Do emotions conquer facts? A CCME model for the impact of emotional information on implicit attitudes in the post-truth era

The study is situated in the post-truth era, characterized by the privileging of emotions over facts and objective evidence. Emotional expressions function not only as personal but also social information, shaping attitudes and public opinion and potentially reinforcing biases. Prior work suggests affective responses interact with cognitive processes and may precede and influence factual processing. However, communication research has lacked clarity on the temporal neurocognitive mechanisms by which emotional media content affects audience attitudes. To probe this, the authors employ the Implicit Association Test (IAT) and event-related potentials (ERPs; N2, N400, LPP) to capture high-temporal-resolution indices of information processing. Research question: Do audiences engage different information-processing mechanisms when reading neutral news versus emotionally framed news? Hypothesis: Emotional discourse depletes attentional resources, altering implicit attitudes; specifically, larger N2 and N400 amplitudes would be observed, reflecting heightened early conflict monitoring and semantic processing, followed by reduced late evaluation due to resource depletion. The study aims to articulate a two-stage Cognitive Conflict Monitoring and Evaluation (CCME) model for emotional information processing in post-truth communication.

The article reviews scholarship on the post-truth era, emphasizing how emotions and biases can outweigh facts in shaping attitudes (e.g., Higgins, 2016; Lewandowsky et al., 2017). Emotions act as social information (Van Kleef, 2009), reproducing social and structural biases and affecting media influence (Boler & Davis, 2018). Neurocognitive studies suggest affect impacts complex cognition such as truth discernment (Van Bavel & Pereira, 2018). ERP components (N2, N400, LPP) are established markers of stages of cognitive and semantic processing and attentional allocation. Prior ERP/IAT research links incongruence with larger N2/N400 and congruence with larger late positive components, yet the speed and staging of negative bias development in media contexts remain underexplored.

Design: 2 (information type: emotional vs. neutral; between-subjects) × 2 (IAT association: compatible/positive vs. incompatible/negative; within-subjects) × 3 (electrode site: Fz, Cz, Pz) with ERP recording during IAT. Participants: 44 right-handed undergraduate/graduate students (experimental/emotional group n=24, 12 males, Mage=22.13±1.94; control/neutral group n=20, 9 males, Mage=23.20±2.19). Normal or corrected vision; no neurological/psychiatric history; non-psychology majors. Baseline affect (PANAS), anxiety (BAI), and depression (BDI) assessed; no significant group differences in positive affect (t(42)=1.314, p=0.196) or negative affect (t(42)=1.224, p=0.228). Informed consent obtained; IRB approved. Materials: Two Chinese-language news articles on genetically modified foods (GMFs): (1) neutral article introducing national compulsory labeling (1369 characters); (2) emotional article consistent in argument but with more emotional expressions (1343 characters). Printed on A4. Procedure: Approx. 10 min total. Pre-rest 3 min; reading task (~4 min) with manipulation-check questionnaire; post-rest 3 min. IAT: Five-phase paradigm. Key mapping for positive vs negative attribute words and GMF-related target words counterbalanced across participants. Formal experimental phases (3rd and 5th) contained 60 trials each; practice phases had 20 trials each. Stimuli: 10 target words and 20 attribute words (10 positive, 10 negative). Valence validated on 7-point Likert; positive vs negative differed significantly (t(38)=18.36, p<0.001). Presentation parameters: Microsoft YaHei font size 34; fixation 500 ms; stimulus 1000 ms; 13-inch screen at ~60 cm, 1920×1280 resolution. Responses via F/J keys, instructed to be fast and accurate. Implemented in E-Prime 3.0. EEG/ERP recording and analysis: Wireless Cognionics Quick-30 32-channel amplifier, 1000 Hz sampling, 0–100 Hz bandpass, referenced to left mastoid during recording. EEGLAB 14.1.1 used for ICA; averages low-pass filtered at 30 Hz; epochs of 2 s. ERP amplitudes analyzed at Fz, Cz, Pz. RTs <100 ms or >1000 ms excluded. Statistical analyses via SPSS 22.0. Target RTs from the 3rd (negative association) and 5th (positive association) phases analyzed by 2 (group) × 2 (association) repeated-measures ANOVA. ERP components examined: N2 (200–350 ms, frontocentral), N400 (~300–450/400 ms, parietal/temporal), LPP (~500–800 ms, late positive).

Behavioral (RT): Significant main effect of association [F(1,42)=80.467, p<0.001, η²=0.646]; RTs were shorter for positive associations (congruent) than negative associations (incongruent). No main effect of group [F(1,42)=0.200, p=0.657, η²=0.005], and no group×association interaction [F(1,42)=1.741, p=0.194, η²=0.038]. Means (ms): Experimental group—positive 576.16±51.35, negative 644.73±55.90; Control group—positive 577.15±81.95, negative 628.14±62.24. Interpretation: All participants showed positive implicit attitudes toward GMFs. ERP—N2: Main effect of association significant [F(1,42)=4.244, p=0.046, ηp²=0.092]; larger N2 for negative (incongruent) vs positive (congruent) associations. No significant effects for group, electrode, or interactions. ERP—N400: Main effect of association significant [F(1,42)=7.490, p=0.009, ηp²=0.151]; larger N400 for negative (incongruent). Main effect of group significant [F(1,42)=13.846, p=0.001, ηp²=0.248]; the emotional-information group showed smaller N400 amplitudes than the neutral-information group. No significant electrode or interaction effects. ERP—LPP: Main effect of association significant [F(1,42)=5.565, p=0.023, ηp²=0.117]; larger LPP for positive (congruent). Main effect of group significant [F(1,42)=4.438, p=0.041, ηp²=0.096]; the emotional-information group showed smaller LPP amplitudes than the neutral-information group. No significant electrode or interaction effects. Overall pattern: Emotional news elicited earlier/greater early conflict monitoring (N2 effects) and reduced semantic integration (smaller N400) and late evaluative processing (smaller LPP) compared to neutral news, despite similar behavioral RT patterns indicating positive implicit attitudes toward GMFs.

Findings address the research question by showing that emotional framing in news alters neurocognitive processing despite comparable behavioral indicators of implicit attitude valence. Incongruent associations reliably elicited greater conflict (larger N2/N400), but exposure to emotional content attenuated N400 and LPP amplitudes, suggesting depleted attentional resources for semantic integration and late-stage evaluation. These results support the proposed CCME model: (1) Early conflict-monitoring stage—emotional arousal drives allocation of attention to discrimination and semantic processing (indexed by stronger N2 and earlier/weaker N400), generating cognitive conflict that may prioritize emotional cues over factual content; (2) Late evaluation stage—prior resource expenditure reduces sustained evaluative processing of target stimuli (indexed by reduced LPP). In a post-truth environment, this mechanism implies that emotional discourse can weaken integration of core facts and foster reliance on heuristics, reinforcing existing beliefs and echo chambers. The work underscores the importance of media literacy and strategies to help audiences distinguish emotions from evidence in news consumption.

Using an IAT-ERP paradigm, the study demonstrates that emotional information in news engages distinct processing mechanisms relative to neutral content: heightened early conflict monitoring and diminished semantic integration and late evaluation. Behaviorally, participants maintained positive implicit attitudes toward GMFs, but neurophysiological evidence indicates that emotional framing can shift resource allocation. The authors formalize a two-stage Cognitive Conflict Monitoring and Evaluation (CCME) model explaining how emotional discourse in post-truth communication depletes attentional resources for factual evaluation and may alter implicit attitudes. Future research should extend to multimodal (e.g., video) emotional framings and employ neuroimaging with higher spatial resolution (fMRI, fNIRS) to map involved circuits and probe interactions with framing and confirmation bias mechanisms.

The study used text-based stimuli on a single controversial topic (GMFs) with a modest student sample, which may limit generalizability. Emotional and objective processing may be confounded by framing and confirmation bias and engagement of reward-related neural pathways (e.g., mPFC, NAcc). The ERP approach provides high temporal but limited spatial resolution; future work should include fMRI or fNIRS and multimodal news formats (e.g., video) to better capture spatial dynamics and ecological validity.