Alleviating academic stress and enhancing interdisciplinary self-efficacy in university libraries: an exploratory mixed-methods study on restorative built-environment interventions.

BACKGROUND: In the digital intelligence era, higher education students-especially those in STEM (Science, Technology, Engineering, and Mathematics) fields-face widespread academic stress, cognitive rigidity, and negative emotional rumination. Traditional psychological interventions are often confined to clinical counseling and questionnaire screenings, which can easily induce a sense of stigma and pathological labeling. Exploring non-invasive, routine physical built-environment interventions is therefore of significant academic value and practical urgency. As a campus "third space," the university library possesses unique physical and cultural advantages for building a non-evaluative "psychological safety sandbox." METHODS: This study adopted an exploratory sequential mixed-methods design, utilizing the Library of Northeastern University at Qinhuangdao (a collaborative innovation node of the Future Learning Center Construction across five provinces) as the empirical field. We conducted a longitudinal tracking study from Fall 2025 to Spring 2026, analyzing the behavior database of 871 registered students (1,493 activity records) and focusing on a cohort of N = 122 subjects with complete behavioral and psychological tracking data. Quantitative data were obtained via the self-developed "Library Restorative Experience and Capability Questionnaire" (LRECQ). Qualitative text analysis was conducted on students' reflective diaries using thematic analysis. RESULTS: An independent-samples t-test indicated that multimodal interventions combining "static support" (quiet, biophilic study pods) and "dynamic interaction" (embodied handicrafts and physical exercise check-ins) significantly increased students' continuous focused reading time, showing a dramatic expansion compared to the pre-test control group (228.49 ± 75.58 min vs. 86.80 ± 95.56 min, t(116) = 8.98, p < 0.001). Reliability testing demonstrated that the LRECQ scale has excellent reliability, extracting two high-order "Liquid Capability" (LCI) factors: "Cognitive Liquefaction and Interdisciplinary Reorganization" (LCI-Cognitive, /Cronbach's α = 0.7240) and "Social Collaboration and Organizational Efficacy" (LCI-Social, Cronbach's α = 0.8500). Thematic analysis revealed that weakly correlated sensory-motor interactions occupied working memory resources, triggering "cognitive decoupling" and interrupting DMN-based stress loops. Furthermore, applying abstract academic schemas to culturally warm objects ("schema adaptation") dissolved the temporal distortion of the "forever fallacy," sparking a surge in adaptive self-efficacy. CONCLUSION: This study confirms that university libraries can function as multimodal built-environment restorative platforms through two empirically validated mechanisms-sensory-motor cognitive decoupling (weakly correlated manual/physical tasks preempt working memory, severing the Default Mode Network from academic rumination loops) and /Cronbach's (abstract professional knowledge liquefies into creative agency when applied in non-evaluative, culturally warm contexts)-together forming a dual-phase restoration-activation cycle. Three generalizable regularities are identified: (1) the sensory-motor interrupt as a universal, non-stigmatizing pattern-break principle; (2) non-evaluative schema liquefaction as a cross-domain capability-building regularity; and (3) a ~23-25% voluntary-sample selection bias consistently inflating naive effect estimates, correctable by DML. These findings provide a theoretically grounded, data-evidenced paradigm for non-invasive campus mental health space governance.