Acute Diesel Exhaust Exposure Causes a Delayed Reduction in Cognitive Control

Urban residents are frequently exposed to high levels of trafﬁc-derived air pollution for short time periods, often (but not exclusively) during commuting. Although chronic air pollution exposure and health effects, including neurological effects on children and older adults, are known to be correlated, causal effects of acute pollution exposure on brain function in healthy young adults remain sparsely investigated. Neuroinﬂammatory accounts suggest effects could be delayed by several hours and could affect attention, especially in social contexts. Using a controlled atmosphere chamber, we exposed 81 healthy young adults to either diluted diesel exhaust (equivalent to polluted roadside environments) or clean air for one hour. Half of each group immediately completed a selective attention task to assess cognitive control; remaining participants completed the task after a 4-hour delay. Cognitive control was signiﬁcantly poorer after diesel versus clean air exposure for those in the delay but not immediate test condition, suggesting an inﬂammatory basis for this acute negative effect of air pollution on cognition. These ﬁndings provide the ﬁrst experimental evidence that acute diesel exposure, comparable to polluted city streets, causes a negative effect on cognitive control several hours later. These ﬁndings may explain commuter mental fatigue and support clean-air initiatives.


Introduction
Outdoor (ambient) air pollution (AAP) is the primary global environmental risk to health that increases mortality worldwide [1] . Although adverse effects on cardiovascular and respiratory systems of AAP are well established [2,3] , emerging evidence indicates that AAP may also be neurotoxic [4,5] , degrading the brain's cellular structures [6] , leading to neurocognitive decline [7] and delaying neurocognitive development [8] . Specifically, lifetime exposure to AAP has been linked to greater risk of substantial cognitive deficits associated with progressive neurodegenerative diseases such as Alzheimer's Disease [9] , Parkinson's Disease [10] , and Multiple Sclerosis [11] . Chronic AAP exposure is also associated with poorer than expected memory and cognitive executive function in clinically healthy older participants [12] . Additional work on the effects of chronic exposure to AAP has focused on children, linking AAP to neuroinflammation [13] and altered neurodevelopment [14] . Cognitive impact of AAP exposure on children has revealed negative effects on psychomotor and sensory processing [15] , and executive cognitive function [16,17,18] ; particularly working memory [19,20,21,22] and attention [23,20] .
Although substantial gains have been made in understanding the impact of chronic AAP exposure on cognition in the aging and in the developing brain, few studies have systematically investigated how acute exposure to AAP affects brain function in healthy young adults. Every day, road [24] and rail users [25] are exposed to relatively short bursts of very high levels of AAP [26,27] , yet the acute impact of these events are unknown. Acute exposure to such AAP could cause immediate, mild degradation in brain function if impact on systemic physiological systems was rapid, e.g., interference with gas exchange in the lungs, leading to system-wide de-oxygenation. Alternatively, if acute AAP effects resulted from slower acting neuroinflammatory responses [28,29] related to the activation of microglial cells and oxidative stress in the brain [30,31] , then functional consequences post-exposure could be delayed by several hours. The aim of the current study was to examine the effects of acute AAP exposure on cognitive function in healthy young adults immediately after AAP exposure and several hours later to distinguish between these two alternative accounts. As the most ubiquitous source of outdoor pollution in urban areas is often traffic-related, our study used a controlled air delivery system to expose participants to air containing a combination of nitrogen oxides (NO X = NO + NO 2 ) and ultrafine particulate matter PM UF or to clean air (control condition). The type of low-quality air we used in our experimental condition mimics the AAP associated with diesel exhaust (DE) [32] and is often studied in health-related AAP studies.
Neuroinflammation, without a specific AAP association, has been linked to both psychomotor slowing [33,34] and higherorder cognitive consequences, including degradation of attention and social-emotional perception [35,36,37,38] . Critically, these consequences are measurable even with very mild inflammation that does not provoke classic signs of sickness behaviour such as fever and social withdrawal [34,36,37] . This raises the possibility that acute exposure to AAP, also not typically associated with sickness behaviour, could nevertheless produce similarly negative effects on cognition. Inhalation of NO X and PM UF , both components of DE, are especially likely to result in neuroinflammatory effects [39] . Previous studies of acute DE exposure on respiratory health indicate a delayed systemic inflammatory response that peaks between four-and six-hours post-exposure [40,41] , with others showing inflammatory changes between 18-and 20-hours post-exposure [42,43] , although one study failed to find any change in four widely studied bio-markers of systemic inflammation during this interval [44] . These findings support the general prediction that any deficits in cognitive function induced by acute exposure to AAP in our study should be observable after a delay of several hours but not immediately after exposure.
Prior studies of the effects of acute air pollution exposure on cognitive and brain function are sparse and show indeterminate results. Although two electroencephalographic (EEG) studies both reported subtle alterations in brain electrical activity (time-frequency changes) during and up to 2 hours after initial DE or indoor air pollution (cooking) exposure [45,46] , the pattern of activity found was inconsistent between studies and neither study measured changes in behavioural function. Another study of acute exposure to AAP reported immediate, negative cognitive effects on adults [47] , but the cognitive assessment they used involved tests designed for clinical neurological diagnosis that may be inappropriate for use on non-clinical populations [48] . However, a study of the effects of recent versus chronic natural exposure to AAP on children reported that high recent (within 48 hours) exposure slowed visual information processing speed [49] . Similarly, short-term (24 hours) ambient levels of NO 2 and elemental carbon, a DE tracer, were negatively associated with attention function in children [23] . Taken together, these studies provide a mixed picture of the timing and type of effects of acute AAP.
Previous work on the cognitive impact of AAP has lacked precision with regards to cognitive assessment. Like other complex functions, cognition is composed of multiple functional subsystems, each underpinned by a complex connection of multiple brain areas [50,51] . A core cognitive function, implicated in many different disorders of mental health including schizophrenia, depression, and anxiety, is control over attention [52,53,54] . Attention in this context refers to a set of mechanisms that prioritise information processing by selectively boosting neural representations of task-relevant stimuli and suppressing representations of task-irrelevant information [55] .
As high-level systems are limited in capacity, the brain uses a proactive control mechanism to plan strategically and enable selective engagement with expected, pertinent information as well as active avoidance of predictable but distracting information. However, a reactive control mechanism is also available for controlling behaviour when sudden, unpredictable events occur. For example, planning to make a coffee requires proactive control to identify and walk towards the kettle, whilst avoiding the biscuit tin, but reactive control might be needed to avoid colliding with a suddenly appearing colleague on the way. These two mechanisms are thought to compete for control over attention with completion of planned tasks dependent on sustained proactive control [56,57] and distraction by unexpected, task-irrelevant events reflecting reactive control.
Proactive control is typically weakened in mental health disorders that are also associated with neuroinflammation, including depression [58] , anxiety [59] , and schizophrenia [60] . Moreover, proactive control is negatively affected by obesity, a condition associated with high levels of inflammation [61] . Easily measured behaviourally, proactive cognitive control is thus a good candidate function to assess in studies of AAP exposure, especially considering its putative sensitivity to acute inflammatory states. Here we measured cognitive control to assess effects of DE exposure. To enhance the sensitivity of our behavioural test, we used emotional face stimuli as distracting stimuli as numerous previous studies show that expressive faces are highly compelling distractors, strongly activating reactive mechanisms [62] and therefore especially demanding of proactive control mechanisms [63] .

Results
Healthy young participants were randomly assigned to one of four groups: clean air 'immediate' (CA-i), diesel exhaust 'immediate' (DE-i), clean air 'delay' (CA-d), or diesel exhaust 'delay' (DE-d) (See Table 3). Prior to air exposure, participants completed a mood measure, rubbed eucalyptus-scented gel under their nose, and were given a hard candy mint to mask any smell or taste of chamber air. During a one-hour air exposure period, participants completed questionnaires providing demographic information; recent sleep quality (PSQI); and depression, anxiety, and stress levels (DASS). CA groups were exposed to low average concentrations of NO X [mean = 19 ppb ± (17)], CO [mean = 281 ppb ± (111)], and PM 2.5 [mean = 0.08 µgm −3 ± (0.08)]; and DE groups exposed to city-street comparable concentrations of NO X [mean = 524 ppb ± (51)] and 2/14 CO [mean = 1784 ppb ± (222)], although PM 2.5 concentrations were below WHO limits [mean = 6.97 µgm −3 ± (3.22)] (See Figure 3). After exposure, participants then completed a second mood measure and side-effects questionnaire. Those in the immediate groups then completed the cognitive control task illustrated in Figure 1. Participants in the delay groups completed another mood and side-effects measure 4 hours later and then completed the cognitive control task. At the end of their session, participants reported which air type they thought they had breathed.

Figure 1.
An illustration of the cognitive task. Each trial began with a fixation cross presented for 200 milliseconds (ms), followed by a spatial cue for 400 ms indicating with 100% reliability the location of the upcoming target. After another fixation cross (350 -850 ms), a target array appeared for 75 ms. The item appearing at the uncued location (distractor) was either a scrambled image or another face. The task was to report the gender of the target face as quickly and accurately as possible. Face images were full colour photos of real people.
The cognitive control task required fast, accurate target face gender identification in the presence of another face (2-face trials) or a non-face distractor (1-face trials). Typically, in this task a face distractor slows response time indicating that it has captured reactive selective attention. Numerous studies of cognitive control show that distractor-induced slowing is exacerbated when the preceding trial has no compelling distractor, as in a 1-face trial, compared to when a distractor is present, as in a 2-face trial [64] . This is due to prior experience of distractor suppression on trial n-1 boosting proactive target processing on trial n, whereas when trial n-1 does not require suppression, proactive control is weakened leading to increased susceptibility to attention capture by an irrelevant distractor on trial n [65] . Differences in response time (RT) for 2-face trials preceded by a 2-face trial (repeat sequence) versus RT on 2-face trials preceded by a 1-face trial (change sequence) serves to inversely index proactive cognitive control, i.e., the capacity to maintain strong selection bias for the target, despite recent sensory events. It is therefore expected that RTs would be slower for change versus repeat sequences, and that this effect would be exacerbated after DE exposure due to loss of proactive cognitive control.
An analysis of variance (ANOVA) of individual mean RT was used to determine the effect of trial sequence type (repeat, change) and air exposure groups  Table 2).
To investigate the interaction of air exposure group and sequence type, RT difference scores (∆RT -Change minus repeat sequences) were compared across air exposure groups; see Figure 2. For the 'delay' conditions, group mean ∆RT for the DE-exposed group (mean = 24, s.d. = 25) was 23 ms larger than that for the CA-exposed group (mean = 1, s.d. = 26), a statistically significant difference [2-tailed, independent sample t-test, t(41) = -2.89, p = 0.012]. The corresponding difference No significant group differences were identified in those tested immediately after exposure, indicating that cognitive control ability is poorer four-hours after exposure to diesel exhaust compared to clean air. between the 'immediate' DE-exposed group (mean = 8, s.d. = 29) and CA-exposed group (mean = 7, s.d. = 13) was negligible and non-significant [2-tailed, independent sample t-test; t(36) = -0.042, p = 1.934], Bayesian statistics confirmed the null hypothesis for this comparison (BF excl = 3.172).
All corresponding analyses of proportion correct responses produced non-significant results. ANOVA of individual proportion correct scores using trial sequence type and air exposure group as between-subjects factors showed only nonsignificant main and interaction effects (F < 1). Acceptance of the null hypothesis regarding proportion correct scores was confirmed using Bayesian analysis (BF excl > 3), suggesting RT differences are unlikely to reflect strategic processes such as speed-accuracy trade-offs.
Significant variations in PM 2.5 concentrations between 'immediate' and 'delay' DE groups led us to investigate whether pollutant concentration for each test time could explain differences in task performance alone. Session average NO, NO 2 , NO X , PM 2.5 and CO concentrations experienced by individual participants in each testing group ('immediate' and 'delay'), irrespective of exposure group, were correlated with their ∆RT (cognitive control measure). In the 'immediate' conditions no significant correlation was identified between ∆RT and pollutant. Conversely, concentrations of all pollutants were significantly positively correlated with ∆RT in the 'delay' conditions, explaining 13-24% of the variance in our data (See Table 1). Together these results argue against the notion that differences in pollutant concentration alone resulted in cognitive control changes, pointing to time of testing as the primary factor to explain the significantly poorer cognitive control identified in the DE-d condition compared to CA-d.

Socio-Emotional Processing
Previous studies of acute inflammation effects on social emotional cognition reported a reduction in the ability to identify emotional expression when viewing the eyes of expressive face photographs [36,37] . These studies raise the possibility that those in the DE-d condition may have experienced more difficulty interpreting the emotional expression on faces even though emotional information was task irrelevant. To investigate whether this potential processing difficulty could have negatively affected cognitive control, responses on 1-face trials with happy versus fearful face target expressions were compared for each pollution group. RT analysis found neither main nor interaction effects to be significant [  > 3). This suggests that differences in perceptual acuity of emotion expression had no significant role in determining two-trial sequence performance.
To further explore a possible influence of face expression on DE effects, individual mean RTs and proportion correct scores in 2-face trials were analysed for face expression congruency effects. Congruent trials, where both faces had the same emotion expression, were compared to incongruent trials, where target and distractor expression differed. For participants sensitive to face expression, the former condition could be considered to have less information than the latter rendering processing easier, an effect that might not benefit those with less facial expression sensitivity.  Table 2).

Pollutant Metric Variability
There was a significant main effect of air-exposure group identified for all pollutants, with DE concentrations higher than CA concentrations in all cases (p < 0.001) (See Figure 3). For PM 2.5 there was also a significant interaction of time of testing and air exposure [F(1, 77) = 31.484, p < 0.001, η p 2 = 0.290, 1-β = 1] such that, although concentrations were similar between CA groups, they were significantly higher for DE-d compared to DE-i (Figure 3e). The DE-d group were exposed to almost double (mean = 8.98, s.d. = 2.52) the concentration of the DE-i group (mean = 4.74, s.d. = 2.33) µgm −3 on average. However, average mass concentrations were not higher than the WHO annual limit for PM 2.5 concentration (10 µgm −3 ) and all below 24-hour limits (25 µgm −3 ) for either DE group (Figure 3f). No temporal differences in pollutant exposures (interactions) were identified for other pollutants.

Subjective Measures
To investigate whether effects of loss of cognitive control in the DE-d condition were related to the subjective feelings of sickness behaviour; affect and arousal or to expectation due to awareness of air-exposure condition, self-report measures were compared among air exposure and testing time groups. A χ 2 test indicated no significant association between pollution group and accuracy of participant's identification of their air-exposure condition [χ 2 (3, N = 81) = 1.520, p = 0.678], supporting the view that awareness of air condition did not account for cognitive control degradation observed in the DE-d group. (See Table  3).
All mood measures were unaffected by pollution group and interaction between pollution group and time of assessment (pre-exposure, immediately post-exposure, delayed post-exposure; see Supplementary Materials for details). The absence of subjective changes in mood suggests internal emotional states provide no basis for explanation for cognitive control degradation observed in the DE-d group and suggests participants were unaware of their exposure condition. Supporting the latter point, analyses of side effects subjectively reported immediately after exposure for immediate and delay exposure groups indicated no effect of air-exposure group. This also held for the delay group when side effects were again assessed 4 hours after exposure. (See Supplementary Materials for details). These data indicate that participants did not experience "sickness behaviour" in any conditions, a side effect often reported in vaccination paradigms that itself may alter executive functioning [66] .

Discussion
The current study investigated the impact of one-hour diesel exhaust exposure on executive brain function in a healthy adult population. We found that participants exposed to DE had significantly lower cognitive control compared to those exposed to clean air four hours after exposure, with no differences between groups when tested immediately after exposure. This study provides the first experimental evidence that control over selective visual attention, a critical cognitive function, is impaired in clinically healthy adults after a single acute air pollution exposure. Not only does this important finding add to the extant research showing a negative correlation between cognitive function and long-term chronic exposure to air pollution, but it also provides evidence that acute air pollution effects are relatively slow acting, consistent with air pollution causing neuroinflammation.
The negative impact of DE exposure identified here is unlikely to be due to "sickness behaviour" as groups did not differ in self-report of physical health or mood alterations after air exposure. Nor are effects easily explained as a psychological demand characteristic resulting from awareness of air exposure condition, as evidence indicates participants were unable to reliably report the condition to which they had been exposed. Although an indirect effect of air pollution on face expression perceptual capacity could potentially account for our findings, an analysis of face expression perception effects on behavioural responses provides no support for this potential confound.
Previous experimental studies on acute effects of air pollution on cognition are sparse. A study using a colour Stroop task that measures task-switching and suppression of automatic responses failed to identify any negative consequences of brief AAP exposure [47] . Shehab and Pope (2019) [47] used a candle to increase PM 2.5 concentrations before cognitive testing immediately after exposure, yielding an average concentration of 41.4 µgm −3 ± (46.1) compared to just 8.98 µgm −3 ± (2.5) in the current study (DE-d group). Consistent with our study, no immediate degradation of attention was found. On the other hand, despite the lower average PM 2.5 concentrations used in the present study, we report a significant impact of AAP exposure on delayed cognitive control over visual selective attention. Consistent with out findings, attention measured using the Stroop task was associated with recent (up to 48-h) PM exposure in children [49] . Likewise, a recent correlational study identified an association in older adults between acute exposure to higher PM 2.5 concentrations and lower global cognitive functioning [67] . On the other hand, Saenen et al., 2016 [49] found no association between recent exposure and attention, using the Continuous Performance test on the same study. The discrepancy in these findings is likely due to the greater sensitivity of selective attention tasks that allow precise response time measures and the ability to examine subtle sequential trial effects that reveal cognitive adaptation to changing events.
The current study was not designed to identify which pollutant in the DE-d condition led to cognitive deficits nor to identify the physiological means by which DE exposure affects brain function. Although the present study utilised relatively low PM concentrations, the concentrations of NO 2 for both DE groups were above WHO guideline limits [68] , although not beyond what might be experienced in a busy city street. Thus NO 2 , other nitrogen oxides, or CO, could be the cause of the cognitive deficits identified here instead of PM. If so, then hypoxia, whereby the body is deprived of adequate oxygen supply at the tissue level [69] , needs to be considered as a potential underlying cause of the cognitive deficits observed here. Acute hypoxia is known 7/14 to cause memory and executive function performance deficits, e.g., when climbing at altitude [70] and can eventually cause neuronal cell death [71] . Milder prolonged (30 min) exposure to hypoxic conditions can slow response time in selective attention task, but these effects peak immediately after exposure and improve significantly within 2 hours, returning to baseline with 24 hours [72] . This implies that, if hypoxia were the basis of the cognitive deficits observed here, performance should have been lower for the DE-i compared to the DE-d group, whereas the reverse was observed.
The PM 2.5 concentrations dropped off over the testing day because of deposition in the chamber, leading to significantly higher concentrations for the DE-d versus DE-i group. Whilst an immediate acute loss of cognitive control in the DE-d group cannot be ruled out, from the available evidence a delayed response is more likely given the necessary time for physiological processes, such as inflammation, to progress. In either case, the cognitive effects of repeated exposure to DE at these levels is of concern.
A candidate process to explain the delayed effect is an inflammatory response that results in modulated neurotransmission. Short-term air pollution exposure has been shown to cause a neuroinflammatory response [40,41,42,43] , and vaccination paradigms show that neuroinflammation negatively impacts attentional processing and therefore, potentially, cognitive control, about 6 hours after vaccination [34,35] . These findings are consistent with the speculation that inflammatory responses to AAP led to the observed cognitive deficits in our DE-d group.
A recent study identified an association between acute exposure to higher PM 2.5 concentrations and lower global cognitive functioning [67] . Importantly, participants in that study taking anti-inflammatory medications showed less decline of cognitive functioning compared to their counterparts, implying a protective effect of anti-inflammatory medications against worsening cognitive performance due to air pollution exposure. However, the Gao et al., (2021) [67] study study did not experimentally manipulate anti-inflammatory dosage, and executive functioning alone was not changed as a result of AAP exposure. Despite these caveats, their results are consistent with the notion that inflammation is a mechanism whereby AAP exposure may cause cognitive deficit.
In summary, this study reveals for the first time that cognitive control can be disrupted after short-term exposure to diesel exhaust. The implications of this result are significant as humans are frequently exposed to high concentrations of air pollutants in their environment [24,25] . Urban citizens including children will typically experience morning and evening peaks in acute exposure, associated with daily commutes. Given the cognitive demands of the commute on drivers and pedestrians, it is reassuring that our results do not find immediate loss of cognitive control after exposure, although we cannot rule-out an immediate effect at higher PM exposures. Instead, the current study finds delayed loss of control, suggesting that cognitive 'dips' might be experienced in the early afternoon and again in the evening. Although the consequences of such 'dips' is unknown, even subtle degradation of cognitive control likely impacts the quality and ease of decision making [73] and places strain on emotion control, potentially degrading mental health. We expect that, all other things being equal, avoiding daily acute exposure (e.g., by shifting work patterns to work from home more often) will be beneficial for cognitive health. Understanding how physical environments impact psychological processes such as cognition is an important component of the emerging picture of how the urban environments, in which most humans now live, affects the health of the species.

Participants
Ninety staff and student participants aged between 18 and 44 years were recruited through a database held by the University of Manchester and via on and offline advertisements. Individuals who reported current neurological, psychiatric, inflammatory, or respiratory disorders (e.g., multiple sclerosis, depression, rheumatoid arthritis, asthma), use of anti-inflammatory medication during the past 7 days, vaccination within the last 14 days, or current smoking were excluded. Data from 9 individuals were excluded from all analyses due to their early withdrawal (N = 2), their depression score was +2.5SDs from group mean (N = 3), their sleep quality was +2.5SDs from group mean (N = 1), or their overall behavioural accuracy was less than 70% (N = 3). Table 3 shows characteristics of the remaining participants. All procedures were approved by the University of Birmingham Science, Technology, Engineering, and Mathematics Ethical Review Committee (reference number ERN_18-1613). All methods were performed in accordance with relevant guidelines and regulations.

Procedure
Daily, prior to participant arrival, a Volkswagen SD1-1.9 Diesel Engine was run for 20 seconds before injecting air into the atmospheric chamber (DE condition) or outside the building (CA condition) by a non-experimenter, keeping both the participant and experimenter blind to the daily condition. After informed consent, participants completed a mood questionnaire, height and weight were measured, eucalyptus-scented gel placed under their nose, and they were given a hard candy mint. Participants wore a non-rebreather nose and mouth mask connected to the atmospheric chamber; air was inhaled from the chamber, but exhaled breath expelled into the ventilated testing room. After 60 minutes of air exposure, during which time participants filled out questionnaires, participants then completed a mood and side-effects questionnaire. In the immediate condition,  Table 3. Descriptive statistics of participant demography. Numbers in parenthesis indicate standard deviation. One-way ANOVAs were conducted for each demographic. No significant group differences were identified giving us confidence that between-group comparisons are valid and appropriate. Importantly, there was no significant χ 2 association [χ 2 (3) = 1.520, p = 0.678] between pollution group and self-reported condition accuracy. This implies participants in all pollution groups were equally blind to their condition, validating the double-blind nature of the study.
participants subsequently took part in the facial identification task (∼ 20 minutes) before being debriefed and paid £25. In the delay condition, participants returned 4 hours later to complete the mood and side-effects questionnaire again, take part in the facial identification task, were debriefed and paid £30.

Facial identification task
Stimuli. A white spatial cue arrow (2°x 1.5°) pointing left or right and a white centrally presented fixation cross (0.5°in diameter) were used. Faces were gathered from the A set of the Karolinska Directed Emotional Faces [74] utilising the frightened (fearful) & smiling (happy) emotional stimuli. Scrambled images were created by splitting face images into 13,984 squares and randomising their position. Each target/distractor image subtended 9°x 12.1°, with the centre of each presented 8.8°of visual angle laterally to the left and right of centre.
Procedure. See Figure 1 for the sequence of displays in each trial. Participants were instructed to respond as quickly as possible identifying the target stimulus gender using the 'a' or 'z' keyboard keys with index and middle fingers of their dominant hand; key assignment to 'male' and 'female' was counterbalanced between participants. The gender of the target and distraction face was incongruent on 90% of the trials. There were 6 blocks containing 66 trials each, for a total of 396. Second fixation cross was the duration the product of a random integer chosen between 20 & 50 by frame rate (17 Hz); the target array consisted of one image either side of the fixation cross, for 75 ms; and a final fixation cross presented for 1,500 ms or until participants responded. The target array comprised a central fixation, distractor image (scrambled, happy, fearful), and target image (happy, fearful) with all stimuli combinations equally likely. After the short (75 ms) presentation of the target array, participants were instructed to identify the gender of the target face as quickly and accurately as possible.

Atmospheric Chamber
The Manchester Aerosol Chamber (MAC) is a purpose-built chamber comprising an 18m 3 FEP Teflon bag in an air-conditioned enclosure and a high-volume clean air filling system incorporating a series of trace gas and particulate scrubbers. See Shao et al. (2021) [75] for details Control of pollutant concentrations was maintained through timed injection of the engine exhaust and the clean air, ensuring similar chemical composition between participants. Cycling between air mix conditions was facilitated by full computer control and monitoring of all necessary chamber conditions. A high-capacity ozone generator served as a cleaning agent during flushing between experiments, occurring daily, with a "harsher" clean carried out weekly.
Participants were fitted with non-rebreather masks attached to the chamber via a plastic pipe, which allowed them to breathe in air from the chamber without forced air pressure. Relative humidity and temperature were measured at several points throughout the chamber and were kept at 50% and 20°C, respectively, to avoid uncomfortable breathing conditions.

Questionnaires
The Depression Anxiety Stress Scale (DASS) [76] measured recent mental health: Participants rated to what extent 42 statements applied to them over the past week on a scale of 0 to 4 (Did not apply at all -Applied to me very much, or most of the time). The Pittsburgh Sleep Quality Index (PSQI) [77] measured participant sleep quality the night prior to testing: This 10-item questionnaire uses objective and subjective questions to determine participant sleep quality over the past month; this was adjusted to refer to the night prior to testing. The Pollution Exposure Lifestyle Questionnaire (PEL) was formulated in house to collect demographic information including postcode and city of previous residence, used to calculate an urbanicity score to ensure no group differences in previous pollution history. A Mood measure contained 4 sliding scales from 0 to 100 asking participants to rate their tenseness, irritation, excitement, and happiness. A Side Effects measure formed of 9-items asking participants to self-report on a scale of 0 to 4 (No, Mild, Severe, and Extreme) changes in cognition, headache, dizziness, nausea, fatigue, shortness of breath, coughing, throat irritation, and any non-specific uncomfortable feeling, after exposure.

Data Analyses
RT data. RTs were excluded from statistical analyses if there was no response, the response was too fast (RT < 150 ms), or the response was incorrect, accounting for 15.3% of data. First trial data from each block were removed, and individual RTs were trimmed if ±2.5 SDs from mean of 1-face trials, and 2-face trials, accounting for 2.3% of data.
To calculate cognitive control, mean RTs and accuracy scores were subjected to a 2x4 mixed ANOVA using sequence type (change and repeat) as the repeated, and pollution group (CA-i, DE-i, CA-d, and DE-d) as the between factor. Socio-emotional processing utilised 2x4 mixed ANOVAs with target emotion (happy, and fearful) as the repeated and pollution group as the between factors for both RT and accuracy on one-face trials. 2x4 mixed ANOVAs were also conducted on two-face trials comparing emotional congruency of stimuli [congruent (target and distractor same emotion), and incongruent (target and distractor different emotion)] as the repeated and pollution group as the between factor. JZS Bayesian statistics were calculated alongside frequentist statistics for these measures.
For air quality measures, concentrations for each pollutant were averaged across the one-hour exposure for each participant. 2x2 between subject's ANOVAs using time of testing ('immediate' and 'delay') and exposure type (diesel exhaust and clean air) were conducted for each pollutant.
Levene's test of Equality of Variances and Mauchly's Test of Sphericity were used to test for assumption violations; adjustments using the Greenhouse-Geisser correction were made where necessary. Alpha (α) values were set at 0.05 throughout. Bonferroni corrections were applied for multiple comparisons where appropriate. All frequentist analyses were conducted using SPSS v.26.0 [78] . All Bayesian analysis were conducted using JASP [79] .