AI-Generated Emotional Background Music for Learning-Related Well-being: Task-Dependent Effects on Cognitive Performance, Workload, and Physiological State

INTRODUCTION: Background music is frequently incorporated into learning environments not only to enhance cognitive engagement but also to regulate learners’ affective state and perceived mental effort, both of which are closely associated with learning-related well-being. However, empirical findings regarding its impact on memory performance remain inconclusive and appear to be contingent upon task characteristics as well as the emotional properties of the auditory stimulus. Recent advances in Artificial Intelligence Generated Content (AIGC) enable the parametric synthesis of music with systematically controllable emotional attributes (e.g., valence, arousal, motivational tone), thereby providing a novel methodological pathway for examining how adaptive auditory environments may support learners’ cognitive functioning, stress regulation, and psychological well-being beyond the constraints of pre-composed music. OBJECTIVES: This study investigates how emotionally differentiated AI-generated background music influences memory performance, subjective workload, and user preference within learning contexts, with particular attention to task-dependent effects across numerical and image-based memory paradigms.  METHODS: Seventeen participants completed both numerical recall and visual memory tasks under four auditory conditions: positive/motivational, soothing, focus-oriented (Low-arousal) AI-generated music, and a silence control. Behavioural accuracy was recorded alongside indicators of electroencephalography (EEG) and heart rate variability (HRV). Subjective workload and affect were assessed using the NASA Task Load Index (NASA-TLX) and the Positive and Negative Affect Schedule (PANAS). RESULTS: Results revealed dissociable, task-dependent patterns across behavioural and psychophysiological measures. Silence yielded the highest accuracy in numerical memory tasks, whereas positive-valence music was associated with enhanced performance in image memory conditions. Distinct categories of AI-generated music also elicited differential neural and autonomic responses, suggesting variations in affective regulation and cognitive load during task execution, with focus-oriented music emerging as the most preferred auditory condition for sustained learning. Although overall behavioural accuracy remained relatively stable across conditions, emotionally parameterised AI-generated music may function as autonomy-supportive cognitive scaffolds CONCLUSION: These findings provide preliminary evidence that AIGC-driven auditory environments may function as autonomy-supportive cognitive scaffolds that dynamically regulate stress and attentional demands during task engagement. By adaptively aligning emotional soundscapes with task characteristics and user state, AI-generated background music holds promise as a well-being-aware intervention capable of promoting cognitive sustainability and reducing perceived mental strain in digitally mediated learning systems.