Associations of Human Cognitive Abilities with Elevated Carbon Dioxide Concentrations in an Enclosed Chamber

: Fifteen participants were exposed in an enclosed environmental chamber to investigate the effects of elevated carbon dioxide (CO 2 ) concentration on their cognitive abilities. Three CO 2 conditions (1500, 3500, and 5000 ppm) were achieved by constant air supply and additional ultrapure CO 2 . All participants received the same exposure under each condition, during which they performed six cognitive tests evaluating human perception, attention, short-term working memory, risky decision-making, and executive ability. Generalized additive mixed effects model (GAMM) results showed no statistically signiﬁcant differences in performance on the reaction time (RT) tests, the speed perception test, and the 2-back test. This suggests that elevated CO 2 concentrations below 5000 ppm did not affect participants’ perception and short-term working memory. However, a significant increase in response time was observed in the visual search (VS) test, the balloon simulation risk test (BART), and the Stroop test at 5000 ppm compared to lower exposure concentrations. The slower responses reﬂected the detrimental effects of elevated CO 2 concentrations on visual attention, risky decision-making, and executive ability. The ﬁndings suggest that the control level of CO 2 concentrations should be tighter in enclosed workplaces where rapid response and operational safety are required.


Introduction
The atmospheric average concentrations of carbon dioxide (CO 2 ) are approximately 400 ppm, and may increase to the range between 794 and 1142 ppm by 2100 according to the Intergovernmental Panel on Climate Change (IPCC) [1]. The indoor CO 2 mainly comes from outdoor CO 2 brought in by fresh air ventilation and CO 2 generated by human metabolism indoors. Given the likelihood of increasing atmospheric CO 2 concentration, the indoor CO 2 concentration could also increase gradually in future. The typical CO 2 concentration levels are generally lower than 1000 ppm in residential buildings [2]. However, in crowded or confined indoor spaces, CO 2 concentrations may exceed 2000 ppm due to insufficient per capita fresh air ventilation [3][4][5]. For people working in a typical indoor workplace, high CO 2 exposure can occur for a few hours. However, in special enclosed workspaces (such as submarines and manned spacecraft), exposure to high CO 2 concentrations could last for several days. Considering that higher indoor CO 2 exposure may lead to detrimental effects on human work performance, it is essential to study the direct CO 2 impacts on human cognitive abilities that are closely associated with productivity and work safety. Table 1 summarizes the studies examining the human cognitive responses to elevated CO 2 concentrations via the addition of pure CO 2 or the manipulation of ventilation rate. For the former, the pure CO 2 was fully mixed with the outdoor air and then sent into the experimental room. For the latter, the outdoor air ventilation rates were reduced to increase the concentrations of CO 2 and other pollutants. The current studies of CO 2 effects on cognitive performance still show a disappointing lack of consistency in results. It seems to be that the more complex the task, the greater CO 2 effects would be evidenced. For example, the performance of simple office tasks was not significantly affected by elevated CO 2 concentration below 5000 ppm [6,7], but the performance of complex tasks decreased significantly with elevated CO 2 concentration [8,9]. As shown in Table 1, the strategic management simulation (SMS) test has been used to evaluate the strategic decisionmaking ability. The exposure to elevated pure CO 2 concentration could significantly reduce the SMS test performance starting at 1000 ppm [10,11]. Rodeheffer et al. [12] observed that the SMS performance declined slightly for the subjects exposed to CO 2 concentration of 2500 ppm compared to 600 ppm, but did not further decline when they were exposed to 15,000 ppm. They speculated that the inconsistent finding could be explained by submariners having developed physiological adaptations to high CO 2 level, because they were routinely exposed to high CO 2 concentrations. Although Snow et al. [13] reported that participants did not exhibit significantly worse cognitive test scores at a CO 2 concentration of 2700 ppm, their cognitive flexibility and executive ability still appeared to be negatively affected relative to the CO 2 exposure at 830 ppm, given lack of learning effect. a. The study used the between-subject design, and the others used the within-subject design. b. The participants were exposed at 35 • C, not a thermal comfort environment. c. The PaCO 2 increased (mean = 47 mmHg) or decreased (mean = 38 mmHg) from the resting level (mean = 30 mmHg). d. Compared the ventilation rate instead of CO 2 concentration.
However, several other studies indicated that pure CO 2 effects on cognitive abilities were not potent. Zhang et al. [6,7] reported that no significant effect of CO 2 concentration on office task performance were observed in the 500-5000 ppm exposure range, regardless if the task difficulty was simple or moderate. Similarly, Liu et al. [14] found that no significant differences were observed in the performance of nine cognitive tests when participants were exposed to CO 2 concentrations up to 3000 ppm, compared with 380 ppm. Bloch-Salisbury et al. [15] reported that neither the partial pressure of carbon dioxide in the arterial blood (PaCO 2 ) increases (mean = 47 mmHg) nor decreases (mean = 38 mmHg) from the resting level (mean = 30 mmHg) would affect the cognitive performance, which was probably attributed to the low statistical power of small sample size. In contrast to the addition of pure CO 2 , an impaired cognitive performance was more commonly observed under restricted ventilation conditions [16][17][18][19]21]. Only one study of astronaut-like subjects indicated no significant changes in SMS and cognition test batteries as concentration rose from 600 ppm to 5000 ppm via decreased ventilation rates [20]. The outcomes conflicted with those of similar studies [10,11], which was likely caused by differing characteristics of the various subject populations.
As summarized above, the previous findings remain controversial regarding the effects of additional pure CO 2 on human cognition. More research evidence is needed to investigate the sole effects of CO 2 exposure on different kinds of cognitive abilities. Particularly, the potential adverse impacts of high CO 2 exposure may affect work performance and safety in confined workplaces, which merit further exploration. Therefore, the main purpose of this study was to examine experimentally the sole effects of CO 2 exposure on several work-related cognitive abilities including perception, attention, short-term working memory, risky decision-making, and executive ability, and to further provide a reference for the exposure limit of CO 2 concentration in the working environments. In this study, fifteen subjects were recruited to take six cognitive tests under three different CO 2 conditions (1500, 3500, and 5000 ppm) in an enclosed environmental chamber. The performance metrics of the cognitive tests were evaluated by statistical models to analyze the effects of elevated CO 2 concentrations from a moderate level to the occupational limit. Based on the observed effects, the effects of test difficulty and physiological regulation on cognitive performance are further discussed.

Participants
The participants were limited to healthy adult males considering that the operators in high-duty workplace environments are mostly male. Therefore, fifteen healthy male college students with an engineering background were recruited for this study. The mean age of the participants was 24.20 ± 2.48 years, and their average body mass index (BMI) was 22.67 ± 2.70 kg/m 2 . They were all right-handed, and had no history of disorders such as color blindness, neurological errors, allergy, and alcohol addiction. The participants were informed that they would be tested under three CO 2 concentration conditions, but were not notified which conditions they were exposed to. Additionally, they were asked to maintain adequate sleep (not less than eight hours per day) before coming to the experimental site. Stimulants such as drugs, perfumes, alcohol, and caffeinated drinks were prohibited before and during the tests. All participants signed consent forms, completed the tests without dropping out, and received compensation after the experiment. The experimental protocol was approved by the Institute Review Board (IRB) of Beihang University.

Environmental Conditions
The experiment was carried out in an environmental chamber with a size of 7.6 m × 3.0 m × 2.1 m, as shown in Figure 1. The conditioned outdoor air was fully mixed with ultrapure CO 2 (≥99.99% purity) and then sent into the chamber. The exhausted air in the chamber was discharged by a fan to external atmosphere. The air temperature was controlled by the heating and cooling units. The fresh air supply rate was constantly set at 250 m 3 /h by the volume regulator, with which the human bioeffluent concentrations were expected to be moderate to low during the tests. The dosing rate of pure CO 2 was controlled by the flowmeter and valve opening to achieve different CO 2 concentrations. were not notified which conditions they were exposed to. Additionally, they were asked to maintain adequate sleep (not less than eight hours per day) before coming to the experimental site. Stimulants such as drugs, perfumes, alcohol, and caffeinated drinks were prohibited before and during the tests. All participants signed consent forms, completed the tests without dropping out, and received compensation after the experiment. The experimental protocol was approved by the Institute Review Board (IRB) of Beihang University.

Environmental Conditions
The experiment was carried out in an environmental chamber with a size of 7.6 m × 3.0 m × 2.1 m, as shown in Figure 1. The conditioned outdoor air was fully mixed with ultrapure CO2 (≥99.99% purity) and then sent into the chamber. The exhausted air in the chamber was discharged by a fan to external atmosphere. The air temperature was controlled by the heating and cooling units. The fresh air supply rate was constantly set at 250 m 3 /h by the volume regulator, with which the human bioeffluent concentrations were expected to be moderate to low during the tests. The dosing rate of pure CO2 was controlled by the flowmeter and valve opening to achieve different CO2 concentrations. Three CO2 exposure conditions were tested [9]: a baseline exposure with concentration at 1500 ppm (referred to as low CO2 condition), and the exposures to concentrations at 3500 ppm and 5000 ppm that were achieved by dosing pure CO2 to supply air (referred to as medium and high CO2 conditions, respectively). The high CO2 condition was based on the current 8 h occupational limit and the safety requirements by the IRB for human-oriented experiments. The low CO2 condition was based on the fact that the CO2 concentrations are typically higher than 1500 ppm in enclosed spaces such as aircraft cabins [5] and submarines [12]. The medium CO2 condition was set between the low and high concentration conditions.
The CO2 concentration was measured by a T6615 sensor (measurement range: 0-10,000 ppm; measurement error: ≤75 ppm or ±10%; sampling interval: 1 min) that was installed near the participants at a height of 1.2 m in the test area. The measured CO2 concentrations under the three conditions were 1626 ± 306 ppm, 3562 ± 259 ppm, and Three CO 2 exposure conditions were tested [9]: a baseline exposure with concentration at 1500 ppm (referred to as low CO 2 condition), and the exposures to concentrations at 3500 ppm and 5000 ppm that were achieved by dosing pure CO 2 to supply air (referred to as medium and high CO 2 conditions, respectively). The high CO 2 condition was based on the current 8 h occupational limit and the safety requirements by the IRB for human-oriented experiments. The low CO 2 condition was based on the fact that the CO 2 concentrations are typically higher than 1500 ppm in enclosed spaces such as aircraft Atmosphere 2022, 13, 891 5 of 17 cabins [5] and submarines [12]. The medium CO 2 condition was set between the low and high concentration conditions. The CO 2 concentration was measured by a T6615 sensor (measurement range: 0-10,000 ppm; measurement error: ≤75 ppm or ±10%; sampling interval: 1 min) that was installed near the participants at a height of 1.2 m in the test area. The measured CO 2 concentrations under the three conditions were 1626 ± 306 ppm, 3562 ± 259 ppm, and 5087 ± 318 ppm. The concentrations of VOCs, CO, and NH 3 under the three exposure conditions were 105-108 ppb (SD < 4), 36-38 ppb (SD < 1), and 17-18 ppb (SD < 1), as measured by a CPR-KA air quality analyzer (measurement range: 0-10 ppm (VOCs), 0-50 ppm (CO), 0-30 ppm (NH 3 ); measurement error: ≤5% full scale; sampling interval: 2 min). This indicated that the confounding effects of other air pollutants were well controlled by the constant fresh air ventilation.
The air temperature and relative humidity were maintained at 25 • C and 65% by the air-conditioning units. The temperatures during the tests were 24.6 ± 1.7 • C, as measured by a LX8013 thermo-hygrometer (measurement range: −20 • C-60 • C; measurement accuracy: ±0.5 • C). The predicted mean vote (PMV) was estimated to be between −0.5 and 0.5 [9], which indicated that the thermal environment in the experimental chamber was neutral. Participants were also allowed to adjust their clothing to ensure they were thermally comfortable during the experiment. In sum, the confounding effects of other environmental factors were controlled to study the effects of additional pure CO 2 exposure on cognitive abilities.

Cognitive Tests
The participants performed six computer-based cognitive tests sequentially under each exposure condition. The performance metrics of the tests were used to evaluate five basic cognitive abilities of perception, attention, short-term working memory, risky decision-making, and executive ability. The details are described as follows: (1) Reaction time (RT) tests (perception) The RT tests [22,23] (duration: 7 min) consist of a simple RT test, discriminative RT test, and choice RT test. In the RT tests, symbols were presented in the center of the screen. The participants were asked to look at the screen and to press the corresponding response buttons on keyboard immediately when they saw the target symbols. Table 2 lists the symbols displayed, target symbols, and response buttons. The accuracy of the simple, discriminative, and choice RT tests are referred to as ACC sim , ACC dis , and ACC cho , respectively. The correct response time (CRT) for the three tests are referred to as CRT sim , CRT dis , and CRT cho , respectively. The higher accuracy and shorter CRT indicate better perception ability. • "J" ( ); "F" ( ); Space bar ("•") a. The target symbol was a red triangle. b. The target symbol was a red square. c. The target symbol was a red circle.
(2) Speed perception test (perception) In the speed perception test [24] (duration: 7 min), a ball moved uniformly and randomly from different directions into the center area of the screen covered with a gray circle. When the ball entered the gray circle, the participants were asked to make a key response when they predicted that the ball reached the center point of the screen. The deviation rate of the predicted position from the center point was used to measure the human perception ability to estimate the speed of a moving object, and the smaller deviation rate indicated better performance. In the VS test [25,26] (duration: 5 min), the participants were asked to find the target character ("L" in red) against other visual distracters ("L" in black and "T" in red). They pressed the "F" button when found the target character, or pressed the "J" button if not. Six performance metrics were calculated to evaluate human attention ability, including the accuracy and average CRT for all trials (referred as ACC total and CRT total , respectively), the accuracy and average CRT for trials with the report of target character (referred as ACC report and CRT report , respectively), the false alert rate, and the missing report rate. The higher accuracy, shorter CRT and lower false/missing report rate indicated stronger attention ability.
(4) 2-back test (short-term working memory) In the 2-back test [27,28] (duration: 7 min), a sequence of characters was presented one by one in the central area of the screen. Additionally, for each character, the participants needed to judge whether the current character matched the previous second character, and responded as soon as possible by pressing the "F" or "J" button ("F"-matching; "J"-not matching). The ACC and the average CRT were used to evaluate the short-term working memory ability. The higher ACC and shorter CRT indicated better short-term working memory.

(5) Balloon analogue risk test (BART) (risky decision-making)
Thirty trials were set for the BART [29,30]. In each trial, the participants could earn credits by clicking the "F" button to pump the balloon up on the screen. Each click incrementally inflated the balloon and credits were added to a count-up counter. However, the earnings for the trial would be lost if the balloon exploded. Specific information regarding the balloon's breakpoint determination was not given to the participants. The participants could decide to stop pumping the balloon by clicking the "J" button at any time prior to the explosion, and collect the credits earned in the trial. The participants were also informed that the total earned credits could be redeemed as cash rewards. The average number of pumps on unexploded balloons, the critical response time of pumps on unexploded balloons (the average time for the last five inflation decisions) and the final response time of pumps on unexploded balloons (the response time of the last inflation decision) were used to evaluate the risk decision-making ability. The higher number of pumps indicated the greater risk-taking propensity, and the longer response time indicated the more hesitant decision-making.

(6) Stroop test (executive ability)
In the Stroop test [31,32] (duration: 2 min), the words "RED" or "GREEN" appeared in the central area of the screen, randomly shown in red or green. The participants were asked to press the "Z" button when the color of the word was red, and to press the "/" button when the color of the word was green, regardless of the meaning of the words. Six metrics were used to evaluated human executive ability, including the accuracy and average CRT for all trials (referred as ACC total and CRT total , respectively), for the trials with consistent word color and meaning (referred as ACC con and CRT con , respectively), and for the trials with inconsistent word color and meaning (referred as ACC incon and CRT incon , respectively). The higher accuracy and shorter CRT indicated the stronger executive ability.

Experimental Procedure
The participants took part in practice sessions before the experimental sessions to reduce the potential learning effect of the cognitive tests. They received detailed visual instructions for each cognitive test from the experimenters during the initial practice session. Then they were asked to practice twice a day for three days during the later practice sessions. During the experimental sessions, the fifteen participants were evenly allocated to three groups, and were exposed to the same CO 2 condition by groups in one day. For each group, the tests under different exposure conditions were carried out during the same time period on three consecutive weekdays. During each exposure condition, RT tests, speed perception test, VS test, 2-back test, BART, and Stroop test were conducted in succession with a total rest interval of approximately 13 min, as shown in Figure 2. The participants were instructed to try their best to complete the cognitive tests, with their dominant hand using a laptop that was placed on the table in front.

Statistical Analysis
Generalized additive mixed effect model (GAMM) analyses [8][9][10][11]13,33] were performed using the open-source statistical package R version 3.6.1 (R Project for Statistical Computing, Vienna, Austria) to study the associations of cognitive test performance metrics with CO2 exposure concentrations, treating the participant as a random effect, as shown in Equations (1) and (2). The level of statistical significance was set at p < 0.05.
where y is the cognitive test performance metric; and * are the fixed intercepts; and are the fixed effects of medium CO2 and high CO2 compared to low CO2, respectively; * is the fixed effect of high CO2 compared to medium CO2; b and * are the random effects of individual differences between participants; and e and * are the residuals.

Results
The detailed GAMM results of the CO2 effects on the cognitive test performance metrics (Table S2) are presented in Table S1. Additional plots shown in

Statistical Analysis
Generalized additive mixed effect model (GAMM) analyses [8][9][10][11]13,33] were performed using the open-source statistical package R version 3.6.1 (R Project for Statistical Computing, Vienna, Austria) to study the associations of cognitive test performance metrics with CO 2 exposure concentrations, treating the participant as a random effect, as shown in Equations (1) and (2). The level of statistical significance was set at p < 0.05. (1) where y is the cognitive test performance metric; β 1 and β * 1 are the fixed intercepts; β 2 and β 3 are the fixed effects of medium CO 2 and high CO 2 compared to low CO 2 , respectively; β * 3 is the fixed effect of high CO 2 compared to medium CO 2 ; b and b * are the random effects of individual differences between participants; and e and e * are the residuals.

Results
The detailed GAMM results of the CO 2 effects on the cognitive test performance metrics (Table S2) are presented in Table S1. Additional plots shown in

Statistical Analysis
Generalized additive mixed effect model (GAMM) analyses [8][9][10][11]13,33] were performed using the open-source statistical package R version 3.6.1 (R Project for Statistical Computing, Vienna, Austria) to study the associations of cognitive test performance metrics with CO2 exposure concentrations, treating the participant as a random effect, as shown in Equations (1) and (2). The level of statistical significance was set at p < 0.05.
where y is the cognitive test performance metric; and * are the fixed intercepts; and are the fixed effects of medium CO2 and high CO2 compared to low CO2, respectively; * is the fixed effect of high CO2 compared to medium CO2; b and * are the random effects of individual differences between participants; and e and * are the residuals.

Results
The detailed GAMM results of the CO2 effects on the cognitive test performance metrics (Table S2) are presented in Table S1. Additional plots shown in        Figure 3 presents the CO2 effects on the performance metrics of the RT tests and the speed perception test. The overall performance of the simple RT test was the best, followed by the discriminative RT test and the choice RT test. This indicated that cognitive performance was worse with increased test difficulty, regardless of the CO2 conditions. The performance metrics of each RT tests and the speed perception test were slightly changed with the CO2 conditions, but the variations were not statistically significant. In sum, elevated CO2 concentrations did not significantly affect human perception.

Attention
The CO2 effects on the performance metrics of the VS tests were depicted in Figure 4. Both the CRTtotal and CRTreport increased significantly with elevated CO2 concentration from 1500 to 5000 ppm, and the CRTreport also had a significant increase when the concentration increased from 3500 to 5000 ppm. It could be inferred that the attention response slowed when exposed to higher CO2 concentrations. The accuracy metrics were marginally worse under the medium CO2 condition, as indicated by slightly lower accuracy and higher missing/false report rate, but the differences were not statistically significant.  Figure 3 presents the CO 2 effects on the performance metrics of the RT tests and the speed perception test. The overall performance of the simple RT test was the best, followed by the discriminative RT test and the choice RT test. This indicated that cognitive performance was worse with increased test difficulty, regardless of the CO 2 conditions. The performance metrics of each RT tests and the speed perception test were slightly changed with the CO 2 conditions, but the variations were not statistically significant. In sum, elevated CO 2 concentrations did not significantly affect human perception.

Attention
The CO 2 effects on the performance metrics of the VS tests were depicted in Figure 4. Both the CRT total and CRT report increased significantly with elevated CO 2 concentration from 1500 to 5000 ppm, and the CRT report also had a significant increase when the concentration increased from 3500 to 5000 ppm. It could be inferred that the attention response slowed when exposed to higher CO 2 concentrations. The accuracy metrics were marginally worse under the medium CO 2 condition, as indicated by slightly lower accuracy and higher missing/false report rate, but the differences were not statistically significant.

Short-Term Working Memory
As shown in Figure 5, no statistically significant difference was observed in the ACC and the CRT between any two CO 2 conditions, which indicated that elevated CO 2 concentrations had no significant effect on short-term working memory. Nevertheless, the response time was slightly longer under the high CO 2 condition.

Risky Decision-Making
As shown in Figure 6, the average number of pumps, critical response time, and final response time on unexploded balloons all increased gradually with elevated CO 2 concentration in the BART. The critical response time had a significant increase as the concentration increased from 1500 to 5000 ppm. The results indicated that the participants slightly tended to make riskier and more hesitant decisions when exposed to higher CO 2 concentrations. Figure 7 shows that the CRT con increased significantly as the CO 2 concentration rose from 1500 to 5000 ppm. The CRT total was also modestly longer at higher CO 2 concentration level. It can be inferred that the participants spent more effort to execute the Stroop test trials with higher CO 2 exposure. Nevertheless, no clear variation trend was found between accuracy metrics and CO 2 conditions.

Executive Ability
In sum, exposure to elevated CO 2 concentrations from 1500 to 5000 ppm had some detrimental effects on attention, risky decision-making, and executive ability, as manifested by a significant increase in response time. Perception and short-term working memory, however, were only marginally affected by the CO 2 conditions.

Discussion
The main finding of this study is that the elevated CO 2 concentrations below 5000 ppm did not affect participants' perception and short-term working memory, but had a detrimental effect, reflected as slower response, on their visual attention, risky decision-making, and executive ability. The following sections discuss four aspects of the results.

Comparison with Previous Studies
Only the findings in the literature on the effects of additional pure CO 2 are compared with our results, because both CO 2 and other air pollutants may contribute to the detrimental impacts on cognition when the target CO 2 concentrations are achieved by manipulating ventilation rates. Figure 8 summarizes the recent findings of impact on the five cognitive abilities by pure CO 2 addition. As shown in Figure 8, the previous studies [6,13,14] found that perception and short-term memory were not significantly affected by the additional pure CO 2 concentration below 3000 ppm, which was consistent with our findings. Our study further extended the concentration limit of nonsignificant effect to 5000 ppm. It is worth noting, however, that perception could be affected by CO 2 concentration much higher than 5000 ppm. For example, Sheehy et al. [34] reported a significantly slower response of choice reaction time test during the inhalation of 5% CO 2 in 50% O 2 .
Furthermore, our results indicated that attention, risky decision-making, and executive ability were significantly impaired as the CO 2 concentration increased from 1500 to 5000 ppm. Similarly, a significant detrimental effect of CO 2 exposure on vigilance was also found at concentrations as low as 3500 ppm using the psychomotor vigilance test (PVT) in our previous study [35]. Snow et al. [13] stated that the increase in CO 2 concentration from 800 to 2700 ppm could lead to reduced cognitive flexibility and executive ability, given lack of learning effect, as examined by the shifting attention test and the Stroop test. However, some scholars have also found inconsistent results. For example, Zhang et al. [6] and Liu et al. [14] reported nonsignificant effects on attention using the D2 test and on execution using the Stroop test, which may be because the concentration range they studied was limited to 3000 ppm. This result indicates that the CO 2 effects on cognitive abilities become more significant as the concentration increases.
Zhang et al. [6] and Liu et al. [14] reported nonsignificant effects on attention using the D2 test and on execution using the Stroop test, which may be because the concentration range they studied was limited to 3000 ppm. This result indicates that the CO2 effects on cognitive abilities become more significant as the concentration increases. The main negative effect on cognition observed in this study is delayed reaction, which is consistent with the summary of Du et al. [36] that indicated accumulation of CO2 Figure 8. Results summary of additional pure CO 2 effects on the five cognitive abilities. * Studies with statistically significant (p < 0.05) changes in performance metrics [6,[10][11][12][13][14]35].
The main negative effect on cognition observed in this study is delayed reaction, which is consistent with the summary of Du et al. [36] that indicated accumulation of CO 2 and other indoor pollutants could reduce the reaction speed of various cognitive tests but leaves the accuracy unaffected. The difficulty of a test determines its capability for observing changes in cognitive performance metrics. Therefore, the nonsignificant effect on performance metrics was possibly due to the relatively low difficulty of the cognitive tests we used, most of which were accomplished with accuracy rates higher than 95%. In contrast, the SMS test let subjects experience the management of a business strategy in a competitive business environment, which is more difficult than other cognitive tests used in the reviewed literature. As assessed by the SMS test, the increased CO 2 exposure from 550 to 1400 ppm could result in a 50% reduction in cognitive function scores [11]. This supports that the speculation that the effect of CO 2 exposure on cognitive performance could be more pronounced with increasing task difficulty and workload.

Potential Mechanism Underlying the Cognitive Performance
The underlying mechanism of CO 2 affecting cognition may be the interaction between passive protection and subjective adaption. With the elevation of CO 2 concentrations, passive protection triggered by physiological regulation, such as the increase in sympathetic activity and deep breath [37,38], would reduce arousal levels and lead to further cognitive decline. This will increase the need for extra mental effort and arousal level to accomplish cognitive tests. According to the fight-or-flight response theory [39], a state of physiological arousal occurs when the human body perceives stress caused by an unfavorable environment. In this case, people will devote more energy to maintaining their cognition through subjective adaption, which helps to counteract the detrimental effect of elevated CO 2 concentration on cognitive performance. The increased arousal level could maintain cognitive performance, but may sacrifice cognitive speed to some extent. As demonstrated in this study, the accuracy metrics of the Stroop test and the VS test were not affected by increasing concentrations, but at the expense of slower responses. The participants also tended to be more hesitant in making risky decisions in the BART at higher concentration. The response times of tests for perception and short-term memory tests were not significantly affected, probably because less additional mental effort was required. In sum, the detrimental effect of high CO 2 concentration on cognitive performance may be induced by passive protection, and then be alleviated when subjective adaption is activated. It can also be inferred that the accuracy metrics may be significantly reduced if the difficulty level or CO 2 concentration further increases, when subjective adaption is not sufficient to meet the demands of extra mental work.

Exposure Limit for CO 2 Concentration in Workplaces
ASHRAE recommends that indoor CO 2 concentrations should be maintained at or below 1000 ppm to ensure adequate ventilation for acceptable indoor air quality. The current 8 h permissible exposure limit is 5000 ppm as regulated by the Occupational Safety and Health Administration (OSHA) and the American Conference of Governmental Industrial Hygienists (ACGIH). However, current indoor CO 2 concentration standards have not taken into account the maintenance or improvement of cognitive function, despite growing evidence pointing to the negative impact of CO 2 on cognitive function. As discussed in this study, exposures to elevated CO 2 concentrations above 1000 ppm have been reported to adversely affect various cognitive abilities, and the effects would become more significant with increasing exposure concentrations and task difficulty. Specifically, the findings of our study suggest delayed responses could occur during the CO 2 exposure at a concentration of 5000 ppm, when performing the cognitive tests of visual attention, risky decision-making, and executive ability. Slow reaction (lapse) has been considered as one of the most important factors leading to human error accidents [40]. Therefore, the current occupational exposure threshold of 5000 ppm may not meet the requirements for rapid response and operational safety in workplaces, especially for those enclosed environments (e.g., submarines, cockpits, space stations), where staff members are routinely exposed to high concentrations of CO 2 . To prevent human error accidents due to cognitive decline, we suggest formulating more stringent occupational regulations on permissible exposure limits for CO 2 .

Limitation of This Study
There are several limitations to consider when interpreting the results of this study. A major limitation is the unbalanced order of CO 2 exposures among the participants, because there was insufficient time to establish different CO 2 conditions in one day. This could lead to potential confounding effects of prior exposures. Another limitation is that the recruited participants were limited to fifteen healthy college-age male students. The sample size was only fifteen as limited by the experimental resource. Because of the relatively small sample size, the statistical power could be lowered [41], which indicates that the effects of p-values modestly higher than 0.05 are also worthy of attention. Only college-age male students were recruited, considering the CO 2 effects could be related to the population type and our target population was the operators in high-duty workplace environments. Therefore, a larger and mixed study population would be desirable in future research. In addition, the assessment of human cognitive function was limited by the performance metrics of six cognitive tests, which covered only certain aspects of cognition. The cognitive tests employed in our study were also relatively easy. Therefore, to further study the cognitive abilities from more perspectives, additional cognitive tests of different types and difficulty levels are suggested to be involved. It is possible to observe a greater effect of CO 2 exposure on human cognition as the increase in test difficulty and mental workload.

Conclusions
In this study, fifteen participants were recruited to perform six classic cognitive tests under three CO 2 conditions (1500, 3500, and 5000 ppm) to investigate the sole effects of CO 2 exposure on five work-related cognitive abilities. The statistical results showed that perception and short-term working memory were only marginally affected by increased CO 2 concentrations. However, attention, risky decision-making, and executive abilities were significantly impaired at 5000 ppm, as indicated by a significant increase in response time, though leaving the accuracy metrics unaffected. It can be inferred that people have to devote more energy to maintaining cognition through subjective adaption, which helps to combat the cognitive decline caused by elevated CO 2 concentrations, but at the cost of slower cognitive speed. The adverse effects of CO 2 on cognitive performance may even become more pronounced as concentrations and task difficulty increased. The research findings suggest that to meet the cognitive requirements for rapid response and operational safety, CO 2 controls in enclosed workplaces should be more stringent than the current permissible exposure limit.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/atmos13060891/s1, Table S1. GAMM results of the CO2 effects on the cognitive test performance metrics; Table S2. Values of the cognitive test performance metrics.

Informed Consent Statement:
Informed consent was obtained from all subjects involved in the study.
Data Availability Statement: Data available on request due to restrictions, e.g., privacy or ethical.