Clinical and cognitive insight

Early models treated insight in schizophrenia as a single construct, but extensive research indicates at least two distinct facets. Clinical insight involves acknowledging mental illness, identifying psychotic experiences as pathological, and recognizing the need for treatment (Lysaker et al., Reference Lysaker, Chernov, Moiseeva, Sozinova, Dmitryeva, Alyoshin and Kostyuk2021). It is typically measured using clinician-rated tools such as the Schedule for the Assessment of Insight – Expanded (SAI-E) and Scale to Assess Unawareness of Mental Disorder (SUMD) (Amador & Kronengold, Reference Amador, Kronengold, Amador and David2004; David, Buchanan, Reed, & Almeida, Reference David, Buchanan, Reed and Almeida1992).

Cognitive insight reflects a metacognitive stance toward personal beliefs. Beck and colleagues developed the Beck Cognitive Insight Scale (BCIS), which assesses Self-Reflectiveness (the tendency to question one’s interpretations) and Self-Certainty (confidence in one’s judgments) (Beck et al., Reference Beck, Baruch, Balter, Steer and Warman2004). Lower self-reflectiveness, especially when paired with higher self-certainty, reliably predicts poorer clinical insight, increased delusional conviction, and reduced medication adherence (Engh et al., Reference Engh, Friis, Birkenaes, Jónsdóttir, Ringen, Ruud and Andreassen2007; Pedrelli et al., Reference Pedrelli, McQuaid, Granholm, Patterson, McClure, Beck and Jeste2004; Riggs, Grant, Perivoliotis, & Beck, Reference Riggs, Grant, Perivoliotis and Beck2012). Thus, self-reflection emerges as necessary but insufficient for complete insight.

Differential links to MTT. In this review, clinical insight refers to clinician-rated awareness of illness/symptoms/need for treatment (e.g., SAI-E, SUMD), whereas cognitive insight refers to metacognitive self-evaluation (e.g., BCIS Self-Reflectiveness and Self-Certainty) (Beck et al., Reference Beck, Baruch, Balter, Steer and Warman2004). Where studies reported both, mental time travel (MTT) measures tended to show more consistent associations with clinical insight than with cognitive insight, although BCIS Self-Reflectiveness often aligns with greater episodic specificity and future-event detail (Beck et al., Reference Beck, Baruch, Balter, Steer and Warman2004).

Relation to metacognition. Cognitive insight, operationalized by the Beck Cognitive Insight Scale (BCIS), reflects a metacognitive appraisal of one’s own interpretations – indexed by Self-Reflectiveness and Self-Certainty (Beck et al., Reference Beck, Baruch, Balter, Steer and Warman2004). Metacognition is a broader construct that includes self-reflectivity and the ability to use that understanding to guide behavior (mastery) and to take perspectives beyond the self (decentration), as indexed by instruments such as the Metacognition Assessment Scale–Abbreviated (MAS-A) (Lysaker et al., Reference Lysaker, Carcione, Dimaggio, Johannesen, Nicolò, Procacci and Semerari2005; Lysaker et al., Reference Lysaker, Chernov, Moiseeva, Sozinova, Dmitryeva, Alyoshin and Kostyuk2021; Semerari et al., Reference Semerari, Carcione, Dimaggio, Falcone, Nicolò, Procacci and Alleva2003). Thus, cognitive insight overlaps with the self-reflective facet of metacognition but does not encompass other metacognitive capacities (e.g., mastery), which have shown independent associations with clinical insight (Bröcker et al., Reference Bröcker, Bayer, Stuke, Giemsa, Heinz, Bermpohl and Montag2017; Lysaker et al., Reference Lysaker, Carcione, Dimaggio, Johannesen, Nicolò, Procacci and Semerari2005).

Autobiographical memory and self-reflection

A coherent sense of self is supported by a reservoir of autobiographical memories (AM), encoding details of what happened, the context in which events occurred, and their significance for personal goals. Vivid, detailed recollections serve as reference points for evaluating current beliefs, imagining counterfactual alternatives, and updating self-knowledge (Fivush, Reference Fivush2011). In schizophrenia, AM is often characterized by ‘over-generalization,’ with patients recalling events only in broad, gist-like terms that lack specific temporal and spatial details – a pattern linked to impairments in executive function and affect regulation (Berna et al., Reference Berna, Göritz, Schröder, Martin, Cermolacce, Allé, Danion, Cuervo-Lombard and Moritz2016; Herold, Lässer, & Schröder, Reference Herold, Lässer and Schröder2023; Mediavilla et al., Reference Mediavilla, López-Arroyo, Gómez-Arnau, Wiesepape, Lysaker and Lahera2021). Because self-reflection fundamentally relies on detailed personal evidence, a compromised AM store restricts the ability to reality-test psychotic interpretations and integrate current experiences into a coherent life narrative.

Mental time travel (MTT)

Mental time travel (MTT) is the capacity to vividly re-experience past events and pre-experience plausible future scenarios by flexibly recombining episodic details into coherent mental scenes (Suddendorf & Corballis, Reference Suddendorf and Corballis2007). Experimental paradigms reveal three consistent deficits in schizophrenia. First, during episodic memory tasks such as the Autobiographical Interview, patients generate significantly fewer internal details (who was present, what occurred, where, and when) compared to controls, despite similar overall verbosity (Berna, Potheegadoo, et al., Reference Berna, Potheegadoo, Aouadi, Ricarte, Allé, Coutelle, Boyer, Cuervo-Lombard and Danion2016; Danion, Huron, Vidailhet, & Berna, Reference Danion, Huron, Vidailhet and Berna2007). Second, in future simulation paradigms like Scene Construction or Prospective Thinking tasks, patients typically produce vague, generic, and less vivid imagined futures, lacking concrete goals (e.g., ‘I might do something outside’) compared with detailed, goal-oriented scenarios offered by controls (e.g., ‘Next Saturday I’ll meet Tom at the café at 2 p.m. to plan our project’) (Raffard et al., Reference Raffard, D’Argembeau, Lardi, Bayard, Boulenger and Van Der Linden2009; Raffard et al., Reference Raffard, D’Argembeau, Lardi, Bayard, Boulenger and Van der Linden2010). Third, temporal-horizon assessments such as the Temporal-Extension Mental Time Travel (TEMT) task indicate a marked ‘foreshortened future’ bias, with patients typically envisioning only the near-term future (days or weeks ahead), whereas controls project months or years ahead; a recent meta-analysis confirms a moderate-to-large pooled effect for this narrowed projection (Amadeo et al., Reference Amadeo, Escelsior, Esposito, Inuggi, Versaggi, Marenco and Gori2024; Casadio et al., Reference Casadio, Patané, Candini, Lui, Frassinetti and Benuzzi2024). Together, these findings suggest that schizophrenia involves impaired event construction in both past and future contexts and a narrowed temporal scope, limiting the autobiographical reference points necessary for accurate self-reflection and insight.

Metacognition: the integrative hub

Metacognition refers to the ability to reflect on, monitor, and flexibly manage one’s own and others’ mental states. In schizophrenia, broad metacognitive impairments – especially in self-reflectivity (understanding one’s own thoughts) and mastery (using that understanding to respond adaptively) – are now considered core features of the disorder (Flavell, Reference Flavell1979). These deficits predict poorer clinical and cognitive insight, independent of neurocognitive performance or symptom severity, and have been shown to partially mediate the relationship between overgeneral autobiographical memory and reduced illness awareness (Davies & Greenwood, Reference Davies and Greenwood2020; Mediavilla et al., Reference Mediavilla, López-Arroyo, Gómez-Arnau, Wiesepape, Lysaker and Lahera2021). In this framework, cognitive insight (BCIS) is treated as a metacognitive appraisal subdomain (self-reflection/self-certainty), whereas metacognition additionally includes mastery and related capacities not captured by BCIS, helping to explain distinct patterns of association with clinical insight.

Neuroimaging studies reveal that metacognitive self-reflection recruits the medial prefrontal cortex, precuneus, and posterior cingulate – key nodes within the default-mode network (DMN), a system in the brain that becomes active during rest, self-focused thinking, remembering the past, and imagining the future (Buckner, Andrews-Hanna, & Schacter, Reference Buckner, Andrews-Hanna and Schacter2008). Disruptions in this network are common in schizophrenia and may explain overlapping problems in metacognition, autobiographical memory, and mental time travel. These same regions also support MTT and show functional disruption in patients with poor insight (Fuentes-Claramonte et al., Reference Fuentes-Claramonte, Martin-Subero, Salgado-Pineda, Santo-Angles, Argila-Plaza, Salavert and Salvador2019; Holt et al., Reference Holt, Cassidy, Andrews-Hanna, Lee, Coombs, Goff and Moran2011; Shan et al., Reference Shan, Liao, Ou, Ding, Liu, Chen and He2020). Thus, metacognition may act as a central mechanism through which impoverished mental time travel simulations (due to degraded autobiographical memory) fail to transform into accurate, treatment-relevant insight.

Cross-sectional findings in established illness

Since 2005, around two dozen studies have shown that poorer performance on mental time travel tasks – such as episodic past recall (e.g., Autobiographical Interview), future-event simulation (e.g., Scene Construction, Future-Event Fluency), and temporal-horizon tasks – is consistently linked to lower insight in individuals with schizophrenia. After controlling for factors like IQ and negative symptoms, correlations between MTT measures and insight typically fall in the small-to-moderate range (r ≈ 0.30–0.45) (Barry, Hallford, Del Rey, & Ricarte, Reference Barry, Hallford, Del Rey and Ricarte2020; Berna et al., Reference Berna, Göritz, Schröder, Martin, Cermolacce, Allé, Danion, Cuervo-Lombard and Moritz2016; MacDougall et al., Reference MacDougall, McKinnon, Herdman, King and Kiang2015; Raffard et al., Reference Raffard, D’Argembeau, Bayard, Boulenger and Van der Linden2010). This relationship holds across diverse samples, including outpatient and forensic populations, and is consistent across multiple measures of insight (e.g., SUMD, SAI-E, BCIS), regardless of whether patients primarily show positive or negative symptoms. When both components were analyzed, associations between MTT measures and clinical insight were generally stronger and more consistent than those with cognitive insight, although higher Self-Reflectiveness frequently accompanied greater autobiographical/future specificity.

Together, these findings suggest that MTT impairments are not just secondary effects or side issues – they are closely tied to the core clinical problem of poor illness awareness.

Early-course and high-risk cohorts

Emerging evidence suggests that the link between MTT and insight is present early in the course of illness and even in individuals at elevated risk. In a study of first-episode patients, Potheegadoo et al. found that distorted perceptions of how close or distant past events felt – along with reduced episodic detail – explained about 11% of the variance in SUMD insight scores (r = .31) (Potheegadoo, Cuervo-Lombard, Berna, & Danion, Reference Potheegadoo, Cuervo-Lombard, Berna and Danion2012). Among high-schizotypy undergraduates, Hazan et al. reported that vague, goal-sparse future narratives and low perceived control in turning-point memories predicted weaker perceived need for help (r = −.28) on the Insight and Treatment Attitudes Questionnaire – ITAQ (Hazan, Reese, & Linscott, Reference Hazan, Reese and Linscott2019).

In an early-course clinical sample, Allé et al. found that fragmented narratives across past and future life stories were linked to higher unawareness of illness (β = −.33) (Allé et al., Reference Allé, d’Argembeau, Schneider, Potheegadoo, Coutelle, Danion and Berna2016). Similarly, Barry et al. showed that in patients within 2 years of diagnosis, generating more specific future events predicted higher BCIS Self-Reflectiveness (r = .39) and lower Self-Certainty (r = −.35) (Barry et al., Reference Barry, Hallford, Del Rey and Ricarte2020). Even in non-clinical populations, less vivid and less detailed ‘delusion-like’ memories predicted stronger conviction in those thoughts (β = −.33) (Berna et al., Reference Berna, Potheegadoo, Aouadi, Ricarte, Allé, Coutelle, Boyer, Cuervo-Lombard and Danion2016). Together, these findings suggest that impoverished scene construction, narrowed time horizons, and disrupted narrative continuity are not late effects of chronic schizophrenia, but early cognitive vulnerabilities that impair the development of accurate insight.

Mediation and moderation analyses

Interim synthesis – strengths, limits, and open questions

Behavioral findings converge on a plausible chain – impoverished MTT performance → weaker metacognitive monitoring → poorer clinical insight – and show that delusional conviction, prominent negative symptoms, and trauma intensify this pathway, whereas depressive mood may blunt it. This proposed pathway is depicted in Figure 1. (Balzan et al., Reference Balzan, Mattiske, Delfabbro, Liu and Galletly2019; Faith et al., Reference Faith, Lecomte, Corbière, Francoeur, Hache-Labelle and Lysaker2020; Luther et al., Reference Luther, Bonfils, Fischer, Johnson-Kwochka and Salyers2020; Vohs et al., Reference Vohs, Lysaker, Francis, Hamm, Buck, Olesek and Breier2014). Yet three key gaps remain: (i) MTT-specific mechanisms are still opaque. Most studies measure global metacognition; only a handful deploy dedicated MTT tasks (scene construction, temporal-horizon) when examining insight. Whether the episodic component uniquely drives insight loss is therefore unproven (Lysaker, Gagen, et al., Reference Lysaker, Gagen, Moritz and Schweitzer2018). (ii) Trauma’s role is only indirectly supported. Childhood adversity fragments autobiographical memory and weakens metacognition, but no published work has tested trauma × MTT or trauma × insight interactions in a single model (Aharon Biram et al., Reference Aharon Biram, Horesh, Tuval-Mashiach and Hasson-Ohayon2024; Takarangi, Smith, Strange, & Flowe, Reference Takarangi, Smith, Strange and Flowe2017). (iii) Causality is unknown. Evidence is almost entirely cross-sectional, so we cannot determine whether MTT deficits cause poor insight, whether poor insight erodes MTT, or whether both stem from a shared neural factor such as default-mode dysconnectivity – an idea awaiting longitudinal or interventional tests (Jun, Miao, & Ying, Reference Jun, Miao and Ying2025). In sum, therapies that jointly enhance episodic specificity and metacognitive mastery (e.g., Metacognitive Reflection and Insight Therapy (MERIT), episodic-future-thinking modules) remain the best-supported clinical options (de Jong et al., Reference de Jong, van Donkersgoed, Timmerman, Aan Het Rot, Wunderink, Arends, van Der Gaag, Aleman, Lysaker and Pijnenborg2019; Martin et al., Reference Martin, Bullock, Fiszdon, Stacy, Martino, James and Lysaker2023), but definitive studies – longitudinal, experimental, and trauma-stratified – are still needed to confirm the direction and specificity of the MTT → metacognition → insight pathway.

Figure 1.

Conceptual pathway from autobiographical memory to clinical insight in schizophrenia.

Task-based fMRI: overlapping circuitry for MTT and insight

Core brain regions jointly implicated in MTT and insight—particularly the medial prefrontal cortex, posterior cingulate cortex, and hippocampus—are summarized in Table 1. Advanced neuroimaging studies increasingly support the idea that MTT and insight rely on overlapping brain networks (DMN) (Østby et al., Reference Østby, Walhovd, Tamnes, Grydeland, Westlye and Fjell2012; Viard et al., Reference Viard, Chételat, Lebreton, Desgranges, Landeau, de La Sayette and Piolino2011). Moving beyond traditional univariate contrasts, multivariate fMRI methods such as spatial independent component analysis (sICA) show that seemingly different tasks – like episodic recall, future simulation, and self-appraisal – activate common functional networks (Xu et al., Reference Xu, Calhoun, Worhunsky, Xiang, Li, Wall and Potenza2015). These shared networks often involve the medial prefrontal cortex, posterior cingulate cortex, and hippocampus, where the same brain regions show synchronized activation patterns across task types (Dafni-Merom et al., Reference Dafni-Merom, Monsa, Benbaji, Klein and Arzy2024).

Table 1.

Core brain regions shared by mental time travel (MTT) and insight in schizophrenia. These three regions – part of the default mode network – are consistently implicated in both autobiographical memory processes and self-reflective functions. Their disruption may underlie the co-occurring deficits in MTT and illness awareness observed in schizophrenia

Dynamic connectivity studies (‘connectotyping’) show that connections among fronto-hippocampal and default-mode regions reconfigure on the order of seconds as people move through different task stages, rather than remaining static (Miranda-Dominguez et al., Reference Miranda-Dominguez, Mills, Carpenter, Grant, Kroenke, Nigg and Fair2014; Vazquez-Trejo et al., Reference Vazquez-Trejo, Nardos, Schlaggar, Fair and Miranda-Dominguez2022). Time-resolved approaches capture brief, recurring interaction ‘states,’ highlighting flexible adaptation of network coupling to cognitive demands. This rapid reconfiguration is particularly relevant for fronto-hippocampal communication that supports self-referential processing, episodic retrieval, and prospection (Campbell et al., Reference Campbell, Madore, Benoit, Thakral and Schacter2018; Molnar-Szakacs & Uddin, Reference Molnar-Szakacs and Uddin2013). In schizophrenia, multiple studies report altered dynamic functional connectivity – reduced time in highly integrated states, longer dwell in weakly connected states, and fewer state transitions – patterns linked to symptom burden and cognitive dysfunction; such instability plausibly undermines metacognitive operations and clinical insight that depend on fluid shifts between internal mentation and external task focus (Lysaker et al., Reference Lysaker, Gagen, Wright, Vohs, Kukla, Yanos and Hasson-Ohayon2019; Sendi et al., Reference Sendi, Zendehrouh, Ellis, Liang, Fu, Mathalon and Calhoun2021; Shan et al., Reference Shan, Liao, Ou, Ding, Liu, Chen and He2020; Weber et al., Reference Weber, Johnsen, Kroken, Løberg, Kandilarova, Stoyanov and Hugdahl2020; You et al., Reference You, Luo, Yao, Zhao, Li, Wang and Li2022).

Other methods, such as non-negative matrix factorization (NMF), isolate partially overlapping brain networks and can support multivariate decoding of task states. These network patterns have even been used to classify what a person is doing – for example, constructing a mental scene, evaluating personal traits, or making a simple decision (Aggarwal & Gupta, Reference Aggarwal and Gupta2018; Anderson et al., Reference Anderson, Douglas, Kerr, Haynes, Yuille, Xie and Cohen2014; Shirer et al., Reference Shirer, Ryali, Rykhlevskaia, Menon and Greicius2012). Importantly, ventromedial prefrontal and frontoparietal regions often carry high weights in both MTT and insight-related tasks, suggesting that these domains rely on shared control/default-adjacent components (Aggarwal & Gupta, Reference Aggarwal and Gupta2018; Anderson et al., Reference Anderson, Douglas, Kerr, Haynes, Yuille, Xie and Cohen2014; Cole et al., Reference Cole, Reynolds, Power, Repovs, Anticevic and Braver2013; Vincent et al., Reference Vincent, Kahn, Snyder, Raichle and Buckner2008). Here, ‘insight-related tasks’ refer to self-/other trait-judgment paradigms linked to SAI-E/BCIS, reality/source-monitoring with confidence reports, error-awareness paradigms (ACC-mediated), and probabilistic ‘jumping-to-conclusions’ tasks, each recruiting vmPFC and frontoparietal control circuitry (Andreou et al., Reference Andreou, Steinmann, Leicht, Kolbeck, Moritz and Mulert2018; Bedford et al., Reference Bedford, Surguladze, Giampietro, Brammer and David2012; Carter, MacDonald III, Ross, & Stenger, Reference Carter, MacDonald, Ross and Stenger2001; Garrison, Fernandez-Egea, Zaman, Agius, & Simons, Reference Garrison, Fernandez-Egea, Zaman, Agius and Simons2017; Hester, Nestor, & Garavan, Reference Hester, Nestor and Garavan2009; Simons, Garrison, & Johnson, Reference Simons, Garrison and Johnson2017; van der Meer et al., Reference van der Meer, de Vos, Stiekema, Pijnenborg, van Tol, Nolen and Aleman2013). Large-scale analyses pooling data across many task paradigms (8 or more) consistently point to the frontoparietal control network as a key integrative hub. Activity in this network predicts performance in tasks requiring episodic future thinking, autobiographical memory, and clinical insight (Albouy, Martinez-Moreno, Hoyer, Zatorre, & Baillet, Reference Albouy, Martinez-Moreno, Hoyer, Zatorre and Baillet2022; Pagnotta, Riddle, & D’Esposito, Reference Pagnotta, Riddle and D’Esposito2024; Vanasse et al., Reference Vanasse, Fox, Fox, Cauda, Costa, Smith and Lancaster2021). Some researchers note that traditional GLM-based fMRI analyses may miss this overlap, because they average out rapid shifts in connectivity that multivariate methods are better equipped to detect (Fang, Poskanzer, & Anzellotti, Reference Fang, Poskanzer and Anzellotti2023; Moeller & Habeck, Reference Moeller and Habeck2006; Salvador et al., Reference Salvador, Verdolini, Garcia-Ruiz, Jiménez, Sarró, Vilella and Voineskos2020). Taken together, these findings support the idea that MTT and insight do not rely on separate, isolated brain systems. Instead, they draw on a partially overlapping and dynamically coordinated neural architecture (Brocas & Carrillo, Reference Brocas and Carrillo2018; Dafni-Merom et al., Reference Dafni-Merom, Monsa, Benbaji, Klein and Arzy2024; Gauthier & van Wassenhove, Reference Gauthier and van Wassenhove2016).

Resting-state connectivity: trait-level links between network integrity, MTT, and insight

Schizophrenia is reliably associated with altered connectivity within the DMN, most notably reduced (hypo-)connectivity between the vmPFC and the PCC (Dong et al., Reference Dong, Wang, Chang, Luo and Yao2018; Peng et al., Reference Peng, Zhang, Zhou, Song, Yang, Hao and Zhang2021), as well as between the PCC and the hippocampus (Dugré et al., Reference Dugré, Dumais, Tikasz, Mendrek and Potvin2021). Meta-analyses estimate these disruptions to be moderate in magnitude (Cohen’s d ≈ 0.4–0.6). At the same time, early or prodromal findings are heterogeneous: while many reports emphasize hypoconnectivity, some studies in CHR/early-course samples describe increased vmPFC–PCC connectivity (hyperconnectivity), particularly among individuals with poorer clinical insight (Clark et al., Reference Clark, Mittal, Bernard, Ahmadi, King and Turner2018; O’Neill, Mechelli, & Bhattacharyya, Reference O’Neill, Mechelli and Bhattacharyya2019), whereas other early-onset cohorts show reductions (e.g., Hilland et al., Reference Hilland, Johannessen, Jonassen, Alnæs, Jørgensen, Barth, Andreou, Nerland, Wortinger, Smelror, Wedervang-Resell, Bohman, Lundberg, Westlye, Andreassen, Jönsson and Agartz2022) Accordingly, the literature search was broadened to include clinical high-risk (CHR)/ultra-high-risk (UHR) and first-episode cohorts; stage and analytic choices (e.g., global signal regression, parcellation, motion handling) likely moderate the direction of observed DMN effects. These DMN connectivity abnormalities are thought to underlie impairments in self-referential processing, autobiographical memory, and metacognition observed in schizophrenia.

Crucially, the weaker these connections, the fewer episodic details patients recall during the Autobiographical Interview, and the lower their clinical insight, as measured by SUMD scores (Fan et al., Reference Fan, Tan, Huang, Chen, Fan, Wang and Tan2022). Connectivity between the anterior hippocampus and vmPFC is especially important: reduced coupling in this pathway predicts both less vivid future-event construction and lower self-reflectiveness on the BCIS.

In addition, graph-theory studies show reduced global efficiency and weaker hub integrity across the DMN and frontoparietal control networks. These network-level disruptions explain unique variance in SAI-E insight scores – even after accounting for symptom severity (Blessing et al., Reference Blessing, Murty, Zeng, Wang, Davachi and Goff2020; Du et al., Reference Du, Pearlson, Yu, He, Lin, Sui, Wu and Calhoun2016; Dugré et al., Reference Dugré, Dumais, Tikasz, Mendrek and Potvin2021; Livingston et al., Reference Livingston, Kiemes, O’Daly, Knight, Lukow, Jelen and Modinos2024; Micheloyannis, Reference Micheloyannis2012; Pan et al., Reference Pan, Liu, Xue, Sheng, Cai, Cheng and Chen2022).

Dynamic connectivity analyses add a time-based perspective. Patients spend less time in a brain state dominated by DMN and control-network activity, and more time in a sensorimotor-dominated state. Notably, shorter ‘high-DMN’ dwell time predicts both poorer future-event fluency and lower insight 6 months later (Du et al., Reference Du, Pearlson, Yu, He, Lin, Sui, Wu and Calhoun2016; Fox et al., Reference Fox, Abram, Reilly, Eack, Goldman, Csernansky and Smith2017; Sendi et al., Reference Sendi, Zendehrouh, Ellis, Liang, Fu, Mathalon and Calhoun2021).

Together, these resting-state findings mirror task-based deficits. They suggest that predominantly weakened connectivity among the vmPFC, precuneus, and hippocampus, while accommodating early-stage reports of hyperconnectivity, disrupts the flow of episodic content into metacognitive systems, ultimately degrading clinical insight (Fornara et al., Reference Fornara, Papagno and Berlingeri2017; Gee et al., Reference Gee, Dazzan, Grace and Modinos2025; Kühn & Gallinat, Reference Kühn and Gallinat2013).

Electrophysiology: millisecond-scale evidence for a predictive-processing link

EEG studies show that insight failures in schizophrenia begin within milliseconds of processing information. Reduced error-related negativity (ERN) in the medial frontal cortex and lower mismatch negativity (MMN) – both markers of disrupted predictive coding – are linked to fewer episodic memory details and poorer clinical insight (higher SUMD scores) (Hamilton, Boos, & Mathalon, Reference Hamilton, Boos and Mathalon2020; Kansal, Patriciu, & Kiang, Reference Kansal, Patriciu and Kiang2014; Perrottelli et al., Reference Perrottelli, Giordano, Brando, Giuliani, Pezzella, Mucci and Galderisi2022). Additionally, a smaller P300 response to self-relevant cues is associated with lower self-reflectiveness on the BCIS. Together, these findings suggest that weak early brain signals related to prediction errors and salience detection deprive metacognitive systems of the moment-by-moment input needed for vivid mental scene construction – ultimately contributing to impaired insight (Lysaker et al., Reference Lysaker, Gagen, Wright, Vohs, Kukla, Yanos and Hasson-Ohayon2019; Lysaker, Pattison, et al., Reference Lysaker, Pattison, Leonhardt, Phelps and Vohs2018). Electrophysiological findings suggest that fast, early brain signals involved in prediction and self-relevance processing are weakened in schizophrenia, contributing to poor memory detail and impaired insight. Importantly, EEG/magnetoencephalography (MEG) studies of mental time travel show that constructing past and future events elicits a parietal late positive component and late frontal monitoring effects, and engages hippocampal–vmPFC theta interactions; these millisecond-scale signatures track episodic detail and temporal distance (Barry, Barnes, Clark, & Maguire, Reference Barry, Barnes, Clark and Maguire2019; Colás-Blanco, Mioche, La Corte, & Piolino, Reference Colás-Blanco, Mioche, La Corte and Piolino2022; Monk, Barnes, & Maguire, Reference Monk, Barnes and Maguire2020).

Multimodal & molecular imaging: converging structural evidence

Findings from structural MRI, diffusion tensor imaging (DTI), and PET all support the same hippocampal–vmPFC circuit implicated in fMRI and EEG studies. Smaller volumes in the anterior hippocampus and ventromedial prefrontal cortex (vmPFC) are linked to both fewer autobiographical memory details and poorer insight (effect sizes d ≈ 0.4–0.5) (Adriano, Caltagirone, & Spalletta, Reference Adriano, Caltagirone and Spalletta2012; Duan et al., Reference Duan, He, Ou, Wang, Xiao, Li and Chen2020; Dugré et al., Reference Dugré, Dumais, Tikasz, Mendrek and Potvin2021). In addition, reduced fractional anisotropy in the uncinate fasciculus – the main white-matter tract connecting the anterior hippocampus and vmPFC – has been shown to partially mediate the relationship between mental time travel abilities and clinical insight (Herold et al., Reference Herold, Lässer, Schmid, Seidl, Kong, Fellhauer and Schröder2013; Kelly et al., Reference Kelly, Jahanshad, Zalesky, Kochunov, Agartz, Alloza and Donohoe2018; Lysaker & Dimaggio, Reference Lysaker and Dimaggio2014; Samartzis, Dima, Fusar-Poli, & Kyriakopoulos, Reference Samartzis, Dima, Fusar-Poli and Kyriakopoulos2014; Von Der Heide, Skipper, Klobusicky, & Olson, Reference Von Der Heide, Skipper, Klobusicky and Olson2013). This supports a structural basis for failures in integrating episodic memory content into metacognitive awareness.

Taken together, these findings suggest that a structurally and chemically weakened hippocampal–vmPFC loop acts as a ‘hardware bottleneck’ through which impoverished MTT content reaches the systems responsible for self-reflection and insight (Lysaker & Dimaggio, Reference Lysaker and Dimaggio2014; McCormick, Ciaramelli, De Luca, & Maguire, Reference McCormick, Ciaramelli, De Luca and Maguire2018). However, longitudinal multimodal studies that incorporate explicit MTT tasks are still needed to confirm the direction and specificity of this circuit’s role in insight.