Rapid modulation in music supports attention in listeners with attentional difficulties

The study addresses which musical properties support sustained attention and whether these effects depend on individual differences in attentional capacity. With the widespread use of background music during knowledge work, prior findings suggest that music can affect cognition and that personality, preference, and familiarity modulate these effects. Optimal stimulation theory proposes that individuals with attentional difficulties (e.g., ADHD) may require higher levels of stimulation for optimal performance. However, most prior work contrasts stimulation (music/noise) with silence rather than comparing different musical properties. The authors hypothesize that specific acoustic amplitude modulations in music—particularly rapid, beta-range modulations—can enhance sustained attention and that benefits will be greater for listeners with more ADHD-like symptoms. The Sustained Attention to Response Task (SART) is used to quantify sustained attention and its time course.

Prior research shows that music serves functional roles in daily life and can influence cognitive performance. Individual differences (e.g., introversion/extraversion, neuroticism) and factors like preference and familiarity modulate cognitive effects of background music. For individuals with ADHD, auditory stimulation (music or noise) can improve performance, consistent with optimal stimulation theory. Nevertheless, earlier studies largely compare stimulation against silence or noise, leaving open which musical features matter. Neural oscillations in alpha, beta, and gamma bands are implicated in attention and top-down control, and rhythmic auditory stimulation can entrain cortical activity, potentially supporting attentional processes. The SART is validated as a sensitive measure of sustained attention with ecological validity.

Design overview: Four experiments assessed behavioral and neural effects of background audio on sustained attention using the SART under controlled acoustic conditions.

Experiment 1 (Online behavioral):

• Participants: Recruited via Amazon Mechanical Turk (MTurk). Ratings: N=62 after screening; SART performance: N=83 after screening (from 160 recruited). Mean age ~45.
• Conditions: Three background audio conditions: AM+ Music (fast amplitude modulations; high arousal), Control Music (slow modulations; low arousal), and Pink Noise. Stimuli were loudness-normalized by ear (rms: 0.059 AM+, 0.061 Control, 0.014 Pink Noise).
• Tasks: 1A: Valence and arousal ratings. 1B: SART with 1200 trials (digits 0–9; 250 ms presentation, 900 ms mask; 1150 ms ITI). Respond to 1–9; withhold to 0 (10% targets). Six blocks of 200 trials; two blocks per audio condition in counterbalanced order.
• Procedures: Volume calibration task; headphone screening; adherence checks. Participants could self-adjust volume (recorded as covariate). Payment contingent on performance.
• Analysis: Ratings analyzed via within-subjects MANOVA (Music: 3 levels). SART accuracy (d′) analyzed with repeated-measures mixed-effects ANOVA: within-subject factors Music (3), Trial Block (1–6); between-subjects factor Presentation Order; listening volume as covariate.

Experiment 2 (fMRI):

• Participants: N=34 undergraduates (mean age 20.4).
• Task/Conditions: SART during three auditory conditions (AM+ Music, Control Music, Pink Noise) in counterbalanced order. ITI=1425 ms (3 TRs; TR=475 ms). ~10 min per condition (~30 min total).
• MRI acquisition: 3T Siemens Skyra. T1 MPRAGE (0.8 mm isotropic). EPI: TR=475 ms, TE=30 ms, 48 slices, 3×3×3 mm³, ~1268 volumes per run.
• Preprocessing: SPM12 + CONN toolbox: motion correction, slice timing, artifact detection, normalization to MNI, 8 mm smoothing; denoising (WM/CSF confounds), bandpass 0.008–0.09 Hz for connectivity.
• Analyses: Within-subject ANOVA comparing overall activity across conditions (voxel- and cluster-wise FDR p<0.05). Trial types (hits, misses, false alarms, correct rejections) modeled separately. Parametric contrasts Hits>False Alarms per condition; seed-based connectivity assessed between conditions.

Experiment 3 (EEG):

• Participants: N=40 undergraduates (mean age 19).
• Task/Conditions: SART (680 trials; ITI 1150 ms) with four auditory conditions: AM+, Silence (within-subject), and either Control Music or Pink Noise (between-subjects).
• EEG recording: 64-channel Brain Vision actiCHamp; sound-attenuated, shielded room.
• Preprocessing: 0.5 Hz high-pass; 60 Hz notch; re-reference to TP9/TP10; ICA ocular artifact correction; bad channels rejected/interpolated.
• Stimulus-brain coupling: Morlet wavelet filtering 1–50 Hz (1-Hz steps); Hilbert phase extraction; phase-locking value (PLV) between EEG and stimulus envelopes computed per frequency and channel. Pink Noise PLV not computed (stimulus generated in real time).
• Behavioral metric: RT coefficient of variation per 10-trial block.

Experiment 4 (Online behavioral; parametric modulation):

• Participants: Recruited via MTurk. N=221 recruited; N=175 after screening (mean age 36). Two sub-experiments: 4A (rate), 4B (depth). Rate: N=82; Depth: N=93 (final datasets). Participants provided Adult ADHD Self-Report Scale (ASRS) scores.
• Stimuli: Two base musical tracks; generated variants with added amplitude modulation at rates 8, 16, 32 Hz and depths low, medium, high. Modulation applied to 200 Hz–1 kHz band; aligned to metrical grid at 120 bpm (2 Hz), so 8/16/32 Hz align with 16th/32nd/64th notes. Conditions validated via modulation spectrum differences; spectral (EQ) content held constant. Peak-normalized to 0.5 (rms 0.080–0.092).
• Procedure: SART total 1080 trials across 4 blocks (~270 trials/block, ~5 min each). Each participant experienced all four conditions within either the rate or depth dimension, including a no-modulation control. Order counterbalanced.
• Analysis: Mixed-effects ANOVA on d′ difference relative to no-modulation, with within-subject factor Modulation (Rate: 8/16/32 Hz; Depth: low/medium/high), within-subject factor Block (early vs late), and ASRS score as continuous covariate. Ratings (valence/arousal) analyzed with within-subjects MANOVA across all six stimuli (No Mod; 16 Hz Low/Med/High; 8 Hz Med; 32 Hz Med).

Experiment 1:

• Ratings (N=62): Significant differences in valence and arousal across AM+ Music, Control Music, Pink Noise: MANOVA F(4,364)=56.63, p<0.001, Wilks’ λ=0.38, partial η²≈0.38. AM+ Music rated positive valence, high arousal; Control Music positive valence, low arousal; Pink Noise low valence, high arousal.
• SART performance (N=83): Significant Music × Trial Block interaction, F(38,38)=2.024, p=0.016, partial η²=0.669; significant Music × Presentation Order interaction, F(10,150)=2.298, p=0.015, ηp²=0.133. Performance benefit was driven by primacy of AM+ Music: participants who heard AM+ first showed higher d′, especially in the first block.

Experiment 2 (fMRI; N=34):

• AM+ Music elicited higher overall BOLD activity than Control Music and Pink Noise (p<0.05 FDR-corrected) across bilateral superior temporal, frontal, parietal, and occipital cortices, encompassing default mode, executive control, and salience networks. Additional clusters included middle frontal gyri, frontal operculum, medial prefrontal cortex, bilateral temporal lobes, and lateral occipital cortex.
• Hits>False Alarms contrasts showed more extensive correct-response-related activity during AM+ Music than other conditions in sensorimotor (SMA, precentral gyrus), salience (anterior cingulate, anterior insula), and visual association cortices, despite similar behavioral hit/FA rates (one-way ANOVAs Fs<1).

Experiment 3 (EEG; N=40):

• Stimulus-brain PLV during AM+ Music showed peaks at 8, 12, 14, 16, 24, and 32 Hz. PLV at 8 Hz strongest frontally; Control Music showed lower, less sharply tuned PLV.
• Effect sizes comparing AM+ vs Control Music: Cohen’s d=3.74 for 4-Hz bins (capturing 8,12,16,20,24,28,32 Hz); d=0.227 for 1-Hz bins, indicating selective coupling at music-related frequencies.
• Over time, AM+ Music showed increasing PLV at 16 Hz (beta) and decreasing PLV at 8 Hz (theta/alpha), with significant increases at left frontal channels (p<0.05 FDR-corrected late vs early). No comparable increase for Control Music.

Experiment 4 (Parametric rate/depth; N=175):

• Ratings across six stimuli showed main effects of Music on valence (F(5,305)=32.7, p<0.001) and arousal (F(5,305)=7.1, p<0.001).
• Modulation rate: Significant three-way interaction between modulation rate (quadratic; peak at 16 Hz), block (higher early), and ASRS score: F(1,80)=4.03, p=0.048, ηp²=0.048. Participants with higher ASRS scores improved more over time with 16 Hz modulation than in other rate conditions.
• Modulation depth: No significant main effects or interactions on performance.

Overall: Rapid (beta-range) amplitude modulations in music preferentially engaged attention-related neural networks, enhanced stimulus-brain coupling at targeted frequencies, and improved sustained attention behaviorally, particularly in listeners with more ADHD-like symptoms, with effects emerging early and strengthening over time at 16 Hz.

The findings support the hypothesis that specific amplitude modulation properties in music, especially beta-range (≈16 Hz), can enhance sustained attention and that benefits vary with individual attentional difficulty. Behaviorally, AM+ Music improved SART performance when presented early, and parametrically added 16 Hz modulation selectively benefited participants with higher ASRS scores over time. Neurally, AM+ Music increased activity across the salience, executive control, default mode, sensorimotor, and visual networks, and strengthened stimulus-brain phase locking at note-rate harmonics and amplitude-modulated frequencies, with a time-dependent increase at 16 Hz. These results suggest an oscillation-based mechanism whereby external acoustic modulations entrain and amplify cortical rhythms associated with attentional control and motor readiness, aligning with optimal stimulation theory for ADHD. Importantly, Experiment 4 indicates that acoustic modulation parameters, rather than arousal or valence per se, primarily drive performance changes, as the condition with lower arousal (16 Hz medium depth) yielded the best outcomes for high-ASRS listeners. Together, the data point to a targeted, frequency-specific pathway by which music can modulate attention-related brain dynamics to support sustained attention, especially in individuals with attentional challenges.

This work demonstrates that adding rapid, specifically beta-range (≈16 Hz), amplitude modulations to music enhances sustained attention and entrains neural activity in attention-related networks. The benefits are most pronounced for individuals with greater ADHD-like symptoms, indicating that personalized, oscillation-targeted music may support cognitive performance. Contributions include: (1) identifying modulation-domain features as key determinants of music’s impact on attention; (2) linking behavioral improvements to increased network activity (salience/executive/sensorimotor/visual) and enhanced stimulus-brain coupling; and (3) showing rate-specific benefits with parametric control of modulation in otherwise identical music. Future research should optimize modulation parameters for individual profiles (e.g., ADHD phenotypes), test adaptive or closed-loop approaches that track brain state over time, examine functional connectivity and cross-frequency coupling mechanisms, and evaluate generalization to real-world tasks and clinical outcomes.

• Initial stimuli (Experiments 1–3) differed in multiple musical and low-level acoustic properties beyond modulation (e.g., tonality, instrumentation, spectral balance), motivating the controlled parametric design in Experiment 4.
• In Experiment 3, stimulus-brain coupling (PLV) could not be computed for Pink Noise because it was generated in real time and not stored.
• Online experiments allowed participant-controlled volume; although measured and used as a covariate, variability in playback settings may introduce noise.
• Order effects were observed in Experiment 1, with benefits strongest when AM+ Music appeared first.
• In Experiment 4, to maximize within-subject comparisons the duration per condition was limited (~5 minutes per block), potentially constraining sensitivity to longer-term effects.
• EEG conditions included both within- and between-subject manipulations (Silence vs Control/Pink), which may reduce direct comparability across all audio conditions.