Skip to main content
NCBI home page
Social Cognitive and Affective Neuroscience logoLink to Social Cognitive and Affective Neuroscience
. 2026 Jul 21;21(1):nsag058. doi: 10.1093/scan/nsag058

Approach motivation sharpens shared neural processing: dissociable beta and alpha synchrony

Yael Caspi 1, Barak Atia 2, Abraham Goldstein 3,✉
PMCID: PMC13445982  PMID: 42478770

Abstract

Emotional narratives invite listeners into shared cognitive–emotional spaces, yet whether and how a listener’s motivational state shapes this convergence remains unknown. Understanding this question is central to accounts of emotional communication, which assume shared processing but rarely examine the individual and contextual factors that modulate it. We recorded magnetoencephalogram (MEG) from 61 participants who listened to anger and fear narratives after an approach or avoidance motivation induction. Source-level intersubject correlation (ISC) of oscillatory envelopes was computed across seven frequency bands in 72 cortical regions, providing spectral specificity unavailable to hemodynamic ISC. Approach motivation enhanced beta-band ISC in left-lateralized temporal networks, and theta ISC bilaterally, consistent with sharpened top-down predictive engagement. Fear narratives produced higher alpha-band ISC than anger narratives in the right-lateralized temporal–parietal regions and the left orbitofrontal cortex, indicating a dissociable sensory-attentional mechanism. Trait behavioral inhibition predicted reduced beta synchrony with the group, particularly when emotional content conflicted with the listener’s motivational orientation. Motivational conflict also uniquely sustained shared processing across the narrative, favoring an engagement account over a fluency account. These findings reveal that shared neural processing during emotional narrative comprehension is simultaneously shaped by motivational state, narrative emotion, and dispositional threat sensitivity through parallel oscillatory channels.

Keywords: intersubject correlation, magnetoencephalography, approach-avoidance motivation, emotional narratives, behavioral inhibition, neural oscillations

Introduction

Emotional narratives are a primary vehicle through which people share and make sense of affective experience, inviting listeners into a shared cognitive–emotional space (Nummenmaa et al. 2012; Nguyen et al. 2019). Intersubject correlation (ISC) of neural activity provides a window into this sharing: the degree to which listeners’ brain responses converge as they hear the same narrative reflects how similarly they process it, with high ISC indicating shared cognitive-affective processing and low ISC idiosyncratic processing shaped by individual context (Hasson et al. 2004; Wilson et al. 2008; Dmochowski et al. 2012; Nastase et al. 2019). Indeed, studies using naturalistic emotional stimuli have shown that affectively engaging narratives and films evoke widespread neural synchrony across listeners, particularly in regions associated with emotional appraisal, language comprehension, and social cognition (Wilson et al. 2008; Dmochowski et al. 2012; Nummenmaa et al. 2012, 2014; Nguyen et al. 2019). Although ISC was developed in functional magnetic resonance imaging (fMRI), it can also be computed on the amplitude envelopes of band-limited magnetoencephalography (MEG) oscillations—a comparatively small literature applied mainly to silent movie viewing (Lankinen et al. 2014; Chang et al. 2015; Lankinen et al. 2018) and continuous speech (Thiede et al. 2020). Whereas fMRI ISC indexes shared slow hemodynamic dynamics, envelope ISC indexes the similarity of oscillatory dynamics within specific frequency bands, allowing shared processing arising from distinct neural mechanisms to be dissociated by frequency. Yet listeners do not approach emotional stories as blank slates. Their motivational state (whether they are oriented toward engagement or withdrawal) may fundamentally shape how they process emotional content, and thus how closely their neural responses align with those of other listeners (Gable and Harmon-Jones 2010; Horan et al. 2014).

Approach and avoidance are fundamental motivational orientations that shape how individuals perceive, attend to, and respond to emotional information (Carver and Scheier 1998; Elliot 2006). Specific emotions carry their own motivational signatures: fear is linked to avoidance, anger to approach (Harmon-Jones et al. 2003; Carver and Harmon-Jones 2009). We organize the congruence predictions with a fluency–engagement contrast, which we treat as a heuristic framework rather than as competing formal theories. Each pole draws on an established tradition. The fluency account extends processing-fluency principles: when a listener’s motivational state aligns with the emotion’s direction, processing is eased, predicting that alignment enhances convergent processing. The engagement account extends predictive-coding views of beta as a sustained top-down signal (Engel and Fries 2010): when state and emotion conflict, the listener cannot readily disengage, and conflict instead sustains shared processing by demanding more effortful, stimulus-driven attention. We use this contrast to organize the exploratory analysis of congruent versus conflicting conditions.

Beyond experimentally induced motivational states, individuals differ in their chronic sensitivity to threat and punishment. The behavioral inhibition system (Gray 1982; Carver and White 1994) reflects a dispositional avoidance bias, a heightened reactivity to aversive cues, novel stimuli, and signals of potential punishment (Gray and McNaughton 2000). This heightened threat sensitivity may bias appraisal of aversive narrative content toward idiosyncratic, threat-focused processing, altering a listener’s neural synchrony with the group. Whether trait inhibition predicts who diverges from shared processing, and whether it interacts with state-level motivation × emotion dynamics, is a central question of the present study.

If motivation and emotion shape shared neural processing, they are unlikely to do so through a single neural mechanism. Beta oscillations (13–30 Hz) support top-down prediction during language processing with a left-hemispheric predominance (Weiss and Mueller 2012; Lewis et al. 2016), alpha oscillations (8–12 Hz) regulate sensory gating and emotional prosodic processing in right-lateralized networks (Jensen and Mazaheri 2010), and theta oscillations (4–8 Hz) support temporal chunking and narrative encoding (Lerner et al. 2011; Giraud and Poeppel 2012). This functional architecture raises the possibility that motivational state and emotional content modulate shared processing through different oscillatory channels, a prediction that has not been tested. Notably, prior work linking motivation to oscillations has focused almost exclusively on resting individual alpha power asymmetry (the frontal alpha asymmetry tradition; Davidson 1998; Harmon-Jones and Gable 2018). Envelope ISC differs fundamentally: it indexes the similarity of oscillatory dynamics across listeners over time, not single-subject activation magnitude. Whether motivational and emotional factors selectively modulate this synchrony in specific frequency bands during naturalistic listening remains an open question.

To address these questions, we recorded MEG while participants listened to emotional narratives after an approach or avoidance motivation induction. Narratives varied in emotion (anger, fear) and coping strategy (approach, avoidance). We computed source-level ISC of oscillatory envelopes across seven frequency bands in 72 cortical regions, enabling spatially and spectrally resolved analysis of shared neural processing. This design allowed us to ask three questions: (i) Does experimentally induced motivational state shape how similarly listeners process emotional narratives, and if so, in which oscillatory channels? (ii) Does the emotional content of the narrative (anger versus fear) modulate shared processing through distinct neural mechanisms? (iii) Does trait behavioral inhibition predict who diverges from shared processing, and does it interact with the state-level motivation × emotion relationship? We predicted that approach motivation would enhance shared processing selectively in the beta band, with greater beta-band ISC in the approach than in the avoidance condition. The rationale follows a three-step chain: beta oscillations carry top-down predictions during language comprehension and signal maintenance of the current cognitive state (Engel and Fries 2010; Lewis et al. 2016); approach motivation sustains engaged, forward-looking processing of content (Gable and Harmon-Jones 2010); and listeners who engage more similarly generate more similar moment-to-moment predictions, whose parallel beta dynamics raise beta ISC. We further predicted that high-BIS individuals would show reduced synchrony, particularly under motivation–emotion conflict. The conflict conditions offer a direct test: if congruence facilitates shared processing, conflict should reduce ISC; if conflict sustains engagement, it should maintain or enhance ISC, with implications for whether shared neural processing reflects ease of processing or depth of engagement. Two predictions were confirmatory and directional: that beta ISC would be greater in approach than avoidance and that behavioral inhibition would predict reduced beta synchrony under motivation–emotion conflict. The emotion-specific alpha effect, the finer decomposition of the inhibition effect, and the fluency-versus-engagement interpretation were exploratory and are labeled as such throughout.

Materials and methods

Participants and design

Sixty-six right-handed native Hebrew speakers (aged 18–33 years, M = 24.25 years) were recruited from Bar-Ilan University. Five were excluded due to missing data, yielding N = 61 (22 men, 39 women; 30 Approach, 31 Avoidance). Sample size was guided by Pajula and Tohka (2016). Participants provided written informed consent and received course credit or 80 NIS. The study was approved by the Ethics Committee, Department of Psychology, Bar-Ilan University (No. 04/18).

The study used a mixed design with Motivation (Approach versus Avoidance) as a between-subjects factor and Emotion (Anger versus Fear) as a within-subjects factor. Participants were randomly assigned to the Approach or Avoidance condition. Each participant listened to two narrative audio clips (∼3 min each): one anger-themed and one fear-themed, both framed with the participant’s assigned motivation. The order of anger and fear narratives was counterbalanced across participants.

Procedure and motivational state induction

Upon arrival, participants completed the Behavioral Inhibition/Behavioral Activation Scales (BIS/BAS; Carver and White 1994) and a demographics questionnaire. Participants lay supine in a magnetically shielded room while MEG was recorded. Each of two blocks began with a 90-s rest, followed by the owl-cheese maze task (Friedman and Förster 2010) to induce the assigned motivational state and one emotional narrative clip (∼3 min). A final resting-state recording concluded the session.

Narrative stimuli

Four narrative audio clips were developed in Hebrew, each approximately 3 min in duration, forming a 2 (Emotion: Anger, Fear) × 2 (Motivation: Approach, Avoidance) matrix. The scripts were based on imagery scripts from Wacker et al. (2003) and described scenarios plausible for university students during the pandemic. All narratives were written in the second person to promote self-referential engagement. The four narratives were as follows: (i) Fear-Approach: receiving a failing grade with a cheating accusation, the protagonist confronts the department head; (ii) Fear-Avoidance: a computer crashes before a deadline, the protagonist withdraws; (iii) Anger-Approach: a power outage threatens a deadline, the protagonist confronts the administration; and (iv) Anger-Avoidance: receiving a failing grade with a cheating accusation, the protagonist withdraws. The exam/cheating scenario thus appeared in both Fear-Approach and Anger-Avoidance, while the deadline/technical-failure scenario appeared in Anger-Approach and Fear-Avoidance, partially crossing narrative content with the experimental conditions. Scripts were narrated by the same male narrator.

The scripts were pre-tested on an independent sample of 70 students. Anger scripts elicited greater anger than fear scripts (M = 6.60 versus 4.07, t(69) = 8.79, P < .001), and fear scripts elicited greater fear than anger scripts (M = 5.93 versus 5.09, t(69) = 2.46, P = .017). No significant differences emerged between approach and avoidance framings in anger or fear ratings. Matched script excerpts illustrating the emotion and motivation manipulations are provided in Supplementary Methods.

All narratives were delivered via a STAX SRS-005 amplifier and SR-003 electrostatic ear speakers coupled by vinyl tubing to silicone earpieces, preventing magnetic interference within the shielded room. Audio onset was marked with two square-wave triggers on the MEG acquisition system.

MEG acquisition

Brain activity was recorded with a 4D-Neuroimaging 248-channel whole-head magnetometer system at a sampling rate of 1017.25 Hz, with an online bandpass filter of 0.1–400 Hz. Environmental noise reduction was achieved using reference coils. Head shape and coil position were digitized using a Polhemus FASTTRACK digitizer. A 10 ms audio delay correction was applied to align neural recordings with stimulus onset. The experiment was presented using PsychoPy 3.1.5.

MEG preprocessing

Line noise, building vibrations, and heartbeat artifacts were removed (Tal and Abeles 2013). Bad channels were identified using a standard deviation threshold of 4000fT and replaced via spherical spline interpolation (mean 0.3 per subject). ICA (FastICA, 25 components) was fitted on 1–40 Hz band-passed data and applied to the unfiltered continuous recording; ocular artifacts were identified using averaged frontal magnetometers as an electrooculogram (EOG) proxy (z-score threshold = 2.0) and muscle artifacts using a correlation threshold of 0.3 (mean 2.6 components removed per subject).

Source reconstruction

Source-level analysis used Linearly Constrained Minimum Variance (LCMV) beamforming. Individualized head models were built by warping a template MRI (MNI Colin27) to each participant’s digitized head shape (single-shell BEM, FieldTrip; see Supplementary Section 2.3). Data covariance was estimated from the broadband continuous recording over the full duration of each narrative clip, yielding a single common LCMV spatial filter per clip rather than separate filters for distinct time windows (Ledoit-Wolf estimation, 5% regularization, max-power orientation, unit-noise-gain normalization). Source voxels (3273 per subject) were parceled into 72 bilateral cortical ROIs using the AAL atlas, and ROI time courses were obtained by averaging across all voxels within each region.

Frequency-band envelope extraction

Source time courses were bandpass filtered (fourth-order Butterworth) into seven bands: delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), low beta (13–20 Hz), high beta (20–30 Hz), full beta (13–30 Hz), and gamma (30–50 Hz). Analytic envelopes were extracted via Hilbert transform, low-pass filtered at 0.8 Hz, downsampled to 100 Hz, and z-scored (following Thiede et al. 2020).

Intersubject correlation

Leave-one-out ISC: For each subject, we computed the Pearson correlation between that subject’s envelope and the mean of all remaining subjects listening to the same clip. Significance was assessed via circular time-shift permutation tests (1000), with region of interest (ROI)-level P-values corrected by false discovery rate (FDR) (Benjamini–Hochberg).

Sliding-window ISC and stimulus feature tracking

ISC was computed in 5-s sliding windows with 1-s steps, yielding ISC time courses that could be related to moment-by-moment narrative properties. Eight acoustic and linguistic stimulus features (four prosodic, four text-derived; see Supplementary Materials for definitions and computational details) were computed from the audio and transcripts and binned into matching 5-s windows.

ISC–feature associations were tested at three levels: (i) Pearson correlations between whole-brain ISC time courses and each feature per band and clip, with effective degrees of freedom corrected for autocorrelation; (ii) multiple regression with all eight features as simultaneous predictors; (iii) cross-clip meta-analysis using Fisher z-transformation and Stouffer’s method, FDR-corrected across features.

Experimental design analysis

Motivation and emotion effects on ISC were tested using 2 × 2 analysis of variance (ANOVA; Emotion × Motivation) per frequency band, and a cross-band mixed ANOVA including Band as an additional factor. Post-hoc comparisons used independent-samples and paired-samples t-tests.

Individual differences: BIS/BAS

BIS and BAS scores (available for 60 of 61 participants) were correlated with per-subject ISC values, with partial correlations controlling for Motivation condition. Clip-specific BIS–ISC correlations were computed to test whether trait inhibition sensitivity differentially predicted neural synchrony across the four experimental conditions. Within-subject emotion contrasts (Fear − Anger ISC) were correlated with BIS separately for each motivation group to test whether BIS moderates emotion-specific processing.

An intersubject representational similarity analysis (IS-RSA) tested whether participants with similar BIS scores showed more similar neural responses (i.e. whether high-BIS individuals converge on an alternative shared processing mode instead of simply diverging from the group).

Speech transcription and alignment

Audio was transcribed and force-aligned using WhisperX (Whisper large-v2, Hebrew). Original Hebrew script was aligned to the automatic speech recognition (ASR) output to correct transcription errors before natural language processing (NLP) feature computation.

Software

All MEG analyses were conducted in MNE-Python. Statistical tests used SciPy and statsmodels. NLP features were computed using sentence-transformers (AlephBERT for Hebrew embeddings) and a Hebrew GPT-Neo model for surprisal. Claude Code (Anthropic) was used to assist in writing code for speech transcription and alignment (section Speech transcription and alignment) and generating publication figures.

Results

Following the maze induction, the Approach and Avoidance groups did not differ in rated task motivation (Approach M = 5.67, Avoidance M = 5.60; t(58) = 0.22, P = .83), indicating equivalent motivational intensity across the two framings and ruling out a differential-arousal or general-engagement account of the motivation effects reported below.

Narrative listening produces robust shared neural dynamics

Listening to emotional narratives elicited significant ISC across all frequency bands (permutation P < .001; Fig. 1). Beta-band ISC was strongest (full-band mean r = .038–.066 across clips; sub-band values ranged .023–.060), followed by alpha (.033–.053) and theta (.007–.022), with 71–72 of 72 regions showing significant beta ISC after FDR correction. ISC magnitudes reflect source-level oscillatory envelope correlations, which are typically lower than sensor-level or fMRI-based ISC values, and are consistent with comparable MEG pipelines (Thiede et al. 2020).

Figure 1.

Color-coded grid (heatmap) of mean ISC values. Rows are frequency bands from theta to gamma; columns are four conditions: Anger and Fear narratives within the Approach group (left) and the Avoidance group (right). Darker blue indicates higher correlation. The beta row is darkest overall, and cells in the two Approach columns are consistently darker than the matching Avoidance columns. Each cell is labeled with its correlation value and significance asterisks.

Open in a new tab

Whole-brain ISC by frequency band and experimental condition. Each cell shows the mean leave-one-out ISC value averaged across 72 cortical ROIs. Columns represent the four conditions (Anger and Fear narratives within Approach and Avoidance groups). Rows represent frequency bands ordered from delta (1–4 Hz) to gamma (30–50 Hz). Asterisks denote permutation significance (***P <.001, **P <.01, *P <.05, ns = not significant; 1000 circular time-shift permutations). Beta-band ISC was the strongest overall, with approach conditions (left) showing consistently higher ISC than avoidance conditions (right).

Shared neural processing is organized by frequency and hemisphere

To characterize the spatial organization of ISC, we averaged per-subject leave-one-out ISC across left-hemisphere and right-hemisphere ROIs (36 each) and entered them into a Band × Hemisphere repeated-measures ANOVA. This revealed a highly significant interaction [F(4,240) = 26.42, P < .0001], indicating that the lateralization of shared processing depends on frequency band (Fig. 2). Beta ISC was left dominant (left: .180, right: .123), whereas alpha (.129 versus .152) and theta (.052 versus .074) were right dominant. At the ROI level, 17 of the top 20 beta ROIs were left lateralized, while 15 of the top 20 alpha ROIs were right lateralized (Supplementary Table S1). This frequency-dependent lateralization frames the subsequent analyses.

Figure 2.

Two rows of glass-brain projections (transparent brain outlines seen from the side, back, and top) showing the 20 regions with the highest ISC. Panel a (beta band) shows dots concentrated in the left hemisphere (17 left, 3 right). Panel b (alpha band) shows dots concentrated in the right hemisphere (5 left, 15 right). Left-hemisphere regions are blue and right-hemisphere regions are red.

Open in a new tab

Spatial lateralization of ISC by frequency band. Glass-brain projections show the 20 ROIs with the highest mean ISC for (a) beta (13–30 Hz) and (b) alpha (8–12 Hz), with ROIs colored by hemisphere (blue = left, red = right). Beta ISC was predominantly left-lateralized (17 of 20 top ROIs), spanning inferior frontal, cingulate, and occipital regions. Alpha ISC was predominantly right-lateralized (15 of 20 top ROIs), concentrated in temporal and parietal regions. This frequency-dependent lateralization was confirmed by a significant Band × Hemisphere interaction (F(4,240) = 26.42, P <.0001).

Motivation and emotion modulate shared processing through distinct lateralized channels

A full factorial ANOVA (Band × Hemisphere × Emotion × Motivation; Supplementary Table S2) tested whether motivational state and emotion modulate ISC. Below we report the theoretically relevant effects.

Motivation shapes ISC through left-lateralized beta networks

Approach motivation enhanced shared neural processing overall [F(1,59) = 7.61, P = .008]. Band-by-band follow-ups (Table 1) confirmed approximately 50% higher ISC for Approach than Avoidance participants in theta and all beta and sub-bands (.059 versus .039 in full beta), with no motivation effect in alpha or delta.

Table 1.

Band-by-band effects of Motivation (Approach > Avoidance) on whole-brain ISC. Asterisks denote P-value significance: * P < 0.05, ** P<0.01.

Band F(1,59) P η²P d
Theta 11.23 .001** .160 0.86
Low beta 6.17 .016* .095 0.64
High beta 5.88 .018* .091 0.62
Beta (full) 5.27 .025* .082 0.59
Gamma 4.38 .041* .069 0.54
Alpha 0.04 .837 .001 0.05
Delta 0.56 .458 .009 0.19
Open in a new tab

Critically, the motivation effect was lateralized. The three-way Band × Hemisphere × Motivation interaction was significant [F(4,236) = 2.48, P = .045, η2p = .040]. The Approach > Avoidance effect was lateralized differently across bands: beta/gamma ISC was enhanced selectively in the left hemisphere, delta in the right, theta bilaterally, and alpha ISC was unaffected in either hemisphere (left P = .658, right P = .666). Thus, the motivational enhancement of shared processing in the beta range was concentrated in the same left-lateralized networks identified in the previous section.

At the ROI level, 24 of 72 regions showed FDR-significant motivation effects on beta ISC (Supplementary Table S3). The largest effects were in the left temporal cortex (inferior and middle temporal gyrus, fusiform gyrus; all d > 1.0). Although most significant ROIs were left lateralized (15 of 24), nine right-hemisphere regions also survived correction. The effect was thus broadly distributed but left dominant, concentrated in temporal regions associated with language comprehension and semantic processing.

Emotion modulates ISC through right-lateralized alpha networks

The emotion effect showed a different pattern. The main effect of Emotion was significant [F(1,59) = 4.43, P = .040], with fear narratives showing higher ISC overall. Planned comparisons showed selectivity to alpha (t = 2.58, P = .013, d = 0.43), with no emotion effect in other bands. The emotion effect was spatially complementary to the motivation effect: hemisphere-split analyses revealed a significant effect in the right hemisphere [F(1,59) = 5.75, P = .020] but not the left [F(1,59) = 1.51, P = .224]. The right-lateralized alpha regions driving this effect (superior and middle temporal cortex, postcentral gyrus, supramarginal gyrus) are associated with prosodic processing and emotional voice perception.

At the ROI level, nine regions showed FDR-significant emotion effects on alpha ISC, the strongest in left orbitofrontal cortex (d ≈ 0.5), with left angular gyrus showing the opposite pattern (higher alpha ISC for anger; d = −0.48). Thus, while the hemisphere-level emotion effect was carried by a broadly distributed pattern across right-hemisphere temporal and parietal regions, the strongest focal effects emerged in left orbitofrontal cortex, a region implicated in emotional appraisal and value-based evaluation of affective content.

Together, these results reveal a dissociation (Fig. 3): motivation selectively modulates beta ISC in left-hemisphere networks, while emotion selectively modulates alpha ISC in right-hemisphere regions, operating through parallel channels that differ in frequency, topography, and functional significance.

Figure 3.

Two bar charts with individual data points overlaid and error bars showing 1 SD. Panel a (beta ISC by motivational state) shows a taller bar for Approach than Avoidance (P = .025). Panel b (alpha ISC by emotion) shows a taller bar for Fear than Anger (P = .012).

Open in a new tab

Dissociation between motivation and emotion effects on ISC. (a) Beta ISC (13–30 Hz) by motivational state: approach participants showed significantly higher beta ISC than avoidance participants. (b) Alpha ISC (8–12 Hz) by emotion type: fear narratives elicited significantly higher alpha ISC than anger narratives. Bars show means across subjects; error bars represent ±1 SE. Individual data points are overlaid. Motivational state selectively modulated beta ISC with no effect on alpha, while emotion selectively modulated alpha ISC with no effect on beta.

Trait behavioral inhibition predicts who diverges from the group

BIS scores negatively correlated with full-beta ISC (r = −.260, P = .045; Fig. 4a), surviving partial correlation for motivation condition. The effect was stronger in high-beta (r = −.287, P = .026); BAS showed no association (all P > .22).

Figure 4.

Panel a is a scatter plot of trait BIS score on the x-axis against mean beta ISC on the y-axis, with points colored by motivation group and a downward-sloping dashed regression line (negative association, r = −0.26). Panel b is a bar chart of the BIS–beta ISC correlation for each of the four clips; all bars are negative, and the anger-avoidance bar is the most negative (r = −0.43).

Open in a new tab

Trait BIS and beta ISC. (a) Scatter plot of BIS score against mean beta ISC averaged across clips, with participants marked by motivation group (circles = approach, triangles = avoidance). The dashed line shows the overall regression (r = −.260, P =.045). (b) Clip-specific BIS–beta ISC Pearson correlations. The BIS–ISC association was strongest for the Anger-Avoidance clip (r = −.429, P =.016), which paired an approach-related emotion (anger) with an avoidance motivational state. * P <.05.

The BIS–ISC relationship was not uniform across experimental conditions (Fig. 4b). It was strongest for the Anger-Avoidance clip, (r = −.429, P = .016), the condition pairing an approach-related emotion with an avoidance motivational state. BIS showed no association with Fear-Avoidance ISC (r = −.036, P = .85), indicating that high-BIS individuals diverged selectively when processing a BIS-incongruent emotion, not when processing a BIS-congruent one. Within the avoidance group, BIS predicted the relative balance between fear and anger ISC (full beta: r = +.500, P = .004): high-BIS individuals showed preserved synchrony for fear narratives but markedly reduced synchrony for anger narratives. This pattern was particularly pronounced in the low-beta sub-band (r = +.592, P < .001) and absent in high beta (r = +.050, P = .788), indicating that the BIS moderation of emotion-specific processing operates through low-frequency beta mechanisms. A median split on BIS confirmed the overall pattern, with high-BIS participants showing lower beta ISC than low-BIS participants (d = −0.39, P = .155), a trend that was significant in the high-beta sub-band (d = −0.64, P = .025).

IS-RSA tested whether participants with similar BIS scores showed more similar neural responses; no significant effects emerged (all P > .13), indicating idiosyncratic divergence.

Motivational conflict produces uniquely sustained engagement

The omnibus Emotion × Motivation interaction was not significant [F(1,59) = 1.99, P = .163], but Fear-Approach showed the highest ISC across every beta-range band (Fig. 1), motivating a planned time-resolved analysis of their distinct temporal profiles.

To test whether shared processing changed over the course of each narrative, we computed per-subject leave-one-out beta ISC separately for the first and second halves of each clip. A one-way ANOVA on the late-minus-early ISC difference revealed a significant effect of Clip [F(3,118) = 4.95, P = .003; Fig. 5; Supplementary Table S4]. Fear-Approach was the only clip showing a significant increase in beta ISC from early to late [t(29) = 2.42, P = .022, d = 0.45], whereas Fear-Avoidance showed a significant decline (t(30) = −3.47, P = .002, d = −0.63), and the remaining clips showed no significant change. Post-hoc comparisons confirmed that Fear-Approach’s temporal increase was significantly larger than every other clip (all P < .04). The sustained rise was not attributable to narrative structure: Fear-Approach had the fewest semantic boundaries and lowest pause rate in its second half, while Fear-Avoidance showed declining ISC despite having more of both, implicating the motivational conflict itself (approach-state listeners processing fear content) rather than low-level stimulus properties.

Figure 5.

Line graph of beta ISC over the duration of the narrative (time in seconds on the x-axis), with four lines for the four conditions, each surrounded by a shaded confidence band, and a vertical dashed line marking the midpoint. In the second half, the Fear-Approach line rises above the others while the Fear-Avoidance line declines; the remaining two stay relatively flat.

Open in a new tab

Time-resolved beta ISC during narrative listening. Sliding-window ISC (5 s windows, 1 s steps) is plotted across the duration of each narrative. The vertical dashed line marks the narrative midpoint. Fear-Approach (dark blue) showed a sustained rise in beta ISC across the second half of the narrative (t(29) = 2.42, P =.022, d = 0.45), while Fear-Avoidance declined and the remaining clips remained stable. Fear-Approach’s temporal increase was significantly larger than every other clip (all P <.04).

Stimulus feature tracking confirms the beta–alpha dissociation

Time-resolved correlations between windowed ISC and eight stimulus features reinforced the beta–alpha dissociation. Beta ISC increased during slow speech and pauses, while alpha ISC increased at semantic boundaries. Multiple regression explained 8%–11% of windowed ISC variance. Full feature-tracking results, including clip-specific correlations and cross-clip meta-analyses, are reported in Supplementary Material (Fig. S1).

Discussion

The present study asked whether motivational state, emotional content, and dispositional threat sensitivity shape shared neural processing of emotional narratives. Three findings emerged: (i) approach motivation enhanced beta ISC in left-lateralized networks (and theta bilaterally); (ii) fear narratives produced higher alpha-band ISC in right-lateralized regions and left orbitofrontal cortex; and (iii) trait behavioral inhibition predicted reduced beta synchrony, particularly under motivation–emotion conflict. These findings indicate that shared processing is a multi-channel phenomenon, each operating through a distinct oscillatory and spatial pathway. Because ISC reflects shared temporal dynamics across multiple co-occurring processes, from sensory entrainment to higher-order integration, we interpret the regional effects below at the network level rather than as one-to-one mappings between ROIs and discrete functions.

Approach motivation sharpens shared narrative processing

Approach motivation produced greater ISC than avoidance selectively in the beta and theta bands. The selectivity of the approach motivation effect to beta and theta oscillations (Engel and Fries 2010; Giraud and Poeppel 2012; Lewis et al. 2016) suggests that approach motivation promotes a shared predictive framework in which listeners generate more similar expectations about what comes next in the narrative. Theta ISC was equally sensitive to motivation, in keeping with its role in temporal chunking and narrative encoding (Lerner et al. 2011).

The ROI-level analysis identified the left temporal regions as the sites where approach motivation most strongly enhanced shared processing. These regions are core components of the ventral language stream involved in lexical-semantic processing and narrative comprehension (Hickok and Poeppel 2007). The left-lateralized beta enhancement places the motivation effect within language networks (Weiss and Mueller 2012).

Moment-to-moment feature tracking provided converging evidence for the predictive account. Beta ISC peaked during slow speech and pauses, moments when external input is reduced, and listeners are likely generating predictions or integrating preceding content. This pattern fits the proposal that beta oscillations reflect a “status quo” maintenance signal that is most prominent when the system is holding a prediction rather than receiving new input (Engel and Fries 2010). Importantly, avoidance motivation did not eliminate shared processing; avoidance listeners still showed significant ISC in most cortical regions. The effect is one of magnitude, not presence or absence, suggesting that avoidance motivation loosens but does not dismantle the shared predictive framework through which listeners engage with emotional narratives.

Fear captures shared attention through alpha

The emotion effect operated through a distinct neural channel. Fear narratives were associated with more convergent alpha processing in right-lateralized regions linked to prosodic and emotional voice processing (Ross 1981; Belin et al. 2000; Jensen and Mazaheri 2010; Ethofer et al. 2012; Klimesch 2012), pointing to a sensory-attentional mechanism distinct from the top-down predictive mechanism associated with motivation and beta. The strongest focal effects fell in the left orbitofrontal cortex (Kringelbach 2005; Rolls 2019), consistent with more uniform appraisal-related processing for fear, with left angular gyrus (Binder et al. 2009) showing the opposite pattern for anger.

Why is fear associated with more convergent processing than anger? One possibility is that fear narratives contain more uniform attentional capture cues (falling pitch, pauses, tension markers) that entrain listeners’ attention similarly. Anger, in contrast, may afford more variable interpretive stances: some listeners may resonate with the protagonist’s confrontational response while others may find it aversive, producing more divergent neural responses. This interpretation is consistent with the broader observation that fear is a more stereotyped emotional response than anger, which varies more with context, agency, and appraisal (Scherer 2009).

When motivation and emotion conflict: sustained engagement, not disengagement

Fear-Approach, the condition pairing a withdrawal-related emotion with an approach motivational state, presented the most striking pattern. A simple alignment model would predict that motivational conflict reduces shared processing. Instead, Fear-Approach produced the highest beta ISC of any condition and was the only clip in which shared processing escalated over time. This pattern favors the engagement over the fluency reading: motivational conflict does not disrupt shared processing but appears to sustain it. Because the omnibus interaction was not significant, this interpretation rests on the planned time-resolved follow-up and should be regarded as suggestive. Listeners in an approach state who encounter fear content may be unable to disengage from the narrative, producing a vigilant, locked-in response that is highly convergent across individuals. This accords with evidence that approach motivation during negative affect sustains engagement and increases left frontal activity (Harmon-Jones et al. 2003; Gable and Harmon-Jones 2010; Threadgill and Gable 2019).

The early-to-late escalation in Fear-Approach, contrasting with the decline in Fear-Avoidance, implicates the motivational conflict itself as the driver, since the Fear-Approach clip had the fewest semantic boundaries and lowest pause rate in its second half.

Trait behavioral inhibition: who falls out of step

The BIS finding addresses a question central to the study of shared affective experience: who shares in shared processing, and who processes differently? High-BIS individuals showed reduced beta ISC: they fell out of step with the group. This effect survived controlling for condition assignment, was specific to the inhibition system (BAS showed no association), and was most pronounced during the Anger-Avoidance narrative. The specificity to Anger-Avoidance is theoretically coherent: anger is an approach-related emotion (Harmon-Jones et al. 2003), avoidance is the induced motivational state, and high BIS amplifies sensitivity to avoidance cues (Gray and McNaughton 2000). This creates a convergence of opposing forces (the emotion pushes toward approach, the state pushes toward avoidance, and the dispositional bias amplifies avoidance) under which shared processing breaks down most dramatically. The within-avoidance pattern adds nuance: high-BIS individuals tracked the fear narrative (a BIS-congruent emotion) through the shared framework but decoupled from it during the anger narrative (a BIS-incongruent emotion), an effect concentrated in the low-beta sub-band. High-BIS listeners thus appear unable to sustain convergent processing when narrative emotion conflicts with their dispositional orientation.

The IS-RSA null result bears on this question. If high-BIS individuals diverged from the group mean but converged with each other, it would suggest an alternative shared processing mode, a BIS-typical way of engaging with the narrative. Instead, the null result indicates that high-BIS listeners diverge idiosyncratically, each processing the narrative through their own threat-sensitive lens (Finn et al. 2020) and not through a common alternative interpretation. The absence of any BAS–ISC association reinforces the asymmetry between inhibition and activation systems, consistent with evidence that inhibition-related processes more strongly modulate neural dynamics (Gray and McNaughton 2000; Amodio et al. 2008).

Limitations and future directions

The most significant limitation concerns stimulus confounding. Each experimental condition involved a different narrative, and although the scenarios were partially crossed (the exam plot appeared in both Fear-Approach and Anger-Avoidance; the paper plot in Anger-Approach and Fear-Avoidance), narrative content was not fully independent of the Emotion × Motivation manipulation. The feature-tracking analyses provide some reassurance: condition-level effects were not explained by moment-to-moment narrative properties, and the Fear-Approach temporal escalation occurred despite that clip having the fewest boundary transitions and lowest pause rate. We note, however, that controlling low-level acoustic and semantic features does not establish that the higher-order construct of emotion was isolated, and that ISC indexes shared processing rather than any specific affective mechanism; the emotion effects in particular should be read as describing where and in which band shared processing differed, not as identifying a determinate emotional mechanism. The manipulations also differ in causal warrant. Because motivation was randomized between subjects, its effect is causally attributable to the induction; the within-subject emotion contrast, with each emotion realized in a single narrative, does not support comparably strong causal claims. The induction also rested on the established maze paradigm rather than a direction-specific in-session check. Future work should administer a validated approach/avoidance measure during the session, employ a fully crossed design that reframes the same scenarios with different emotional and coping content, and include a within-subjects motivation manipulation, to remove stimulus-specific confounds and strengthen causal inference.

ISC measures convergence, not depth of processing (Simony et al. 2016); linking motivation-dependent ISC to behavioral outcomes would strengthen interpretation. Replication in other languages would help determine whether the motivation effect generalizes beyond Hebrew-specific language processing.

The sample of approximately 30 participants per group is adequate for detecting ISC effects (Pajula and Tohka 2016) but limits power for individual-difference analyses, particularly the condition-specific BIS correlations that involve subgroup analyses. The BIS–ISC associations are theoretically coherent and internally consistent across multiple analytic approaches, but the moderate effect sizes warrant replication in larger and more diverse samples. Finally, the present study examined shared processing at the group level using leave-one-out ISC. Dyadic approaches, testing whether pairs of listeners who share motivational states show higher neural synchrony than mismatched pairs, would extend the present findings from group-average convergence to the interpersonal alignment that characterizes real-world emotional communication.

Conclusion

How similarly we process emotional narratives depends on our motivational state, the emotion being conveyed, and our dispositional sensitivity to threat. Approach motivation sharpens shared predictive processing in left-lateralized beta networks, with parallel theta enhancement that is not lateralized; fear captures shared sensory-attentional processing in right-lateralized alpha networks; and trait behavioral inhibition predicts who diverges, particularly under motivation–emotion conflict. Contrary to a simple alignment model, motivational conflict sustains shared processing, suggesting that the interplay between listener state and narrative emotion is more nuanced than congruence alone would predict. These findings reveal that shared neural processing during emotional narrative comprehension is a multi-channel phenomenon shaped by the convergence and conflict of motivational, emotional, and dispositional forces acting through parallel neural pathways.

Supplementary Material

nsag058_Supplementary_Data
nsag058_supplementary_data.docx (130KB, docx)

Contributor Information

Yael Caspi, Department of Psychology and Gonda Multidisciplinary Brain Research Center, Bar-Ilan University, Ramat Gan 52900, Israel.

Barak Atia, Department of Psychology and Gonda Multidisciplinary Brain Research Center, Bar-Ilan University, Ramat Gan 52900, Israel.

Abraham Goldstein, Department of Psychology and Gonda Multidisciplinary Brain Research Center, Bar-Ilan University, Ramat Gan 52900, Israel.

Author contributions

Yael Caspi (Data curation [equal], Formal analysis [equal], Investigation [equal], Methodology [equal], Writing—review & editing [equal]), Barak Atia (Data curation [equal], Investigation [equal], Software [equal]), and Abraham Goldstein (Conceptualization [equal], Formal analysis [equal], Funding acquisition [equal], Methodology [equal], Software [equal], Supervision [equal], Writing—original draft [equal])

Supplementary material

Supplementary material is available at SCAN online.

Conflicts of interest

None declared.

Funding

This work was partly supported by the Israel Science Foundation grant 726/15 to A.G.

Data availability

The data and analysis code underlying this article will be shared on reasonable request to the corresponding author.

References

  1. Amodio DM, Master SL, Yee CM  et al.  Neurocognitive components of the behavioral inhibition and activation systems: implications for theories of self-regulation. Psychophysiology  2008;45:11–9. 10.1111/j.1469-8986.2007.00609.x [DOI] [Google Scholar]
  2. Belin P, Zatorre RJ, Lafaille P  et al.  Voice-selective areas in human auditory cortex. Nature  2000;403:309–12. 10.1038/35002078 [DOI] [Google Scholar]
  3. Binder JR, Desai RH, Graves WW  et al.  Where is the semantic system? A critical review and meta-analysis of 120 functional neuroimaging studies. Cerebral Cortex  2009;19:2767–96. 10.1093/cercor/bhp055 [DOI] [Google Scholar]
  4. Carver CS, Harmon-Jones E.  Anger is an approach-related affect: evidence and implications. Psychol Bull  2009;135:183–204. 10.1037/a0013965 [DOI] [Google Scholar]
  5. Carver CS, Scheier MF.  On the Self-Regulation of Behavior. New York: Cambridge University Press, 1998. [Google Scholar]
  6. Carver CS, White TL.  Behavioral inhibition, behavioral activation, and affective responses to impending reward and punishment: the BIS/BAS scales. J Pers Soc Psychol  1994;67:319–33. 10.1037/0022-3514.67.2.319 [DOI] [Google Scholar]
  7. Chang WT, Jääskeläinen IP, Belliveau JW  et al.  Combined MEG and EEG show reliable patterns of electromagnetic brain activity during natural viewing. Neuroimage  2015;114:49–56. 10.1016/j.neuroimage.2015.03.066 [DOI] [Google Scholar]
  8. Davidson RJ.  Anterior electrophysiological asymmetries, emotion, and depression: conceptual and methodological conundrums. Psychophysiology  1998;35:607–14. 10.1017/s0048577298000134 [DOI] [Google Scholar]
  9. Dmochowski JP, Sajda P, Dias J  et al.  Correlated components of ongoing EEG point to emotionally laden attention- a possible marker of engagement?. Front Hum Neurosci  2012;6:112. 10.3389/fnhum.2012.00112 [DOI] [Google Scholar]
  10. Elliot AJ.  The hierarchical model of approach-avoidance motivation. Motiv Emot 2006;30:111–116. [Google Scholar]
  11. Engel AK, Fries P.  Beta-band oscillations—signalling the status quo?  Curr Opin Neurobiol  2010;20:156–65. 10.1016/j.conb.2010.02.015 [DOI] [Google Scholar]
  12. Ethofer T, Bretscher J, Gschwind M  et al.  Emotional voice areas: anatomic location, functional properties, and structural connections revealed by combined fMRI/DTI. Cereb Cortex  2012;22:191–200. 10.1093/cercor/bhr113 [DOI] [Google Scholar]
  13. Finn ES, Glerean E, Khojandi AY  et al.  Idiosynchrony: from shared responses to individual differences during naturalistic neuroimaging in the human brain. NeuroImage  2020;10215:1045116828–.60.e6. [Google Scholar]
  14. Friedman R, Förster J.  Implicit affective cues and attentional tuning: an integrative review. Psychol Bull  2010;136:875–93. 10.1037/a0020495 [DOI] [Google Scholar]
  15. Gable PA, Harmon-Jones E.  The motivational dimensional model of affect: implications for breadth of attention, memory, and cognitive categorisation. Cogn Emot  2010;24:322–37. 10.1080/02699930903378305 [DOI] [Google Scholar]
  16. Giraud AL, Poeppel D.  Cortical oscillations and speech processing: emerging computational principles and operations. Nat Neurosci  2012;15:511–7. 10.1038/nn.3063 [DOI] [Google Scholar]
  17. Gray JA.  The Neuropsychology of Anxiety: An Enquiry into the Functions of the Septo-Hippocampal System. Oxford: Oxford University Press, 1982. [Google Scholar]
  18. Gray JA, McNaughton N.  The Neuropsychology of Anxiety: An Enquiry into the Functions of the Septo-Hippocampal System, 2nd edn. Oxford: Oxford University Press, 2000. [Google Scholar]
  19. Harmon-Jones E, Gable PA.  On the role of asymmetric frontal cortical activity in approach and withdrawal motivation: an updated review of the evidence. Psychophysiology  2018;55:e12879. 10.1111/psyp.12879 [DOI] [Google Scholar]
  20. Harmon-Jones E, Sigelman J, Bohlig A  et al.  Anger, coping, and frontal cortical activity: the effect of coping potential on anger‑induced left frontal activity. Cogn Emot  2003;17:1–24. 10.1080/02699930302278 [DOI] [Google Scholar]
  21. Hasson U, Nir Y, Levy I  et al.  Intersubject synchronization of cortical activity during natural vision. Science  2004;303:1634–40. 10.1126/science.1089506 [DOI] [Google Scholar]
  22. Hickok G, Poeppel D.  The cortical organization of speech processing. Nat Rev Neurosci  2007;8:393–402. 10.1038/nrn2113 [DOI] [Google Scholar]
  23. Horan WP, Wynn JK, Mathis I  et al.  Approach and withdrawal motivation in schizophrenia: an examination of frontal brain asymmetric activity. PLoS One  2014;9:e110007. 10.1371/journal.pone.0110007 [DOI] [Google Scholar]
  24. Jensen O, Mazaheri A.  Shaping functional architecture by oscillatory alpha activity: gating by inhibition. Front Hum Neurosci  2010;4:186. 10.3389/fnhum.2010.00186 [DOI] [Google Scholar]
  25. Klimesch W.  Alpha-band oscillations, attention, and controlled access to stored information. Trends Cogn Sci  2012;16:606–17. 10.1016/j.tics.2012.10.007 [DOI] [Google Scholar]
  26. Kringelbach ML.  The human orbitofrontal cortex: linking reward to hedonic experience. Nat Rev Neurosci  2005;6:691–702. 10.1038/nrn1747 [DOI] [Google Scholar]
  27. Lankinen K, Saari J, Hari R  et al.  Intersubject consistency of cortical MEG signals during movie viewing. Neuroimage  2014;92:217–24. 10.1016/j.neuroimage.2014.02.004 [DOI] [Google Scholar]
  28. Lankinen K, Saari J, Hlushchuk Y  et al.  Consistency and similarity of MEG-and fMRI-signal time courses during movie viewing. Neuroimage  2018;173:361–9. 0.1016/j.neuroimage.2018.02.045 [Google Scholar]
  29. Lerner Y, Honey CJ, Silbert LJ  et al.  Topographic mapping of a hierarchy of temporal receptive windows using a narrated story. J Neurosci  2011;31:2906–15. [Google Scholar]
  30. Lewis AG, Schoffelen JM, Schriefers HBastiaansen MCM.  A predictive coding perspective on beta oscillations during sentence-level language comprehension. Front Hum Neurosci  2016;10:85. [Google Scholar]
  31. Nastase SA, Gazzola V, Hasson U  et al.  Measuring shared responses across subjects using intersubject correlation. Soc Cogn Affect Neurosci  2019;14:667–85. 10.1093/scan/nsz037 [DOI] [Google Scholar]
  32. Nguyen M, Vanderwal T, Hasson U.  Shared understanding of narratives is correlated with shared neural responses. NeuroImage  2019;184:161–70. 10.1016/j.neuroimage.2018.09.010 [DOI] [Google Scholar]
  33. Nummenmaa L, Glerean E, Viinikainen M  et al.  Emotions promote social interaction by synchronizing brain activity across individuals. Proc Natl Acad Sci USA  2012;109:9599–604. 10.1073/pnas.1206095109 [DOI] [Google Scholar]
  34. Nummenmaa L, Saarimäki H, Glerean E  et al.  Emotional speech synchronizes brains across listeners and engages large-scale dynamic brain networks. Neuroimage  2014;102 (Pt 2):498–509. 10.1016/j.neuroimage.2014.07.063 [DOI] [Google Scholar]
  35. Pajula J, Tohka J.  How many is enough? Effect of sample size in inter-subject correlation analysis of fMRI.  Comput Intell Neurosci  2016;2016:2094601. 10.1155/2016/2094601 [DOI] [Google Scholar]
  36. Rolls ET.  The orbitofrontal cortex and emotion in health and disease, including depression. Neuropsychologia  2019;128:14–43. 10.1016/j.neuropsychologia.2017.09.021 [DOI] [Google Scholar]
  37. Ross ED.  The aprosodias: functional-anatomic organization of the affective components of language. Archives of Neurology  1981;38:561–69. [Google Scholar]
  38. Scherer KR.  Emotions are emergent processes: they require a dynamic computational architecture. Phil Trans R Soc B  2009;364:3459–74. 10.1098/rstb.2009.0141 [DOI] [Google Scholar]
  39. Simony E, Honey CJ, Chen J  et al.  Dynamic reconfiguration of the default mode network during narrative comprehension. Nat Commun  2016;7:12141. 10.1038/ncomms12141 [DOI] [Google Scholar]
  40. Tal I, Abeles M.  Cleaning MEG artifacts using external cues. J Neurosci Methods  2013;217:30–7. [Google Scholar]
  41. Thiede A, Glerean E, Kujala T  et al.  Atypical MEG inter-subject correlation during listening to continuous natural speech in dyslexia. Neuroimage  2020;216:116799. 10.1016/j.neuroimage.2020.116799 [DOI] [Google Scholar]
  42. Threadgill AH, Gable PA. Negative affect varying in motivational intensity influences scope of memory. Cogn Emot  2019;33:332–45. [Google Scholar]
  43. Wacker J, Heldmann M, Stemmler G.  Separating emotion and motivational direction in fear and anger: effects on frontal asymmetry. Emotion  2003;3:167–93. 10.1037/1528-3542.3.2.167 [DOI] [Google Scholar]
  44. Weiss S, Mueller HM.  “Too many betas do not spoil the broth”: the role of beta brain oscillations in language processing. Front Psychol  2012;3:201. 10.3389/fpsyg.2012.00201 [DOI] [Google Scholar]
  45. Wilson SM, Molnar-Szakacs I, Iacoboni M.  Beyond superior temporal cortex: intersubject correlations in narrative speech comprehension. Cereb Cortex  2008;18:230–42. 10.1093/cercor/bhm049 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

nsag058_Supplementary_Data
nsag058_supplementary_data.docx (130KB, docx)

Data Availability Statement

The data and analysis code underlying this article will be shared on reasonable request to the corresponding author.

Articles from Social Cognitive and Affective Neuroscience are provided here courtesy of Oxford University Press

RESOURCES