Recently, much etiological and pathophysiological evidence has shown that hyperarousal is “a state of relatively increased arousal in physiological, cortical, and cognitive-emotional domain” (Dressle and Riemann, 2023). In insomnia disorders, hyperarousal includes three functions: (1) as a predisposing factor, (2) as a state marker for insomnia, and (3) as a pathophysiological component that potentially leads to or causes various insomnia-related symptoms. Hyperarousal during insomnia often leads to cognitive dysfunctions, such as conflict control (Ding et al., 2023) and cognitive flexibility (Ding et al., 2024; Uddin, 2021), and triggers negative emotions such as anxiety and depression (Ding et al., 2024; Dopheide, 2020). As a tool for measuring the level of arousal in insomnia disorders, the Pre-Sleep Arousal Scale (PSAS) includes two dimensions of cognitive pre-sleep arousal and physical pre-sleep arousal (Nicassio et al., 1985) and effectively assesses the sleep quality of individuals with insomnia disorders (Gong et al., 2016; Schneider et al., 2019). In recent years, based on the hyperarousal model for insomnia disorders, researchers have conducted relevant studies on insomnia disorders using integrated intervention strategies (Gong et al., 2016; Schneider et al., 2019).

With the burgeoning interest in using music to treat insomnia (Chang et al., 2012; Dickson and Schubert, 2019; Luo et al., 2024; Majeed et al., 2021), studies have shown that various methods of music therapy, such as passive listening therapy and music performance therapy, can effectively alleviate symptoms associated with insomnia (Chang et al., 2012; Lai et al., 2014) and regulate negative emotions of mental disorders (Dickson and Schubert, 2019; Huang et al., 2017; Lai et al., 2014). With the in-depth exploration by more music therapy researchers, the methods and strategies of music therapy are constantly being optimized and updated (Lai et al., 2014; Umemoto et al., 2021; Zhu et al., 2023). Aesthetics-based music listening, as an effective intervention method, has been proven to be the most widely used in music therapy (Liu et al., 2021b,c; Liu et al., 2023; Susino, 2023). The integration of music therapy with other psychological therapies is becoming a new trend and is being widely applied (Miranda, 2021; Susino, 2023; Umemoto et al., 2021; Zhu et al., 2023). It is worth noting that musical works using the treatment for insomnia have characteristics such as smoothness, slowness, and relaxation (Chang et al., 2012; Lai et al., 2014; Luo et al., 2024), which perfectly align with the requirements of mindfulness meditation training (Liu et al., 2021a).

Mindfulness meditation, as an economical, safe, and effective intervention strategy, has been widely used to enhance the psychological wellbeing of healthy individuals (Hernandez-Ruiz and Dvorak, 2020; Liu et al., 2021a) and provide the psychological treatment for patients with mental disorders, such as depression, anxiety, and insomnia (Dopheide, 2020; Gong et al., 2016; Rusch et al., 2018; Tomaselli, 2014; Wittmann and Schmidt, 2014). Previous studies have shown that mindfulness meditation “can serve as an auxiliary treatment to medication for sleep complaints” and may mildly improve cognitive dysfunction, regulate anxiety, and alleviate depression in individuals with insomnia disorders (Dressle and Riemann, 2023; Dopheide, 2020; Gong et al., 2016). Mindfulness meditation has also been integrated with other therapies, such as cognitive behavioral therapy and family therapy (Smith et al., 2019), resulting in positive and promising clinical effects (Edinger et al., 2021; Ziegler et al., 2019).

Recent research has found that mindfulness-based music listening (MBML), as an effective intervention strategy in music therapy, has a significant regulatory effect on negative emotions (Liu et al., 2021a,c). MBML intervention strategies primarily focus on the behavioral performance of individual attention processing and different emotional states under individual emotional–cognitive interaction (Liu et al., 2021a,c; Wardle-Pinkston et al., 2019). Conflict control is “the ability of the brain to monitor conflicts in the process of information processing and measures the ability of inhibitory control at the cognitive level” (Liu et al., 2021a). Conflict control in emotion–cognition interactions (Raschle et al., 2017) effectively reveals the improvement effect of mindfulness and music listening on individual physical and mental health and cognitive function (Liu et al., 2021c).

Previous studies have shown that the integration of Questionnaire-Stroop Task-ERP technology (Liu et al., 2021a,c) may provide a practice approach for measuring the arousal level of individuals with insomnia disorders in physiological, cortical, and cognitive-emotional domains (Dressle and Riemann, 2023; Ding et al., 2023; Ding et al., 2024). Previous studies have shown that the emotional face-word Stroop task is an effective paradigm for detecting emotion–cognition interactions and attentional control processing under different emotional states (Liu et al., 2021a). Under consistent or inconsistent conditions of Stroop trials, task accuracy (ACC) and reaction times (RTs) reflect the behavioral and neural correlates of conflict control in different emotional states (Liu et al., 2021a; Xue et al., 2015).

From the perspective of emotion–cognition integration (Dennis, 2010), event-related potentials (ERPs) effectively reveal individual behavior and ERP correlates of emotion–cognition interaction, providing an effective measure for evaluating different emotional states and cognitive functions in the emotional face-word Stroop task (Raschle et al., 2017; Weth et al., 2015). Previous research has indicated that emotions significantly impact individual attention processing and cognitive function (Raschle et al., 2017), and different emotional states affect individual conflict control (Liu et al., 2021a). Different ERP indicators can effectively evaluate the behavior and neural correlation in both healthy and unhealthy individuals (Liu et al., 2020a,b, 2021c). Previous studies found, compared to healthy sleepers, individuals with insomnia disorders exhibited lower accuracy, slower reaction times, and lower N450 amplitude in the color-word Stroop task, along with a larger P3 amplitude in the two-choice oddball task (Ding et al., 2023). In addition, studies have shown that the left dorsolateral prefrontal cortex plays an important role in conflict control for individuals with insomnia disorders (Ding et al., 2024).

Previous studies have explored emotional–cognitive interactions using conflict control tasks and found that the N3, P3, and LPC of EEG indicators reflect attentional processing and emotional regulation in conflict control (Liu et al., 2021a). In research on attentional processing and emotional regulation, N3 is a useful variable for exploring the emotion–cognition processing of visual stimuli, characterized by a negative amplitude occurring approximately 250–350 ms after the stimulus onset (Truman and Mudrik, 2018). N3 is associated with semantic violations in conflict control, and the N3 effect reflects attentional processing in the recognition of the target stimulus and the classification of semantic matching or mismatch (Maguire et al., 2013). Previous studies have found that a larger N3 effect occurs during the initial classification of targets in a neutral mood state (Draschkow et al., 2018; Truman and Mudrik, 2018).

P3 is associated with higher cognitive resources and is a positive component that occurs approximately 300–600 ms after the appearance of a stimulus. In emotion–cognition interactions, the P3 effect is an effective EEG indicator reflecting positive or negative emotional states in emotion–cognition interactions (Liu et al., 2020a,b, 2021a). Larger P3 amplitudes reflect the consumption of more cognitive resources in negative emotional states, while smaller P3 amplitudes indicate the consumption of fewer cognitive resources in calm emotional states (Ding et al., 2023; Liu et al., 2021a; Yuan et al., 2007). The late positive component (LPC) is a key indicator for evaluating emotional–cognitive interaction effects (Liu et al., 2021c; Mengfan et al., 2020). As a positive component, the LPC exists approximately 600–1,000 ms after stimulus onset. Previous studies have found that negative emotions induce a greater LPC effect than neutral or positive emotions (Liu et al., 2020a,b, 2021a).

In summary, this study was based on the hyperarousal model of insomnia disorder theory (Dressle and Riemann, 2023; Riemann et al., 2010), using the emotional face-word Stroop task (Liu et al., 2021a) and a combination of questionnaire-behavioral performance-EEG technology to explore the behavioral and neural correlates of MBML intervention strategies on emotional regulation and conflict control in young Chinese adults with ID. The within-participant (pre-test, mid-test, and post-test in emotional face-word Stroop task) and between-participant (mindfulness-based music listening group, MMG; the wait-list control group, WCG) differences in the conflict control were examined to illustrate the effect of different emotional states evoked by music on conflict control in young adults with ID. The purpose of this study was to explore the effects of MBML on cognitive function in Chinese young adults with insomnia disorders. Based on previous studies, we hypothesize the following:

Three hundred college students with insomnia disorders (56.5% female students, M_age = 21.53, SD_age = 1.72) were recruited through campus advertising and evaluated for the severity of their insomnia disorder using the Pre-Sleep Arousal Scale (PSAS) (Nicassio et al., 1985). Then, sixty college students with the highest PSAS scores (58.3% female students, M_age = 20.92, SD_age = 1.09) were included in this study and assessed their sleep arousal levels (Riemann et al., 2010) using the PSAS (Nicassio et al., 1985). According to the study requirements, participants were randomly divided into two groups: the mindfulness-based music listening group (MMG, N = 30) and the wait-list control group (WCG, N = 30). They were required to avoid taking substances or drugs that affect attention. All participants did not take any psychotropic drugs for 7 days before the experiment and reported normal hearing and speech, normal or corrected-to-normal vision, and no other psychological disorders, except insomnia. Before the experiment began, all participants were required to read the experimental instructions and provide informed consent. This study was approved by the Southwest University Ethics Committee (IRB No. H22117).

In this study, the revised emotional face-word Stroop task (Liu et al., 2021a) was used to explore the effects of conflict control and attentional processing on different affective states among young adults with ID. The visual stimuli of the experimental task consisted of 40 face images (10 sad faces for male students and 10 sad faces for female students; 10 happy faces for male students and 10 happy faces for female students) [extracting from the Chinese Affective Picture System (Wang and Luo, 2005)] and emotional words (sad and happy), forming four stimulus conditions: sad congruency, sad incongruency, happy congruency, and happy incongruency (Carretié et al., 1997; Liu et al., 2019a, 2021a; Truman and Mudrik, 2018). In the Stroop task (Figure 1), the stimulus for 1,000 ms was presented after a fixation point for 500 ms, followed by a blank screen interval for 500 ms. In the experimental task, the participants were asked to press the buttons as quickly as possible, with pressing “1” button for congruent conditions and pressing “2” buttons for incongruent conditions. The stimuli were randomly presented.

To assess facial and semantic processing of young adults with ID, respectively, the experimental task was divided into two sub-tasks of emotional faces and emotional words. The stimuli were randomly presented and repeated twice and formed 160 trial experimental sub-tasks. The two sub-task orders were counterbalanced across the participants. There is an exercise module of 80 trials before the Stroop task of 320 trials. Each image was identical in size (300 × 260 pixels), resolution (96 dots per inch), brightness, and background. The image stimuli are presented in the center of a 21-inch computer display screen. In EEG data collection, the participant was asked to avoid eye blinking as much as possible during the completion of Stroop tasks to reduce experimental artifacts.

The current study adopted a double-blind design, which recruited experimental assistants for collecting experimental data, and college students with insomnia disorders who participated in the experiment. In this study, the experimental data consist of three parts: questionnaire data, task data, and ERP data. To minimize researcher bias, questionnaire data, task data, and ERP data were analyzed by different individuals. Participants were randomly assigned to the MMG or the WCG group. The experiment was divided into three sections: the pre-test phase (the baseline measurement), the mid-test phase (negative emotion induction), and the post-test phase (the mindfulness-based music listening intervention). All participants were required to complete the PANAS, PSAS, TMS, and MAAS scales, as well as the Stroop task (pre-test) prior to the task to access their baseline level. In the negative emotion induction phase, all participants were asked to watch an 8-min sad movie clip, rate their emotional states using the PANAS again, and complete the Stroop task (the mid-test). During the mindfulness-based music listening intervention phase, the MMG received 10 min of mindfulness meditation training and listened to calm music for 5 min, while the WCG sat still for 15 min. All participants completed the PANAS, TMS, MAAS, and PSAS and finally completed the Stroop task (post-test). The EEG data were recorded throughout the experiment. The entire experiment lasted approximately 1 h.

Independent sample t-tests were used to evaluate the between-group differences in age and sex. Within- and between-group differences in PANAS, PSAS, TMS, and MAAS scores were evaluated using repeated-measures analysis of variance (ANOVA). A repeated-measures ANOVA [3 (test: Pre-test, mid-test, post-test) × 2 (group: MMG; WCG) × 4 (condition: Sad congruency, sad incongruency, happy congruency, and happy incongruency)] was used for the ACC and RTs of emotional faces and emotional words in the experimental task, with task and mood state as within-participant factors and group as between-participant factors. SPSS (version 28.0) was used to analyze the experimental data. The multiple comparisons of post-hoc t-tests were adjusted using Bonferroni, while the p-value for sphericity was adjusted using the Greenhouse–Geisser method.

A tin electrode mounted on an elastic cap from 32 scalp sites (Neuroscan, Charlotte, NC, United States) was used to record brain electrical activity, with the fronto-central aspect (REF) as the reference electrode and the medial frontal aspect (GRD) as ground electrode. An electrode placed infraorbital near the left eye was used to record vertical electrooculogram, and all inter-electrode impedances were maintained below 5 kΩ. MATLAB R2023a was used to process ERP data.

Based on the individual’s grand ERP averages of correct trials, we create the emotional face and word stimuli. The data from 256 to 1,000 Hz were sampled, and high- and low-pass filtering at 0.1 Hz and 45 Hz was performed, respectively, with selecting the left and right mastoids as the reference sites. Data were epoched from 300 ms prior to stimulus onset to 1,000 ms after the presentation and were baseline-corrected to the pre-stimulus interval. Trials were excluded if they included electrooculogram (EOG) artifacts (ocular movements and eye blinks); artifacts owing to amplifier clipping, bursts of electromyographic activity, or peak-to-peak deflections exceeding ±80 μV also were excluded from averaging before independent component analysis (ICA).

No differences in trial counts of between- or within-group for the emotional faces and emotional words sub-tasks were found. Moreover, components of EOG artifacts and head movements in the ICA results were removed after visual inspection. Based on previous studies and emotion–cognition interactions of the topographical distribution of the grand-averaged ERP activities (Liu et al., 2020a,b, 2021a), their time epochs were selected for analysis: Three time windows from 250 to 350 ms [N3 component, (Liu et al., 2021a; Maguire et al., 2013)] and 300 to 600 ms [P3 component, (Liu et al., 2020a,b; Madsen et al., 2019)] and a late time window from 600 to 1,000 ms [late positive component: LPC, (Liu et al., 2019b, 2021a)]. These ERP component latencies were assessed relative to the onset of the visual stimulus, which included four conditions in the emotional faces and emotional words sub-tasks (sad congruency, sad incongruency, happy congruency, and happy incongruency).

The following regions of interest (ROIs) (see Liu et al., 2021c) were selected: frontal (F3, Fz, F4), frontal–central (FC3, FCz, FC4), central (C3, Cz, C4), central–parietal (CP3, CPz, CP4), parietal (P3, Pz, P4), and occipital (O1, OZ, O2) regions. For the conflict control task, repeated measures of ANCOVA [3 (test: post-test, mid-test, post-test) × 4 (condition: sad congruency, sad incongruency, happy congruency, and happy incongruency) × 2 (group: MMG, WCG) × 6 (electrode: frontal, frontal–central, central, central–parietal, parietal, and occipital sites)] were conducted on the amplitudes of N3, P3, and LPC, with group as a between-participant factor and task, condition, and electrode as within-participant factors. SPSS (version 28.0) was used to analyze the data. The multiple comparisons of post-hoc t-tests were adjusted using Bonferroni, while the p-value for sphericity was adjusted using the Greenhouse–Geisser method.

Participants’ demographic information and questionnaire results are presented in Table 1 and Figure 2. No significant between-group differences were observed in age or sex (all ps > 0.05).

Repeated-measures ANOVA on PANAS scores (Figure 2) indicated that there was no main effect of the group (p > 0.05). A main effect of the test was recorded (F (2, 57) = 37.35, p < 0.001, η2 p = 0.39), and the post-hoc t-test showed that compared to the pre- and post-test, the NA score was highest and the PA score was lowest in the mid-test. In the post-test phase, compared to WCG, MMG has higher PA scores and lower NA scores, there were no between-group differences in the pre- and mid-test. The PANAS results indicated that the simulated video significantly induced all participants’ negative emotions (ranked in descending order of scores: Upset >distressed>scared >terrifying) in the mid-test phase, and MBML significantly improves negative emotions in young adults with ID in the post-test.

There was a main effect of the test on PSAS scores (Figure 2) (F (1, 58) = 21.42, p < 0.001, η2 p = 0.27), and the post-test score was lower than the pre-test. A main effect of the group was found (F (1, 58) = 13.54, p < 0.001, η2 p = 0.19), and the scores for MMG were lower than those for WCG. The result of PSAS suggested that MBML reduced cognitive and physical arousal levels in young adults with ID, particularly with a more significant improvement in somatic arousal.

There was a main effect of the test on TMS scores (Figure 2) (F (1, 58) = 7.43, p = 0.008, η2 p = 0.11), and the post-hoc t-test found that the TMS scores for MMG were higher than that of WCG in the post-test, with no between-group difference in pre-test (p > 0.05). From Cronbach’s alpha of TMS (0.70), there is a problem of low validity in the structural validity of the Chinese version of TMS (Chung and Zhang, 2014; Ireland et al., 2019). To make up for this limitation, we mainly used the results of MAAS to evaluate the mindfulness state of participants induced by MBML.

Repeated-measures ANOVA on MAAS scores found no main effect of test and group (all p > 0.05), there was an interaction between test and group, the post-hoc t-test showed that the post-test was higher than the pre-test for the MMG score (p = 0.005), and a score of the MMG was higher than those of the WCG in the post-test (p = 0.074); no insignificant difference between the pre- and post-test for the WCG was found in MAAS (p > 0.05). The result of MAAS indicated that mindfulness-based music intervention improved the levels of mindfulness state among the MMG.

The experimental task in this study consisted of two sub-tasks: The Emotional Faces Sub-task for detecting facial recognition and the Emotional Words Sub-task for evaluating semantic processing.

The ERP results for the emotional face-word Stroop task under the three mood states are presented in Table 4 and Figure 5.

Consistent with existing studies (Liu et al., 2019b, 2021a,c; Luo et al., 2024; Liu et al., 2020a,b), MBML intervention effectively regulated negative emotions induced by an 8-min simulated video and improved positive emotion among young adults with ID. The results of combining questionnaire tasks and ERP indicators suggest that the effective integration of mindfulness meditation and calm music significantly promotes attention processing of conflict control (Ziegler et al., 2019) and cognitive function in young adults with ID (Ding et al., 2024; Liu et al., 2021c; Loo et al., 2020). Based on our questionnaire results and existing research findings (Marchand, 2012; Zhao et al., 2021), MBML reduced levels of cognitive and somatic arousal in individuals with ID.

In the emotional-face sub-task, the Stroop effect on accuracy and response time was significant. Negative emotional states induced by the 8-min simulated video inhibited the processing of happy or sad faces in conditions of congruency or incongruency (lower ACC and slower RTs), whereas the MBML intervention effectively promoted the processing of happy or sad faces in conditions of congruency or incongruency (higher ACC and faster RTs) among young adults with ID. In the emotional word sub-task, there was no significant between-group difference in ACC, but there was a significant Stroop effect on RTs. In the pre-test phase, the RTs of the MMG were faster than those of the WCG, which may be a cue to the higher positive emotions of the MMG compared to those of the WCG (see Table 1). However, there were no significant between-group differences in the negative emotion induction phase of the mid-test. In the intervention phase of the post-test, the MBML improved the RTs of young adults with ID for emotional words. Our results are consistent with those of previous studies (Ding et al., 2023; Liu et al., 2021a; Luo et al., 2024; Liu et al., 2020a,b) in that negative emotions have an inhibitory effect on individual conflict control and semantic processing.

In terms of conflict control performance in the emotional face-word Stroop task, compared with the pre-test (baseline phase), lower ACC and slower RTs were found during the emotional faces sub-task, and slower RTs were recorded in the emotional words sub-task in the mid-test (negative emotion induction phase). In the post-test (MBML intervention phase), higher ACC and faster RTs were observed in the emotional faces sub-task, and faster RTs were found in the emotional word sub-task. In the emotion-cognition interaction, our task and questionnaire results indicated that the negative emotions induced by the 8-minute simulated video had a negative impact on conflict control performance among young adults with ID. This indicates that negative emotions (upset, distressed, scared, terrified) exacerbate the impaired conflict control in young adults with ID. However, in the post-test phase, MBML intervention improved conflict control in young adults with ID, enhanced the levels of cognitive and somatic arousal, and induced positive emotions. Our ERP results in the emotional face-word Stroop task confirmed the behavioral performance in attention processing of conflict control for ID, exhibiting greater N3 effects and smaller P3 and LPC effects.

Our ERP results indicated that the N3, P3, and LPC are EEG indicators that reflect the impact of emotional states on conflict control. Compared to the pre-test phase, although the amplitudes of N3, P3, and LPC showed no statistically significant differences, the P3 and LPC amplitudes decreased for all participants in the mid-test phase. In the intervention phase of the post-test, compared with the WCG, the MBML intervention induced a larger N3 effect in the frontal region and a smaller N3 effect in the occipital region. This indicates that N3, as an EEG indicator reflecting an individual’s emotional state, shows smaller effects with positive emotions and larger effects with negative emotions (Draschkow et al., 2018; Liu et al., 2021a; Maguire et al., 2013; Truman and Mudrik, 2018).

Our ERP results indicate that negative emotions induce smaller P3 effects, whereas larger P3 effects are associated with positive emotions. In the post-test intervention phase, MBML significantly reduced the P3 amplitude in five regions (frontal, frontal-central, central, central–parietal, and occipital regions), which is consistent with previous studies (Ding et al., 2023; Liu et al., 2020a,b, 2021a). Our ERP results reveal that the P3 and LPC are effective indicators for detecting MBML intervention strategies, which can effectively evaluate the conflict control performance of individuals with ID in emotion–cognition interactions. Compared to the WCG, the smaller LPC effect indicates that MBML effectively regulates negative emotions and promotes conflict control in individuals with ID.

Compared to the WCG, the smaller LPC effect indicates that MBML promotes their attention processing of conflict control in the post-test intervention phase. Studies have shown that LPC is associated with higher cognitive resources (Liu et al., 2020a,b) and that listening to calm music in a state of mindfulness can effectively reduce the consumption of higher cognitive resources and alleviate individual cognitive loads (Guo and Koelsch, 2015; Liu et al., 2021a). Based on these advantages in the emotion–cognition interaction, our ERP results support the positive role of MBML intervention strategies in regulating negative emotions among young adults with ID and improving facial and semantic processing in conflict control (Raschle et al., 2017; Zhu et al., 2010).