Pain-Related Fear Modulates Attentional Bias for Threat Processing in Chronic Pain: An ERP Study.

Chronic pain is characterized by a threat-related attentional bias that prioritizes threatening stimuli at the expense of contextually relevant information. 1–4 Importantly, this cognitive pattern represents a key mechanistic pathway linking pain-related fear (PRF) to maladaptive avoidance and pain chronicity. 5–8 The clinical significance of this mechanism is underscored by recent evidence that targeted modulation of prefrontal fear-related circuits can attenuate pain sensitization, highlighting PRF as a promising therapeutic target. 9 , 10 In epigastric pain syndrome (EPS), where symptoms lack identifiable pathology, pain-related fear may extend beyond the fear of pain itself to encompass heightened concern about what the pain signifies—including fear of progression and deteriorating health. This broader manifestation represents a clinically salient form of PRF in EPS, yet its impact on attentional processing remains unknown. Existing research has relied almost exclusively on pain-specific cues, overlooking evidence that attentional biases in chronic pain extend to socially salient stimuli such as emotional faces. The present study departs from this tradition by employing fearful, sad, and happiness faces to examine whether PRF modulates attention in a threat-graded manner, thereby providing a more ecologically valid investigation of the cognitive architecture underlying PRF. The Fear-Avoidance Model (FAM) provides the dominant theoretical framework for understanding how PRF perpetuates chronic pain. 11–14 According to this model, when pain is appraised as threatening—a process heavily influenced by catastrophic misinterpretations—it triggers a cascade of fear, hypervigilance, and avoidance behaviors that ultimately exacerbate disability and maintain the chronic pain cycle. 12 , 14 Within this framework, pain catastrophizing amplifies the perceived threat value of pain, while kinesiophobia drives the behavioral avoidance that prevents disconfirmation of these threat beliefs. Consistent with this conceptualization, kinesiophobia and pain catastrophizing are frequently examined together as key FAM constructs. 15 Such fear, often acquired through interoceptive experiences, 16 , 17 promotes hypervigilance and behavioral avoidance. 11 , 18 , 19 Despite the centrality of PRF in the FAM, critical questions remain regarding its precise modulation of attention. The empirical landscape reveals considerable complexity. First, the relationship between pain-related fear and attentional bias may be nonlinear, with biases most pronounced at low to moderate levels of fear. 6 Second, PRF appears to exhibit temporal specificity, selectively modulating early, automatic stages of attention (vigilance) while exerting distinct effects on later, controlled stages (disengagement). 20–23 Third, while elevated PRF has been associated with threat hypervigilance, 24–26 the nature of its relationship with inhibitory control remains debated: some evidence indicates that PRF may impair disengagement from threat, 4 , 18 whereas other findings point to more nuanced, context-dependent effects. 27 , 28 Critically, whether this reflects a generalized processing deficit or a threat-specific alteration remains unresolved, as prior research has not directly tested PRF-related effects across stimuli that systematically vary in threat salience. Neuroscientific evidence provides convergent insight into the temporal dynamics of threat processing. Impaired cognitive control in chronic pain has been linked to dysfunction within shared pain–cognition networks, including the dorsal anterior cingulate–prefrontal circuit. 29 , 30 Event-related potential (ERP) studies have elucidated distinct processing stages: early components such as the N1 have been associated with rapid threat capture, whereas later components such as the N2 have been implicated in conflict processing and attentional disengagement. 31 Fear conditioning paradigms have demonstrated that conditioned threat signals can alter attentional allocation toward pain-related threat signals and modulate disengagement. 17 , 32 Pain-related fear, by consuming cognitive resources through sustained vigilance, 33 , 34 may influence the neural efficiency required for effective top-down control. Yet, the precise manner in which PRF modulates each of these processing stages—and whether such modulation is generalized across emotional contexts or specific to high-threat stimuli—remains to be established. The present study directly addresses this gap by systematically comparing ERP responses to emotional faces of varying threat intensities in individuals with high versus low levels of PRF. In EPS, the clinical presentation of PRF may take a specific form. Unlike acute pain conditions where fear centers on the immediate painful sensation, the diagnostic uncertainty inherent to EPS—where symptoms are recurrent and unpredictable yet lack identifiable structural pathology—may shift the focus of fear from the pain itself to its potential implications: disease progression, loss of control, and deteriorating health. Longitudinal evidence from chronic visceral pain demonstrates that pain reduction precedes psychological improvement, not the reverse, 35 consistent with the proposal that pain experience may contribute to subsequent emotional distress. This pathway is likely amplified in EPS where the absent of biomarkers preclude diagnostic reassurance. This pattern aligns with evidence that pain-related fear operates as a transdiagnostic construct across chronic conditions, 36 is elevated in visceral pain populations such as endometriosis, 37 and is associated with health threat-related interpretation bias. 38 , 39 Within the FAM, PRF is distinct from yet functionally linked to pain catastrophizing and kinesiophobia: catastrophizing magnifies the perceived threat value of pain, while kinesiophobia drives the fear-based avoidance of movement. PRF is thought to arise when catastrophic interpretations of pain fuels a generalized fear response—a core tenet of the FAM that provides the theoretical justification for our multidimensional operationalization. Accordingly, we adopted a composite index integrating catastrophizing (assessed via the PCS) and kinesiophobia (assessed via the TSK) to capture the full spectrum of pain-related fear. To comprehensively capture how PRF-driven attentional bias operates, two methodological gaps must be addressed. First, existing research has relied heavily on pain-specific cues, overlooking evidence that individuals with chronic pain exhibit attentional biases toward a broader range of negative stimuli, including fearful faces, disgust-related cues, and negative words. 1 , 40–42 Emotional faces represent ecologically potent threat signals that capture attention rapidly and automatically, yet their processing in the context of PRF remains largely unexamined. Second, threat has typically been treated as a unitary construct, with limited consideration of whether PRF differentially modulates the processing of stimuli varying in threat salience. To address these gaps, the present study employed fearful and sad faces as high-threat and low-threat stimuli, respectively. Fear was selected as an evolutionarily grounded, high-arousal threat signal, whereas sadness was chosen as a low-arousal negative emotion 43–45 that is particularly relevant to chronic pain: it enhances pain unpleasantness via disruption of emotion regulation circuitry 46 and is psychologically intertwined with somatic pain in these patients. 47 This distinction is important, as fear and sadness engage partially dissociable motivational systems, 48 and show differential associations with pain sensitivity. 49–51 In sum, departing from traditional paradigms that rely on pain-specific cues, the present study employed emotional faces varying in threat salience—high-threat (fearful), low-threat (sad), and non-threat (happiness)—within a three-stimulus oddball paradigm combined with ERPs. This approach enables sensitive detection of attentional differences 52–54 and enhanced ecological validity for capturing how PRF modulates the socio-affective signals encountered in daily life. 55 , 56 Consistent with the FAM’s emphasis on threat processing, we hypothesized that PRF-related group differences would be most pronounced for high-threat (fearful) stimuli. This threat-specificity hypothesis follows directly from the FAM’s core prediction that fear amplifies the salience of threatening information. By characterizing the temporal dynamics and threat-specificity of these effects, the findings aim to provide preliminary neurophysiological evidence to inform future research on attentional processing in PRF.

No significant between-group differences were observed in age, gender, residence, or educational level (all ps >0.05), confirming that the groups were well matched on key background characteristics. Consistent with the grouping strategy, the high-PRF group scored significantly higher than the low-PRF group on the two scales constituting the composite index —the PCS (Z = –4.575, p <0.001) and TSK (Z = –4.598, p <0.001)—as well as on the FoP-Q-SF (Z = –4.204, p <0.001), which served as an independent validation measure. These results confirm that the composite successfully discriminated individuals with markedly divergent levels of fear of progression, establishing the high- and low-PRF groups as representing meaningfully distinct profiles on the core construct under investigation. ( Table 2 ). Table 2 Characteristics of Patients Stratified by PRF Level Variable Low-PRF Group (n = 15) High-PRF Group (n = 14) Statistic p Age,years (Q1,Q3) 46.00 (42.00, 48.00) 46.00 (37.75, 56.25) Z = –0.241 0.813 a Gender, n (%) χ 2 =1.007 0.450 b  Male 7 (46.7%) 4 (28.6%)  Female 8 (53.3%) 10 (71.4%) Residence, n (%) — 0.169 b  Rural 1 (6.7%) 4 (28.6%)  Urban 14 (93.3%) 10 (71.4%) Education level, n (%) — — b Elementary school 0 (0.0%) 2 (14.3%) Junior high school 8 (53.3%) 3 (21.4%) Senior high school 6 (40.0%) 8 (57.1%) Associate degree 1 (6.7%) 0 (0.0%) Postgraduate 0 (0.0%) 1 (7.1%) FOPQ (Q1,Q3) 33.00 (31.00, 34.00) 37.00 (36.00, 39.00) Z = –4.204 <0.001 a PCS total score (Q1,Q3) 23.00 (20.00, 25.00) 29.50 (28.00, 33.00) Z = –4.575 <0.001 a Rumination (Q1,Q3) 6.00 (6.00, 8.00) 10.00 (8.75, 12.25) Z = –3.848 <0.001 a Magnification (Q1,Q3) 6.00 (5.00, 7.00) 6.00 (5.00, 7.25) Z = –0.157 0.880 a Helplessness (Q1,Q3) 9.00 (8.00, 12.00) 14.50 (12.00, 16.00) Z = –3.555 <0.001 a TSK total score (Q1,Q3) 34.00 (32.00, 35.00) 41.00 (39.00, 43.75) Z = –4.598 <0.001 a Activity avoidance (Q1,Q3) 22.00 (19.00, 22.00) 26.50 (25.00, 30.25) Z = –4.208 <0.001 a Somatic focus (Q1,Q3) 12.00 (10.00, 13.00) 15.00 (13.75, 16.25) Z = –3.128 0.001 a PCS transformation score –0.54 (–0.77, −0.38) –0.04 (–0.15, 0.23) Z = –4.575 <0.001 a TSK transformation score –0.18 (–0.29, −0.12) 0.24 (0.12, 0.40) Z = –4.598 <0.001 a Composite transformation score –0.83 (–0.89, –0.56) 0.27 (0.07, 0.54) Z = –4.583 <0.001 a Note : Continuous variables are presented as median (Q1, Q3), and categorical variables are presented as n (%). Group comparisons were performed using the Mann–Whitney U -test (a) and chi-square test (b). Given the small sample size and the presence of zero cells in the education variable, between-group comparisons were not conducted for education level; only descriptive statistics are reported. Characteristics of Patients Stratified by PRF Level Note : Continuous variables are presented as median (Q1, Q3), and categorical variables are presented as n (%). Group comparisons were performed using the Mann–Whitney U -test (a) and chi-square test (b). Given the small sample size and the presence of zero cells in the education variable, between-group comparisons were not conducted for education level; only descriptive statistics are reported. For the target-level analysis, the main effect of target was significant, F(1, 27) = 110.547, p <0.001, η 2 = 0.804, indicating that RTs were longer for deviant stimuli than for standard stimuli (see Figure 2 ). The main effect of group was not significant, F(1, 27) = 0.476, p = 0.496, η 2 p = 0.017. The group × target interaction was not significant, F(1, 27) = 0.001, p = 0.980, η 2 p = 0.000. Figure 2 Reaction time (RT) and accuracy (ACC) for standard and deviant stimuli in pain patients with high- and low-PRF. ( A ) Accuracy rates for standard and deviant stimuli in patients with high and low-PRF. ( B ) Reaction times for standard and deviant stimuli in patients with high- and low-PRF. ( C ) Accuracy rates for all participants in response to deviant stimuli (fear, sad, happy) versus the neutral standard stimulus. ( D ) Reaction times for all participants in response to deviant stimuli (fear, sad, happy) versus the neutral standard stimulus. *** p< 0.001. A grouped bar graph set showing accuracy and reaction time for standard and deviant stimuli and emotions. Image A: Grouped bar graph labeled ACC. X-axis: H-FOP and L-FOP with Deviant and Standard Stimulus. Y-axis: ACC, range 0.0 to 1.0. H-FOP: Deviant ~0.90, Standard ~1.00. L-FOP: Deviant ~0.88, Standard ~0.99. Brackets labeled ***. Image B: Grouped bar graph labeled RT. X-axis: H-FOP and L-FOP with Deviant and Standard Stimulus. Y-axis: RT, range 0 to 800. H-FOP: Deviant ~620, Standard ~480. L-FOP: Deviant ~640, Standard ~500. Brackets labeled ***. Image C: Bar graph labeled ACC. X-axis: Neutral, Fear, Happy, Sad. Y-axis: ACC, range 0.0 to 1.0. Values: Neutral ~1.00, Fear ~0.95, Happy ~0.88, Sad ~0.86. Bracket labeled ***. Image D: Bar graph labeled RT. X-axis: Neutral, Fear, Happy, Sad. Y-axis: RT, range 0 to 800. Values: Neutral ~490, Fear ~610, Happy ~620, Sad ~630. Bracket labeled ***. Reaction time (RT) and accuracy (ACC) for standard and deviant stimuli in pain patients with high- and low-PRF. ( A ) Accuracy rates for standard and deviant stimuli in patients with high and low-PRF. ( B ) Reaction times for standard and deviant stimuli in patients with high- and low-PRF. ( C ) Accuracy rates for all participants in response to deviant stimuli (fear, sad, happy) versus the neutral standard stimulus. ( D ) Reaction times for all participants in response to deviant stimuli (fear, sad, happy) versus the neutral standard stimulus. *** p< 0.001. For the emotion-type analysis, the main effect of emotion type was significant, F(3, 81) = 78.452, p <0.001, η 2 p = 0.744, indicating that RTs differed across the four emotion types. Post hoc comparisons showed that RTs were shorter for neutral faces than for fearful, happy, and sad faces. The main effect of group was not significant, F(1, 27) = 0.365, p = 0.551, η 2 p = 0.013. The group × emotion type interaction was not significant, F(3, 81) = 0.153, p = 0.928, η 2 p = 0.006 (see Table 3 ). Table 3 Behavioral Performance in Response to Emotional Faces in Patients with High- and Low- PRF Reaction Time Accuracy H-PRF L-PRF H-PRF L-PRF Neutral 502.213±68.222 500.014±83.562 0.996±0.004 0.994±0.007 Fear 631.182±114.761 621.936±86.932 0.946±0.064 0.962±0.052 Happy 636.102±114.865 623.875±84.445 0.902±0.114 0.897±0.089 Sad 635.040±103.208 634.113±75.149 0.886±0.086 0.887±0.085 Group effect F =0.038, p =0.846, η 2 p =0.001 F =0.026, p =0.872, η 2 p =0.001 Emotion effect F =40.126, p =0.000, η 2 p =0.822 F =19.282, p =0.000, η 2 p =0.408 Group × Emotion Interaction Effects F =0.200, p =0.895, η 2 p =0.023 F =0.497, p =0.487, η 2 p =0.017 Behavioral Performance in Response to Emotional Faces in Patients with High- and Low- PRF For the target-level analysis, the main effect of target was significant, F(1, 27) = 57.843, p <0.001, η 2 p = 0.682, indicating that accuracy was lower for deviant stimuli than for standard stimuli (see Figure 2 ). The main effect of group was not significant, F(1, 27) = 1.167, p = 0.290, η 2 p = 0.041. The group × target interaction was not significant, F(1, 27) = 0.171, p = 0.682, η 2 p = 0.006. For the emotion-type analysis, the main effect of emotion type was significant, F(3, 81) = 22.573, p <0.001, η 2 p = 0.455, indicating that accuracy differed across the four stimulus types. Post hoc comparisons showed that accuracy was highest for neutral faces and that accuracy for fearful faces was higher than that for happy and sad faces. The main effect of group was not significant, F(1, 27) = 0.746, p = 0.395, η 2 p = 0.027. The group × emotion type interaction was not significant, F(3, 81) = 0.068, p = 0.997, η 2 p = 0.003 (see Table 3 ). All ERP results reported below are based on difference-wave amplitudes (deviant minus standard), indexing neural sensitivity to emotional deviation rather than absolute ERP responses. Mean difference-wave amplitudes for the N1, P2, N2, and P3 components across emotional conditions and groups are summarized in Table 4 . Table 4 Amplitude Differences (N1, P2, N2, P3) Between Emotional Deviants (Fear, Happy, Sad) and Standard Stimuli in Patients with High- and Low-PRF ( \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}$\bar x \pm s$\end{document} ) Fear Happy Sad H-PRF L-PRF H-PRF L-PRF H-PRF L-PRF N1 −0.25±1.63 0.65±1.34 0.52±1.36 −0.17±1.29 0.61±1.50 −0.20±1.15 P2 2.53±1.93 4.82±2.70 1.38±1.23 1.57±2.26 2.39±1.62 2.19±2.95 N2 −1.19±2.19 0.72±3.22 −0.64±2.74 0.00±2.46 0.60±3.29 0.62±2.69 P3 3.43±2.62 6.20±4.15 2.81±2.58 2.84±2.77 2.82±4.15 3.45±2.82 Amplitude Differences (N1, P2, N2, P3) Between Emotional Deviants (Fear, Happy, Sad) and Standard Stimuli in Patients with High- and Low-PRF ( \documentclass[12pt]{minimal} \usepackage{wasysym} \usepackage[substack]{amsmath} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage[mathscr]{eucal} \usepackage{mathrsfs} \DeclareFontFamily{T1}{linotext}{} \DeclareFontShape{T1}{linotext}{m}{n} {linotext }{} \DeclareSymbolFont{linotext}{T1}{linotext}{m}{n} \DeclareSymbolFontAlphabet{\mathLINOTEXT}{linotext} \begin{document}$\bar x \pm s$\end{document} ) The N1 difference-wave amplitude was analyzed at electrode O1 (80–130 ms). The main effect of Emotion was not significant, F(2, 54) = 0.008, p = 0.992, η 2 p = 0.000. The main effect of Group was also not significant, F(1, 27) = 0.300, p = 0.588, η 2 p = 0.011. A significant group × emotion interaction was observed, F(2,54) = 4.612, p = 0.014, η 2 p = 0.146 (see Figure 3 ). Simple effects revealed no between-group differences within any single emotion condition (all p s > 0.05). Descriptively, the low-PRF group showed larger N1 amplitudes than the high-PRF group under fearful emotion (0.65 vs −0.25), whereas the high-PRF group showed larger N1 amplitudes than the low-PRF group under happy (0.52 vs −0.17) and sad emotion (0.61 vs −0.20). The interaction effect on N1 is illustrated in Figure 4a . Figure 3 Difference-wave amplitudes (N1, P2, N2, P3) for emotional deviant stimuli in patients with High- and Low-PRF. Panels ( a – d ) illustrate amplitude differences across three emotional conditions (fearful, happy, sad) for each component. Panels ( e – h ) depict amplitude differences between high- and low-PRF groups for each component. * p < 0.05, *** p < 0.001, ns, non-significant. A set of eight bar charts showing difference wave amplitudes by emotion and by group. Image A: Bar chart with labels Fear, Happy, Sad; Mean range 0.0-0.8. Heights: Fear ~0.2, Happy ~0.15, Sad ~0.18. ns comparisons. Image B: Same labels; Mean range 0.0-5.0. Heights: Fear ~3.7, Happy ~1.5, Sad ~2.3. Triple asterisk Fear-Happy, asterisk Fear-Sad, ns Happy-Sad. Image C: Same labels; Mean range -0.3-1.5. Zero line shown. Heights: Fear ~-0.2, Happy ~-0.3, Sad ~0.6. ns comparisons. Image D: Same labels; Mean range 0-7. Heights: Fear ~4.8, Happy ~2.8, Sad ~3.1. Triple asterisk Fear-Happy, asterisk Fear-Sad, ns Happy-Sad. Image E: Labels L PRF, H PRF; Mean range 0.0-1.2. Heights: L PRF ~0.1, H PRF ~0.3. ns comparison. Image F: Same labels; Mean range 0-4. Heights: L PRF ~2.8, H PRF ~2.1. ns comparison. Image G: Same labels; Mean range -0.6-1.8. Zero line shown. Heights: L PRF ~0.4, H PRF ~-0.4. ns comparison. Image H: Same labels; Mean range 0-6. Heights: L PRF ~4.1, H PRF ~3.0. ns comparison. Figure 4 Interaction effects between PRF level and emotion type on N1, P2, and P3 difference-wave amplitudes. ( a ) For the N1 component, a significant interaction was observed. Simple effects analysis revealed no significant differences between the high- and low-PRF groups for fearful, happy, or sad faces. ( b ) For the P2 component, a significant interaction was observed. Simple effects analysis showed that the low-PRF group exhibited significantly larger P2 amplitudes than the high-PRF group in response to fearful faces, whereas no significant between-group differences were found for happy or sad faces. ( c ) For the P3 component, a significant interaction was also observed. Simple effects analysis indicated that the low-PRF group exhibited significantly larger P3 amplitudes than the high-PRF group in response to fearful faces, whereas no significant between-group differences were found for happy or sad faces. A three part multi line graph comparing L PRF and H PRF mean microvolt by emotion for N1, P2, P3. Image A: Multi-line graph with X-axis ′Emotion′ (Fear, Happy, Sad) and Y-axis ′Mean (microvolt)′ from -1.2 to 1.3. L PRF (solid line, circles) peaks at Fear (0.65), dips at Sad (-0.20). H PRF (dashed line, triangles) peaks at Sad (0.61), dips at Fear (-0.25). Lines cross between Fear and Happy. Error bars included. Image B: Graph with X-axis ′Emotion′ and Y-axis ′Mean (microvolt)′ from -1.0 to 7.0. L PRF values: Fear (4.8), Happy (1.4), Sad (2.2); H PRF values: Fear (2.5), Happy (1.3), Sad (2.3). L PRF highest at Fear, lowest at Happy. H PRF peaks at Fear, lowest at Happy. Largest error bars at Fear for L PRF, overlap at Happy and Sad. Image C: Graph with X-axis ′Emotion′ and Y-axis ′Mean (microvolt)′ from -2.0 to 10.0. L PRF values: Fear (6.2), Happy (3.0), Sad (3.5); H PRF values: Fear (3.5), Happy (3.0), Sad (3.0). Groups meet at Happy (3.0). L PRF highest at Fear, lowest at Happy. H PRF peaks at Fear, lowest at Sad. Largest error bars at Fear, overlap at Happy. Difference-wave amplitudes (N1, P2, N2, P3) for emotional deviant stimuli in patients with High- and Low-PRF. Panels ( a – d ) illustrate amplitude differences across three emotional conditions (fearful, happy, sad) for each component. Panels ( e – h ) depict amplitude differences between high- and low-PRF groups for each component. * p < 0.05, *** p < 0.001, ns, non-significant. Interaction effects between PRF level and emotion type on N1, P2, and P3 difference-wave amplitudes. ( a ) For the N1 component, a significant interaction was observed. Simple effects analysis revealed no significant differences between the high- and low-PRF groups for fearful, happy, or sad faces. ( b ) For the P2 component, a significant interaction was observed. Simple effects analysis showed that the low-PRF group exhibited significantly larger P2 amplitudes than the high-PRF group in response to fearful faces, whereas no significant between-group differences were found for happy or sad faces. ( c ) For the P3 component, a significant interaction was also observed. Simple effects analysis indicated that the low-PRF group exhibited significantly larger P3 amplitudes than the high-PRF group in response to fearful faces, whereas no significant between-group differences were found for happy or sad faces. Although no individual comparison reached significance, the significant interaction confirms that the two groups’ N1 response profiles across emotions differed reliably. The low-PRF group exhibited a relatively more positive N1 only to fear, while the high-PRF group exhibited more positive N1 to happy and sad expressions. This pattern suggests that PRF levels may modulate the sensitivity of early attentional processing to distinct emotional cues. Consistent with our hypothesis that PRF-related group differences would be most pronounced for high-threat stimuli, the largest numerical group difference emerged for fearful faces; however, the absence of significant pairwise effects indicates that PRF-related modulation at the N1 stage is subtle and warrants cautious interpretation. The mean P2 difference-wave amplitude during the 190–250 ms time window, averaged across electrodes F3, F7, and Fz, was submitted to the mixed-design ANOVA. The main effect of Group was not significant, F(1,27) = 1.487, p = 0.233, η 2 p = 0.052 (see Figure 3 ). The main effect of emotion was significant, F(2,54) = 11.546, p < 0.001, η 2 p = 0.300. Bonferroni-adjusted post hoc comparisons showed that fearful faces (M = 3.71, SD = 2.60) elicited significantly larger P2 amplitudes than happy faces (M = 1.48, SD = 1.81), t(28) = 5.230, p < 0.001, Cohen’s d = 1.00. The difference between fearful and sad faces (M = 2.29, SD = 2.37) was marginally significant ( p = 0.053), whereas the difference between happy and sad faces was not significant ( p = 0.271). A significant group × emotion interaction was observed, F(2,54) = 4.200, p = 0.020, η 2 p = 0.135. Simple effects analysis showed that under the fear condition, the low-PRF group exhibited significantly larger P2 amplitudes (M = 4.82, SD = 2.70) than the high-PRF group (M = 2.53, SD = 1.93), t(27) = 2.614, p = 0.014, Cohen’s d = 0.97. No significant group differences emerged under the happy [t(27) = 0.289, p = 0.775] or sad conditions [t(27) = −0.222, p = 0.826]. The interaction effect on P2 is illustrated in Figure 4b . These findings indicate that the modulation of P2 by PRF level is emotion-specific, manifesting only during the processing of fearful faces. The enhanced P2 to fearful expressions in the low-PRF group suggests heightened sensitivity in the early perceptual encoding of threat-related emotional cues. This finding directly supports our threat-specificity hypothesis: PRF-related group differences were indeed most pronounced for high-threat stimuli, with the low-PRF group showing greater early perceptual engagement with fearful faces. The mean N2 difference-wave amplitude during the 280–350 ms time window, averaged across electrodes F3 and Fz, showed no significant main effect of Group, F(1, 27) = 1.250, p = 0.273, η 2 p = 0.044; no significant main effect of Emotion, F(2, 54) = 1.439, p = 0.246, η 2 p =0.051; and no significant Group × Emotion interaction, F(2, 54) = 1.275, p = 0.288, η 2 p = 0.045 (see Figure 3 ). Simple effects analysis revealed no significant group differences under fear, happy, or sad conditions ( p = 0.074, 0.509, and 0.989, respectively). Although the low-PRF group exhibited descriptively larger N2 amplitudes than the high-PRF group under the fear condition (0.72 vs −1.19), and this difference approached significance, the overall interaction did not reach the statistical threshold. Prior research has yielded inconsistent findings regarding N2 modulation by pain-related fear, and the present null results are consistent with this variability, suggesting that conflict monitoring processes indexed by the N2 were not significantly modulated by PRF level in the current paradigm. The mean P3 difference-wave amplitude during the 400–550 ms time window, averaged across electrodes F3, Fz, F4, and Cz, showed no significant main effect of Group, F(1, 27) = 1.239, p = 0.275, η 2 p = 0.044. The main effect of emotion was significant, F(2,54) = 7.391, p = 0.001, η 2 p = 0.215 (see Figure 3 ). Bonferroni-adjusted post hoc comparisons showed that fearful faces (M = 4.87, SD = 3.71) elicited significantly larger P3 amplitudes than happy faces (M = 2.83, SD = 2.63), t(28) = 3.793, p = 0.002, Cohen’s d = 0.63. Fearful faces also elicited significantly larger P3 amplitudes than sad faces (M = 3.14, SD = 3.47), t(28) = 2.767, p = 0.030, Cohen’s d = 0.48. No significant difference was observed between happy and sad faces ( p = 1.000). The Group × Emotion interaction was also significant, F(2,54) = 3.331, p = 0.043, η 2 p = 0.110. Simple effects analysis showed that under the fear condition, the low-PRF group exhibited significantly larger P3 amplitudes (M = 6.20, SD = 4.15) than the high-PRF group (M = 3.43, SD = 2.62), t(27) = 2.127, p = 0.043, Cohen’s d = 0.79. No significant group differences emerged under the happy [t(27) = 0.025, p = 0.980] or sad conditions [t(27) = 0.482, p = 0.634]. The interaction effect on P3 is shown in Figure 4c . These findings indicate that the modulation of P3 by PRF level is emotion-specific, with the low-PRF group showing enhanced late-stage cognitive processing selectively for threatening faces. Although the group difference for fearful faces yielded a large effect size (Cohen’s d = 0.79), it did not survive Bonferroni correction. Nonetheless, this pattern converges with the P2 results in providing further support for the threat-specificity hypothesis: PRF-related group differences in resource allocation were restricted to high-threat stimuli, with the low-PRF group mobilizing greater late-stage processing resources in response to fearful faces. Table 5 summarizes the main effects and interactions for all ERP components. Grand-average ERP waveforms and topographic maps for the P2 and P3 components in response to fearful and happy faces are shown in Figure 5 . Table 5 Summary of Main and Interaction Effects (Group × Emotion) on N1, P2, N2, and P3 Amplitudes Differences ERP Effect F η 2 p p N1 Emotion 0.008 0 0.992 Group 0.300 0.011 0.588 Emotion × Group 4.612 0.146 0.014 N2 Emotion 1.439 0.051 0.246 Group 1.250 0.044 0.273 Emotion × Group 1.275 0.045 0.288 P2 Emotion 11.546 0.300 <0.001 Group 1.487 0.052 0.233 Emotion × Group 4.200 0.135 0.020 P3 Emotion 7.391 0.215 0.001 Group 1.239 0.044 0.275 Emotion × Group 3.331 0.110 0.043 Figure 5 P2 and P3 responses to fearful and happy faces in patients with high- and low-PRF. ( A ) Grand-average ERP waveforms at representative electrode sites showing the P2 and P3 components elicited by fearful and happy deviant stimuli in the high- and low-PRF groups. ( B ) Topographic maps depicting voltage distributions at the peak latencies of the P2 and P3 components: (a) P2 topography for high-PRF patients, (b) P2 topography for low-PRF patients, (c) P3 topography for high-PRF patients, and (d) P3 topography for low-PRF patients. Solid and dashed lines in ( A ) represent responses to fearful and happy faces, respectively. A line graph and scalp maps of P2 and P3 responses for high and low PRF, fear and happy. Image A displays a line graph of event-related potential waveforms with shaded time windows labeled P2 and P3. The x-axis shows time in milliseconds from -200 to 800, with a downward arrow at 0 labeled Target onset. The y-axis indicates amplitude in microvolts from -10 to 10. The legend includes four traces: H PRF Fear, L PRF Fear, H PRF Happy, L PRF Happy. In the P2 window around 200 ms, H PRF Fear reaches ~3, L PRF Fear ~6, H PRF Happy ~2, L PRF Happy ~4. In the P3 window (350-450 ms), H PRF Fear peaks near -7, H PRF Happy near -6, L PRF Fear near -3, L PRF Happy near -4, then all traces return toward 0 to 3 by 600-800 ms. Image B shows four groups of scalp topographic maps in a 2x2 grid of boxed triplets labeled Neutral, Fear, Happy emotion. Each map features electrode dots and contour lines with a color scale from -10 to 10. The Fear emotion maps exhibit the strongest central positive region compared to Neutral and Happy emotions, which are more moderate and evenly distributed. Summary of Main and Interaction Effects (Group × Emotion) on N1, P2, N2, and P3 Amplitudes Differences P2 and P3 responses to fearful and happy faces in patients with high- and low-PRF. ( A ) Grand-average ERP waveforms at representative electrode sites showing the P2 and P3 components elicited by fearful and happy deviant stimuli in the high- and low-PRF groups. ( B ) Topographic maps depicting voltage distributions at the peak latencies of the P2 and P3 components: (a) P2 topography for high-PRF patients, (b) P2 topography for low-PRF patients, (c) P3 topography for high-PRF patients, and (d) P3 topography for low-PRF patients. Solid and dashed lines in ( A ) represent responses to fearful and happy faces, respectively. Given the modest sample size, we acknowledge that the study may be underpowered for detecting smaller interaction effects, and the findings should be interpreted as preliminary and in need of replication.

Visceral pain differs from somatic and externally generated pain in both psychophysiological principles and neurobiological mechanisms. 57 It is more likely than somatic pain to evoke pain-related fear and heightened unpleasantness, making it a clinically relevant model for investigating pain-related fear. 58–61 Epigastric Pain Syndrome (EPS) is a chronic pain condition characterized by visceral hypersensitivity, 62 , 63 with psychological factors closely linked to its onset and symptomatology. 64 , 65 To examine how pain-related fear modulates attentional bias more specifically, this study focused within the EPS patient group. This study employed a cross-sectional, quasi-experimental between-groups design. Participants were classified into high- and low-PRF groups based on pre-existing psychological characteristics, precluding random assignment to groups. The required sample size was determined using an a priori power analysis conducted in G*Power 3.1.9.7 for the Group × Emotion interaction in a 2 (Group: high-PRF vs low-PRF) × 3 (Emotion: fearful, sad, happy) mixed-design ANOVA. The analysis was specified under the F-test family using “NOVA: Repeated measures, within-between interaction”, with Cohen’s f = 0.25, 66 α = 0.05, statistical power (1 − β) = 0.80, two groups, three repeated measurements, a correlation of 0.50 among repeated measures, and a nonsphericity correction of ε = 1.00. The analysis indicated that a minimum total sample size of 28 participants was required. Given that the assumed effect size represents a conventional medium effect, the power estimate should be regarded as approximate and the findings as preliminary. Based on this criteria, 30 patients meeting the diagnostic criteria for EPS were recruited from the First Affiliated Hospital of Jiangxi University of Traditional Chinese Medicine. Participants were classified into high- and low-PRF groups based on a composite index derived from the TSK-17 and PCS-13. We chose these two scales for grouping because they have well-established clinical cutoffs (TSK-17: >37; 67 PCS-13: ≥30 68 ), whereas no widely accepted clinical threshold currently exists for the FoP-Q-SF 69 in chronic pain populations. For each scale, the per-item deviation from the clinical cutoff was computed as (Total Score − Cutoff)/Number of Items. This transformation centers each scale on its clinical threshold and adjusts for differences in scale length, yielding a score that reflects each participant’s per-item deviation from the respective cutoff. The two deviation scores were summed to form the composite index. Participants were rank-ordered by this index; the top 15 and bottom 15 were assigned to the high- and low-PRF groups, respectively (n = 15 each). To empirically validate that the composite index derived from TSK and PCS also captures variance shared with the fear of progression, we compared the two groups on the FoP-Q-SF. A Mann–Whitney U -test confirmed a significant between-group difference, verifying that the groups were effectively discriminated on fear of progression as directly measured by the FoP-Q-SF. Accordingly, the groups are referred to as “high-PRF” and “low-PRF” throughout the manuscript, reflecting that they represent meaningfully distinct levels of fear of progression. One participant from the high-PRF group were subsequently excluded due to baseline EEG amplitude exceeding ±100μV, resulting in a final sample of 29 EPS patients (14 high-PRF, 15 low-PRF). All participants were right-handed and provided written informed consent prior to participation. To minimize potential confounding arising from the non-randomized design, key demographic variables—including age, gender, and education—were compared between groups and entered as covariates in analyses where significant group differences emerged. Electroencephalographic (EEG) data were recorded using a Brain Products (BP) EEG system equipped with a 32-channel electrode cap arranged according to the international 10–20 system. The experimental paradigm was programmed and presented using E-Prime 3.0. Visual stimuli consisted of emotional facial expressions selected from the Chinese Facial Affective Picture System (CFAPS). 70 An initial pool of 270 facial images was independently rated by 15 participants using a 7-point emotional valence scale. Based on these ratings, 60 emotional facial stimuli were selected, including 10 fearful, 10 sad, and 10 happy expressions. In addition, 240 neutral facial images were selected as standard stimuli. All images were standardized to a resolution of 300×260 pixels and matched for color and contrast. Stimuli were presented on a 24-inch monitor with a resolution of 1920×1080 pixels. The arousal and valence ratings for the selected emotional faces are presented in Table 1 . Table 1 Recognition and Intensity of Emotional Facial Expressions Fearful Happy Sadness F Arousal 6.57±0.43 5.71±0.20 6.44±0.35 99.37*** Valence 3.72±0.49 5.31±0.52 3.78±0.54 165.01*** Note : *** p< 0.001. Recognition and Intensity of Emotional Facial Expressions Note : *** p< 0.001. During online acquisition, EEG signals were recorded using the Cz electrode as the reference, bandpass-filtered between 0.5 and 100 Hz, and sampled at 500 Hz. Scalp electrode impedances were maintained below 10 kΩ. Offline preprocessing was conducted in MATLAB (R2021b) using the EEGLAB toolbox (2021.1). Data were re-referenced to the average of the bilateral mastoids and further bandpass-filtered between 0.1 and 40 Hz. Event-related potentials (ERPs) were segmented into epochs extending from 200 ms before stimulus onset to 800 ms after picture onset, with the pre-stimulus interval serving as the baseline. Analyses focused on mean amplitudes of the N1 (80–130 ms), P2 (190–250 ms), N2 (280–350 ms), and P3 (400–550 ms) components. Trials exhibiting excessive baseline drift or peak-to-peak amplitudes exceeding ±100 μV were excluded from further analysis. As a result, one participant from the high-PRF group was excluded, leaving a final sample of 29 participants for EEG analyses. Behavioral reaction times were constrained to a window of 100–1000 ms. Trials without behavioral responses or with excessively fast responses (<100 ms) were excluded from behavioral analyses. After providing written informed consent, participants completed demographic questionnaires as well as the PCS, TSK, and fear of progression questionnaires. They were seated approximately 0.8 m from the monitor, with horizontal and vertical viewing angles maintained within 5°. Throughout the experiment, participants were instructed to keep their head and body movements to a minimum. An oddball paradigm was employed to assess attentional processing of threat-relevant stimuli. Standard stimuli (neutral faces) were presented on 80% of trials, whereas deviant stimuli (emotional faces: sad, fearful, or happy) were presented on 20% of trials. Participants were instructed to press the “F” key in response to standard stimuli and the “J” key in response to deviant stimuli. The experiment consisted of three blocks, each comprising 100 trials presented in a randomized order. Each trial began with a fixation cross (“+”) displayed for 300 ms, followed by a blank screen (300–800 ms), and then an emotional face presented for 1000 ms, during which participants were required to make their response. Prior to the formal experiment, participants completed a practice session and were required to achieve an accuracy rate of at least 80% before proceeding. The presentation order of emotional faces (fearful, sad, and happiness) was fully randomized across participants within each experimental block to minimize order effects. The experimental procedure is illustrated in Figure 1 . Figure 1 Experimental flowchart. Flowchart of an experiment with emotional face stimuli and EEG setup. A flowchart illustrates an experimental setup involving emotional face stimuli. The top section shows a sequence starting with ′Instruction′, followed by a fixation cross displayed for 300 milliseconds, then a blank screen for 300 to 800 milliseconds. Next, a neutral face is shown for 1000 milliseconds, followed by either a fearful, sad, or happy face. The sequence ends with another blank screen for 300 to 800 milliseconds. Below, instructions indicate participants should wear an EEG cap, sit comfortably, adjust the distance to 0.8 meters and ensure viewing angles are not greater than 5 degrees. The text includes ′Main Reading Guide Language′, ′Sign informed consent form′ and ′Wear an EEG cap, adjust to a comfortable sitting position, fix the adjustment distance at 0.8m and ensure that the horizontal and vertical viewing angles are not greater than 5 degrees′. Experimental flowchart. Statistical analyses were performed using SPSS 24.0. Demographic variables and questionnaire scores were compared using chi-square tests and Mann–Whitney U -tests. Behavioral accuracy (ACC) and reaction time (RT) data were analyzed using two mixed-design ANOVAs. First, to examine the target effect, separate 2×2 mixed-design ANOVAs were conducted for accuracy and RT, with group (high-PRF, low-PRF) as the between-subjects factor and target (standard, deviant) as the within-subjects factor. Second, to examine behavioral performance across different emotion types, separate 2×4 mixed-design ANOVAs were conducted for accuracy and RT, with group (high-PRF, low-PRF) as the between-subjects factor and emotion type (neutral, fearful, happy, sad) as the within-subjects factor. Where significant interactions were observed, follow-up comparisons were conducted. For ERP amplitudes, difference waves were calculated separately for fearful, happy, and sad faces by subtracting the ERP elicited by the standard stimuli from the ERP elicited by each emotional deviant condition. For each ERP component, mean difference-wave amplitudes were analyzed using a 2×3 mixed-design ANOVA, with group (high-PRF, low-PRF) as the between-subjects factor and emotion (fearful, happy, sad) as the within-subjects factor. When a significant interaction was observed, follow-up independent-samples t tests were conducted to examine between-group differences, with Bonferroni correction applied to control for multiple comparisons.

This study employed a three-stimulus oddball paradigm combined with ERP techniques to investigate the temporal dynamics through which PRF is associated with emotional face processing in patients with chronic pain. Behaviorally, all patients showed longer reaction times and higher error rates for deviant emotional stimuli, consistent with the possibility that the task effectively engaged inhibitory control demands. 71 , 72 ERP analyses revealed that PRF-related attentional modulation was emotion-specific: significant group differences emerged selectively for high-threat (fearful) faces at the P2 and P3 stages, while the N1 interaction, though significant, did not yield significant pairwise effects, and the N2 showed no PRF-related modulation. These findings provide preliminary evidence that PRF is associated with altered perceptual and late-stage processing of high-threat signals, while sparing responses to low-threat and non-threat stimuli, consistent with the threat-specificity hypothesis. Our findings may contribute to a more nuanced understanding of how PRF relates to attentional processing in chronic pain. The observed threat-specific pattern should be interpreted with caution and requires further validation in larger samples. The N1 component (approximately 80–130 ms) reflects early attentional orienting. 73–77 In the present study, the significant Group × Emotion interaction, in the absence of significant main effects, indicates that the two groups differed reliably in their early attentional response profiles across emotional categories, though the effect was subtle and did not yield significant pairwise differences for any single emotion. The largest numerical group difference emerged for fearful faces, which is partially consistent with the threat-specificity hypothesis. However, this pattern diverges from the classic threat superiority effect observed in healthy populations, wherein high-arousal threat signals typically capture early attention. Instead, the diverging response profiles align with evidence that chronic pain patients exhibit altered sensitivity to a broader range of socio-emotional signals, including low-arousal negative expressions such as sadness. 78 , 79 Prior research indicates that both negative valence 80 , 81 and stimulus arousal 81–84 influence attentional allocation in chronic pain, and regions such as the inferior frontal operculum (IFO) may differentially influence the processing of emotional faces with varying arousal levels. 85 Notably, the high-PRF group exhibited descriptively larger N1 amplitudes to sad faces. Sadness has been shown to amplify pain unpleasantness via disruption of emotion regulation circuitry and is closely linked to somatic pain perception in chronic pain populations. 86 Thus, an exaggerated early attentional response to sadness in high-PRF individuals may reflect a maladaptive pattern, wherein sensitivity to signals of social disconnection or distress compounds the burden of chronic pain. Given the absence of significant pairwise effects, these N1 findings should be interpreted with caution; future studies with larger samples are needed to clarify whether this interaction reflects reliable PRF-related modulation of early perceptual processing. The P2 component (approximately 190–250 ms), which has been associated with early perceptual encoding and evaluation, 87–89 revealed a significant Group × Emotion interaction driven by enhanced P2 amplitudes in the low-PRF group selectively for fearful faces. This finding directly supports the threat-specificity hypothesis: PRF-related differences in early perceptual processing emerge specifically for high-threat signals. Low-PRF patients showed an efficient “threat-priority” processing strategy, selectively allocating enhanced perceptual resources to high-threat (fearful) faces. This pattern may reflect a potential precise resource allocation and has been associated with lower sensitivity to pain-related threat. 90–92 In contrast, the attenuated P2 response to fearful faces in the high-PRF group may reflect reduced mobilization of perceptual resources for high-threat stimuli, possibly involving fear generalization, altered threat appraisal, or disrupted memory encoding. 8 , 86 , 93 , 94 Because both groups showed comparable P2 responses to happy and sad faces, this attenuation does not appear to reflect a generalized perceptual deficit, but rather a specific alteration in the evaluation of high-threat signals. Such a mechanism may contribute to broader emotional regulation disturbances via dysregulation of emotion-motivation pain circuits, 95 as well as to the attenuated placebo analgesia responses previously reported in such individuals. 96 The N2 component (approximately 280–350 ms), which has been linked to conflict detection and monitoring, 97–101 showed no significant main effects or interaction involving PRF. This finding diverges from prior evidence, where elevated fear is associated with sustained vigilance that imposes chronically heightened cognitive load during conflict processing, manifesting as enhanced N2 amplitudes. 102–104 This enhancement is thought to reflect increased load on the dACC-centered conflict monitoring system 105–107 in response to ambiguity from imprecise early perceptual encoding. 108–110 While potentially compensatory, such engagement consumes resources and may impair subsequent executive control. 111 , 112 Its absence here may reflect a resource ceiling: chronic pain patients already operate under elevated cognitive load from sustained threat processing, and the significant P2 alterations—indicative of atypical early perceptual encoding—would further tax conflict monitoring, leaving no additional resources to mobilize. Reduced N2 habituation to threat-related stimuli in chronic pain 113–115 may further attenuate differential neural responding to emotional deviants. Together with the significant PRF effects at P2 and P3, this null N2 finding suggests a stage-specific pattern: PRF modulates early perceptual evaluation and late-stage resource allocation, while its impact on intermediate conflict monitoring may be obscured under cognitive saturation. Whether PRF modulates N2 activity thus requires further investigation. The P3 component (approximately 400–550 ms), which indexes processing resources mobilized toward motivationally salient stimuli to support late-stage inhibitory control and conflict resolution, 99 , 116–120 showed a significant Group×Emotion interaction, with the low-PRF group exhibiting larger amplitudes than the high-PRF group selectively for fearful faces, and no group differences for sad or happy faces. This pattern converges with the P2 findings and indicates that PRF-related modulation of late-stage processing is selective to high-threat signals. The low-PRF group’s enhanced P3 to fearful faces suggests greater allocation of late-stage resources to the most motivationally salient stimuli, facilitating adaptive evaluation of cues with maximal threat relevance for chronic pain populations. 26 , 121 Conversely, the high-PRF group’s attenuated response to fearful faces, in the absence of a significant group main effect, points to reduced resource engagement that emerges specifically for high-threat rather than reflecting a generalized deficit. This pattern suggests that elevated PRF is characterized by a failure to recruit adequate inhibitory control resources when confronted with the most salient threat cues, consistent with the proposal that pain-related fear disrupts adaptive allocation of cognitive resources to motivationally relevant stimuli. Fearful faces, as the most threat-relevant stimuli, may impose disproportionate cognitive load on high-PRF individuals, thereby suppressing P3 amplitudes and undermining effective mobilization of late-stage processing resources. 122 , 123 Such high-threat-selective inefficiency in resource mobilization may represent a key neurocognitive correlate of elevated PRF in chronic pain. Across the four ERP components, a converging pattern emerged: PRF-related differences were confined to fearful faces at the P2 and P3 stages, while N1 showed no significant pairwise effects and N2 revealed no PRF modulation. This dissociation indicates that PRF does not exert a global influence across processing stages, but selectively shapes perceptual evaluation and late-stage resource allocation for high-threat signals. We propose a motivational resource allocation account: elevated PRF is characterized not by a pervasive deficit, but by a selective failure to upregulate processing resources when confronted with the most salient threat cues. Consistent with the FAM, the present findings suggest that fear-driven amplification of threat salience is more circumscribed than previously assumed—affecting primarily high-threat stimuli at specific processing stages rather than operating uniformly. The enhanced P2 and P3 to fearful faces in low-PRF individuals further suggests that adaptive threat processing involves the capacity to mobilize additional resources selectively for high-threat signals. This account is inferred from cross-sectional data and awaits validation through longitudinal and experimental designs. Several alternative explanations warrant consideration. First, the observed group differences may not be specific to PRF but could reflect broader emotional distress. General anxiety and depression are known to influence attentional processing of emotional stimuli and frequently co-occur with elevated PRF. Although we controlled demographic variables, unmeasured psychological factors may partially account for the observed effects. Future studies should include measures of general anxiety and depression to disentangle PRF-specific effects from those of broader negative affectivity. Second, the absence of a healthy control group limits our ability to determine whether the observed neurocognitive profile is specific to chronic pain pathology or represents a general characteristic of individuals with elevated PRF. Third, the limitations of modest sample size should be noted. Although a priori power analysis indicated adequacy for detecting a medium-sized interaction effect, the power estimate is approximate given the reliance on a conventional effect size. Replication in larger, well-powered samples is therefore warranted. The exclusive focus on EPS patients enhances internal validity for visceral pain but restricts generalizability to other chronic pain populations. The rank-based median-split grouping approach—classifying participants into the upper and lower halves of the PRF composite-score distribution—retained the full sample but may have attenuated between-group differentiation relative to more restrictive extreme-group approaches based on the upper and lower quartiles. While the present findings suggest potential clinical directions, the following implications should be considered preliminary given the study’s cross-sectional design and modest sample size. The stage-specific, high-threat-selective pattern observed here raises the possibility that interventions targeting multiple processing stages may be more effective when tailored to the processing of threatening cues, rather than targeting general cognitive function. Specifically, perceptual discrimination training emphasizing high-threat cues could recalibrate early threat assessment, 124 , 125 cognitive restructuring and mindfulness-based strategies, by promoting reappraisal of pain-related threat, may reduce cognitive load that high-threat signals impose on high-PRF individuals; and targeted exposure with inhibition training could enhance late-stage control efficacy. 126 , 127 Interventions such as Pain Neuroscience Education (PNE) and Cognitive Behavioral Therapy (CBT) may contribute to pain management by improving threat appraisal and conflict resolution efficiency. 128–131 However, the translation of these ERP-derived insights into clinical protocols requires prospective studies demonstrating that changes in these neurocognitive markers predict meaningful functional outcomes. 127 , 132

This study provides preliminary ERP evidence that pain-related fear (PRF) in patients with epigastric pain syndrome is associated with threat-specific alterations in attentional processing. Across four ERP components, PRF-related group differences emerged selectively for high-threat (fearful) faces at the P2 and P3 stages, suggesting that elevated PRF is characterized by reduced mobilization of perceptual and late-stage processing resources specifically when confronted with the most salient threat cues. This converging pattern indicates that PRF does not exert a global influence across all processing stages, but rather selectively shapes the allocation of cognitive resources to high-threat signals. These findings highlight the potential value of interventions tailored to the processing of personally threatening cues, though prospective studies are required to establish causality and clinical utility. Future research incorporating healthy controls and longitudinal designs will be essential to validate and extend these observations.