiagnostic characteristics, sharing partially overlapping neural features with those of other neurodevelopmental disorders, such as autism spectrum disorder, bipolar disorder, and attention-deficit/hyperactivity disorder[76]. Adopting this perspective helps elucidate shared pathophysiological pathways across disorders and provides a scientific basis for developing intervention strategies that integrate both universal and targeted approaches[24,59].

Transdiagnostic neural foundations and disease specificity

Across psychiatric disorders, evidence increasingly points to a shared neural architecture centered on the dorsal anterior cingulate, bilateral anterior insula, and large-scale default-mode, frontoparietal, and salience networks[91]. In a transdiagnostic structural meta-analysis of 193 voxel-based morphometry studies comprising 15892 individuals, convergent gray matter loss was localized to the dACC and bilateral insula, suggesting that it is a common substrate for impaired cognitive control and salience regulation[92,93]. Complementing this, meta-analytic connectivity work revealed shared hypoconnectivity and hyperconnectivity among the DMN, salience, and frontoparietal networks across diagnoses, whereas task-functional magnetic resonance imaging analyses of 283 experiments revealed overlapping abnormalities in the prefrontal, insular, and midcingulate control circuitry[91]. However, disease specificity remains detectable: Lower-order sensory–motor dysconnectivity appears disproportionately prominent in patients with schizophrenia compared with patients with mood disorders[76,94]. This pattern supports a model in which transdiagnostic control–salience vulnerability is modified by disorder-specific perceptual and motor network pathology[93]. The shared and relatively distinctive findings across schizophrenia, bipolar disorder, and autism spectrum disorder are summarized in Table 5.

Table 5 Shared and relatively distinctive findings across disorders.

Ref.	Domain	Schizophrenia	Bipolar disorder	Autism spectrum disorder
[10,22,105]	Conscious access	Often impaired in masking and related tasks	Less consistently studied in this specific form	Not central in the same psychosis-oriented literature
[33,111-113]	Preattentive deviance detection	Robust MMN abnormalities	Abnormalities reported, often less central	Atypical prediction and sensory processing are also reported
[76,114,115]	Thalamocortical dysconnectivity	Prominent and repeatedly implicated	Present but often less sensory weighted	Sensory and social differences are common but framed differently
[23,67,116]	Metacognition and insight	Confidence, source monitoring, and insight often disturbed	Insight varies with mood state	Self-monitoring difficulties differ in structure and clinical meaning

MMN: Mismatch negativity.

Brain state classification and the development of transdiagnostic biomarkers

Machine learning-based brain state classification studies have opened new avenues for developing transdiagnostic biomarkers[95]. Models integrating functional connectivity, graph theory metrics, and cortical gradient features can distinguish between unconscious states, psychedelic states, and various neuropsychiatric disease states with high mean balanced accuracy, indicating that these distinct brain states possess quantifiable and discriminable neural signatures. Notably, the transferability of these classification models across different datasets varies, and their performance is highly dependent on the specific state and the feature set employed. These findings suggest that future biomarker development may require customizing classification schemes for specific clinical phenotypes or states[3,76,95].

Neuromodulation intervention strategies targeting cognitive circuits

Neuromodulation interventions may exert their effects by targeting the interaction between conscious and unconscious processes[12,13,15,19,20]. For instance, dynamic causal modeling of eye gaze processing has revealed that abnormal effective connectivity from the posteromedial prefrontal cortex to the posterior superior temporal sulcus in schizophrenia patients is closely associated with social functional impairment and symptom severity. These specific connectivity patterns could serve as potential targets for interventions combining neuroregulation techniques with social cognitive training, aiming for precise correction of aberrant circuits[6,24,54,59]. Furthermore, clinical translation research based on MMN indicates that while MMN holds great promise for detecting preattentive processing abnormalities, optimizing recording algorithms and establishing standardized protocols are necessary for it to become a stable and valuable tool for clinical investigation in psychopathology[1,26]. Neuromodulation findings are symptom specific rather than uniformly positive. With respect to medication-resistant auditory verbal hallucinations, a meta-analysis revealed no significant pooled benefit of either repetitive transcranial magnetic stimulation (rTMS) or transcranial direct current stimulation (tDCS) over sham therapy[96]; in contrast, for negative symptoms, excitatory left dorsolateral prefrontal protocols appear more promising, with network meta-analytic support for high-frequency rTMS and anodal tDCS[97] and a more recent meta-analysis showing a modest benefit of tDCS, especially when negative symptoms were the primary treatment target[98].

Insights from placebo effects and the neural mechanisms of psychosocial interventions

Placebo research has demonstrated that the psychosocial context exerts measurable neurobiological effects rather than serving as a mere epiphenomenon[99]. In pain paradigms, expectancy increases prefrontal activity while attenuating responses in the thalamus, insula, and anterior cingulate cortex, with placebo analgesia additionally recruiting rostral anterior cingulate–periaqueductal gray coupling as a top-down antinociceptive mechanism[99,100]. Related work has shown that anterior prefrontal regions can suppress ventral striatal prediction–error signaling, thereby stabilizing treatment expectations[101]. In patients with depression, placebo-associated improvement is correlated with μ-opioid activation in the subgenual cingulate, nucleus accumbens, thalamus, and amygdala[102]. Additionally, psychosocial interventions such as CBT modify anterior cingulate cortex (ACC)-prefrontal circuitry across disorders and normalize amygdala–dorsolateral prefrontal cortex/ACC connectivity in psychosis[103]. Together, these findings suggest that psychosocial treatments act, in part, through expectancy, learning, and top-down regulatory mechanisms first clarified by placebo neuroscience[102].

The self-awareness neural circuit as a transdiagnostic intervention target

Research on the neural mechanisms of self-awareness offers higher-order cognitive targets for transdiagnostic intervention. Studies indicate that a specific paralimbic network demonstrates activity specificity and causal involvement in self-awareness[39]. Regions within this network interact primarily through gamma-band synchronization[8,54,104]. This gamma synchrony develops progressively from infancy through childhood and adolescence into adulthood and is finely regulated by neurotransmitters such as dopamine via GABAergic interneurons. In developmental disorders such as autism, attention-deficit/hyperactivity disorder, and schizophrenia, the functional integrity of this network and the development of self-awareness significantly differ[39,60,85]. Therefore, interventions targeting this neural network synchrony may provide novel avenues for ameliorating shared deficits in self-integration and self-regulatory functions across multiple disorders[54,59].

FUTURE PROSPECTS AND OUTLOOK

Recent research has begun to clarify the neural mechanisms and circuit bases underlying conscious and unconscious functions in patients with schizophrenia[8,105]. Abnormalities in conscious function have often been linked to dysregulated activity in the prefrontal-parietal and DMNs, whereas deficits in unconscious function have often been linked to thalamocortical circuits, the limbic system, and sensory cortices[15]. The interaction between these domains may involve structures and mechanisms such as the claustrum, orbitofrontal cortex, and neural synchronization, and an imbalance in this interaction may lead to the fragmentation of conscious experience and distortions in self-perception. From a transdiagnostic perspective, schizophrenia shares partially overlapping neural features with conditions such as autism spectrum disorder and bipolar disorder, yet each disorder retains a unique neural signature. Normative modeling and machine learning classification studies offer promising new directions for developing transdiagnostic biomarkers, while the combination of neuromodulation and psychosocial interventions has the potential to produce therapeutic effects by targeting the interactive mechanisms of conscious and unconscious functions[76,95]. Future research should delve into the following directions: First, more precise neurocomputational models that integrate multiscale neurobiological data should be developed to comprehensively dissect the interactive mechanisms of conscious and unconscious functions from the molecular level to the circuit level[15,19,47,76]. Second, intervention strategies based on brain state classification require further validation of their specificity and transferability, particularly their potential for application across transdiagnostic populations. Finally, interventions targeting neural network synchrony, such as neurofeedback and transcranial magnetic stimulation, may open new avenues for improving the integration of consciousness in individuals with schizophrenia[54]. In summary, the study of conscious and unconscious functions in patients with schizophrenia is undergoing a paradigm shift from correlational analyses to the exploration of causal mechanisms[7,8]. This shift provides new perspectives for understanding the development of self and its disruption in neurodevelopmental disorders[9,85]. By integrating multidisciplinary approaches, future research can provide more effective intervention strategies for schizophrenia and other psychiatric conditions[1,5,24].

METHODOLOGICAL COMPLEMENTARITY AND LIMITS

The main paradigms discussed in this review are complementary but not interchangeable. Visual masking is useful for studying access thresholds and late amplification, but performance may be influenced by general visual deficits, attention, and task demands[28,30,106-108]. MMN and related electrophysiological markers provide strong temporal resolution and translational value, but they reflect composite processes whose interpretation depends on paradigm design and deviant type[74]. Metacognitive paradigms are essential for studying explicit self-evaluations, yet their interpretation can be distorted if first-order performance is not equated between groups[67]. Future work will benefit from multiparadigm designs that combine psychophysics, electrophysiology, computational modeling, and neuroimaging among the same participants.

CONCLUSION

The available evidence does not support a simple preserved implicit vs impaired explicit dichotomy in patients with schizophrenia[9,109]. Rather, it suggests a graded and domain-dependent pattern affecting conscious access, preattention prediction, self-monitoring, and metacognitive evaluation[15,17,43]. Some isolated subliminal effects remain measurable in selected paradigms, whereas mismatch responses, sensory gating, habituation, and corollary discharge are often abnormal[15,21,85]. Across these domains, convergent work has linked schizophrenia to altered recurrent amplification, predictive updating, inhibitory control, and thalamocortical and large-scale network disconnection[54,76,110].

ACKNOWLEDGEMENTS

The authors express their gratitude to the Affiliated Mental Health Center of Jiangnan University for its institutional support.