The Effects of Air-Conditioning and Road-Traffic Noise on Perceived, Cognitive, and EEG Responses in a University Classroom

1. Introduction

• Air-conditioning and road traffic are common but acoustically distinct classroom sounds, and their effects can vary with sound characteristics, level, task, and outcome.Source: §1 · pp. 1–2

1.1 Cognitive performance as an indicator of acoustic effects

• Stable accuracy or response time can coexist with changed perception, workload, or neural activity, so performance alone cannot establish that conditions were experienced alike.Effort, attentional shielding, and task difficulty can help keep performance stable despite sound exposure.Source: §1 · p. 2, §1.1 · p. 3

1.3 Study aims and research questions

• The study asks how responses vary across noise sources and levels, whether domains covary within participants, and whether perceived and EEG responses distinguish conditions performance does not.It compares air-conditioning and road-traffic noise at three levels with a no-added-noise control, measuring all three domains in the same participants.Source: §1 · p. 2, §1.3 · pp. 4–5

1.2 Subjective and EEG responses and their consistency with performance

• Subjective evaluations and task-period EEG provide complementary perspectives, but prior work leaves within-participant consistency and EEG discrimination of performance-indistinguishable conditions unresolved.Multidomain measurement has precedents, but collecting measures together does not establish whether they agree.Source: §1.2 · pp. 3–4

2.1. Overview

• Each participant completed the same protocol under different noise types and levels, enabling within-person comparisons while reducing between-subject variability.Source: §2.1 · pp. 5–6
• Questionnaires, predefined cognitive tasks, and task-period EEG jointly measured perceived experience, task performance, and brain activity, with cross-domain associations also examined.Source: §2.1 · p. 6

2.2.1 Acoustic stimuli

• Air-conditioning and road-traffic noise were each presented at 55, 60, and 65 dBA, alongside a no-added-noise control.Source: §2.2.1 · p. 6

2.2 Experimental measurement

• Acoustic appraisal used loudness, pleasantness, and arousal ratings, while NASA-TLX assessed six perceived-workload dimensions.Source: §2.2.1 · p. 6

2.2.3 Cognitive performance

• Four tasks progressed from focused attention to divided attention, decision making, and information extraction, yielding accuracy, response time, and a performance index.Source: Figure 1 · p. 6, §2.2.3 · p. 7

2.2.4 Neural responses

• EEG was recorded continuously during tasks, segmented by synchronized markers, and analyzed as five frequency bands aggregated across four functional scalp regions.Processing included preprocessing and artifact reduction before task-specific spectral estimation.Source: §2.2.4 · pp. 7–8, Figure 2 · p. 7, Figure 3 · p. 8

2.3.1 Participant selection

• Of 22 participants completing the protocol, 16 university students remained after EEG quality screening for the multimodal analyses.Source: §2.3.1 · p. 9, Figure 5 · p. 9

2.3.2 Experiment setting

• In a Hong Kong university classroom, six recorded-noise conditions used a fixed speaker, with windows closed and a no-added-recording control.The classroom's existing air-conditioning unit did not generate the experimental air-conditioning stimulus.Source: §2.3.2 · p. 10, Figure 6 · p. 10

2.4. Data analysis

• The analysis proceeded from deriving indicators, to testing acoustic-condition effects within each response domain, to examining cross-domain associations.The authors reported exact p values, effect sizes, and 95% confidence intervals.Source: §2.4 · p. 11

2.4.1 Stage 1: Data processing

• Perceived outcomes retained three acoustic-appraisal ratings and overall Raw and Weighted TLX scores, with six workload dimensions as supporting measures.Cognitive outcomes included task-specific accuracy, mean response time, and a performance index integrating accuracy and speed; EEG outcomes were control-normalized band power by task and ROI.Source: §2.4.1 · p. 11

2.4.2 Stage 2: Statistical analysis

• Perceived and cognitive outcomes used a two-stage within-subject analysis, while EEG band-power changes were modeled by task and ROI with participant random intercepts.EEG models included noise type, noise level, frequency band, and their interactions; significant interactions were decomposed with estimated marginal means and simple-effect comparisons.Source: §2.4.2 · p. 12

2.4.3 Stage 3: Cross-domain association

• Repeated-measures correlations tested within-participant associations between selected perceived measures and relative EEG power, and between task-specific PI and corresponding EEG power.Ratings were paired with EEG from the same condition, and each task’s PI with EEG recorded during that task; correlations were calculated separately by ROI.Source: §2.4.3 · pp. 12–13

3.1.1 Acoustic appraisal

• All six noise conditions were rated louder, less pleasant, and more arousing than control; all three ratings varied with level, while noise type affected loudness only.RT was rated slightly quieter than AC at 55 dBA but louder at 60 and 65 dBA; pleasantness fell as level rose, and arousal rose with level, especially under RT.Source: §3.1.1 · p. 13, Figure 7 · p. 13, Table 4 · p. 14

3.1.2 Perceived workload

• Raw and weighted TLX scores were significantly higher under noise than control, by 9.59 and 10.17 points, respectively.Temporal demand and effort were also significantly higher; overall TLX did not vary significantly with noise type, level, or their interaction.Source: §3.1.2 · p. 14, Figure 8 · p. 14, Table 5 · p. 15
• Among workload dimensions, mental demand was higher under RT than AC noise, and physical demand showed a noise-type-by-level interaction.Other individual dimensions did not differ significantly between noise and control; variability between participants was considerable.Source: §3.1.2 · p. 14, Figure 9 · p. 15

• No significant differences in PI, accuracy, or mean response time were found across the tested acoustic conditions in any task.PI did not differ between noise and control for Tasks 1–4; none of the tests among noise conditions was significant, and the distributions largely overlapped.Source: §3.2 · pp. 15–16, Figure 10 · p. 16

3.3.1 Spectral characteristics

• Across all four tasks, the 32-channel spectra showed the typical decline in power with frequency and largely overlapped across noise conditions.The authors note that visual differences alone do not establish statistical significance.Source: §3.3.1 · p. 16, Figure 11 · p. 17

3.3.2 Effects of noise type and level

• Significant noise-related EEG effects varied by task and ROI, and no interaction involving frequency band was significant.Omnibus tests also found significant main effects of frequency band, noise type, and noise level, as well as type-by-level interactions; the plotted colors encode −log10(p), not effect size or direction.Source: §3.3.2 · p. 17, Figure 12 · p. 19
• Task 1 showed type-by-level interactions in all four ROIs, with AC-versus-RT differences reversing across sound levels in some regions.In the posterior-attention ROI, the normalized change was larger under AC at 55 and 65 dBA, but smaller under AC at 60 dBA.Source: §3.3.2 · pp. 17–18
• In Task 2, noise level affected central-motor and temporal ROIs, with distinct contrasts across 55, 60, and 65 dBA.The central-motor change was larger at 65 than 60 dBA; the temporal change at 60 dBA was smaller than at both 55 and 65 dBA.Source: §3.3.2 · p. 18
• In Task 3, noise level affected posterior-attention and central-motor ROIs, while temporal-ROI change was larger under RT than AC noise.The posterior-attention level effect was confined to RT noise, where change was larger at 65 than at 55 or 60 dBA; central-motor change was larger at 55 than 60 dBA.Source: §3.3.2 · p. 18
• In Task 4, RT noise produced a larger midfrontal change than AC, and temporal-ROI change at 65 dBA exceeded that at 55 and 60 dBA.Source: §3.3.2 · p. 18

3.4.1 Associations between perceived responses and EEG

• Higher loudness and lower pleasantness covaried with relative EEG band power, while workload had no significant associations.Loudness was associated with higher theta and alpha and lower gamma; pleasantness with lower theta and higher gamma. Fig. 13 displays the correlations.Source: §3.4.1 · p. 19, Figure 13 · p. 20

3.4.2 Associations between cognitive performance and EEG

• Only 2 of 80 performance–EEG correlations were significant, and they occurred in different tasks and bands.Task-specific coefficient patterns were mostly nonsignificant and are reported descriptively; Fig. 14 displays the correlations.Source: §3.4.2 · p. 20, Figure 14 · p. 21
• No consistent within-participant association between cognitive performance and EEG power was found, and the association strengths were not directly compared with those for perceived responses.The contrast does not establish that EEG is more closely coupled to perception than to performance.Source: §3.4.2 · pp. 20–21

4.1 Effects of noise source and level within each domain (RQ1)

• Performance did not differ significantly across sources or levels, while appraisal varied with level, workload was higher under noise, and EEG differed in selected task and region combinations.Appraisal differences between sources were limited to loudness; EEG power comparisons were noise conditions against one another, not against control.Source: §4.1 · p. 22

4.2 Consistency across response domains (RQ2)

• EEG covaried mainly with loudness and pleasantness, rarely with arousal or performance, and not with workload; sparse associations do not establish that these responses are unrelated to neural activity.The correlations were not adjusted for source or level and may partly reflect a shared response to sound level.Source: §4.2 · pp. 22–23

4.5 Limitations and future research

• Interpretation is limited by the EEG measures: relative-band shifts may reflect aperiodic as well as oscillatory activity, and scalp gamma is susceptible to muscle artifacts.ROI labels indicate scalp locations, not neural sources; future work could separate periodic from aperiodic components, verify gamma artifact handling, and test arousal with peripheral indicators.Source: §4.2 · p. 23, §4.5 · pp. 24–25

4.3 Differences beyond cognitive performance (RQ3)

• Appraisal, workload, and EEG distinguished some conditions that performance did not, but the measures did not agree on which conditions differed.The proposed effort, attention-shielding, and arousal accounts are compatible with the findings, but the design did not test them directly; EEG covaried with rated arousal in only 2 of 20 tests.Source: §4.3 · pp. 23–24

4.4 Implications for assessing environmental noise in classrooms

• Classroom noise assessments should pair performance tests with perceived responses; EEG adds information about task processing but is better treated as a research tool than an acoustic-quality criterion.The inference that sound level may matter more than source for perceived quality is a hypothesis: playback did not change ventilation, air quality, or thermal conditions.Source: §4.4 · p. 24

5. Conclusions

• Cognitive performance did not differ significantly between acoustic conditions in any of the four tasks.The within-participant study tested no-noise control and air-conditioning and road-traffic noise at 55, 60, and 65 dBA.Source: §5 · p. 25
• Responses were only partly consistent: EEG covaried mainly with loudness and pleasantness, rarely with arousal or performance, and not with workload.Noise was rated louder, less pleasant, and more arousing than control; workload was also higher under noise.Source: §5 · p. 25
• Perceived responses should accompany performance assessments, and nonsignificant performance differences should not be treated as evidence that acoustic conditions are equivalent.The findings should be confirmed in larger samples, with future work testing performance equivalence and separating within-condition from between-condition covariance.Source: §5 · p. 25