Emotion-cognition dysregulation in major depression: Multidimensional biases, neural circuit imbalance, and translational opportunities

Abstract

Major depressive disorder is increasingly conceptualized as a disorder of emotion-cognition dysregulation. Convergent evidence shows reliable negative biases in recognition and attention (facilitated capture by threat, impaired disengagement), overgeneral and negatively weighted autobiographical memory, blunted processing of positive/rewarding stimuli, and preferential reliance on maladaptive regulation strategies (rumination, suppression). At the systems level, task-based and resting-state functional magnetic resonance imaging support the presence of a limbic-prefrontal imbalance - amygdala/insula hyperreactivity with reduced top-down control from dorsolateral/ventromedial prefrontal cortex and anterior cingulate - accompanied by large-scale network disruption (default-mode hyperconnectivity and weakened coupling to salience and frontoparietal control networks). Electrophysiology adds temporal specificity: Enhanced early components to negative faces (e.g., N170), reduced late positive potential to appetitive cues, diminished prefrontal theta power, altered gamma activity, and perturbed theta-gamma coupling. Mechanistic contributors include monoaminergic imbalance, hypothalamic-pituitary-adrenal-axis activation, inflammatory signaling, and excitatory-inhibitory disequilibrium, moderated by cognitive style, early adversity, sex, and age. Translational opportunities span composite behavioral-neural signatures for diagnosis and stratification, risk forecasting, and treatment selection/monitoring; partial normalization of biases and circuits has been observed with cognitive behavioral therapy, selective serotonin reuptake inhibitor/serotonin-norepinephrine reuptake inhibitor, and noninvasive neuromodulation (repetitive transcranial magnetic stimulation/transcranial direct current stimulation), especially with imaging-guided targeting. Priorities include standardized paradigms, longitudinal multimodal designs, and mechanistic modeling to qualify biomarkers and enable mechanism-guided, precision interventions.

Key Words: Major depressive disorder; Emotion-cognition dysregulation; Amygdala-prefrontal circuitry; Attentional bias; Emotion regulation; Biomarkers

Core Tip: This review reframes major depression as an emotion–cognition dysregulation syndrome characterized by a limbic-prefrontal imbalance and large-scale network disruption. By integrating behavioral signatures (negative attention/memory bias, impaired reappraisal) with neural markers (amygdala/insula hyperreactivity, weakened prefrontal control, default mode network-frontoparietal control network/salience network dyscoupling), it outlines composite biomarkers for diagnosis, stratified risk prediction, and mechanism-guided therapies (e.g., imaging-informed neuromodulation). The field now needs standardized tasks and longitudinal multimodal designs to operationalize precision psychiatry.

INTRODUCTION

Major depressive disorder (MDD) is among the leading causes of disability worldwide and profoundly impairs affective experience, cognitive functioning, and social adaptation[1]. Beyond persistent low mood and loss of interest, converging evidence shows pervasive disturbances in the processing of emotional information in individuals with MDD[2]. These disturbances encompass heightened sensitivity and attentional fixation on negative stimuli, blunted responsiveness to positive stimuli, and reliance on maladaptive emotion-regulation strategies (e.g., rumination), jointly constituting a cognitive substrate that sustains symptoms and confers risk for relapse[3,4]. Emotion-cognition processing refers to the cascade of operations by which individuals perceive, interpret, memorize, and regulate emotional information. As a bridge between external stimuli and internal experience, it is implemented by a distributed network centered on the prefrontal cortex (PFC), amygdala, anterior cingulate cortex (ACC), and hippocampus[5,6]. In MDD, an imbalance between reduced prefrontal top-down control and amygdala hyperreactivity gives rise to biased processing at multiple stages, thereby amplifying negative affect and reinforcing maladaptive cognitive schemas[7,8]. Rapid advances in cognitive neuroscience and multimodal neuroimaging now allow increasingly precise delineation of the circuits and molecular mechanisms underlying these behavioral abnormalities[9,10]. Systematic characterization of multidimensional emotion-cognition abnormalities and their neural bases is therefore essential both for deepening mechanistic understanding and for developing targeted interventions within a precision-psychiatry framework. Despite robust evidence for the centrality of emotion-cognition abnormalities in MDD, the literature remains fragmented, with many studies confined to single processing stages (e.g., attention or memory) and limited cross-stage integration[2,6]. Moreover, a substantial portion of the work is descriptive rather than mechanistic, with insufficient examination of how these abnormalities map onto symptom spectra and longitudinal illness course. Bridging mechanistic insights and clinical application thus remains a key challenge for the field.

THEORETICAL FRAMEWORK OF EMOTION-COGNITION PROCESSING

Emotion-cognition processing is a multistage, dynamically unfolding sequence of operations. It begins with emotion recognition, namely the perceptual discrimination of external affective cues such as facial expressions and prosody, whose deviations can propagate forward to bias later stages[11,12]. Selective attention then allocates limited cognitive resources to emotional information; in MDD, a robust attentional bias toward negative material tends to capture resources automatically and sustain their engagement[2,13]. Emotional memory governs the encoding, consolidation, and retrieval of affect-laden content; patients commonly exhibit enhanced memory for negative stimuli alongside attenuated memory for positive material, a profile that strengthens and stabilizes maladaptive self-schemas[14,15]. Finally, emotion regulation comprises higher-order strategies - such as cognitive reappraisal and response suppression - through which individuals modulate affective responses. People with MDD rely disproportionately on maladaptive strategies (e.g., rumination, suppression) and struggle to implement adaptive strategies effectively, thereby failing to mitigate negative affect[16,17].

At the neural level, these stages are orchestrated by coordinated interactions between limbic structures (e.g., amygdala, insula) and cognitive control systems (notably the PFC and ACC). The amygdala supports rapid detection and initial appraisal of affective significance, whereas the PFC implements top-down modulation of emotional responses[18,19]. Prominent theoretical accounts generally converge on this imbalance: Gross’s process model of emotion regulation identifies disrupted coordination between emotion generation and cognitive control as a mechanistic core of affective disorders[20], while Beck’s cognitive model posits that deeply embedded negative schemas bias processing across stages, from perception through memory and regulation[19]. More recent interaction models emphasize bidirectional coupling, whereby mood state persistently shapes attention and memory, while cognitive deficits in turn maintain negative affect, establishing a self-reinforcing loop[8,21]. Together, these perspectives provide a conceptual scaffold for understanding emotion-cognition abnormalities in MDD and delineate candidate targets for psychological and biological interventions.

Despite their explanatory value, current models largely derive from experimental psychology and neuroimaging extrapolations and remain insufficiently validated in heterogeneous clinical populations[19,20]. Moreover, the accounts differ in emphasis - some foreground regulatory processes, others cognitive schemas - yet none fully integrates the breadth of multidimensional abnormalities while accommodating interindividual variability. Advancing toward a unified framework will require multimodal evidence and computational modeling capable of linking mechanisms across levels and predicting person-specific trajectories.

BEHAVIORAL CHARACTERISTICS OF EMOTION-COGNITION PROCESSING IN MDD

In emotion recognition, a substantial body of work shows that individuals with MDD are more likely to misclassify neutral or ambiguous facial expressions as negative, alongside reduced accuracy and prolonged reaction times for identifying positive emotions. This pattern indicates that bias emerges at the earliest stage of processing and is thought to reflect heightened amygdala sensitivity to threat coupled with insufficient prefrontal regulation[22]. Aberrant recognition - marked by heightened threat appraisal and blunted sensitivity to positive cues - maps directly onto co-morbid anxiety and anhedonia in MDD. Threat-biased evaluations inflate perceived environmental costs, while reduced positive salience undermines reward anticipation and consummatory pleasure. Clinically, this profile predicts elevated distress, avoidance, and lower engagement with rewarding activities, suggesting value for early stratification and behavioral activation targeting[6,23].

Attentional processing likewise exhibits a marked negative bias. On dot-probe and emotional Stroop tasks, patients demonstrate facilitated engagement with negative stimuli and impaired disengagement even when task demands require shifting attention. Eye-tracking studies converge on longer dwell times on negative images relative to healthy controls[13,24,25]. The persistence and automaticity of this attentional capture allow negative information to dominate cognitive resources and, in turn, to sustain dysphoric mood. Difficulties disengaging from negative stimuli and sustained attentional capture provide a mechanistic bridge to rumination and mood persistence. Prolonged allocation of attentional resources to dysphoric material constrains cognitive flexibility and amplifies negative affect, thereby maintaining symptom severity. Clinically, this phenotype indicates targets for attentional bias modification and cognitive control training to reduce perseverative negative thinking[26-28].

In the memory domain, patients show enhanced recall for negative material and attenuated memory for positive content across free recall and recognition paradigms. A characteristic feature is overgeneral autobiographical memory, in which specific, detail-rich recollections are replaced by vague, gist-level summaries[15,29,30]. This mnemonic profile has been linked to hippocampal and prefrontal abnormalities and is thought to reinforce negative self-schemas. A negative recall bias and overgeneral memory diminish access to specific positive experiences and problem-solving templates, reinforcing hopelessness and relapse vulnerability. By narrowing autobiographical specificity, overgeneral memory sustains global negative self-schemas and impairs emotion regulation. Clinically, memory-focused interventions such as memory specificity training can improve autobiographical specificity and reduce depressive symptoms[29,31,32].

With respect to emotion regulation, people with MDD report and exhibit reduced use of adaptive strategies such as cognitive reappraisal and greater reliance on avoidance, suppression, or rumination. Converging behavioral and neuroimaging evidence indicates diminished dorsolateral prefrontal activation during regulation attempts, insufficient down-regulation of amygdala responses, and heightened default mode network (DMN) activity[33-35]. A strategy profile favoring suppression and diminished reappraisal links to greater symptom chronicity, functional impairment, and anhedonia. Reduced top-down reframing capacity perpetuates negative affect, while suppression elevates physiological load without resolving distress. Clinically, this suggests prioritizing reappraisal-based cognitive-behavioral therapy (CBT) and behavioral activation; when appropriate, augment with approaches that enhance prefrontal control[36-39].

Negative attentional bias is among the most robust and replicable behavioral signatures of MDD, supported by meta-analytic evidence across diverse paradigms and large cohorts[13,21,40]. In contrast, abnormalities in emotional memory and cognitive reappraisal show more modest and variable effect sizes, emerging primarily in subgroups characterized by recurrent illness, high rumination, or comorbid anxiety. These differences suggest that not all reported impairments reflect core transdiagnostic mechanisms, but instead map onto partially dissociable emotion-cognition pathways influenced by task features and patient characteristics. Taken together, we argue that these abnormalities form a stage-to-stage causal cascade rather than isolated deficits. Early perceptual misinterpretation and preferential attentional capture of negative cues increase their encoding priority, resulting in negative memory enhancement and overgeneral autobiographical recall, which subsequently fuels rumination and regulatory failure[6,23,29]. This reinforcing negativity cycle provides a mechanistic account of why depressive symptoms are persistent and relapse-prone despite transient mood improvements[13,27]. A schematic representation of these cross-stage abnormalities is provided in Figure 1.

Figure 1 Dysfunctional multi-stage emotion-cognition processing model in major depressive disorder. This schematic illustrates the cascading abnormalities in emotion-cognition processing in major depressive disorder, which evolves from a typical sequential hierarchy into a self-reinforcing pathological system. The model is initiated by external stimuli (e.g., facial expressions, social cues). Core processing stages (red boxes) are disrupted by specific biases and their associated neural substrates (blue boxes): Negative interpretation bias (amygdala/insula hyperreactivity), preferential attentional capture of negative cues (dysregulation of frontal eye fields and the ventral attention network), overgeneral autobiographical memory (hippocampal-DMN dysfunction), and reduced cognitive reappraisal capacity (dorsolateral prefrontal cortex/anterior cingulate cortex hypoactivation). Solid arrows represent the primary, forward flow of maladaptive information processing. Dashed arrows denote critical reinforcing interactions and feedback loops (e.g., failed top-down regulation amplifying initial perceptual bias, memory biases directing attention, and rumination impairing memory specificity). These cross-stage interactions collectively form a reinforcing cycle of negativity, which underlies the persistence of negative, self-focused thought and contributes to symptom chronicity and relapse vulnerability. DMN: Default mode network; dlPFC: Dorsolateral prefrontal cortex; ACC: Anterior cingulate cortex.

NEURAL MECHANISMS UNDERLYING ABNORMAL EMOTION–COGNITION PROCESSING

Neuroimaging studies implicate a set of cortical-subcortical circuits in the emotion-cognition abnormalities observed in MDD[41]. Structural magnetic resonance imaging (MRI) frequently reports reduced gray-matter volume in prefrontal regions - most notably the dorsolateral PFC - as well as in the hippocampus, changes that align with impairments in cognitive control and emotion regulation[41]. Task-based functional MRI (fMRI) further demonstrates amygdala hyperreactivity to negative stimuli accompanied by insufficient prefrontal down-regulation, yielding a characteristic imbalance in emotion generation vs top-down control[42,43]. Resting-state investigations complement these findings, revealing heightened activity/connectivity within the DMN together with weakened coordination with frontoparietal control and salience networks[44,45]. Taken together, these results support a model of disrupted equilibrium between limbic and control systems in MDD.

Electrophysiological evidence refines the temporal dynamics of these processes. Event-related potential (ERP) studies indicate amplified early components (e.g., enhanced N170) to negative facial expressions, suggesting selective sensitivity at initial perceptual stages[46,47]. At later stages, the late positive potential is attenuated for appetitive or positive cues, consistent with diminished motivational engagement and closely linked to anhedonia[48]. Electroencephalography (EEG) studies show that gamma oscillatory activity and cross-frequency coupling mechanisms are disrupted in MDD, suggesting impaired integration of emotional information and working memory processes[49,50]. Multimodal approaches have become increasingly prominent. Studies integrating structural MRI, functional MRI, and electrophysiology report aberrant structure-function coupling in MDD, indicating cross-scale integration deficits that may constitute a systems-level substrate of the disorder[51]. In parallel, emerging technologies - including portable functional near-infrared spectroscopy (fNIRS), mobile EEG, and imaging-guided, individualized targeting for transcranial magnetic stimulation - are opening avenues for ecologically valid monitoring and precision intervention[52,53].

Notwithstanding the breadth of supportive evidence, findings are not uniformly consistent. Amygdala hyperreactivity has been consistently reported in well-powered fMRI studies and meta-analytic syntheses, particularly during implicit negative processing[21,54]. However, this effect becomes weaker or absent in explicit emotion evaluation tasks and in medicated or remitted cohorts, suggesting state dependence and task sensitivity[55,56]. In contrast, fronto-limbic connectivity alterations show greater trait-like stability and may represent more reliable neural markers across illness stages[57]. These discrepancies likely arise from heterogeneity in task paradigms, clinical subtypes, and analytical pipelines, reinforcing the need to interpret neural findings through stratified mechanistic models rather than assuming uniformity across studies. We propose a “reactivity-connectivity trade-off” framework to reconcile such divergent evidence: Limbic hyperactivation may be context-dependent - expressed predominantly under conditions of implicit threat or heightened stress reactivity - whereas disrupted prefrontal-amygdala connectivity persists and may represent the primary pathophysiological substrate, particularly in chronic or recurrent depression[55,57]. This framework reconciles seemingly contradictory findings by suggesting that neural abnormalities in MDD may shift along a state-trait continuum, where reactivity differences reflect task features (implicit vs explicit emotion processing), illness stage, and clinical subtype heterogeneity, including anxious vs melancholic presentations[56]. Beyond amygdala-connectivity discrepancies, heterogeneity likely reflects differences in task demands (implicit vs explicit emotion processing; passive viewing vs regulation), illness phase and subtype composition (e.g., anxious vs melancholic; chronic/recurrent), and analytic flexibility (preprocessing pipelines, region of interest definitions, statistical thresholds), which jointly modulate observed effect sizes[21,54-57].

PUTATIVE MECHANISMS AND MODULATING FACTORS

At the neurotransmitter level, converging evidence implicates insufficiency of monoaminergic systems-serotonin, dopamine, and norepinephrine - together with impaired inhibitory control along prefrontal-amygdala pathways, in shaping negative information bias and deficits in emotion regulation[16,18,42]. Within the reward circuit, reduced dopaminergic signaling is closely linked to anhedonia and diminished motivation, aligning behavioral phenotypes with mesocorticolimbic dysfunction[58]. Beyond monoamines, disruption of the excitatory-inhibitory balance likely further compromises the efficiency and stability of emotion-regulation networks. Neuroendocrine and inflammatory processes provide an additional layer of vulnerability. Chronic stress-related activation of the hypothalamic-pituitary-adrenal axis and heightened proinflammatory signaling have been repeatedly associated with MDD; alterations in stress-responsive circuitry, elevations in circulating cytokines, and suppression of neurotrophic factors collectively impair hippocampal and prefrontal plasticity and thereby exacerbate emotion-cognition dysfunction[59]. Existing models tend to examine monoaminergic imbalance, network dysregulation, or excitatory-inhibitory disturbances in isolation, leading to theoretical fragmentation. We propose that network-level dysregulation represents the convergence point, whereby serotonergic and dopaminergic signaling abnormalities impair cognitive-control gating, inflammation disrupts neuroplasticity, and excitatory-inhibitory imbalance undermines temporal coordination of neural information flow[23,60]. This cross-level integration may clarify why distinct biological perturbations yield shared emotion-cognition phenotypes in MDD.

Psychosocial factors confer cognitive risk and interact with biological mechanisms. Negative automatic thoughts, pessimistic attributional style, and a tendency toward rumination constitute a cognitive vulnerability profile for depression[61]. Personality traits such as high neuroticism and perfectionism may amplify threat sensitivity and perpetuate self-critical cycles, increasing affective burden[62]. Adverse childhood experiences can recalibrate the hypothalamic-pituitary-adrenal axis and induce lasting epigenetic modifications, substantially elevating adult depression risk and potentially shaping the long-term development of emotion-cognition circuits[63]. Individual differences are salient across sex and age. Women with MDD more often exhibit heightened emotional sensitivity and stronger rumination tendencies[62]. Adolescents, whose prefrontal-limbic connectivity is still maturing, appear particularly susceptible to interference from negative affect, whereas older adults more commonly display affective blunting and slowed processing[64]. Heterogeneity across clinical subtypes is also evident, as recent neuroimaging-multiomics studies have delineated biologically distinct subtypes within MDD[65].

Importantly, abnormalities in emotion-cognition processing show partial plasticity. Pharmacotherapy may modify negative information processing and enhance cognitive flexibility, sometimes preceding overt symptom change[42,66]. Psychological interventions such as CBT can restructure maladaptive appraisals and strengthen regulatory strategy use, with accompanying gains in prefrontal-amygdala circuit efficiency[33,34]. Noninvasive neuromodulation [e.g., repetitive transcranial magnetic stimulation (rTMS)/transcranial direct current stimulation] modulates cortical excitability and downstream network dynamics, improving regulation capacity when appropriately targeted[67]. At the molecular frontier, exosome-mediated delivery of brain-derived neurotrophic factor has been shown to promote neuroplasticity, highlighting a potential avenue for refractory depression[59].

Taken together, mechanistic studies underscore the multidimensional nature of MDD but also reveal gaps in cross-level integration. While neurotransmitter and inflammatory mechanisms are well supported in animal models, findings in clinical populations remain heterogeneous with variable replicability[68,69]; psychosocial evidence often relies on self-report, limiting causal inference[61,63]. Future work should prioritize multi-level, multimethod integration to delineate the relative contributions and temporal sequencing of these mechanisms in driving emotion–cognition abnormalities.

CLINICAL RELEVANCE AND TRANSLATIONAL VALUE

To move beyond feasibility, we foreground two implemented and reusable pathways that link objective markers to clinical decisions. First, recent work demonstrates fNIRS-based machine-learning models can predict early antidepressant response using prefrontal hemodynamic features, enabling pre-treatment triage and reducing trial-and-error in pharmacotherapy[52]. Second, fMRI-guided individualized rTMS targeting - for example, using network-informed sites anti-correlated with subgenual ACC - has been deployed as a practical workflow to improve target precision in MDD neuromodulation[67]. Beyond these implemented examples, abnormalities in emotion-cognition processing are not merely epiphenomenal to MDD; they align closely with illness severity, longitudinal course, and treatment response. Negative attentional bias correlates with rumination and co-occurring anxiety symptoms, deficits in the processing of positive or reward-related information track anhedonia, and impairments in cognitive control map onto psychomotor retardation and real-world functional decline[13,21,48]. These associations suggest that behavioral signatures reflect core disease mechanisms.

With ongoing advances in multimodal assessment, emotion-cognition metrics are increasingly positioned as candidate biomarkers. Composite models that integrate behavioral tasks, ERP components, and neuroimaging features have demonstrated improvements in diagnostic discrimination[52], while aberrant emotion processing predicts later onset in high-risk cohorts[63]. For instance, recent multimodal fusion approaches integrating behavioral, electrophysiological, and fMRI features provide clinically meaningful gains in both diagnostic classification and treatment-response prediction. Distinct processing phenotypes also index partially dissociable neural mechanisms, supporting biologically informed subtyping for treatment selection[67]. A connectivity-based classifier has differentiated antidepressant-responsive biotypes[56], whereas multimodal models integrating fMRI and behavioral markers enable individualized targeting[65].

At the therapeutic level, emotion-cognition indices show increasing promise for prognosis and personalization. Objective markers such as ERP or fNIRS can provide early signals of treatment response that are more sensitive than symptom scales[53]. However, evidence for attention bias modification training and cognitive bias modification reflects selective but modest clinical benefit, with inconsistent durability and high variability in adherence, highlighting the need for improved engagement strategies and phenotype-based targeting[70]. Meanwhile, connectivity-based stratification demonstrates stronger predictive validity, differentiating subgroups preferentially responsive to rTMS and dopaminergic augmentation[58,66]. Patients with pronounced attentional bias benefit more from attention bias modification training/cognitive bias modification or CBT targeting disengagement control[71,72], whereas reward-processing deficits guide dopaminergic or activation-based therapies[73,74]. These findings collectively indicate a spectrum of evidence across therapeutic avenues - ranging from preliminary behavioral modification approaches to more robust, mechanism-informed neuromodulation strategies.

Expanding this mechanistic perspective, different treatments preferentially act on partially dissociable nodes within the emotion-cognition circuit: CBT engages top-down cognitive control, strengthening dorsolateral PFC and ACC regulation of emotion[75]; selective serotonin reuptake inhibitors/serotonin-norepinephrine reuptake inhibitors normalize bottom-up affective reactivity and reward-related signaling via amygdala-striatal modulation[76]; and rTMS/transcranial electric stimulation recalibrate network-level integration across the DMN, salience network, and frontoparietal control networks[77]. Recent consensus and systematic reviews underscore the evolving parameterization of neuromodulation protocols in depression[78]. This neural circuit imbalance is illustrated in Figure 2. Building on this mechanistic dissociation, we conceptualize these therapeutic modalities as collectively repairing distinct but interlocking components of a maladaptive emotion-cognition circuit. Accordingly, specific emotion-cognition phenotypes - such as exaggerated attentional capture, impaired reappraisal, or anhedonia - can be used as matching rules for mechanism-guided, personalized intervention allocation. This framework positions behavioral-neural signatures from correlational markers to actionable treatment targets suited for precision psychiatry.

Figure 2 Neural circuit imbalance model of emotion-cognition dysfunction in major depressive disorder. This model delineates three convergent neural disruptions and their integrative pathways leading to clinical dysfunction. The core disruptions are: (1) Limbic hyperreactivity (amygdala, insula), which intensifies the bottom-up processing of negative salience; (2) Prefrontal regulatory deficits (dorsolateral, ventromedial, and anterior cingulate cortices), which compromise top-down cognitive control; and (3) Large-scale network dysregulation, manifesting as default mode network hyperconnectivity and weakened coupling between the frontoparietal control network and salience network. vmPFC: Ventromedial prefrontal cortex; dlPFC: Dorsolateral prefrontal cortex; ACC: Anterior cingulate cortex; DMN: Default mode network; FPCN: Frontoparietal control network; SN: Salience network; MDD: Major depressive disorder.

Despite these opportunities, most findings lack large-scale prospective validation[42]. Limited cross-site reproducibility and platform heterogeneity constrain generalizability and hinder clinical adoption[66]. To overcome economic and infrastructural limitations, several scalable solutions have emerged. First, portable and low-cost neurophysiology platforms (e.g., fNIRS, EEG headsets) allow deployment in primary-care or community settings, reducing reliance on high-resource hospital imaging[79]. Second, harmonized task protocols and interoperable preprocessing pipelines will improve reproducibility and enable cross-site machine-learning generalization. Third, hybrid telehealth-enabled models that combine digital phenotyping with brief neurocognitive testing can support decentralized monitoring and stratification, extending access to underserved populations[80]. As regulatory standards evolve, these converging advances may allow emotion-cognition biomarkers to be delivered as scalable, equitable clinical tools rather than limited research resources. Continued multi-center collaboration and prospective validation will be critical for transitioning these advances from research settings into routine clinical care.

LIMITATIONS AND FUTURE DIRECTIONS

Although substantial progress has been made across behavioral, neural, and translational domains, critical gaps remain that limit mechanistic insight and clinical implementation. At the behavioral level, most studies examine single-stage abnormalities (e.g., attention or memory) in isolation, with insufficient evaluation of dynamic interactions and causal sequencing across stages. At the neural level, findings remain constrained by methodological heterogeneity, limited longitudinal designs, and insufficient integration of multiscale evidence linking brain dynamics to symptom trajectories. At the translational level, external validation and real-world feasibility remain limited, particularly regarding the scalable deployment of multimodal biomarkers. Addressing these cross-domain limitations will be essential for advancing emotion-cognition science toward clinical precision.

Progress will require coordinated advances on methodological, theoretical, and translational fronts. First, standardization and multi-center collaboration are essential to improve reproducibility and enable pooled inference across sites and populations[41]. Second, theoretical frameworks should integrate multi-level evidence and leverage computational psychiatry - including dynamic/effective connectivity and generative modeling - to explain emotion-cognition dynamics and predict individual trajectories[60]. Computational psychiatry frameworks now map learning, connectivity, and inference dynamics to clinical phenotypes[81]. Third, translational research must close the laboratory-clinic gap by deploying streamlined, low-cost, and ecologically valid tools (e.g., portable neurophysiology and digital phenotyping) and by establishing interoperable pipelines for large-scale prospective validation and regulatory-grade biomarker qualification[82]. The integration of digital phenotyping represents a promising direction for bridging laboratory-based emotion-cognition markers with real-world clinical utility. Smartphone and wearable sensors can continuously capture behavioral and emotional patterns associated with depression severity, such as mobility, sleep-wake cycles, and social interaction rhythms[79,80]. These passive indicators may enable earlier detection of worsening attentional or regulatory deficits and support stratification of emotion-cognition phenotypes to guide personalized interventions[83]. Furthermore, mobile platforms provide the infrastructure for just-in-time adaptive interventions that trigger targeted cognitive or emotional training when risk signals emerge[84]. Digital phenotyping pipelines are maturing, showing promise in scalable detection and monitoring of depression[85]. Future studies combining digital phenotyping with neurobehavioral measures (e.g., ERP, fMRI) are expected to accelerate closed-loop, precision monitoring and intervention in routine depression care.

CONCLUSION

Emotion-cognition abnormalities in MDD span recognition, attention, memory, and regulation, yielding a characteristic pattern of amplified negative bias, reduced responsiveness to positive/rewarding information, and impaired regulatory control. These phenotypes align with a core neural imbalance between limbic systems (e.g., amygdala) and prefrontal control circuitry (e.g., dorsolateral PFC, ACC), and are further shaped by monoaminergic signaling, neuroendocrine and inflammatory pathways, genetic liability, cognitive style, and early environmental adversity. As the field matures, emotion-cognition measures are moving from descriptive correlates to candidate biomarkers with potential utility in objective diagnosis, biologically informed subtyping, risk stratification, treatment selection, and response monitoring. Realizing this promise will depend on longitudinal designs, rigorous multimodal integration, computational modeling, and pragmatic translational studies. Ultimately, a mechanism-based approach centered on emotion-cognition circuits could shift clinical practice from symptom palliation toward mechanistic repair, advancing precision prevention, diagnosis, and treatment in MDD.