Emotion-Matched Music Increases Cardiovascular Engagement Without Elevating Perceived Exertion During Self-Paced Cycling

Technical Summary: Emotion-Matched Music Increases Cardiovascular Engagement Without Elevating Perceived Exertion During Self-Paced Cycling

Problem Statement
Exercise adherence remains a significant public health challenge, with dropout rates often reaching 50% within the first six months. A primary barrier to adherence is the subjective experience of effort and fatigue, which often limits performance and motivation even when individuals possess the physiological capacity for higher intensity. While listening to music is a recognized intervention to modulate exercise experience, existing research has predominantly focused on self-selected or preferred music. This reliance on self-selection confounds the effects of emotional alignment with familiarity and explicit preference. Furthermore, current emotion-aware music recommendation systems are largely evaluated on prediction accuracy or user satisfaction, with few studies assessing whether emotion-matched music meaningfully alters physiological engagement or perceived exertion during real-world physical activities. The study addresses the gap in understanding whether music matched to an individual's emotional profile—derived from their habitual exercise listening but not directly self-selected—can decouple physiological load from perceived effort.

Methodology
The study employed a between-subjects controlled design involving 149 adults (aged 18–45) who completed a 30-minute self-paced cycling session. Participants were allocated to one of two conditions:

1. Control Group ($n=76$): Listened to generic workout music (a non-personalized pop/dance radio stream).
2. Emotion-Matched Group ($n=73$): Listened to music derived from the Phiona® system. Participants provided a sample playlist of music they typically use for successful exercise. The system analyzed the emotional characteristics of these tracks and generated a 30-minute playlist of new tracks that matched the derived emotional profile, without the participants explicitly selecting these specific tracks.

Data Collection and Measures:

• Physiological: Heart rate (HR) and Heart Rate Variability (HRV, operationalized as the standard deviation of HR) were recorded continuously using a wrist-worn EmotiBit photoplethysmography sensor.
• Performance: Objective metrics including distance, average speed, resistance, cadence, and active energy expenditure were captured via an Apple Watch integrated with a Peloton stationary bike.
• Subjective: Ratings of Perceived Exertion (RPE) using the Borg 6–20 scale were collected pre- and post-exercise. Affective responses were measured using selected items from the Brunel Mood Scale (e.g., sleepiness, liveliness, energetic).

Key Results

• Physiological Engagement: Participants in the emotion-matched condition exhibited significantly higher mean heart rates ($M=106.00$ bpm) compared to the control group ($M=98.60$ bpm), $t(116.30)=2.37, p=.020$, indicating greater sustained cardiovascular engagement. HRV was also higher in the emotion-matched group, though this effect was sensitive to analytic specifications. Maximum HR did not differ between groups.
• Perceived Exertion (RPE): There was no significant difference in post-exercise RPE between conditions after controlling for baseline. However, a moderation analysis revealed a significant interaction between condition and mean HR. In the control group, higher mean HR predicted higher RPE. In the emotion-matched group, mean HR did not predict RPE. This indicates that the typical positive coupling between physiological load and perceived effort was attenuated in the emotion-matched condition.
• Performance Output: No statistically significant differences were found in distance, speed, cadence, or active energy expenditure between groups. Resistance was directionally higher in the emotion-matched group but the effect was not robust across outlier treatments. A composite total-work score showed no significant difference.
• Affective Responses: The emotion-matched group reported significantly lower post-exercise sleepiness and higher levels of liveliness and energetic affect compared to the control group, controlling for baseline levels.

Key Contributions

1. Decoupling Load and Effort: The study demonstrates that emotion-matched music can increase objective physiological engagement (mean HR) without a corresponding increase in perceived exertion. This suggests a partial dissociation between interoceptive signals and effort appraisal.
2. Beyond Self-Selection: By using a system-derived emotional profile rather than direct self-selection, the study isolates the effect of emotional alignment from familiarity and explicit preference, showing that emotional resonance alone can influence exercise regulation.
3. Validation of Affective Systems: The research provides empirical evidence linking emotion-aware music systems to objective physiological and behavioral outcomes, moving beyond standard evaluation metrics of prediction accuracy or subjective satisfaction.
4. Mechanism of Action: The findings support the hypothesis that emotional congruence may alter the relationship between internal physiological demand and subjective effort, potentially by reducing the salience of fatigue-related cues or shifting affective-perceptual evaluation, rather than by increasing mechanical output.

Significance and Claims
The paper claims that emotion-matched music functions as an affective-motivational intervention that enables individuals to sustain higher physiological engagement without increasing the subjective cost of effort. The authors posit that this pattern is consistent with psychobiological models of exercise where behavior is regulated around a subjective effort ceiling. By modifying how effort is experienced rather than directly increasing mechanical output, such systems may support sustained engagement and improve the experiential quality of exercise. The authors conclude that these findings illustrate a method for evaluating emotion-based music selection using behavioral and physiological outcomes in ecologically valid contexts, suggesting that future emotion-aware systems should be assessed on their capacity to influence behavioral regulation and physiological responses in real-world settings. The study remains modest, noting that the lack of robust changes in overt mechanical output suggests a shift in tolerance for internally demanding work rather than enhanced productivity.