Abstract
Ensuring high Indoor Environmental Quality (IEQ) in educational settings is crucial for Students performance. Within IEQ, the combined effects of different domains are highly relevant, yet the interaction between acoustics and Indoor Air Quality (IAQ) requires further exploration. While laboratory studies allow precise environmental control, their representativeness of a real classroom is often unclear due to the inability to recreate all real-world stimuli. Therefore, this study examines the combined effects of acoustics and IAQ on calculation tasks, comparing results when experiments are performed in a laboratory versus real classrooms with the same participants. Nearly 200 students participated in the laboratory experiment, which included five auditory conditions: quiet, playground noise, birdsongs, and mechanical ventilation at 160 L/s and 200 L/s. A split plot design was employed where half of the students were exposed to good IAQ while the remaining experienced poorer IAQ. Results revealed main effects only for listening conditions, heavily mediated by individual noise sensitivity. These findings partially contradict previous classroom results. A comparison between settings indicates pupils were generally faster in the laboratory. Interestingly, the most favourable condition was mechanical ventilation in the lab, but playground noise in the classroom. These results illustrate the lab-dilemma regarding discrepancies between controlled and real-world conditions. While precise causes were not definitively identified, they are hypothesised to stem from individual characteristics, highlighting a strong need for better characterisation of student-related aspects in future studies.
1 Introduction
According to the literature, people spend, on average, approximately 90% of their daily time inside buildings (; ). As a result, it is important to ensure that the quality of such environments, hereafter referred to as Indoor Environmental Quality (IEQ) is adequate to effectively support the activities being carried out and the comfort of the occupants. Ensuring a good IEQ becomes even more critical in educational settings, since students typically spend inside between 27% and 45% of their day (), and suboptimal IEQ conditions have been associated with hindered cognition (; ) and comfort ().
IEQ is defined and assessed across four domains (): thermal quality, acoustic quality, indoor air quality (IAQ), and visual quality. Each of these domains independently could influence occupants’ wellbeing, comfort, and cognitive abilities (; ; ). Only in recent years, however, attention shifted towards understanding the interactions among these domains rather than focusing only on a single domain. The new aim is to identify potential combined and cross-modal effects (; ; ; ; ; Yang and Moon, 2018; 2019; ; ; ; ; ; ) between the domains. For definitions and a theoretical framework of combined and cross-modal effects, please refer to .
With regard to educational environments, a significant gap remains in the literature. While studies addressing cognitive performance are available, very few have examined the combined effects of IAQ and acoustic conditions, particularly with respect to school-aged students ().
Additionally, studies on combined effects have generally been conducted in controlled environments such as laboratories, where environmental conditions can be easily manipulated. However, this may result in reduced ecological validity compared to real contexts (; ). This discrepancy between real and simulated settings is known in the literature as the lab-dilemma (; ). Existing research has primarily focused on increasing the realism of laboratory settings, particularly through the use of virtual reality head-mounted displays (). Nevertheless, even minor variations, such as working in small groups or individually, have been shown to lead to differences in cognitive abilities (e.g., working memory performance ()).
To the best of the Authors’ knowledge, this is the first study to directly compare the effects of acoustic stressors on cognitive performance in laboratory and classroom settings, using the same experimental task and acoustic conditions. By isolating the role of the testing environment, the study provides novel evidence on whether, and under which conditions, findings obtained under controlled conditions can be generalized to real classroom context. Such evidence would be valuable for assessing the extent to which laboratory-based findings are representative of, and informative about, everyday classroom conditions.
Given these research gaps, this study would like to address the following two research questions.
Do IAQ and acoustics have a combined effect on the calculation performance of Grade VII and VIII students in a classroom-like laboratory environment, and if so, what is the nature of this effect?
Does conducting a laboratory experiment change the calculation performances of the students compared to a classroom?
This work is structured into two parts, each addressing one of the research questions. Part 1 presents the methodology and results of the laboratory experiment, along with a discussion of the findings. Part 2 compares the results obtained in Part 1 with those from a classroom experiment (reported and discussed in () conducted with the same panel of students. An explanation on the effect of sounds and IAQ on cognitive abilities as well as a summary of the results of the classroom experiment will be provided in the next section.
2 Background
2.1 Effects of sound stimuli on cognition
The literature has proposed several mechanisms to explain the influence of auditory stimuli on students’ cognitive performance. Broadly, Stimuli typically classified as noise (e.g., chatter, traffic noise), may impair cognitive performance by triggering distraction mechanisms, capturing attention and reducing resources for the task at hand (; ). Conversely, positively perceived stimuli, such as natural sounds, may support performance due to their capacity to restore students’ cognitive resources (; ). However, distraction and restoration alone do not provide a unified explanation for why the same auditory stimulus may produce beneficial, detrimental, or negligible effects across individuals. For this reason, in several literature studies (; ; ) the arousal theory was adopted as a broader framework. Arousal is defined by the American Psychological Association Dictionary of Psychology as “a state of excitement or energy expenditure linked to an emotion. Usually, arousal is closely related to a person’s appraisal of the significance of an event or to the physical intensity of a stimulus. Arousal can either facilitate or debilitate performance.” (https://www.apa.org/science/about/psa/2012/02/emotional-arousal). According to the Yerkes-Dodson Law (Yerkes and Dodson, 1908; ), cognitive performance varies as a function of arousal following an inverted U-shaped curve: performance peaks at a moderate level of arousal, while both lower and higher levels of arousal are associated with poorer performance (see Figure 1). Auditory stimuli can modulate individual arousal levels by increasing the baseline value, with the magnitude of this increase depending on both the type of stimulus (; Yang et al., 2018) and its intensity (sound level) (). This framework was therefore selected because it can accommodate both the beneficial, the detrimental effects of sound stimuli reported in previous studies, while also accounting for individual differences in responsiveness to auditory stimuli.
FIGURE 1
2.2 Effects of IAQ on cognition
With regard to the effects of IAQ on cognitive performance, similar considerations apply to those discussed in relation to the Yerkes-Dodson law. The literature indicates that a deterioration in air quality, due to reduced ventilation or increased concentrations of indoor pollutants such as bioeffluents (Wargocki and Wyon, 2007; Zhang et al., 2017a; ; ), can lead to an increase of the arousal levels, which in turn may result in a decline in cognitive performance. It should be noted that indoor CO2 concentration can be used as a proxy IAQ since it provides indications of a lack of ventilation and the increase of the bioeffluents produced by people, however for complex tasks even pure CO2 concentration could be treated as a pollutant affecting performances (). In such circumstances, a CO2 concentration of approximately 1,000 ppm is commonly indicated by standards and guidelines (; ) as the threshold separating good or improved IAQ from poor or reduced IAQ, particularly in the presence of sensitive individuals such as children ().
2.3 Combined effects of acoustics and IAQ
Few studies have investigated the combined effects of IAQ and acoustics on cognitive performance, and to the Authors’ knowledge none have focused on school-aged children. Two studies involving adult students have been identified (; ). examined the influence of carpet presence and two traffic noise levels during arithmetic tasks while studied the combined exposure to bioeffluents and listening conditions on calculation task. No significant combined effects were found in , whereas did. In the latter, a mechanism grounded in arousal was proposed. Specifically, it has been hypothesised that the potential increase in arousal induced by poor IAQ may add to the arousal elicited by auditory stimuli. In addition, a deterioration in air quality might also heighten the perceived difficulty of the task, thereby altering the shape of the Yerkes–Dodson curve. For a detailed representation of the two curves, the reader is referred to and .
2.4 Outline of findings in the classroom experiment
To provide a contextual baseline for comparing the laboratory and field settings, the core methodology and key findings of the classroom study are briefly summarised below; for full methodological details and extended discussions, readers are referred to ().
The field experiment investigated the effects of ventilation-related acoustic conditions on students’ mathematical performance and subjective perception within their standard learning environments. Class size ranged from 14 to 26 students, with the majority of classrooms accommodating approximately 20 pupils. The physical spatial arrangement was split across testing sites: half of the classrooms featured traditional row layouts with desks facing the blackboard, whereas the remaining half utilised a collaborative group-desk configuration. All classrooms were naturally ventilated, with windows oriented towards inner courtyards or quiet perimeter zones isolated from vehicular traffic. To standardise environmental conditions and isolate the spaces from external acoustic disturbances (which were mainly restricted to transient corridor movement), windows and doors remained closed throughout the experimental sessions. Testing was conducted during regular lecture hours under the continuous supervision of the classroom teacher, with students wearing headphones for the duration of the task.
Although formal architectural dimensions and acoustic measurements (such as reverberation time) were not recorded on-site, experimental rigor was maintained by implementing the exact same mathematical task, counterbalancing protocol, and statistical analysis pipeline as detailed in the present laboratory study (with the exception of the IAQ component, which was absent in the field setting).
Regarding performance outcomes, the listening conditions exerted a statistically significant effect on response time, whereas accuracy remained unaffected. Specifically, pupils with low noise sensitivity demonstrated faster response times under anthropogenic acoustic conditions (children’s playground noise) compared to natural sounds (birdsongs) or mechanical ventilation noise. These findings suggest that the heightened arousal induced by anthropogenic noise may have shifted pupil performance closer to the optimal threshold predicted by the Yerkes–Dodson law.
3 Part 1: laboratory experiment
3.1 Methods
3.1.1 Participants and facilities
A total of 189 students participated in the experiment (M = 12.8 years, SD = 1.4 years) with the 48.2% being female. The students belonged to the same 12 classes (6 grade VII and 6 grade VIII) previously involved in a study by . Written informed consent in accordance with Helsinki Declaration was obtained from the legal guardians of the students. The entire experiment was internally reviewed by Eurac Research’s research offices and complies with the Eurac Research Core Ethics and Integrity Principles. Participants were asked to carry out the experiment in a laboratory setting at the Facade System Interaction Lab (FSIL) at Eurac Research in Bolzano, Italy (; ). The laboratory comprises two adjacent rooms (Room A and Room B), each with a volume of 96 m3 (8 × 4 × 3 m), featuring a removable façade and equipped with independent HVAC systems. For the purpose of the study, the rooms were furnished to resemble a classroom, and a highly glazed façades were selected (Figure 2).
FIGURE 2
3.1.2 Experimental stimuli
Students were divided into two groups, each exposed to a different IAQ condition. Room A was consistently assigned to the good IAQ condition, with a CO2 concentration fixed at 800 ppm, while Room B was consistently assigned to the poor IAQ condition, with a CO2 concentration fixed at 2000 ppm. Accordingly, each group of students was exposed to the IAQ condition associated with the assigned room throughout the experimental sessions. The two target CO2 concentrations were achieved using gas cylinders and were maintained by balancing the CO2 exhaled by the participants with the amount extracted by the room’s HVAC system. To avoid alterations in the background acoustic conditions, the ventilation system was set to the same airflow rate of 100 L/s for both conditions. The balancing procedure was performed solely by adjusting the percentage of fresh air supplied to the room. All participants, regardless of the IAQ condition to which they were exposed, completed the experiment while listening to five distinct sound stimuli hereafter referred as listening conditions. These conditions were selected to be representative of both naturally ventilated and mechanically ventilated classroom environments. In the present study, students were exposed to the ambient noise of the laboratory, namely, quiet, playground activity in the schoolyard, birdsongs, and finally two mechanical ventilation conditions referred to as MV_160 and MV_200. The difference between the two mechanical ventilation conditions lies in the airflow rate at which the HVAC system was set, namely, 160 L/s for MV_160 and 200 L/s for MV_200. These settings are representative of an optimal and an over-ventilation scenario, respectively, with the FSIL room intended to be used as an actual classroom. All the listening conditions were recorded with a binaural headset (Head Acoustics BHS II). Birdsongs and playground noise were recorded inside a room with the left ear oriented towards an open window. The room was facing a school courtyard, and the building was surrounded by trees. The two mechanical ventilation stimuli, by contrast, were recorded inside the FSIL facility at the two test air flow rates. Playground, MV_160, and birdsongs were reproduced through equalised headphones (Sony MDR-ZX310) at a sound level of 47 dB(A) at the left ear, whereas the MV_200 condition was presented at 52 dB(A). The ambient noise level of the laboratory was 36 dB(A). Playback levels were set using a binaural head and torso simulator (Head Acoustics HMS II.3 LN HEC) and calculating the equivalent sound level over 60 s. Figure 3 show the spectral content of the listening conditions at the participant’s ears.
FIGURE 3
3.1.3 Calculation task
Students completed 24 equations for each listening condition. Each equation was presented with two response buttons (true/false) and always consisted of two-digit additions or subtractions (e.g., 14 + 17 = 31). Half equations were additions and half of the problems were correct. Additionally, 12 trials were classified as easy to solve (additions without carrying and subtractions without borrowing), while the remaining 12 were considered difficult to solve (with carrying/borrowing). All combinations of result accuracy (true or false), difficulty level, and type of equation (addition or subtraction) were explored. To ensure that students were not exposed to identical questions or fixed sequences, a total of five equivalent lists were constructed. A strict time limit of 20 s was enforced for each equation. Accuracy and response times for each trial were recorded. The task was administered via the online platform Gorilla. sc (). More details on the calculation task could be found in and .
3.1.4 Experimental procedure
Figure 4 presents the experimental procedure employed. Initially, participants were provided with instructions regarding the experiment and were subsequently assigned to one of the two CO2 exposure groups. Group allocation was based on participants’ calculation proficiency and noise sensitivity. These pieces of information were obtained from a previous experiment (): calculation proficiency was assessed through a dedicated task (), while noise sensitivity was measured using the Italian version of the reduced Weinstein Noise Sensitivity Questionnaire (). Both variables were dichotomised into high and low levels using a median split. Participants were then evenly distributed across the two CO2 exposure groups to ensure a balanced representation of all combinations of noise sensitivity and mathematical proficiency. After being assigned to one of the two groups, participants completed a calculation task across all five listening conditions. The order of the conditions was counterbalanced across participants. Specifically, participants were assigned to different experimental pipelines using the Gorilla Experiment Builder platform (gorilla.sc). Each pipeline represented a specific presentation order of the listening conditions, which was counterbalanced using a Latin Square design. Within each main pipeline, sub-pipelines were created to account for all possible combinations of equivalent stimulus lists assigned to each listening condition. In each list equations were presented randomly.
FIGURE 4
3.1.5 Data cleaning
All students identified as neurodivergent or with reported hearing impairments were excluded from the analysis. Subsequently, a data cleaning process based on the single equation was carried out. First, outlier response times were removed from the dataset, in accordance with the procedure described in (). This resulted in the exclusion of all response times shorter than 1.5 s. Time-out responses (response time = 20s) were instead simply considered as incorrect answers. Then, for each participant, listening conditions with fewer than 50% of trials completed per difficulty level were removed to ensure a sufficient amount of data for each listening condition, while those with accuracy rates below 50% were removed, as this corresponds to the chance level of performance. Finally, participants with fewer than two valid listening conditions remaining were also excluded. At the end of the procedure, the original dataset (consisting of 22,680 observations: 189 participants x 24 trials x 5 listening conditions) was reduced by 29.87%. A more detailed table with the exclusion rate for each condition is presented in the Supplementary Table S1.
An exclusion rate of nearly 30% of the dataset may appear excessive. However, this level of data removal reflects the difficulty experienced by students in maintaining sustained engagement with the task. Moreover, following the data-cleaning procedure, the final dataset comprised 165 participants, which is in line with sample sizes reported in comparable studies in the literature (; ).
3.1.6 Data analysis
R software (ver. 4.5.2) () was employed in the analysis and a generalised linear mixed model (GLMM) was used to investigate the accuracy with the lme4 package (ver 1.1–37) (). Accuracy was modelled using a binomial distribution, as a function of task difficulty, listening condition, and air quality condition, including all two-way interactions among these three factors. Calculation proficiency, gender, block order, noise sensitivity, and the interaction between noise sensitivity and listening condition were included as covariates in the model. Participants were modelled as a random effect. Unlike the procedure previously adopted for participant grouping, noise sensitivity was categorised into three levels in the statistical model, in line with the analyses reported in (). Specifically, noise sensitivity was included as a categorical variable divided into tertiles, namely, low, medium, and high noise sensitivity. Calculation proficiency, by contrast, was included as a dichotomous variable obtained by applying a median split across participants. Finally, block order was included as a numeric variable ranging from 1 to 5, representing the progression of participants through the experimental blocks.
A similar model to the one described for accuracy was used to investigate the response time of the correct answers only (72% of the accuracy dataset), using a Gamma distribution with a logarithmic link. Equations 1 and 2 shows the models expressed in Wilkinson (Wilkinson and Rogers, 1973) notation.
3.2 Results
No effect of the environmental stressors was identified on accuracy, for either air quality, listening conditions, or interactions. A significant effect of the difficulty (χ2 (1) = 464.95, p < 0.001), the proficiency level (χ2 (1) = 26.45, p < 0.001), and gender (χ2 (1) = 4.35, p = 0.04) was instead found. A post hoc analysis was run to investigate the significant differences between the difficulty levels and showed higher accuracy with easy equations (M = 87%) rather than with the difficult (M = 74%) ones (z = 21.56, p < 0.001). Furthermore, boys (M = 82%) were more accurate than girls (M = 80%) (z = 2.09, p = 0.04). More proficient students (M = 85%) were more accurate in the calculation task than less proficient students (M = 76%) (z = 5.14, p < 0.001).
The analysis of response times revealed significant effects of task difficulty (χ2 (1) = 1,252.82, p < 0.001), gender (χ2 (1) = 55.87, p < 0.001), block order (χ2 (4) = 346.17, p < 0.001), listening condition (χ2 (4) = 17.7, p = 0.001) and the interaction between listening condition and noise sensitivity (χ2 (8) = 18.21, p = 0.02). Post hoc comparisons showed that participants responded faster to easy equations (5.91 s) than to difficult ones (7.58 s) (z = 35.77, p < 0.001), and that males (5.99 s) exhibited shorter response times than females (6.92 s) (z = 3.75, p < 0.001). Furthermore, response times decreased progressively with increasing block order, with all pairwise comparisons across blocks reaching statistical significance (all ps < 0.04). Regarding the interaction, post hoc analyses identified only two significant differences for mid sensitive to noise students. The differences were between birdsongs (6.54) and MV_200 (5.68) (birdsongs > MV_200 z = 3.55, p = 0.01), and between playground (6.48 s) and MV_200 (playground > MV_200, z = 3.46, p = 0.01). Figure 5 depicts the changes in response time as a function of listening condition for the three noise sensitivities. For the full statistical details regarding the model results please see the Supplementary Material.
FIGURE 5
3.3 Discussion
The results revealed a pattern fully consistent with previous studies (; ), showing for the calculation task no significant effect of environmental stressors on accuracy, but a clear influence on response time. The absence of accuracy-related effects linked to the listening conditions has been addressed in earlier work (); here, only a summary of the possible explanations is provided. These may include the relatively low sound pressure levels employed in the present study compared to those uses in the literature reporting the presence of significant effects, or the modality of test presentation—visual rather than auditory—which likely did not engage auditory-specific interference mechanisms such as noise masking (Wilson et al., 2012; ; ; ), that might made the task more challenging. Regarding the effect of CO2 on cognitive performance, the literature broadly supports the notion that improvements in air quality, typically achieved through enhanced ventilation, are associated with better outcomes in tasks involving mental arithmetic (Wargocki and Wyon, 2007; ; ). However, this was not the case in the present study. Two main hypotheses may explain this absence of effect. First, pure CO2 was used to manipulate air quality conditions. Existing research on adult populations (e.g., university students or office workers) indicates that exposure to pure CO2 alone may not elicit measurable changes in cognitive performance (Zhang et al., 2017b; ), at concentrations commonly found in indoor environment (). Although comparable data for children are currently lacking, it is plausible that a similar mechanism applies here. Second, a split-plot design was adopted in the experiment. The choice was selected to accommodate the logistical needs of school groups visiting the laboratory but may have compromised the statistical power of the study since the chosen designs have a lower power compared to within-subject design when the population is the same (; ). A post hoc power analysis using G*Power () with the following settings: repeated measurements, f = 0.1 (small effect size), power = 0.8, correlation between measurements = 0.1, confirmed it by indicating that at least 247 participants would have been required to detect a CO2 effect. All other effects observed on accuracy, namely, higher accuracy among male participants and those with higher proficiency, are in line with prior findings in the literature (; ; ). In contrast, a significant effect of listening condition was observed for response time, with the MV_200 condition yielding significantly faster responses compared to the other conditions. An hypothesis based on arousal is here proposed. This facilitative effect might be attributed to increased arousal caused by the auditory stimulus. MV_200 had the highest sound level among all tested conditions, which may have enhanced arousal more effectively. As discussed in the introduction and in previous literature () higher sound pressure levels are positively associated with increased arousal, potentially bringing participants closer to the peak of the arousal–performance curve, especially in very calm environments like the laboratory was. Interestingly, these findings diverge from those observed in () and reported in the Introduction. Although the results refer to two different subgroups, namely, participants with low noise sensitivity in the classroom experiment and those with medium noise sensitivity in the present study, a systematic difference attributable to the influence of the testing environment remains evident. This effect underscores the well-known lab-dilemma and should not be overlooked. The identification of these differences motivated the second part of this work, leading to the comparison between laboratory-based and field-based studies discussed in the following section.
4 Part II: comparison between laboratory and real classrooms
4.1 Methods
4.1.1 Comparison between the experiments
The first step in comparing the classroom and laboratory settings was to examine whether there were overall differences in accuracy and response time between the two experiments. Comparison was made including only listening conditions and participants common to the two experiments, leading to a further reduction of the dataset. In total, 118 participants and four listening conditions (quiet, playground, MV_160 and birdsongs) were included. For each listening condition, the mean accuracy and the median response time were calculated for each participant. A Wilcoxon test was used to assess differences in the distributions of the two variables between lab and classroom for each listening condition. No differences in accuracy distributions were found, whereas in the laboratory consistently shorter response times were observed compared to classrooms, with the difference ranging from 0.9 to 1.6 s depending on the listening condition.
4.1.2 Investigating the differences on response time
The differences found in the reduced dataset on response times due to the effect of the listening conditions were further investigated by comparing the results of the two dataset. Moreover, in order to include in the analysis also the effect of the students’ individual characteristics, they were divided into two groups based on their difference in response time between classroom and laboratory, calculated in the quiet condition as a proxy measure. Quiet is the baseline condition qualifying the environment acoustically without alterations arising from superimposed sound stimuli. Therefore, the difference in response time in quiet between the classroom and the laboratory was calculated for each participant. The distribution of the differences can be found in Figure 6. Students were categorised using a median split based on the distribution of their response-time differences between the two environments. Group G1 consisted of students with difference values below the median, namely, those who responded faster in the classroom compared to the laboratory. Conversely, group G2 comprised students with values above the median, representing those who responded faster in the laboratory compared to the classroom. Since the laboratory setting presented significantly fewer external disturbances than a typical classroom, this grouping reflects individual differences in environmental performance interaction. Specifically, pupils assigned to G2 performed better in a quiet and controlled environment, whereas pupils in G1 demonstrated the opposite pattern, performing relatively faster in the classroom setting. Finally, a GLMM was run to investigate how the response time is influenced by the listening conditions (playground, MV_160, birdsongs), the Group (G1, G2), the location (laboratory, classroom) and all the two-way and three-way interactions. Participants were included as a random factor. The full model is present in Equation 3. The model was run for all the listening condition except for quiet since it was used to divide students in two groups.
FIGURE 6
4.2 Results
The three-way interaction between listening conditions, groups and location was found to be significant (χ2 (3) = 9.36, p = 0.009). Therefore, two sub-models were created, one for each group (G1, G2), to examine the effect of listening conditions, the location, and the two-way interaction between these two factors on response time. Model is reported in Equation 4.
The sub-model for Group G1 showed a significant interaction between listening condition and location (χ2 (2) = 21.64, p < 0.001). Pairwise comparisons conducted on the interaction revealed significant differences between the two locations only for the listening conditions MV_160 (z = 6.31, p < 0.001) and birdsongs (z = 2.51, p = 0.02). Comparisons at a fixed location showed differences in response time in the laboratory only between playground and MV_160 (z = 3.09, p = 0.006), with MV_160 exhibiting shorter response times. As regards the classroom, MV_160 was significantly slower than playground (z = 3.48, p = 0.002) and birdsongs (z = 2.62, p = 0.02). The sub-model for group G2 revealed only a significant main effect of location (χ2 (1) = 431.65, p < 0.001), with consistently longer response times in the classroom than in the laboratory across all sound stimuli. Figure 7 presents a graphical representation of these comparisons.
FIGURE 7
4.3 Discussions
The overall comparison between the laboratory and classroom settings revealed consistently shorter response times for the test administered in the laboratory than in the classroom. This difference was observed across all listening conditions, with mean response times in the laboratory being between 0.9 and 1.6 s shorter than those recorded in the classroom, depending on the listening condition. These findings draw attention to the so-called lab-dilemma, as they indicate a measurable discrepancy between classrooms and laboratory settings in relation to the effects of acoustic stimuli on cognitive performance. The observed differences might be related to a lower level of distraction. Owing to structural constraints, the number of students per laboratory room was limited to eight, which likely reduced distractions arising from classmates. Similar evidence has been reported in the literature; for instance (), showed enhanced working memory performance when children worked individually rather than in a typical classroom setting, an effect attributed to reduced distractions. Although the task used in the present study differs, this finding may still be applicable, as previous studies have established a strong relationship between working memory and mathematical performance (; ). It should be noted that all students did the laboratory task after the classroom one, which may in itself have contributed to reduce the response time. Further details are provided in the limitations section. Additionally, a divergent trend was observed regarding the effects of specific sound stimuli. While anthropogenic noise (playground activity) proved most favourable in the classroom, mechanical ventilation noise (MV_160) was more beneficial in the laboratory for group G1. Dividing students into performance-based subgroups (G1 and G2) helped isolate these distinct behavioural profiles. In G1, classroom performance under MV_160 was worse than under other conditions; however, this pattern reversed in the laboratory. Conversely, G2 exhibited only a main location effect without significant differences between sound stimuli, indicating that their response pattern across acoustic conditions was stable across settings. These findings suggest that caution should be exercised when extrapolating laboratory results to real classrooms. To explore potential underlying mediators, individual noise sensitivity was evaluated. Ideally, participants in group G2 might be either so sensitive or so insensitive to noise that no significant differences were observed between sound stimuli. Additional analyses comparing noise sensitivity distributions between groups (Wilcoxon test) revealed no significant differences, indicating that noise sensitivity alone does not account for the subgroup distinction. Other individual traits influencing performance across settings—and how the environment modulates them—warrant further investigation. Overall, the findings suggest that, despite efforts to recreate in laboratory real-world conditions as closely as possible, (i) results may not be fully replicable between settings, neither in absolute response times nor in the relative impact of specific listening conditions, particularly for subgroups resembling G1. Group G2 appeared more stable across acoustic conditions regardless of environment. (ii) Individual student characteristics play a fundamental role in the generalisability of laboratory findings to real-world contexts, requiring dedicated future study.
5 Limitations
This study presents several limitations. First, the split-plot (between-subjects) design for the IAQ factor limited the statistical power to detect combined effects. This structure was dictated by logistical constraints, including school schedule availability and laboratory capacity. A fully within-subjects design would likely have provided greater sensitivity.
Second, regarding IAQ implementation, pure CO2 was injected to achieve target concentrations, and unmeasured pollutants such as volatile organic compounds (VOCs) and particulate matter (PM) were not tracked. Exposure to pure CO2 without human bioeffluents may elicit weaker physiological responses than real-world bioeffluent accumulation (Zhang et al., 2017a), Furthermore, while rooms were flushed prior to testing at maximum capacity (720 L/s), ventilation efficiency and complete air mixing throughout the space were not formally evaluated across all vertical planes. Head-level horizontal uniformity was confirmed across five measurement points (the four room corners and a point in the middle of the 8 m long wall, but vertical thermal or concentration gradients could not be ruled out. In addition, although the ventilation systems were equipped with full filtration and, as noted, the environments were thoroughly flushed beforehand, it cannot be excluded that residual concentrations of other unmeasured pollutants were responsible for the lack of observed differences.
Third, the laboratory sessions were conducted 5 months after the classroom experiment. Although problem sets were randomized and parallel mathematical equations were created, potential practice or order effects cannot be entirely ruled out.
Fourth, thermal–hygrometric and lighting parameters were not strictly matched between the classroom and laboratory settings, which may have introduced subtle environmental comfort differences. Additionally, target CO2 room assignments were kept fixed between Room A and Room B due to equipment setup constraints.
Finally, while explanations are grounded in Yerkes–Dodson arousal theory, direct physiological measurements (e.g., skin conductance, heart rate variability) were not collected in this study.
6 Conclusions and further works
A laboratory experiment was designed to assess the combined effects of air quality and acoustics on students’ calculation abilities. No combined effects were found on calculation performance; only the effects of the acoustic conditions for mid sensitive to noise individuals were observed. The absence of combined effects may be influenced by the use of pure CO2 and to the between-subjects design, both of which reduced the predictive power of the experiment. Additionally, uncontrolled ambient variables or unmeasured IAQ parameters, such as VOCs, may have further affected performance. On the acoustic side, better performance was observed under the noisier mechanical ventilation condition (200 L/s), which is likely to be the condition that most effectively engaged the students. This is because higher sound levels could possibly elicit greater arousal and so better performance; however, this premise warrants confirmation in future studies through direct arousal monitoring. However, it is difficult to generalise these results, as they may have been influenced by the laboratory setting itself. Indeed, the comparison between the laboratory experiment and a previous in-field (classroom) experiment showed consistently shorter response times in the laboratory. Moreover, the listening condition ordering differed from the one observed in the classroom, and the difference was found to depend on individual student response times in the quiet condition. Future research could address the gaps left by this study by exploring a wider range of combinations between sound and indoor air quality (IAQ) conditions. Moreover, dedicated efforts should focus on validating the proposed arousal hypothesis by measuring physiological or psychological arousal during task execution. Finally, a more detailed characterisation of (i) the participants with both physiological or behavioural measurements, and (ii) the environment is needed. Indeed, more comprehensive information on participant characteristics is essential, as these may interact with the environmental conditions and influence where individuals lie on the arousal–performance curve.
Statements
Data availability statement
The aggregated data supporting the conclusions of this article will be made available by the authors, without undue reservation.
Ethics statement
The requirement of ethical approval was waived by Eurac Research's research offices and complies with the Eurac Research Core Ethics and Integrity Principles for the studies involving humans. The studies were conducted in accordance with the local legislation and institutional requirements. Written informed consent for participation in this study was provided by legal guardians of the participants.
Author contributions
MP: Writing – review and editing, Conceptualization, Investigation, Methodology, Writing – original draft, Formal Analysis. CV: Methodology, Conceptualization, Supervision, Writing – review and editing. ST: Methodology, Supervision, Conceptualization, Writing – review and editing. FB: Writing – review and editing, Conceptualization, Methodology, Supervision. NP: Writing – review and editing, Supervision, Conceptualization, Methodology.
Funding
The author(s) declared that financial support was received for this work and/or its publication. The work was part of a University of Ferrara PhD project funded by Eurac Reasearch. In addition the publication of this paper was made possible thanks to the Open Access Fund of the University of Ferrara.
Conflict of interest
The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
Generative AI statement
The author(s) declared that generative AI was not used in the creation of this manuscript.
Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.
Publisher’s note
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.
Supplementary material
The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/fbuil.2026.1869553/full#supplementary-material
References
1
AhmedR.MumovicD.BagkerisE.UcciM. (2022). Combined effects of ventilation rates and indoor temperatures on cognitive performance of female higher education students in a hot climate. Indoor Air32, e13004. 10.1111/ina.13004
2
AllenJ. G.MacNaughtonP.SatishU.SantanamS.VallarinoJ.SpenglerJ. D. (2016). Associations of cognitive function scores with carbon dioxide, ventilation, and volatile organic compound exposures in office workers: a controlled exposure study of green and conventional office environments. Environ. Health Perspect.124, 805–812. 10.1289/ehp.1510037
3
AlvarA.FrancisA. L. (2024). Effects of background noise on autonomic arousal (skin conductance level). JASA Express Lett.4, 013601. 10.1121/10.0024272
4
Anwyl-IrvineA. L.MassonniéJ.FlittonA.KirkhamN.EvershedJ. K. (2020). Gorilla in our midst: an online behavioral experiment builder. Behav. Res. Methods52, 388–407. 10.3758/s13428-019-01237-x
5
ASHRAE EPIDEMIC TASK FORCE (2021).
6
Ashrae terminology IEQ (2025). Available online at: https://terminology.ashrae.org/?entry=indoor%20environment%20quality%20(IEQ) (Accessed ) (Accessed March 6, 2025).
7
BabichF.PinottiR.GazzinR.VisentinC.LolliniR. (2024). From single tests to a test-chain: a comprehensive approach for evaluating the interaction between the building envelope and the IEQ. E3S Web Conf.523, 01001. 10.1051/e3sconf/202452301001
8
BalazovaI.ClausenG.WyonD. P. (2007). “The influence of exposure to multiple indoor environmental parameters on human perception, performance and motivation,” in CLIMA 2007.WellBeing Indoors: Proceedings Congress (Helsinky: FINVAC).
9
BatesD.MächlerM.BolkerB.WalkerS. (2015). Fitting linear mixed-effects models using lme4. J. Stat. Softw.67. 10.18637/jss.v067.i01
10
BluyssenP. M. (2017). Health, comfort and performance of children in classrooms - new directions for research. Indoor Built Environ.26, 1040–1050. 10.1177/1420326X16661866
11
BluyssenP. M.ZhangD.KurversS.OvertoomM.Ortiz-SanchezM. (2018). Self-reported health and comfort of school children in 54 classrooms of 21 Dutch school buildings. Build. Environ.138, 106–123. 10.1016/j.buildenv.2018.04.032
12
BluyssenP. M.KimD. H.EijkelenboomA.Ortiz-SanchezM. (2020). “Workshop with 335 primary school children in the Netherlands: what is needed to improve the IEQ in their classrooms?,” in Building and Environment (Elsevier Ltd). 10.1016/j.buildenv.2019.106486
13
BradleyM. M.LangP. J. (2000). Affective reactions to Acoustic stimuli. Psychophysiology37, 204–215. 10.1111/1469-8986.3720204
14
CaviolaS.GerottoG.DanielaL.MammarellaI. C. (2016). AC-FL PROVE DI FLUENZA NELLE ABILITÀ DI CALCOLO PER IL SECONDO CICLO DELLA SCUOLA PRIMARIA.
15
CaviolaS.VisentinC.BorellaE.MammarellaI.ProdiN. (2021). Out of the noise: effects of sound environment on maths performance in middle-school students. J. Environ. Psychol.73, 101552. 10.1016/j.jenvp.2021.101552
16
ChinazzoG.AndersenR. K.AzarE.BarthelmesV. M.BecchioC.BelussiL.et al (2022). Quality criteria for multi-domain studies in the indoor environment: critical review towards research guidelines and recommendations. Build. Environ.226, 109719. 10.1016/j.buildenv.2022.109719
17
CohenJ. (2013). Statistical Power Analysis for the Behavioral Sciences. Routledge. 10.4324/9780203771587
18
CullingJ. F.StoneM. A. (2017). Energetic Masking and Masking Release, 41–73. 10.1007/978-3-319-51662-2_3
19
DuB.TandocM. C.MackM. L.SiegelJ. A. (2020). Indoor CO2 concentrations and cognitive function: a critical review. Indoor Air30, 1067–1082. 10.1111/ina.12706
20
Education at a Glance 2025 (2025). OECD Publishing. 10.1787/1c0d9c79-en
21
European Committee for Standardization (2019). “CEN EN 16798-1:2019 energy performance of buildings - ventilation for buildings - part 1: indoor environmental input parameters for design and assessment of energy performance of buildings addressing indoor air quality,” in Thermal Environment, Lighting and Acoustics - Module M1-6.
22
FanY.CaoX.ZhangJ.LaiD.PangL. (2023). Short-term exposure to indoor carbon dioxide and cognitive task performance: a systematic review and meta-analysis. Build. Environ.237, 110331. 10.1016/j.buildenv.2023.110331
23
FaulF.ErdfelderE.BuchnerA.LangA.-G. (2009). Statistical power analyses using G*Power 3.1: tests for correlation and regression analyses. Behav. Res. Methods41, 1149–1160. 10.3758/BRM.41.4.1149
24
Friso-van den BosI.van de Weijer-BergsmaE. (2020). Classroom versus individual working memory assessment: predicting academic achievement and the role of attention and response inhibition. Memory28, 70–82. 10.1080/09658211.2019.1682170
25
GuptaA.TorrianiG.TorresinS.PistoreL.PellegattiM.PiazzaL.et al (2025). Thermal comfort and perceived air quality (PAQ) using automatic ceiling fans in residential buildings. Energy Build.330, 115317. 10.1016/j.enbuild.2025.115317
26
HollemanG. A.HoogeI. T. C.KemnerC.HesselsR. S. (2020). The ‘Real-World Approach’ and its problems: a critique of the term ecological validity. Front. Psychol.11, 721. 10.3389/fpsyg.2020.00721
27
HughesR. W.VachonF.JonesD. M. (2007). Disruption of short-term memory by changing and deviant sounds: support for a duplex-mechanism account of auditory distraction. J. Exp. Psychol. Learn. Mem. Cogn.33, 1050–1061. 10.1037/0278-7393.33.6.1050
28
HydeJ. S.FennemaE.LamonS. J. (1990). Gender differences in mathematics performance: a meta-analysis. Psychol. Bull.107, 139–155. 10.1037/0033-2909.107.2.139
29
JiangJ.LiuF.JiangZ.ZhangJ.MaC. (2025). Experimental evaluation of short-term thermal experiences on students’ thermal comfort and cognitive performance. Build. Environ.285, 113613. 10.1016/j.buildenv.2025.113613
30
KlepeisN. E.NelsonW. C.OttW. R.RobinsonJ. P.TsangA. M.SwitzerP.et al (2001). The National Human Activity Pattern Survey (NHAPS): a resource for assessing exposure to environmental pollutants. J. Expo. Sci. Environ. Epidemiol.11, 231–252. 10.1038/sj.jea.7500165
31
KorsaviS. S.MontazamiA.MumovicD. (2020). The impact of indoor environment quality (IEQ) on school children’s overall comfort in the UK; a regression approach. Build. Environ.185, 107309. 10.1016/j.buildenv.2020.107309
32
LidestamB.HolgerssonJ.MoradiS. (2014). Comparison of informational vs. energetic masking effects on speechreading performance. Front. Psychol.5, 639. 10.3389/fpsyg.2014.00639
33
MiriM.FaubelC.Demarquet AlbanU.Martinez-MolinaA. (2025). Impact of indoor environmental quality on students’ attention and relaxation levels during lecture-based instruction. Buildings15, 2813. 10.3390/buildings15162813
34
MohebianZ.Farhang DehghanS.DehghanH. (2018). Evaluation of the combined effects of heat and lighting on the level of attention and reaction time: climate chamber experiments in Iran. Sci. World J.2018, 5171582. 10.1155/2018/5171582
35
PayneE. A.LoiN. M.ThorsteinssonE. B. (2020). The restorative effect of the natural environment on university students’ psychological health. J. Environ. Public Health2020, 4210285. 10.1155/2020/4210285
36
PellegattiM.TorresinS.VisentinC.BabichF.ProdiN. (2023). Indoor soundscape, speech perception, and cognition in classrooms: a systematic review on the effects of ventilation-related sounds on students. Build. Environ.236, 110194. 10.1016/j.buildenv.2023.110194
37
PellegattiM.VisentinC.TorresinS.BabichF.ProdiN. (2026a). The effect of ventilation-related sound stimuli on 11–14 years students’ cognition and soundscape assessments. Appl. Acoust.245, 111206. 10.1016/j.apacoust.2025.111206
38
PellegattiM.VisentinC.TorresinS.BabichF.WargockiP.ProdiN. (2026b). Combined and cross-modal effects of acoustic and indoor air quality conditions on sensory and cognitive responses of university students. Build. Environ.288, 114005. 10.1016/j.buildenv.2025.114005
39
PengP.NamkungJ.BarnesM.SunC. (2016). A meta-analysis of mathematics and working memory: moderating effects of working memory domain, type of mathematics skill, and sample characteristics. J. Educ. Psychol.108, 455–473. 10.1037/edu0000079
40
Pichora-FullerM. K.KramerS. E.EckertM. A.EdwardsB.HornsbyB. W. Y.HumesL. E.et al (2016). “Hearing impairment and cognitive energy: the framework for understanding effortful listening (FUEL),” in Ear and Hearing (Lippincott Williams and Wilkins), 5S–27S. 10.1097/AUD.0000000000000312
41
ProverbioA. M.De BenedettoF.FerrariM. V.FerrariniG. (2018). When listening to rain sounds boosts arithmetic ability. PLoS One13, e0192296. 10.1371/journal.pone.0192296
42
R Core Team (2025). R: A Language and Environment for Statistical Computing.
43
RatcliffR. (1993). Methods for dealing with reaction time outliers. Psychol. Bull.114, 510–532. 10.1037/0033-2909.114.3.510
44
REHVA (2021). REHVA COVID-19 Guidance Document How to Operate HVAC and Other Building Service Systems to Prevent the Spread of the Coronavirus (SARS-CoV-2) Disease (COVID-19) in Workplaces.
45
SchweikerM.AmpatziE.AndargieM. S.AndersenR. K.AzarE.BarthelmesV. M.et al (2020). Review of multi‐domain approaches to indoor environmental perception and behaviour. Build. Environ.176, 106804. 10.1016/j.buildenv.2020.106804
46
ScottJ. L.KraemerD. G.KellerR. J. (2009). Occupational hazards of carbon dioxide exposure. J. Chem. Health Saf.16, 18–22. 10.1016/j.jchas.2008.06.003
47
SeneseV. P.RuotoloF.RuggieroG.IachiniT. (2012). The italian version of the weinstein noise sensitivity scale: measurement invariance across age, sex, and context. Eur. J. Psychol. Assess.28, 118–124. 10.1027/1015-5759/a000099
48
SeyedrezaeiM.AwadaM.Becerik-GerberB.LucasG.RollS. (2023). Interaction effects of indoor environmental quality factors on cognitive performance and perceived comfort of young adults in open plan offices in North American Mediterranean climate. Build. Environ.244, 110743. 10.1016/j.buildenv.2023.110743
49
ShuS.MaH. (2019). Restorative effects of classroom soundscapes on children’s cognitive performance. Int. J. Environ. Res. Public Health16, 293. 10.3390/ijerph16020293
50
SörensenL. K. A.BohtéS. M.SlagterH. A.ScholteH. S. (2022). Arousal state affects perceptual decisionmaking by modulating hierarchical sensory processing in a large-scale visual system model. PLoS Comput. Biol.18, e1009976. 10.1371/journal.pcbi.1009976
51
TeigenK. H. (1994). Yerkes-dodson: a law for all seasons. Theory Psychol.4, 525–547. 10.1177/0959354394044004
52
ThompsonV. A.CampbellJ. I. D. (2004). A power struggle: Between-vs. within-subjects designs in deductive reasoning research. Psychologia47, 277–296. 10.2117/psysoc.2004.277
53
TorresinS.PernigottoG.CappellettiF.GasparellaA. (2018). Combined effects of environmental factors on human perception and objective performance: a review of experimental laboratory works. Indoor Air28, 525–538. 10.1111/ina.12457
54
ToyinboO. (2023). Indoor environmental quality, pupils’ health, and academic Performance—A literature review. Buildings13, 2172. 10.3390/buildings13092172
55
ValletW.van WassenhoveV. (2023). Can cognitive neuroscience solve the lab-dilemma by going wild?Neurosci. Biobehav. Rev.155, 105463. 10.1016/j.neubiorev.2023.105463
56
VisentinC.ProdiN. (2021). How reliable are 11- to 13-Year-Olds’ self-ratings of effort in noisy conditions?Front. Built Environ.7, 688016. 10.3389/fbuil.2021.688016
57
WangC.ZhangF.WangJ.DoyleJ. K.HancockP. A.MakC. M.et al (2021). How indoor environmental quality affects occupants’ cognitive functions: a systematic review. Build. Environ.193, 107647. 10.1016/j.buildenv.2021.107647
58
WargockiP.WyonD. (2007). The effects of outdoor air supply rate and supply air filter condition in classrooms on the performance of schoolwork by children (RP-1257). HVAC&R Res.13, 165–191. 10.1080/10789669.2007.10390950
59
WilkinsonG. N.RogersC. E. (1973). Symbolic Description of Factorial Models for Analysis of Variance. Available online at: https://www.jstor.org/stable/2346786 (Accessed April 14, 2026).
60
WilsonR. H.TrivetteC. P.WilliamsD. A.WattsK. L. (2012). The effects of energetic and informational masking on the words-in-noise Test (WIN). J. Am. Acad. Audiol.23, 522–533. 10.3766/jaaa.23.7.4
61
YangW.MoonH. J. (2018). Combined effects of sound and illuminance on indoor environmental perception. Appl. Acoust.141, 136–143. 10.1016/j.apacoust.2018.07.008
62
YangW.MoonH. J. (2019). Combined effects of acoustic, thermal, and illumination conditions on the comfort of discrete senses and overall indoor environment. Build. Environ.148, 623–633. 10.1016/j.buildenv.2018.11.040
63
YangW.MakitaK.NakaoT.KanayamaN.MachizawaM. G.SasaokaT.et al (2018). Affective auditory stimulus database: an expanded version of the International Affective Digitized Sounds (IADS-E). Behav. Res. Methods50, 1415–1429. 10.3758/s13428-018-1027-6
64
YerkesR. M.DodsonJ. D. (1908). The relation of strength of stimulus to rapidity of habit-formation. J. Comp. Neurology Psychol.18, 459–482. 10.1002/CNE.920180503
65
ZhangX.WargockiP.LianZ.ThyregodC. (2017a). Effects of exposure to carbon dioxide and bioeffluents on perceived air quality, self-assessed acute health symptoms, and cognitive performance. Indoor Air27, 47–64. 10.1111/INA.12284
66
ZhangX.WargockiP.LianZ.XieJ.LiuJ. (2017b). Responses to human bioeffluents at levels recommended by ventilation standards. Procedia Eng.205, 609–614. 10.1016/j.proeng.2017.10.415
Summary
Keywords
acoustics, calculation, indoor air quality, lab-dilemma, students
Citation
Pellegatti M, Visentin C, Torresin S, Babich F and Prodi N (2026) Combined effects of acoustics and indoor air quality on calculation performance: a laboratory experiment and a lab-dilemma investigation. Front. Built Environ. 12:1869553. doi: 10.3389/fbuil.2026.1869553
Received
30 April 2026
Revised
16 August 2026
Accepted
25 August 2026
Published
25 September 2026
Volume
12 - 2026
Edited by
Fan Zhang, Griffith University, Australia
Reviewed by
Bhagyashree Ambore, RNS Institute of Technology, India
Ravi Pratap Singh Jadon, Delhi Technological University, India
Updates
Copyright
© 2026 Pellegatti, Visentin, Torresin, Babich and Prodi.
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
*Correspondence: Matteo Pellegatti, matteo.pellegatti@unife.it
Disclaimer
All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article or claim that may be made by its manufacturer is not guaranteed or endorsed by the publisher.