Metacognitive Modulation of Cognitive-Emotional Dynamics Under Social-Evaluative Stress: An Integrated Behavioural–EEG Study

Abstract

Background/Objectives: Decision-making under socially evaluative stress engages a dynamic interplay between cognitive control, emotional appraisal, and motivational systems. Contemporary models of multi-level co-regulation posit that these systems operate in reciprocal modulation, redistributing processing resources to prioritise either rapid socio-emotional alignment or deliberate evaluation depending on situational demands. Methods: Adopting a neurofunctional approach, a novel dual-task protocol combining the MetaCognition–Stress Convergence Paradigm (MSCP) and the Social Stress Test Neuro-Evaluation (SST-NeuroEval), a simulated social–evaluative speech task calibrated across progressive emotional intensities, was implemented. Twenty professionals from an HR consultancy firm participated in the study, with concurrent recording of frontal-temporoparietal electroencephalography (EEG) and bespoke psychometric indices: the MetaStress-Insight Index and the TimeSense Scale. Results: Findings revealed that decision contexts with higher socio-emotional salience elicited faster, emotionally guided choices (mean RT difference emotional vs. cognitive: −220 ms, p = 0.026), accompanied by oscillatory signatures (frontal delta: F(1,19) = 13.30, p = 0.002; gamma: F(3,57) = 14.93, p ≤ 0.001) consistent with intensified socio-emotional integration and contextual reconstruction. Under evaluative stress, oscillatory activity shifted across phases, reflecting the transition from anticipatory regulation to reactive engagement, in line with models of phase-dependent stress adaptation. Across paradigms, convergences emerged between decision orientation, subjective stress, and oscillatory patterns, supporting the view that cognitive–emotional regulation operates as a coordinated, multi-level system. Conclusions: These results underscore the importance of integrating behavioural, experiential, and neural indices to characterise how individuals adaptively regulate decision-making under socially evaluative stress and highlight the potential of dual-paradigm designs for advancing theory and application in cognitive–affective neuroscience.

1. Introduction

Decision-making in contexts of heightened socio-emotional salience constitutes a multi-layered process in which cognitive evaluation and affective appraisal are continuously interwoven. Classical models have conceptualised this interplay within the framework of dual-process theories, distinguishing between an intuitive, heuristic-based system (System 1) operating rapidly and automatically, and a deliberative, rule-based system (System 2) characterised by slower, more effortful reasoning [1,2]. While influential, this dichotomy has increasingly been viewed as overly static and reductive. Contemporary perspectives argue that decision-making is better understood as an emergent property of dynamic, context-dependent interactions between multiple neurocognitive subsystems [3,4,5]. Within this revised view, affective and cognitive contributions are not simply competitive but are mutually modulatory. Emotional signals can bias attentional allocation, evidence accumulation, and threshold setting in decision formation, while cognitive control processes can recalibrate or suppress affective influences when such modulation supports goal attainment. Crucially, this interplay unfolds on a continuum shaped by the temporal structure, social meaning, and motivational relevance of the decision context [3,6,7,8].

Socially evaluative stress represents a paradigmatic condition in which this dynamic coupling is tested to its limits. Evaluation by others—explicitly communicated or implicitly inferred—constitutes a high-relevance socio-emotional cue that can potentiate limbic–prefrontal reactivity and reorganise decision priorities [9,10].

Under such conditions, empirical evidence suggests a bias towards emotionally resonant, high-salience options, potentially reflecting an adaptive mechanism for maintaining social attunement or mitigating perceived interpersonal threat. This shift may occur at the expense of extended deliberation, as the cost–benefit calculus is recalibrated in favour of immediacy and affective congruence [11].

Recent integrative models conceptualise this phenomenon within the framework of multi-level co-regulation, wherein cognitive control, emotional appraisal, and motivational systems jointly contribute to adaptive functioning under pressure. Rather than operating in isolation, these systems dynamically redistribute processing resources in real time, prioritising either rapid socio-emotional alignment or more deliberate evaluation depending on the contextual affordances and the individual’s regulatory profile. Neurocognitively, this involves flexible coordination across prefrontal executive regions, anterior cingulate cortex, limbic structures (including amygdala and ventral striatum), and posterior association cortices implicated in social cognition and contextual integration [11,12,13,14].

From this perspective, investigating decision-making under socio-emotional and evaluative stress requires not only behavioural measurement but also precise mapping of the underlying neural dynamics. Oscillatory brain activity offers a particularly sensitive index of these processes, capable of capturing the temporal interplay between rapid affective appraisals and controlled cognitive elaboration as they unfold within the decision window.

Delta (1–4 Hz) oscillations have been associated with large-scale integrative processes, motivated attention, and the mobilisation of neural resources in contexts of heightened motivational or affective relevance. In socially or emotionally charged decision contexts, frontal delta is often implicated in the rapid integration of affectively salient cues into ongoing cognitive evaluations, while temporo-parietal delta has been linked to contextual reconstruction and the binding of complex situational features into coherent mental representations [15,16,17,18].

Theta (4–8 Hz) activity, particularly in the midline frontal regions, has been linked to conflict monitoring, working memory maintenance, and the allocation of cognitive control [19,20,21]. Frontal Theta supports the detection of discrepancies between competing response tendencies and the upregulation of control mechanisms to resolve them. Posterior and temporo-parietal Theta, in contrast, is associated with episodic retrieval, semantic integration, and the assimilation of contextual information into decision-making processes [15,19,22]. In dynamic decision contexts, the interplay between frontal control-related Theta and posterior context-related Theta is considered a hallmark of adaptive regulatory functioning [17].

Alpha (8–12 Hz) oscillations, traditionally interpreted as an index of cortical inhibition [23,24,25], have more recently been reconceptualised as markers of functional gating in both sensory and higher-order cortical systems [26,27,28]. Alpha desynchronisation in task-relevant regions reflects the release of inhibition and the facilitation of information processing, whereas synchronisation in task-irrelevant areas may support the suppression of distracting inputs. In the context of decision-making under socio-emotional load, alpha modulations are often observed as a reflection of attentional focusing and the selective gating of affectively or contextually salient information [29,30].

Beta (13–30 Hz) oscillations have been linked to top-down maintenance of cognitive sets, sensorimotor integration, and the stabilisation of current task-relevant representations [31,32,33]. In decision-making paradigms, beta activity may index the maintenance of a chosen response framework and the suppression of alternative action plans. In socio-emotional contexts, beta modulations can also be associated with the suppression or reinforcement of affectively driven response tendencies, reflecting the balance between motor preparation and regulatory control [29].

Gamma (30–100 Hz) oscillations are widely recognised as markers of feature binding and rapid integrative processing across distributed cortical sites [31,34,35]. In decision-making contexts, gamma activity supports the synthesis of perceptual, mnemonic, and affective components into a unified decisional construct. Frontal gamma has been associated with the binding of abstract rules and social–emotional contingencies, whereas temporo-parietal gamma is implicated in perspective-taking, social simulation, and the projection of decision consequences.

To sum up, oscillatory brain activity plays a central role in coordinating and integrating information across distributed neural systems. Neurophysiological measures—particularly EEG—offer a direct, temporally precise window into the neural mechanisms underpinning decision-making and stress regulation. Region-specific modulations (e.g., frontal vs. temporo-parietal) and frequency-specific signatures (e.g., delta for motivated attention, theta for control allocation, gamma for integrative binding) map the allocation of processing resources in ways not accessible through behavioural or self-report data alone.

By providing a temporal framework for neural communication, oscillations enable the flexible alignment of cognitive and emotional processes with moment-to-moment demands. In the context of socio-emotional and evaluative challenges, they offer a sensitive index of how processing resources are dynamically redistributed across functional networks. Different frequency bands are associated with distinct, yet interacting, operations, and their spatial modulation reflects the neural strategies engaged to balance rapid affective appraisal with more deliberate cognitive evaluation.

Understanding these oscillatory signatures is therefore essential for a comprehensive account of how decision-making unfolds under socio-emotional and evaluative stress. However, despite their potential to bridge multiple levels of analysis, investigations in this field have often been constrained by methodological compartmentalization. Research streams have tended to focus either on behavioural performance (e.g., choice frequencies, reaction times—RTs), electrophysiological activity (EEG), or subjective experience (e.g., perceived stress, emotional valence) with limited attempts to integrate these layers into a coherent explanatory framework. While each approach has yielded valuable insights, their isolated application risks underestimating the inherently multi-layered nature of adaptive regulation in complex social–emotional contexts. Behavioural indices capture the overt expression of decision-making tendencies—such as the propensity towards emotionally driven versus cognitively analytical choices—and their modulation under varying affective loads. Yet behavioural data alone cannot fully illuminate the regulatory mechanisms that guide these tendencies, particularly when decisions unfold in the presence of social evaluation, a condition known to amplify affective weighting and alter decisional thresholds.

Subjective self-report measures complement behavioural indices by providing access to internal experiential states, including perceived stress intensity, metacognitive awareness of stress, and affective stability. However, these measures are inherently retrospective and self-referential, subject to biases in introspection and reporting, and may lag behind the rapid, dynamic changes that occur during real-time decision-making under pressure.

Crucially, these three domains—behavioural, neurophysiological and subjective—are not parallel but isolated streams; rather, they form a mutually constraining system in which each level influences, and is influenced by, the others. Integrating these perspectives aligns with multi-level co-regulation frameworks [18,36,37], which conceptualise cognitive–emotional functioning as the product of coordinated adjustments across multiple representational and control systems.

The challenge, therefore, is methodological: to design paradigms that elicit meaningful variability across all three levels, enabling their convergence to be observed and quantified. The present study addresses this challenge through the conceptual coupling of two complementary paradigms—the MetaCognition–Stress Convergence Paradigm (MSCP) and the Social Stress Test—NeuroEval (SST-NeuroEval)—each of which captures different but interconnected facets of decision–stress interaction.

Conceptually, the two paradigms are mutually informative, in that each targets the reciprocal coupling between socio-emotional appraisal and cognitive control processes, but through orthogonal manipulations. The MSCP focuses on how varying decisional demands shape stress-related cognitive–emotional dynamics, whereas the SST-NeuroEval examines how escalating evaluative stress modulates regulatory engagement and cognitive–emotional allocation over time. Their integration within a single experimental framework allows for the joint exploration of content-related effects, such as those determined by the socio-emotional load of decision scenarios, and temporal effects, as revealed by the unfolding of evaluative stress from anticipation to direct exposure.

This dual-paradigm design advances existing approaches by simultaneously manipulating decisional content and evaluative stress trajectories, enabling convergence across behavioural, subjective, and oscillatory indices that isolated paradigms cannot capture.

From a theoretical perspective, we hypothesise that decision scenarios characterised by higher socio-emotional salience will favour rapid, emotionally guided choices over more deliberate cognitive–analytical strategies. Likewise, the transition from anticipatory to reactive phases of evaluative stress is expected to amplify subjective stress and increase reliance on emotional over controlled responses. When the two paradigms are considered jointly, these tendencies are predicted to manifest as stable cognitive–emotional regulatory profiles, whereby individuals with a predisposition toward emotionally driven decisions will exhibit reduced cognitive control and heightened socio-emotional engagement under evaluative load.

Accordingly, the present study aims to systematically examine how decision-making orientation interacts with socio—evaluative stress across behavioural, subjective, and oscillatory indices, providing an integrated account of cognitive—emotional regulation under conditions of ecological stress.

To test these predictions, we operationalized regulatory dynamics across multiple levels. Behaviourally, we examined decision latencies as an index of cognitive versus emotional decision strategies. Subjective stress was assessed through self-reported measures capturing the perceived intensity of evaluative pressure. At the neurophysiological level, oscillatory patterns were analysed as markers of cognitive–emotional integration: heightened socio-emotional engagement was expected to be reflected by increased frontal delta and gamma power and enhanced temporo-parietal theta activity, whereas the escalation of evaluative stress was anticipated to evoke elevated frontal delta and theta activity during early orienting and conflict-monitoring stages, accompanied by transient modulations in gamma power at peaks of evaluative demand.

2. Materials and Methods

2.1. Sample

The experimental population consisted of twenty adult participants employed within the same private organisation operating in the field of strategic consulting and organisational development (north of Italy). Participants (balanced in gender) were recruited through purposive internal sampling to ensure homogeneity of the professional context in line with the aims of the study. The mean age was 31.3 years (SD = 9.10; range: 21–53 years), and the average length of formal education was 17.5 years. All individuals reported being in good general health and demonstrated intact sensory functions, as verified during a preliminary anamnesis-based screening. Eligibility was restricted to full-time employees of the organisation who were able to understand and comply with the experimental instructions and had no uncorrected sensory deficits. Exclusion criteria comprised documented neurological or psychiatric disorders, use of psychotropic medication in the three months preceding data collection, and clinical evidence of burnout or severe chronic stress.

All participants provided written informed consent without compensation. The study was approved by the Ethics Committee of the Department of Psychology (approval code: 2021 PhDTD), Catholic University of The Sacred Heart, Milan, Italy, and was conducted in adherence to the guidelines outlined in the Declaration of Helsinki (2013) and according to the General Data Protection Regulation—Reg. UE 2016/679 and its ethical guidelines.

2.2. Experimental Procedure

The experimental protocol was conducted in a controlled environment, naturally lit and free from distracting stimuli, within a dedicated room at the company premises. Participants were welcomed individually and seated at a computer workstation positioned approximately 70cm from the display, where the experimental tasks were presented. Following a brief introduction to the general aims and procedures, the neurophysiological recording apparatus was applied.

Before the experimental phase commenced, a 120 s neurophysiological baseline was recorded (60 s with eyes closed and 60 s with eyes open) to establish a stable reference free from cognitive or emotional stimulation.

Prior to the commencement of the tasks, participants completed a battery of validated psychometric instruments, including the Perceived Stress Scale (PSS-10), the Emotion Regulation of Others and Self scale (EROS), the General Decision-Making Style inventory (GDMS), and the Balanced Empathy Emotional Scale (BEES). These measures provided individual profiles of emotion regulation, perceived stress, decision-making style, and empathic balance.

The experimental phase consisted of two functionally integrated components.

The MSCP served as the profiling phase for decision-making tendencies and metacognitive orientation [8]. Participants were presented with four realistic decision-making scenarios—two set in an organisational context and two in an emergency medical context—each requiring the selection between two alternatives, one predominantly emotion-driven and the other analytical-rational. Each decision was followed by a series of five-point Likert-scale items probing the underlying motivations and the degree of emotional and cognitive engagement. A maximum of twenty seconds was allowed for each decision to encourage spontaneous responding. The decision-making patterns identified in the MSCP provided an individualised index of cognitive–emotional decision style, which was subsequently used to interpret the neurophysiological and behavioural responses recorded in the second task.

The SST-NeuroEval was an adaptation of the Trier Social Stress Test [38] designed to elicit progressive evaluative pressure [18,34,37,39,40,41]. Participants completed five oral trials of increasing difficulty, each preceded by a video sequence depicting a virtual evaluation panel whose facial expressions became progressively more negative and unsupportive. Participants were instructed to prepare and deliver a short speech relevant to a given prompt, maintaining eye contact with the virtual panel for the full duration of the trial and aiming to convince the panel of their suitability. The structure was calibrated to provoke moderate stress responses both during the preparation and delivery phases. In this way, the individual differences observed in the MSCP were tested under conditions of high social–evaluative stress, enabling analysis of the extent to which decision-making style and metacognitive awareness modulated the observed physiological and behavioural responses.

Upon completion of the SST-NeuroEval, a purpose-designed metacognitive and metaperceptual questionnaire (VAS-Prep and VAS-Esp) was administered to assess subjective stress awareness, temporal perception, perceived emotional control, and the perceived effectiveness of self-regulatory strategies during the preparation and performance phases of the stress task.

The entire experimental session lasted approximately fifty minutes per participant, including arrival, instrumentation, baseline acquisition, task performance, psychometric assessment, and final debriefing.

2.3. Behavioural Data Acquisition and Processing

Behavioural data were digitally recorded throughout both experimental phase—the MetaCognition–Stress Convergence Paradigm (MSCP) and the SST-NeuroEva—via automated logging by the task administration platform. In the MSCP, each response was pre-classified as “emotional” or “cognitive” according to predefined coding criteria. For each participant, and within each decision-making domain (organisational and medical emergency), the arithmetic means of the scores assigned to emotional responses and, separately, to cognitive responses was calculated. These values constituted the EmoScore_Org and EmoScore_Med indices for the emotional component, and the CogScore_Org and CogScore_Med indices for the cognitive component. In parallel, mean RTs for each response type and domain—RT_Emo_Org, RT_Cog_Org, RT_Emo_Med, RT_Cog_Med—were computed as the arithmetic mean of the completion times (i.e., RTs) for all scenarios within the same category.

In the SST-NeuroEval, performance evaluation integrated quantitative and qualitative indices. Stress Regulation (StressReg) was estimated based on the promptness of speech initiation: for each oral trial, the latency between the beginning of the exposure phase and the actual start of performance was converted into a score on a five-point scale, with higher values indicating greater promptness. The final StressReg score was computed as the mean of the values obtained across the five trials. Stress Tolerance (StressTol) was calculated by relating the overall mean preparation time to the preparation times for the final two trials, and then averaging the two resulting ratios to obtain a single composite index.

Conceptually, the two indices capture complementary facets of adaptive functioning under evaluative stress. StressReg primarily reflects an individual’s ability to promptly engage with performance demands, indexing regulatory efficiency and readiness to act under social pressure. In contrast, StressTol provides a measure of persistence and stability across repeated exposure to evaluative stress, capturing how individuals sustain cognitive and emotional resources when facing cumulative demands. Empirical findings from related paradigms [18,34,37,39,40,41] suggest that higher StressReg scores are typically associated with more effective top-down control and faster mobilisation of attentional and motivational resources, while higher StressTol scores indicate resilience mechanisms supporting consistent performance despite escalating social–evaluative challenges.

Subjective measures, collected via visual analogue scales (0–100), were obtained separately for the preparation phase (VAS-Prep) and the exposure phase (VAS-Exp). In addition, two indices were derived from the metacognitive–metaperceptual questionnaire: Stress Metacognition (MetaStress), calculated as the mean of responses to items concerning perceived ability to identify and manage external, cognitive, and physiological sources of stress; and Temporal Awareness (MetaTime), obtained as the mean of responses to items assessing perception and management of time across the two phases of the task.

2.4. Electrophysiological Data Acquisition and Processing

EEG activity was recorded using an integrated, non-invasive wearable system. Recordings spanned the entire experimental session (baseline, MSCP, SST-NeuroEval), amounting to approximately 35 min of continuous acquisition per participant. Cortical activity was acquired through the Muse™ Headband (version 2; InteraXon Inc., Toronto, QC, Canada), a four-channel dry-electrode EEG device positioned according to the international 10–20 system, with active electrodes located over frontal (AF7, AF8) and temporo-parietal (TP9, TP10) regions, and reference electrodes placed at Fpz. Signals were sampled at 256 Hz, with a 50 Hz notch filter applied to reduce mains interference. Data were transmitted in real time via Bluetooth to a mobile device running the Mind Monitor application, which performed a fast Fourier transform (Hamming window, 10% length, 0.5 Hz resolution) to extract the spectral power density for the canonical EEG frequency bands: delta (1–4 Hz), theta (4–8 Hz), alpha (7.5–13 Hz), beta (13–30 Hz), and gamma (30–44 Hz). Power spectral density values were expressed as the logarithm of the absolute spectral power, typically ranging from −1 to +1.

Spectral power was selected as the primary feature due to its validated sensitivity to stress-related modulation and compatibility with the device’s resolution, while acknowledging that future analyses may integrate additional indices (e.g., instantaneous frequency, connectivity metrics) to enhance granularity.

To control for interindividual variability, resting baseline was used to normalise task-related data. Movement-related artefacts were monitored using the integrated accelerometer and, when detected, affected segments were excluded from the analysis. Additional visual inspection ensured removal of segments with eye blinks or muscular artefacts. The use of the Muse device facilitated ecological data collection but its limited spatial resolution constitutes a methodological constraint, preventing advanced source localization. EEG processing considered regional averages for the frontal and temporo-parietal sites, allowing for the evaluation of region-specific modulations in spectral power across task conditions.

For subsequent analyses, EEG spectral power values were averaged within each experimental phase and condition. This allowed the derivation of condition-specific neurophysiological profiles for each participant, which were then correlated with behavioural and psychometric measures.

2.5. Data Analyses

Statistical analyses were performed using Jamovi software (version 2.6.26). For EEG data, repeated-measures analyses of variance (ANOVAs) were conducted separately for each of the five canonical frequency bands—delta (1–4 Hz), theta (4–8 Hz), alpha (8–13 Hz), beta (13–30 Hz), and gamma (30–44 Hz). In the MSCP, the within-subject factors included Region of Interest (ROI: frontal vs. temporo-parietal) and scenario type (4 decision-making scenarios: 2 organisational—S1, S2—2 medical emergency—S3, S4). In the SST-NeuroEval, separate repeated-measures ANOVAs were run for the preparation and exposure phases, each including ROI (frontal vs. temporo-parietal) and trial number (5 preparation—Prep1–5—or 5 exposure trials—Exp1–5) as within-subject factors.

When Mauchly’s test indicated a violation of sphericity, degrees of freedom were adjusted using the Greenhouse-Geisser epsilon. In cases of statistically significant main effects or interactions, simple effects were explored using Bonferroni-corrected pairwise comparisons. Effect sizes were reported as partial eta squared (η²), with the significance threshold set at α = 0.05.

Behavioural indices derived from both the MSCP (EmoScore_Org, EmoScore_Med, CogScore_Org, CogScore_Med, RT_Emo_Org, RT_Cog_Org, RT_Emo_Med, RT_Cog_Med) and SST-NeuroEval (StressReg, StressTol, VAS-Prep, VAS-Exp, MetaStress, MetaTime) were analysed in parallel using repeated-measures ANOVAs where applicable.

Additionally, exploratory Pearson’s correlation analyses were performed to examine associations between behavioural outcomes, psychometric profiles, and EEG measures. Correlation models included the subscales and total scores of the administered self-report instruments (PSS-10, EROS, GDMS, BEES) alongside the behavioural indices from both experimental tasks. All significance values are reported as two-tailed, with p-values below 0.05 considered statistically. Bonferroni corrections were systematically applied to all pairwise comparisons and correlation analyses to control for multiple testing.

3. Results

3.1. Behavioural Results

3.1.1. MSCP Behavioural Results

In the MSCP, a main effect was found for decision type (F(1,19) = 9.72, p = 0.006, η² = 0.338), with a higher emotional than cognitive choices across domains. RTs showed a main effect of decision type (F(1,19) = 5.84, p = 0.026, η² = 0.235), with faster RTs for emotional than cognitive choices. The decision type × domain interaction was significant (F(1,19) = 6.21, p = 0.022, η² = 0.246). Post hoc Bonferroni-corrected comparisons indicated that, in both the organisational domain (p = 0.012) and the medical domain (p = 0.038), mean RTs for emotional decisions were significantly shorter than those for cognitive decisions (Figure 1a).

3.1.2. SST-NeuroEval Behavioural Results

In the SST-NeuroEval, subjective stress ratings (VAS-Prep vs. VAS-Exp) showed a main effect of phase (F(1,19) = 11.38, p = 0.003, η² = 0.375), with Bonferroni-corrected comparisons indicating higher VAS-Exp than VAS-Prep scores (p = 0.003) (Figure 1b).

3.2. EEG Results

3.2.1. MSCP EEG Results

Delta band. A main effect of ROI was observed (F(1,19) = 13.30, p = 0.002, η² = 0.042), with higher temporo-parietal than frontal delta power. A main effect of scenario was also found (F(2.14,40.68) = 23.20, p ≤ 0.001, η² = 0.194). Post hoc Bonferroni-corrected comparisons indicated higher values in S1 than in S2 (p ≤ 0.001) and S3 (p ≤ 0.001), and in S4 than in S2 (p ≤ 0.001). The ROI × scenario interaction was significant (F(3,57) = 21.40, p ≤ 0.001, η² = 0.245). In the frontal ROI, S1 was higher than S2 (p ≤ 0.001); S3 (p ≤ 0.001) and S4 were higher than S2 (p ≤ 0.001). In the temporo-parietal ROI, S1 was higher than S2 (p = 0.004) and S3 (p ≤ 0.001); S3 and S4 was higher than S2 (p = 0.003; p ≤ 0.001) (Figure 2).

Theta band. A main effect of ROI was found (F(1,19) = 46.60, p ≤ 0.001, η² = 0.041), with higher temporo-parietal than frontal theta power. A main effect of scenario was also observed (F(3,57) = 87.60, p ≤ 0.001, η² = 0.519). Post hoc tests showed higher values in S1 than S3 (p ≤ 0.001) and S4 (p ≤ 0.001), in S2 than S3 (p ≤ 0.001) and S4 (p ≤ 0.001), and in S3 than S4 (p ≤ 0.001).

The ROI × scenario interaction was significant (F(3,57) = 15.10, p ≤ 0.001, η² = 0.118). In the frontal ROI, theta power was higher in S1 than in S3 (p ≤ 0.001) and S4 (p ≤ 0.001), and in S2 than in S3 (p ≤ 0.001) and S4 (p ≤ 0.001). In the temporo-parietal ROI, theta power was higher in S3 than in S4 (p ≤ 0.001). Additionally, theta power was significantly lower in the temporo-parietal ROI than in the frontal ROI for S3 (p ≤ 0.001), and higher in the temporo-parietal ROI than in the frontal ROI for S4 (p ≤ 0.001).

Gamma band. A main effect of scenario was detected (F(3,57) = 14.93, p ≤ 0.001, η² = 0.163), with post hoc comparisons showing higher values in S1 than in S2 (p ≤ 0.001) and S3 (p ≤ 0.001), and in S4 than in S2 (p ≤ 0.001).

The ROI × scenario interaction was significant (F(2.12,40.37) = 13.70, p ≤ 0.001, η² = 0.194). In the frontal ROI, gamma power was higher in S4 than in S2 (p ≤ 0.001) and in S1 than in S3 (p ≤ 0.001). In the temporo-parietal ROI, gamma power was higher in S1 than in S2 (p ≤ 0.001), in S3 (p ≤ 0.001), and in S4 (p = 0.022). Additionally, gamma power was significantly higher in the temporo-parietal ROI than in the frontal ROI for S4 (p = 0.001) and significantly higher in the frontal ROI than in the temporo-parietal ROI for S1 (p ≤ 0.001).

No other significant results were found (all p < 0.05).

3.2.2. SST-NeuroEval EEG Results

Delta band. A main effect of preparation was observed (F(4,76) = 68.23, p ≤ 0.001, η² = 0.597). Post hoc Bonferroni-corrected comparisons indicated higher values in Prep2 than in Prep1 (p ≤ 0.001), Prep3 (p ≤ 0.001), and Prep5 (p ≤ 0.001); Prep3 than Prep1 (p ≤ 0.001); Prep4 than Prep1 (p ≤ 0.001), Prep3 (p ≤ 0.001), and Prep5 (p ≤ 0.001).

A main effect of exposure was also found (F(4,76) = 113.77, p ≤ 0.001, η² = 0.624). Post hoc comparisons showed higher values in Esp1 than in Esp3 (p ≤ 0.001), Esp4 (p ≤ 0.001), and Esp5 (p ≤ 0.001); Esp2 than Esp1 (p ≤ 0.001), Esp3 (p ≤ 0.001), Esp4 (p ≤ 0.001), and Esp5 (p ≤ 0.001); Esp3 than Esp5 (p ≤ 0.001); and Esp4 than Esp5 (p ≤ 0.001).

Theta band. A main effect of exposure was found (F(2.42,45.99) = 44.31, p ≤ 0.001, η² = 0.459). Post hoc comparisons indicated higher values in Esp2 than in Esp1 (p ≤ 0.001) and Esp4 (p ≤ 0.001); Esp3 than Esp1 (p ≤ 0.001) and Esp4 (p ≤ 0.001); and Esp5 than Esp1 (p ≤ 0.001), Esp2 (p ≤ 0.001), and Esp4 (p ≤ 0.001).

Gamma band. A main effect of preparation was detected (F(2.60,49.41) = 23.54, p ≤ 0.001, η² = 0.347). Post hoc comparisons revealed higher values in Prep1 than in Prep4 (p ≤ 0.001), Prep2 than in Prep4 (p ≤ 0.001), Prep3 than in Prep4 (p ≤ 0.001), and Prep5 than in Prep4 (p ≤ 0.001). A main effect of exposure was also observed (F(2.81,53.38) = 46.60, p ≤ 0.001, η² = 0.507). Post hoc comparisons showed higher values in Esp4 than in Esp1 (p ≤ 0.001), Esp2 (p ≤ 0.001), Esp3 (p ≤ 0.001), and Esp5 (p ≤ 0.001); and in Esp5 than in Esp1 (p = 0.003) and Esp2 (p = 0.016) (Figure 3).

No other significant results were found (all p < 0.05).

3.3. Correlations Results

EmoScore_Org correlated positively with VAS-Exp (r = 0.448, p = 0.048). CogScore_Org correlated positively with MetaTime (r = 0.463, p = 0.040). RTs for emotional responses in the organisational domain correlated negatively with MetaStress (r = −0.518, p = 0.019). RTs for emotional responses in the medical domain correlated negatively with VAS-Exp (r = −0.462, p = 0.040).

PSS-10 scores correlated positively with VAS-Prep (r = 0.528, p = 0.017) and VAS-Exp (r = 0.560, p = 0.010). StressReg correlated negatively with EROS-Intrinsic_Affect_Worsening (r = −0.588, p = 0.006). StressTol correlated negatively with EROS-Intrinsic_Affect_Worsening (r = −0.493, p = 0.027). GDMS rational style correlated positively with MetaStress (r = 0.585, p = 0.007). GDMS dependent style correlated positively with VAS-Prep (r = 0.606, p = 0.005). GDMS avoidant style correlated positively with VAS-Prep (r = 0.680, p < 0.001).

4. Discussion

The present findings provide converging evidence that individual differences in decision-making orientation and stress regulation are systematically reflected across behavioural, subjective, and neurophysiological indices. By integrating a metacognitive decision-making paradigm with a socially evaluative stress task, the study offers a multilevel perspective on how cognitive–emotional processes unfold under controlled yet ecologically valid conditions. This discussion addresses the behavioural, EEG, and correlational patterns observed, situating them within current theoretical frameworks and identifying their potential implications for applied settings.

4.1. Behavioural Adaptation and Decision-Making Under Stress

Behavioural results from the MSCP indicated a clear predominance of emotional over cognitive–analytical choices, accompanied by significantly shorter RTs for emotional responses. This pattern suggests preferential activation of decision-making processes grounded in the rapid retrieval of affectively salient representations, consistent with dual-process models of decision-making [1,2]. Within such models, immediate access to socio-emotional heuristics is supported by limbic–frontal networks, which can substantially enhance response speed, even at the expense of extended analytical deliberation.

These observations align with recent literature emphasising that decision-making under temporal pressure or social–evaluative stress emerges from a dynamic interplay between bottom-up affective processes and top-down cognitive control mechanisms [3,4,5]. Under stress, neural pathways involved in the rapid appraisal of socially and biologically salient cues—particularly those connecting the amygdala, orbitofrontal cortex, and anterior cingulate cortex—may assume a predominant role, favouring decisions aimed at immediate threat reduction or optimisation of emotional resonance with the social context [11,42,43,44].

In the SST-NeuroEval, the systematic increase in perceived stress from the preparation to the exposure phase confirms the paradigm’s effectiveness in eliciting a socially evaluative load with high ecological validity. This increase is consistent with contemporary models of social–evaluative stress [11,34,45,46], which posit that confrontation with an evaluative audience—whether real or virtual—triggers co-activation of limbic–hypothalamic and fronto-parietal circuits, producing measurable effects on both physiological responses and cognitive self-regulation processes.

Taken together, these behavioural findings support the view that evaluative pressure and task demands can substantially influence the trajectory of cognitive and emotional strategy selection. They align with integrative models conceptualising decision-making under stress as a multilevel co-regulation process between affective and cognitive systems, dynamically modulating decision priorities and balancing accuracy–speed trade-offs [18,36,47].

Within this framework, the observed predominance of rapid emotional responses may be interpreted not as a simple reduction in analytical capacity, but rather as an adaptive strategy aimed at preserving behavioural responsiveness under conditions of social and temporal pressure.

4.2. EEG Correlates of Cognitive–Emotional Processing in Decision-Making (MSCP)

The EEG analysis conducted within the MSCP revealed an oscillatory architecture markedly modulated by the variables scenario and ROI, with distinct profiles in the Delta, Theta, and Gamma frequency bands. The most relevant differences emerged between scenarios with high socio-emotional salience (S1, S4), scenarios with neutral valence and lower affective urgency (S3), and intermediate scenarios characterised by a combination of pragmatic constraints and interpersonal relevance (S2).

Delta activity exhibited significant frontal increases in S1 and S4 compared to the other conditions, with marked differences relative to S2 and S3. In S3, an inverse topographical pattern was observed, with predominance in temporo-parietal regions over frontal regions, indicating a functional reconfiguration as a function of scenario characteristics. In the temporo-parietal ROI, delta power was significantly elevated in S1, S4, and S2 compared to S3.

This pattern is consistent with models linking frontal delta to motivated attention, selective orienting, and rapid decision synthesis in contexts of high affective pressure [15,16,17,48], while temporo-parietal delta is associated with contextual encoding and semantic–narrative reconstruction. The shift from frontal predominance (S1, S4) to temporo-parietal predominance (S3) reflects a flexible resource allocation mechanism [41] whereby oscillatory input is redistributed according to the interplay between affective urgency and contextual complexity.

Frontal theta power was significantly higher in S1 and S2 than in S3 and S4, whereas temporo-parietal Theta showed a marked increase in S3, exceeding that observed in S4 and, in some comparisons, in S1 and S2.

Frontal theta has been consistently associated with conflict monitoring, executive control, and the concurrent maintenance of multiple competing decision representations [21]—processes that are central in scenarios with strong normative conflict (S1) and pragmatic–interpersonal tension (S2). The temporo-parietal increase in S3 is consistent with the role of posterior Theta in mnemonic retrieval, contextual encoding, and semantic integration [24,49]. The inversion of ROI predominance between S3 and S4 illustrates a dynamic reorganisation of oscillatory networks in response to decision pressure and the semantic nature of the scenario.

Frontal gamma was significantly increased in S1 and S4 compared to S2 and S3, whereas temporo-parietal gamma showed higher values in S1, S4, and S2 relative to S3. The ROI-specific contrast was particularly evident: frontal predominance occurred in conditions with greater socio-emotional integrative demands, while temporo-parietal predominance was more pronounced in scenarios requiring extended representational construction.

These findings are consistent with models assigning gamma a pivotal role in multimodal feature binding and high-frequency interregional synchronisation [31,50]. Frontal Gamma enhancement in S1 and S4 indicates rapid integration of perceptual, moral, and emotional inputs into a unified decision, whereas temporo-parietal Gamma enhancement reflects engagement of mechanisms for interpersonal simulation and projection of decision consequences.

To sum up, the ROI × scenario distribution in Delta, Theta, and Gamma demonstrates that the MSCP does not elicit a uniform increase in oscillatory activity but rather a system of selective and topographically targeted modulations. These configurations align with multi-level co-regulation models [41,51,52,53], whereby frontal cortical networks—responsible for decision control and synthesis—and temporo-parietal networks—specialised in contextual integration and semantic representation—operate within a dynamic resource reallocation regime, adjusting in real time to the interplay between decision urgency, emotional salience, and contextual complexity.

4.3. EEG Correlates of Stress Regulation (SST-NeuroEval)

The EEG analysis of the Social Stress Test—NeuroEval (SST-NeuroEval), similarly, revealed marked modulations in delta, theta, and gamma frequency bands as a function of the experimental phase (preparation vs. exposure) and trial progression. The observed patterns delineate the transition from anticipatory cognitive–emotional regulation processes to reactive behavioural engagement under evaluative stress.

Delta activity displayed a distinctive temporal profile across both preparation and exposure phases. During preparation, power increased from Prep2 and was maintained at Prep4, relative to the initial (Prep1) and terminal (Prep5) stages. This suggests that Delta mobilisation is not immediate but emerges after an initial alignment of cognitive–emotional resources to the impending evaluative situation, peaking at an intermediate stage before declining as direct exposure approaches.

During exposure, Delta power was elevated in the first two trials (Esp1, Esp2) and progressively decreased thereafter. This pattern indicates strong early engagement of attentional orienting mechanisms and low-frequency processing systems that register and appraise socially relevant cues at the moment of maximal evaluative impact. The subsequent decrease may reflect habituation or partial reallocation of resources towards higher-frequency bands, in response to reduced novelty or progressive adaptation to social demands.

From a theoretical perspective, these findings align with models linking Delta to motivated attention and to widespread cortical synchronisation under social stress conditions [54,55]. In preparation, the intermediate delta increase may represent peak activation of fronto-limbic and fronto-parietal circuits involved in performance readiness; in exposure, the initial Delta peak likely reflects maximal alertness and environmental registration, followed by oscillatory rebalancing as the situation becomes more familiar.

Theta showed significant increases in specific exposure trials (Esp2, Esp3, Esp5) compared to phases of lower engagement. The temporal profile suggests oscillatory peaks coinciding with moments of heightened demand for cognitive control and emotional regulation. This modulation is consistent with the role of frontal theta in conflict monitoring and management of cognitive–affective load [20,56], and may also reflect context updating processes when the social stressor retains dynamic and unpredictable features [17,57,58].

Gamma activity exhibited a biphasic pattern. In preparation, power was reduced in Prep4 relative to all other stages, suggesting a transient disengagement of integrative mechanisms immediately prior to direct exposure. During exposure, gamma power markedly increased in Esp4, exceeding all other phases.

These findings accord with the established role of gamma in rapid multimodal integration and perceptual–decisional synthesis [35]. The pre-exposure reduction may index a strategic suspension of integrative processing, followed by a potentiated reactivation during peak evaluative engagement.

Overall, the SST-NeuroEval EEG data indicate that delta, theta, and gamma bands are modulated in a phase-specific manner, delineating an oscillatory cycle that parallels the shift from anticipatory regulation to reactive engagement under stress.

4.4. Integrated Behavioural–Neurophysiological Relationships

The integration of behavioural, psychometric, and EEG results provides converging evidence that individual differences in decision-making tendencies and stress regulation capacities manifest coherently across multiple measurement levels. Correlational analyses revealed systematic associations linking task-derived indices, self-reported measures, and oscillatory activity patterns, highlighting the functional interplay between cognitive–emotional style and stress responsivity.

From a behavioural perspective, a higher emotional decisions in organisational scenarios was associated with elevated subjective stress ratings during the exposure phase of the SST-NeuroEval. This suggests that a preferential orientation towards affect-laden decision strategies may amplify perceived stress when confronted with evaluative social demands. Conversely, higher cognitive–analytical decision scores were related to greater temporal awareness of stress, pointing to a more deliberate monitoring of stress dynamics across task phases.

RTs further supported this relationship: faster latencies for emotional decisions were negatively correlated with stress metacognition measures, indicating that rapid affective responses may be accompanied by reduced explicit appraisal of stress states. These findings align with dual-process models in which fast, heuristic-based decision-making relies less on explicit metacognitive monitoring and more on prepotent emotional evaluations [11,59].

Self-report measures corroborated these patterns. Higher PSS-10 scores were consistently associated with increased subjective stress ratings in both preparation and exposure phases, reinforcing the ecological validity of the SST-NeuroEval as a stress-induction paradigm. Stress regulation and tolerance indices were inversely related to self-reported deterioration in intrinsic affect (EROS), in line with models linking effective emotion regulation under stress to preserved affective stability [60].

Oscillatory activity patterns also converged with these behavioural and psychometric profiles. Participants with higher rational decision-making tendencies (GDMS) displayed stronger associations with stress metacognition scores, suggesting a shared reliance on controlled, reflective processing. In contrast, higher dependent and avoidant decision styles correlated with increased subjective stress in preparation, implying a heightened anticipatory affective load.

Together, these multi-domain relationships underscore the value of integrating behavioural indices, subjective reports, and neurophysiological signatures to capture the dynamic coupling between decision orientation and stress regulation. Such integration not only enhances theoretical understanding of cognitive–emotional interaction under evaluative pressure but also supports the development of personalised intervention models that target both decision-making style and stress resilience.

5. Conclusions: Theoretical and Applied Implications, Limitations, and Future Directions

In conclusion, from a theoretical perspective, these results refine current models of decision-making under stress by demonstrating that the integration of task-specific decision contexts and social–evaluative demands produces distinct oscillatory “signatures” that can be reliably mapped onto decision orientation and stress responsivity. The observed convergence across measurement levels underscores the necessity of approaching cognitive–emotional regulation as a multi-domain construct that cannot be adequately characterised by a single level of analysis.

Applied implications emerge for contexts where decision-making is embedded within socially evaluative or high-stakes environments, such as organisational leadership, emergency response, or clinical decision-making under time pressure. Beyond these, potential applications extend to educational and training settings, where stress regulation and decision style profiling may inform targeted curricula and resilience-building programmes. In organisational domains, these findings could inform resilience-oriented training for leaders and teams exposed to evaluative pressure. In clinical practice, the integration of behavioural, subjective, and EEG indices may support diagnostic refinement in stress-related disorders and guide personalised interventions targeting cognitive–emotional regulation.

Profiling individuals based on their decision orientation, stress metacognition, and oscillatory markers may inform personalised training protocols aimed at optimising decision quality and resilience under stress.

However, certain limitations must be acknowledged. First, the study employed a relatively small and homogeneous preliminary sample, which constrains the generalisability of the findings and may have limited the detection of subtler interaction effects. Furthermore, although the present study was not designed for systematic gender comparisons, future larger-scale investigations may explore potential gender-related effects in stress responsivity and cognitive–emotional regulation. Furthermore, only spectral power was considered, while other digital features could provide additional granularity in future studies.

Second, while the EEG Muse Headband allowed for ecologically flexible acquisition, its reduced spatial resolution precludes fine-grained source localisation. Finally, the correlational nature of the multi-domain analyses precludes strong causal inferences regarding the directionality of the observed associations.

Future research should extend this work by employing larger and more diverse samples, integrating higher-density EEG or multimodal neuroimaging to refine spatial characterisation of the observed oscillatory patterns, and adopting counterbalanced or adaptive task sequencing to disentangle order-related effects. Longitudinal or intervention-based designs could also determine whether targeted modulation of decision-making strategies and stress regulation capacities produces measurable shifts in oscillatory dynamics and behavioural outcomes.

In conclusion, the present study advances the understanding of cognitive–emotional regulation under stress by demonstrating a consistent multi-level correspondence between decision-making orientation, subjective stress appraisal, and oscillatory brain activity. This integrated framework not only enriches theoretical models of decision-making under evaluative pressure but also offers practical pathways for developing tailored strategies to enhance resilience, adaptability, and performance in socially demanding contexts.