Neurotoxic effects of home radon exposure on oscillatory dynamics serving attentional orienting in children and adolescents

Abstract

Radon is a naturally occurring gas that contributes significantly to radiation in the environment and is the second leading cause of lung cancer globally. Previous studies have shown that other environmental toxins have deleterious effects on brain development, though radon has not been studied as thoroughly in this context. This study examined the impact of home radon exposure on the neural oscillatory activity serving attention reorientation in youths. Fifty-six participants (ages 6–14 years) completed a classic Posner cuing task during magnetoencephalography (MEG), and home radon levels were measured for each participant. Time-frequency spectrograms indicated stronger theta (3–7 Hz, 300–800 ms), alpha (9–13 Hz, 400–900 ms), and beta responses (14–24 Hz, 400–900 ms) during the task relative to baseline. Source reconstruction of each significant oscillatory response was performed, and validity maps were computed by subtracting the task conditions (invalidly cued – validly cued). These validity maps were examined for associations with radon exposure, age, and their interaction in a linear regression design. Children with greater radon exposure showed aberrant oscillatory activity across distributed regions critical for attentional processing and attention reorientation (e.g., dorsolateral prefrontal cortex, and anterior cingulate cortex). Generally, youths with greater radon exposure exhibited a reverse neural validity effect in almost all regions and showed greater overall power relative to peers with lesser radon exposure. We also detected an interactive effect between radon exposure and age where youths with greater radon exposure exhibited divergent developmental trajectories in neural substrates implicated in attentional processing (e.g., bilateral prefrontal cortices, superior temporal gyri, and inferior parietal lobules). These data suggest aberrant, but potentially compensatory neural processing as a function of increasing home radon exposure in areas critical for attention and higher order cognition.

1. Introduction

Attentional orienting and reorienting are imperative to daily functioning as we must be able to attend to important stimuli within our environment, and subsequently shift our attention elsewhere as competing stimuli emerge. This process must be quick and dynamic to serve efficient and complete processing of our surroundings. Attention develops rapidly throughout childhood and early adolescence, which is perhaps unsurprising given the notable development in neural substrates implicated in attentional processing (Abundis-Gutiérrez et al., 2014; Flores et al., 2010; Konrad et al., 2005). Using neurocognitive tasks involving valid and invalid cues that direct a participant correctly or incorrectly to the location of a target (i.e., the Posner paradigm and Attentional Network Task; Petersen and Posner, 2012; Posner, 1980; Fan et al. 2002), multiple neuroimaging studies have detected neural networks important for attentional reorientation. For example, research using fMRI has repeatedly implicated the superior parietal lobe, the temporoparietal junction, and the intraparietal sulcus in the process of attentional reorientation (Petersen and Posner, 2012; Posner, 2012; Vossel et al., 2006). Along with parietal activation, frontal cortices, including the frontal eye fields (FEF) as well as inferior and middle frontal gyri, have been shown to play a strong role in orienting and reorienting (Petersen and Posner, 2012; Posner, 2012; Vossel et al., 2006).

Similar findings have been demonstrated using magneto- and electroencephalographic (M/EEG) methods probing the temporally-sensitive dynamics (e.g., Gómez et al. 2008). For instance, in healthy adult populations, neural oscillatory activity related to attentional reorienting in the theta (4–8 Hz), alpha (8–14 Hz), and beta (14–24 Hz) bands has been detected in the superior parietal and intraparietal sulcus, along with the FEF, inferior frontal gyrus, and prefrontal cortices (Arif et al., 2020a; Proskovec et al., 2018a). Generally speaking, adult studies tend to show stronger increases in theta power (i.e., stronger oscillations) during attentional reorientation, and more dynamic conditional differences in alpha and beta power that vary by time and location (e.g., Proskovec et al. 2018). Among children and adolescents, recent work has identified that the neural dynamics underlying attentional reorientation are developmentally sensitive. Specifically, during reorienting, youths showed stronger theta activity across multiple areas of the right prefrontal cortex and stronger alpha/beta responses over the left motor cortex and right cuneus as a function of increasing age (Picci et al., 2023). These findings compliment a breadth of literature reporting age-related alterations in attention- and high order cognition-related neural dynamics (e.g., Killanin et al. 2020, Taylor et al. 2021, 2020), suggesting that these oscillatory responses are robustly sensitive to developmental processes during childhood and adolescence.

The maturational sensitivity of the functional brain dynamics serving attentional processing is of great importance when considering that youths are regularly exposed to a wide array of environmental toxins that could impact these developmental trajectories (Bearer, 1995; Perera et al., 2006). A growing literature evidences a myriad of negative outcomes in the neurodevelopment of children connected to various exposures, ranging from lead and other toxic metals (Sanders et al., 2015; Silver et al., 2016) to inhaled air pollutants (Chiu et al., 2016; Peterson et al., 2015). For instance, studies have demonstrated a relationship between increased exposure to a broad range of toxins and increased risk for anxiety and mood disorders (Bornschein et al. 2006), ADHD (Myhre et al., 2018), and learning and neurocognitive disorders (Sanders et al., 2015). Even low level toxin exposure has been related to impaired cognitive functioning and lowered intelligence in youth, including decreased attention and memory functioning (Chiu et al., 2016; Liu and Lewis, 2014; Perera et al., 2006; Sanders et al., 2015). Despite this growing body of research detailing the neurocognitive impacts of environmental toxins on the developing brain, the effects of some of the most common household toxins have not yet been characterized, most notably radon.

Radon is a ubiquitous gas that forms from the decay of naturally occurring uranium (Clement et al., 2010; Darby et al., 2005; Kang et al., 2019; Vogeltanz-Holm and Schwartz, 2018), and it can accumulate to hazardous levels in homes (Laquatra and Laquatra, 2018; Riudavets et al., 2022; Sethi et al., 2012; Vogeltanz-Holm and Schwartz, 2018). In fact, 1 in 15 homes in the United States is estimated to have radon levels exceeding the action limit recommended by the Environmental Protection Agency, which is an indoor concentration equal to or greater than 4.0 pCi/L (United States Environmental Protection Agency, 2016). Further, in the state of Nebraska, more than half of radon tests have results above 4.0 pCi/L (NDHHS, 2024). Despite the guidance from the EPA and Department of Health and Human Services, the public’s knowledge of radon and its hazardous effects is severely lacking. Past research has linked radon exposure to multiple health concerns, including lung cancer (Sethi et al., 2012; Vogeltanz-Holm and Schwartz, 2018) and neurodegenerative diseases in adulthood (Gómez-Anca and Barros-Dios, 2020; Zhang et al., 2022). Further, recent work in children demonstrated robust links between home radon exposure and increases in specific biomarkers of inflammation (Taylor et al., 2022) that are regularly associated with detriments to neurocognitive development in youths (Ehrlich et al., 2021; Loftis et al., 2020; Miller et al., 2009). Despite such clear connections between radon exposure and health consequences ranging from chronic inflammation to cancer and neuro-degenerative disease, few studies have investigated other deleterious effects that radon may have, including in the realm of neurocognitive development. Particularly lacking is research involving youths, despite children being highly vulnerable to the effects of toxic exposures, and thus potentially more vulnerable to the damaging effects of radon (Kendall et al., 2021).

In the present study, we investigated the impact of home radon exposure on the neural oscillatory dynamics serving attentional reorienting in typically developing children and adolescents. To measure the neural processing underlying attentional reorienting, youths completed a classic Posner paradigm during high-density MEG. Home radon test kits were used to measure indoor radon concentrations in the family home. We hypothesized that the effects of radon exposure would be most notable in the theta, alpha, and beta range given their known developmental sensitivity in this paradigm and other higher order cognitive tasks. We predicted that chronic home radon exposure would be related to aberrations in task performance and neural oscillations within attention networks, such that children with higher radon exposure would exhibit both behavioral and neural decrements in attentional reorienting. Further, given the known exceptional sensitivity of these developing networks to environmental toxins, we explored potential interactions between age and radon exposure to see whether radon may be modulating expected developmental trajectories of these neural dynamics.

2. Methods and materials

2.1. Participants

We studied 56 healthy children aged 6 to 14 years old (mean: 10.90 ± 2.38, 22 males, 52 right-handed) who were recruited from the local community. The sample comprised a subset of children recruited for an ongoing NIH-funded longitudinal study (R01-MH121101). All participants were typically developing, without any history of head trauma, neurological or psychiatric disorders, or other conditions affecting brain function. Participants were excluded according to general MEG/MRI exclusionary criteria such as the presence of metal implants, dental braces, permanent retainers, and/or any type of ferromagnetic nonremovable devices. Inclusion/exclusion were confirmed through participant interviews involving the child and parent. After a complete description of the study, written informed consent was obtained from the legally authorized representative of each participant, and participants provided assent. All procedures were approved by the Institutional Review Board.

2.2. Radon data collection

Families were provided with a commercial short-term home radon testing kit (https://www.radon.com/). The test kit was a standard carbon-based envelope that hangs on an interior wall on the lowest livable level of the home for three to seven days. After the testing period, the envelope is sealed and dropped in the mail for processing at the commercial lab. Parents were given the test kit along with instructions from the commercial vendor for proper exposure. We instructed families to leave the kit exposed for approximately four days. Our lab and the family each received a copy of the home radon results. In the case that a result exceeded the EPA action limit for mitigation (4 pCi/L), the principal investigator (BKT) contacted the family to ensure they understood the results and provided additional information on radon safety and local resources.

In addition to completing the radon test kit, parents completed a questionnaire probing information about how long the child had lived in the home, information about the construction of their home, and other details to help characterize home radon exposure. We used the information about how long the child had lived in the home to compute a radon exposure index. Specifically, each child’s individual radon exposure index was computed as the natural log of the measured home radon concentration (in pCi/L) multiplied by the duration that the child had lived in the home (in years), plus one to account for the natural log transform (see Eq. (1) below; Taylor et al., 2022, 2024). This radon exposure index was used in subsequent analyses exploring the associations between chronic home radon exposure and neural dynamics.

$$\text{Radon exposure index}=\ln (([\text{time lived in home}]\ast [\text{home radon concentration}])+1)$$ (1)

2.3. Experimental paradigm

Participants performed a modified Posner cued-attention task (Fig. 1; Posner, 1980) during MEG recording. Participants were instructed to maintain fixation on a centrally presented crosshair throughout the task. Each trial began with the presentation of only the crosshair for 1750 ms (±250 ms). Next, a green bar, serving as the cue, was presented either to the left or right of the crosshair for 100 ms. This cue was presented on each side (left or right) an equal number of times, and could be either valid (i.e., presented on the same side as the subsequent target; 50 % of all trials) or invalid (i.e., opposite side relative to the target). After 100 ms, the cue disappeared and 200 ms later (i.e., 300 ms after cue onset) the target stimulus appeared on either the left or right side of the crosshair for 1200 ms. The target consisted of a box with an opening on either the bottom (50 % of trials) or top surface. Participants were instructed to respond as to whether the opening was on the bottom (right index finger) or the top (right middle finger) of the box. Each target variant appeared an equal number of times on the left and right sides of the crosshair and was preceded by an invalid or valid cue an equal number of times. Each trial lasted 3250 ms (±250 ms) and there was a total of 200 trials (100 valid, 100 invalid), resulting in a total run time of approximately 11 min.

Fig. 1.

Posner cueing task and epoch definition. A fixation cross was presented for 1750 (±250) ms, followed by a cue (green bar) presented in the left or right visual hemifield for 100 ms. After 200 ms, the target stimulus (box with opening) appeared in either the left or right visual hemifield for 1200 ms. Participants responded as to whether the opening was on the bottom or top of the target with their index and middle fingers, respectively. The cue was valid (presented on the same side as the subsequent target) 50 % of the time. To evaluate the responses involved in attentional reorientation (cue-locked), the neuromagnetic data were defined with the onset of the cue as 0 ms and the baseline was defined as −700 to −100 ms preceding cue onset.

2.4. MEG data acquisition

Recordings were conducted in a one-layer magnetically shielded room with active shielding engaged. With an acquisition bandwidth of 0.1–330 Hz, neuromagnetic responses were sampled continuously at 1 kHz using an Elekta MEG system with 306 magnetic sensors (Elekta, Helsinki, Finland). MEG data from each participant were individually corrected for head motion and subjected to noise reduction using the signal space separation method with a temporal extension (tSSS; Taulu et al., 2005; Taulu and Simola, 2006). Briefly, preceding MEG recording, four coils were attached to the participant’s head and localized, together with the three fiducial points and scalp surface, with a 3D digitizer (Fastrak 3SF0002; Polhemus Navigator Sciences, Colchester, VT). Once the participant was positioned for MEG recording, an electric current with a unique frequency label (i.e., 322 Hz) was fed to each of the coils. This induced a measurable magnetic field and allowed each coil to be localized in reference to the sensors throughout the recording session. Total drift (i.e., the overall distance that any individual’s head moved throughout the session) was quite minimal in the present study (median = 0.58 cm, M = 0.88 cm, SD = 0.79, range = 0.07 to 3.56 cm) and was effectively accounted for using tSSS.

2.5. Structural MRI acquisition, processing, and coregistration with MEG data

Since the head positioning coil locations were also known in head coordinates based on the 3D digitizer information acquired prior to recording (see Sections 2, 4), all MEG measurements could be transformed into a common coordinate system. With this coordinate system, each participant’s MEG data were coregistered with their structural T1-weighted neuroanatomical data prior to source space analyses using BESA MRI (Version 2.0; BESA GmbH, Gräfelfing, Germany). Structural T1-weighted MR images were acquired using a Siemens Prisma 3 T MRI scanner with a 32-channel head coil and a MP-RAGE sequence with the following parameters: TR =2400 ms; TE =1.94 ms; flip angle = 8°; FOV = 256 mm; slice thickness = 1 mm (no gap); voxel size = 1 × 1 × 1 mm. All structural MRI data were aligned parallel to the anterior and posterior commissures and transformed into standardized space, along with the functional images, after beamforming.

2.6. MEG time–frequency transformation and statistics

Cardiac and ocular artifacts (e.g., blinks, eye movement) were removed from the data using signal-space projection (SSP), which was accounted for during source reconstruction (Uusitalo and Ilmoniemi, 1997). MEG data were then analyzed with respect to the attentional cue to evaluate the oscillatory dynamics associated with attentional reorientation. First, the continuous magnetic time series was divided into epochs of 3000 ms duration, with the onset of the cue being defined as 0 ms and the baseline being defined as −700 to −100 ms preceding cue onset. Given our task and epoch design, the target onset occurred at 300 ms. Note that the baseline period was selected to prevent motor responses from the previous trial from “contaminating” the baseline. Epochs containing artifacts were rejected based on a fixed threshold method, supplemented with visual inspection. In brief, for each individual, the distribution of amplitude and gradient values was computed across all trials, and those trials containing the highest amplitude and/or gradient values relative to the full distribution were rejected by selecting a threshold that excluded extreme values. Importantly, these thresholds were set individually for each participant, as interindividual differences in variables such as head size and proximity to the sensors strongly affects MEG signal amplitude. Additionally, we visually inspected the data to identify trials contaminated with other types of artifacts, such as those produced by muscle tension, and rejected such trials. On average, 64.86 valid and 62.96 invalid trials per participant remained after artifact rejection. We determined whether the number of trials remaining varied by condition, or as a function of any of the predictors of interest (i.e., age, radon exposure, and their interaction) using a repeated measures ANCOVA (2-way within-person factor of Condition: valid, invalid). We did detect a main effect of age (F(1, 46) = 6.03, p = .02, ${\mathrm{\eta }}_{\mathrm{p}}^2=0.12$), such that older children tended to have more trials than younger children. There were no other significant main effects or interactions by condition (ps = 0.28 to 0.55). Because of this potential confounding effect of the number of trials on age-related effects of interest, we do covary the square root of the number of trials per person in all subsequent analyses of neural dynamics to adjust for differences in signal-to-noise ratio that may be present in the data (Gross et al., 2013; Junghöfer et al., 2000; Vincent, 1992).

Artifact-free epochs were transformed into the time–frequency domain using complex demodulation with a resolution of 1 Hz and 50 ms, and the resulting spectral power estimations per sensor were averaged across all trials to generate time–frequency plots of mean spectral density. These sensor-level data were then normalized with respect to baseline power, which was calculated as the mean power per spectral bin between −700 and −100 ms prior to cue onset. Of note, this normalization was performed separately for each 1 Hz by 50 ms bin within each sensor-level spectrogram using the corresponding baseline data.

The time–frequency windows used for imaging were determined by statistical analysis of the sensor-level spectrograms across all trials (valid + invalid), gradiometers, and participants. Each data point (i.e., 1 Hz by 50 ms bin) in the spectrogram was initially evaluated using a mass univariate approach based on the general linear model. To reduce the risk of false positive results while maintaining reasonable sensitivity, a two-stage procedure was followed to control for Type 1 error. In the first stage, paired-sample t-tests against baseline were conducted on each data point and the output spectrogram of t values was thresholded at p < .05 to define time–frequency bins containing potentially significant oscillatory deviations across all participants. In stage two, time–frequency bins that survived this threshold were clustered with temporally and/or spectrally neighboring bins that were also significant, and a cluster value was derived by summing all of the t values of all data points in the cluster. Nonparametric permutation testing was then used to derive a distribution of cluster values and the significance level of the observed clusters (from stage one) were tested directly using this distribution (Ernst, 2004; Maris and Oostenveld, 2007). For each comparison, 1000 permutations were computed to build a distribution of cluster values. Based on these analyses, only the time–frequency windows that contained significant oscillatory events across all trials were subjected to the beamforming (i.e., imaging) analysis. Thus, a data-driven approach was utilized for selecting the time–frequency windows to be imaged.

2.7. MEG source imaging and statistics

Cortical networks were imaged through an extension of the linearly constrained minimum variance vector beamformer (Gross et al., 2001; Hillebrand et al., 2005; Van Veen et al., 1997), which applies spatial filters to time–frequency sensor data to calculate voxel-wise source power for the entire brain volume. Such images are typically referred to as pseudo-t maps, with units (pseudo-t) that reflect noise-normalized power differences (i.e., active vs. passive) per voxel. Following convention, the source power in these images was normalized per participant using a separately averaged pre-stimulus noise period (i.e., baseline) of equal duration and bandwidth (Hillebrand et al., 2005). MEG preprocessing and imaging used the Brain Electrical Source Analysis (version 7.0) software.

Normalized source power was computed for the selected time–frequency bands over the entire brain volume per participant at 4.0 × 4.0 × 4.0 mm resolution. Each participant’s functional images were transformed into standardized space using the transform that was previously applied to the structural images and then spatially resampled. The resulting 3D maps of brain activity were examined for potential outliers, identified as maps containing clusters with pseudo-t values exceeding ±3.0 standard deviations about mean. Such outliers are often the result of artifacts that could not be removed in our rigorous preprocessing scheme and interfere with accurate source reconstruction during beamforming. The remaining maps after outlier exclusions were grand averaged across both conditions and participants to assess the anatomical basis of the significant oscillatory responses identified through the sensor-level analysis. To identify the effect of cue validity on neural oscillatory responses, we created validity effect maps by subtracting the valid whole-brain maps from the invalid whole-brain maps.

To assess associations between chronic everyday home radon exposure, age, and neural oscillatory dynamics, we computed whole-brain linear regressions (one for each of the oscillatory bands of interest) with the radon exposure index, age, and their interaction as predictors of interest, and the validity effect map of each oscillatory band as the dependent measure of interest. Signal-to-noise ratio, computed as the square-root of the total number of trials per person, was included as a covariate of no interest to adjust for potential differences in the number of trials across individuals. To account for multiple comparisons, a significance threshold of at least p < .005 was used for the identification of significant clusters in all whole-brain statistical maps, accompanied with a stringent cluster (k) threshold of at least 5 contiguous voxels of 4 × 4 × 4 mm (i.e., at least 320 mm³). These analyses, performed in SPM 12, yielded F maps showing significant clusters with main effects of age, radon exposure, and their interaction. From these significant clusters, pseudo-t values were then extracted from the peak voxels of each significant cluster per participant. We report the standardized beta coefficients for significant associations identified in F maps for interpretability of the effects. Radon exposure-by-age interactions were decomposed post-hoc by splitting the sample in half by age (i.e., younger versus older youth) and exploring the correlations between radon exposure and the pseudo-t values extracted from the peak of each cluster. Of note, because the current investigation is focused on the effects of radon exposure, only the main effects of radon and the radon-by-age interaction are reported in the main text. Main effects of age are reported in the Supplementary Materials.

Finally, we conducted a follow-up analysis to examine whether the significant neural oscillatory effects identified in the primary analysis were related to behavioral task performance. Specifically, we tested whether oscillatory activity measured at the peak of each identified cluster in the primary analysis mediated the relationship between chronic radon exposure or the radon-by-age interaction as applicable, and task performance (i.e., reaction time and accuracy). Because traditional tests of indirect effects (e.g., the Sobel test) often violate the assumption of normality, we utilized asymmetrical confidence intervals which best represent the true distribution of the indirect effect (i.e., the product of coefficients from the “a” and “b” paths). Thus, we examined the 95 % confidence intervals of bias-corrected bootstrapped confidence intervals based on 1000 bootstrapped samples to more robustly detect any potential relationships between brain activity and behavior (Austin and Tu, 2004; Efron and Tibshirani, 1986; Fritz and MacKinnon, 2007), which provide a robust estimate of mediation effects and are asymmetrical (Fritz and MacKinnon, 2007). Mediation analyses were conducted in JASP version 17.1.

3. Results

3.1. Behavioral analysis

Six participants were excluded from all analyses due to low accuracy (less than 50 % correct) in the Posner task. The remaining 50 participants performed well, accurately responding to an average of 81.90 % (SD = 11.82 %) of the valid trials and 78.88 % (SD = 12.39 %) of the invalid trials. Using two repeated measures ANCOVAs (2-level within person factor of Condition: valid, invalid), we determined whether there were any associations between the measures of interest (i.e., age, radon exposure, and their interaction) and both accuracy and reaction times during the task. We found a main effect of age on accuracy (F(1, 46) = 8.72, p = .005, ${\mathrm{\eta }}_{\mathrm{p}}^2=0.16$), and on reaction time (F(1, 46) = 5.09, p = .03, ${\mathrm{\eta }}_{\mathrm{p}}^2=0.10$), such that older children generally performed better and responded faster than younger children. There were no other main effects or interactions (accuracy: ps = 0.32 to 0.93; reaction time: ps = 0.20 to 0.88).

3.2. Home radon exposure

On average, the radon kits were exposed in participants’ homes for 115.32 h (SD = 27.36, range =72–176 h). The raw result of the radon tests ranged from 0.3 to 33.3 pCi/L with an average of 7.12 pCi/L (SD = 7.71; median = 4.30 pCi/L; Skewness = 2.03, SE = 0.34; Kurtosis = 3.91, SE = 0.66). Only 44 % of the total sample had home radon concentrations below the EPA action limit of 4.0 pCi/L. Every family also reported how long the child had lived in their current home. Each child had been exposed to the recorded radon level for an average of 3.83 years (SD = 3.39). As stated in the Methods, we computed a radon exposure index for each participant by combining the recorded home radon concentration and the duration of time that the child had lived in their home. The resultant radon exposure index had an average of 2.49 (SD = 1.31), ranging from 0 to 5.94.

3.3. Sensor-level analysis

Statistical analysis of the cue-locked time-frequency spectrograms revealed significant clusters of theta (3–7 Hz), alpha (9–13 Hz), and beta (14–24 Hz) oscillatory activity in gradiometers near the occipital and parietal cortices across all participants and conditions (p < .05, corrected; Fig. 2). Significant theta activity began shortly after the onset of the target stimulus (300 ms = target onset) and tapered off about 500 ms later (i.e., from 300 to 800 ms). In both the alpha and beta range, significant activity emerged roughly 100 ms after target onset and continued for about 500 ms before dissipating (i.e., from 400 to 900 ms).

Fig. 2.

Spectrograms across all trials (valid and invalid). Time-frequency decomposition and permutation-corrected statistical analyses indicated three time-frequency bins with significant responses (p < .05, corrected) relative to baseline (marked by black boxes). These included beta activity (14–24 Hz) and alpha activity (9–13 Hz) from 400 to 900 ms, as well as theta activity (3–7 Hz) from 300 to 800 ms. The statistical analyses included all gradiometers, with the sensors most clearly showing the responses shown above (i.e., MEG1923 for alpha and beta, and MEG0713 for theta).

3.4. Beamformer analysis

To identify the brain regions generating the significant sensor-level oscillations, these time frequency windows were imaged using a beamformer for each condition (valid, invalid) and across all trials combined. These source level images were then examined for outliers, with significant artifactual or excessively noisy data excluded. This left 45 participants with evaluable data in the beta band, 42 in alpha, and 39 in theta. Importantly, the number of children included in analyses did not significantly differ by time-frequency band assessed (χ² = 3.73, p = .16). The resulting maps were grand averaged across participants, and these responses are shown in Fig. 3. Strong oscillations were observed across all trials in the beta band (14–24 Hz) from 400 to 900 ms in the bilateral visual cortices and left parietal. Additionally, strong oscillations were seen in the alpha band (9–13 Hz) from 400 to 900 ms in bilateral visual cortices. In contrast, a strong theta response (3–7 Hz) appeared across the frontal and parietal cortices during the 300 to 800 ms window. In general, across all three cortical oscillatory bands, greater recruitment of cortical areas was evident in the invalid relative to the valid condition. This result was expected and is consistent with previous findings (Picci et al., 2023).

Fig. 3.

Source reconstructions by trial type. Grand-averaged source reconstructions were made per time-frequency bin across participants. (Left panel) Source reconstructions for combined conditions (all trials) showed alpha and beta oscillations largely in the parietal and occipital cortices and theta activity in a widespread cortical network spanning frontal and parietal cortices. (Right panels) Source reconstructions per trial type (valid, invalid) were averaged across participants. Differences between conditions can be seen as a general increase in the strength and extant of activity during the invalid condition across all three imaged oscillatory bands.

To determine the effect of home radon exposure on the oscillatory activity underlying attentional reorientation, subtraction maps were first calculated by subtracting whole-brain voxel-wise validly cued maps from invalidly cued maps. The resulting “validity effect” maps for each participant were then subjected to whole-brain linear regressions with age, radon, and the age-by-radon interaction as predictors of interest. Signal-to-noise ratio, computed as the square-root of the number of included trials per person, was included as a covariate of no interest in all analyses. In the following sections, we report the main effects of radon and the radon-by-age interaction effects. Main effects of age are reported in the Supplementary Materials.

3.5. Main effects of radon

Our whole-brain linear regressions revealed multiple main effects of radon on the neural oscillatory dynamics serving attentional reorienting, controlling for the effects of age and signal-to-noise ratio. All main effects are illustrated in Fig. 4, and the statistical details for each cluster are reported in Table S2. In the theta band, there was a main effect of radon exposure index in the left dorsolateral prefrontal cortex (dlPFC; β = −0.474, p < .001), such that there was a stronger reverse validity effect in children with greater radon exposure. As radon exposure increased, the theta response in the valid condition increased (β = 0.405, p = .011), whereas power in the invalid condition remained relatively stable (β = −0.211, p = .203). Additionally, we identified a main effect of radon exposure on alpha oscillations in the right cerebellum (β = −0.530, p < .001). Here, participants with higher radon exposure showed a stronger validity effect than those with lower exposure. Lastly, in the right anterior cingulate cortex (ACC), greater radon exposure was associated with a stronger reverse validity effect in the beta band (β = 0.416, p = .003). This effect was driven by a weaker beta response as a function of radon during invalid trials (β = 0.299, p = .023), while beta power did not vary as a function of radon for the valid trials (β = −0.185, p = .220).

Fig. 4.

Main effects of radon exposure on the neural validity effect within the theta (top row), alpha (middle row), and beta (bottom row) bands. (Left) Statistical F maps showing clusters for which radon exposure was significantly related to the neural validity effect: left dlPFC in theta, right cerebellum in alpha, and right ACC in beta. (Right) Scatterplots showing the associations between radon exposure and the neural validity effect. All scatterplots show residualized pseudo-t values after regressing out the effects of age and signal-to-noise ratio (defined as the square root of the total number of trials included in the analysis). All reported β values are standardized. Note: “ACC” = anterior cingulate cortex; “DLPFC” = dorsolateral prefrontal cortex; “L” = left; “R” = right.

3.6. Radon-by-age interaction

We also found multiple regions where age and radon exposure had an interactive effect on the oscillatory dynamics underlying attentional reorienting. All statistical details for each cluster that survived our strict thresholding can be found in Table S3. In the alpha band, our whole-brain linear regression revealed significant age-by-radon interactions in the right inferior parietal lobule (IPL) and the left cuneus (Fig. 5). In the right IPL, older youth exhibited a stronger reverse validity effect as a function of increased radon exposure (r = −0.544, p = .011). Conversely, in the left cuneus, younger children tended to show an increased validity effect as radon exposure increased (r = 0.646, p = .002). Notably, older youth with minimal radon exposure showed almost no recruitment of this region; however, recruitment of the region increased indiscriminately during both valid and invalid trials with increasing radon exposure (rs = −0.466 and −0.470, p = .033 and 0.032, respectively).

Fig. 5.

Age-by-radon interaction effect on the neural validity effect within the alpha band. Statistical F maps showing the left cuneus (Left) and right IPL (Right) clusters in which age and radon had a significant interactive effect on the neural validity effect. Note: “IPL” = inferior parietal lobule; “L” = left; “R” = right.

In the beta band, we found significant age-by-radon interaction effects in the right superior temporal gyrus (STG; Fig. 6A, B), right post-central gyrus, and right parahippocampal gyrus. In these regions, younger children showed stronger reverse validity effects as a function of increasing radon exposure (rs = 0.530 to 0.602, ps = 0.003 to 0.011), whereas older children did not show any significant change in their neural validity effect as a function of radon exposure (rs = −0.365 to −0.219, ps = 0.086 to 0.315). Notably, in all cases, younger children exhibited greater recruitment of these regions (i.e., stronger beta responses) during valid trials as a function of increasing radon exposure (rs = −0.682 to −0.483, ps < 0.001 to 0.034). Similarly, we noted an age-by-radon interaction effect in the left STG (Fig. 6D, E) where older youth exhibited a stronger reverse validity effect as a function of increasing radon exposure (r = 0.616, p = .002). Here, older youth with greater radon exposure tended to show less recruitment of this region (i.e., weaker beta responses) during invalid trials (r = 0.587, p = .003), whereas younger children showed the opposite pattern (r = −0.591, p = .004).

Fig. 6.

Age-by-radon interactions on the neural validity effect within the beta band. A,D: Statistical F maps showing representative clusters for which there was a statistically significant age-by-radon exposure interaction. B,C, and E: Scatterplots showing the differential association between radon exposure and the neural validity effect for older versus younger youths within the right STG (B), left IPL (C), and left STG (E). The scatterplots show residualized pseudo-t values after regressing out the effect of signal-to-noise ratio (defined as the square root of the total number of trials included in the analysis). All reported β values are standardized. Notes: “IPL” = inferior parietal lobule; “L” = left; “PFC” = prefrontal cortex; “R” = right; “STG” = superior temporal gyrus.

In contrast, we did detect several instances in which greater radon exposure was associated with stronger neural validity effects in specific age groups. For instance, older children exhibited stronger neural validity effects in the right lateral prefrontal cortex (PFC; Fig. 6A) and right intraparietal sulcus (IPS; Fig. 7) as a function of greater radon exposure (rs = −0.538 and −0.774, p = .008 and <0.001, respectively). These effects were driven by weaker beta responses in the lateral PFC during valid trials (r = 0.448, p = .032) and stronger beta responses in the IPS during invalid trials (r = −0.480, p = .024). We observed a similar effect in the left inferior frontal gyrus (IFG) where younger children exhibited a stronger validity effect with greater radon exposure (r = −0.682, p < .001). This effect was driven by stronger beta responses in this region during invalid trials (r = −0.510, p = .011).

Fig. 7.

Age-by-radon interaction effect on the neural validity effect within the beta band. (Left) Statistical F map showing the right IPS cluster in which age and radon had a significant interactive effect on the neural validity effect. (Right) Scatterplot showing the differential association between radon exposure and the neural validity effect within the right IPS between older and younger youths. The scatterplot shows residualized pseudo-t values after regressing out the effect of signal-to-noise ratio (defined as the square root of the total number of trials included in the analysis). All reported β values are standardized. Note: “IPS” = intraparietal sulcus; “R” = right.

Finally, the left IPL showed a relatively unique pattern, such that older children tended to show consistent recruitment of the region during invalid trials, but increasing recruitment (i.e., stronger beta responses) during valid trials as radon exposure increased (Fig. 6C, D). Conversely, younger children with minimal radon exposure showed almost no recruitment of the region, but recruitment increased during both valid and invalid trials with increasing radon exposure (rs = −0.382 and −0.621, ps = 0.079 and 0.002, respectively).

3.7. Relationships to behavior

Lastly, we performed a set of mediation analyses to identify relationships between home radon exposure, the neural validity effect, and behavioral metrics of task performance. Above, we found 10 clusters where the neural validity effect was significantly related to radon exposure. In this set of analyses, we investigated whether the neural validity effects in these regions mediated the relationship between radon exposure or the radon-by-age interaction, and task reaction time (RT) and accuracy validity effects (i.e., RT for invalid trials – RT for valid trials; accuracy for invalid – accuracy for valid trials). For each oscillatory response and effect, we tested a separate model including all significant clusters of response-related activity, controlling for age. Interestingly, we did not detect any statistically significant indirect effects whereby aberrations in neural dynamics mediated the relationships between radon exposure or the radon-by-age interaction and behavioral validity effects.

4. Discussion

The present study examined the impact of chronic home radon exposure on the neural oscillatory dynamics serving attentional reorientation during a dynamic period of neurodevelopment in youths. We found that radon exposure modulated oscillatory responses in the theta, alpha, and beta range across a distributed network of regions implicated in attentional processing. These oscillatory responses are known to be very sensitive to neurodevelopmental changes during childhood and adolescence (Killanin et al., 2020; Picci et al., 2023; Taylor et al., 2020, 2021). During this developmental window, the brain undergoes a multitude of structural and functional changes, and while this plasticity is essential for neurocognitive maturation, it also creates opportunities for environmental toxins to negatively impact proliferating and immature systems (Bearer, 1995; Perera et al., 2006). We have shown that neural oscillatory activity across a myriad of brain regions critical for attentional processing is coupled to radon exposure during this developmentally sensitive window, adding to the existing literature which points to the negative consequences of toxin exposure on neurodevelopment. Importantly, we extended these critical findings and revealed radon-related modulations to the developmental trajectory of neural processing serving attentional reorientation. We discuss each of our findings and their implications below.

We noted widespread moderating effects of radon exposure on the neural oscillatory dynamics serving attentional reorienting. To better understand the impact of radon exposure, it is first important to contextualize the expected trajectory of development in these attentional networks. A long-standing body of literature characterizing the normative development of attentional orienting has established that youths tend to exhibit a strong neural validity effect, but that throughout adolescence and into adulthood, this effect becomes progressively weaker (Brodeur and Enns, 1997; Konrad et al., 2005; Schul et al., 2003). This maturation-related nulling of validity effects is often interpreted as an indication that the neural substrates have become increasingly efficient and less susceptible to the types of conflict posed in tasks like the classical Posner cuing paradigm. That said, in the present study, we actually observed increasing validity effects as a function of radon exposure in a number of regions, some of which differentially impacted younger versus older youths. There were also some regions where we observed an opposing pattern of change as a function of radon exposure, namely a “reverse validity effect.” Such patterns were driven by stronger oscillatory responses in the valid condition as a function of increasing radon exposure, with minimal to no changes in power in the invalid condition as radon exposure increased.

Reverse validity effects have been seen in both developmental and aging samples (Arif et al., 2020a; Picci et al., 2023). For example, Arif et al. (2020a) found that throughout adulthood, increasing age was associated with reverse validity effects evident in the alpha band in the left superior parietal cortices extending into the IPS, as well as in the beta band in the FEF. They proposed older adults exhibited an inefficiency in visuospatial disengagement, having exhausted resources as cognitive demand increased when attentional reorientation was necessary (Arif et al., 2020a). This is in line with the compensation-related utilization of neural-circuits hypothesis (i.e., CRUNCH) which clarifies the way in which older adults compensate to match performance of their younger counterparts (Reuter-Lorenz and Cappell, 2008). Specifically, under low-cognitive demand, older adults increase recruitment of neural substrates to improve performance, but under higher cognitive demand, they reach a resource ceiling which results in inefficient processing (Reuter-Lorenz and Cappell, 2008). The same explanation may apply to our sample in that the youths who experienced greater radon exposure displayed an inefficiency in attentional reorientation. For the easier (valid) trials, youth with greater exposure tended to recruit more neural resources or increase recruitment of the same substrates in order to match the performance of youth who have less exposure. In the more difficult trials (invalid), cognitive demand increases with the need for attentional reallocation, and those with greater exposure were unable to meet this demand in the way their peers with lower exposure were.

Perhaps our most important findings were the age-by-radon interaction effects, which revealed aberrations in the normative development of the cross-spectral neural oscillatory dynamics supporting attentional reorienting in multiple regions within the orienting attention network (e.g., right STG, IPS, postcentral gyrus, lateral PFC, and IPL; Petersen and Posner, 2012). Within this network, we saw overall exacerbated validity effects as a function of increasing radon exposure. Generally, older youths, who we would expect to have more matured networks, exhibited stronger validity effects in multiple regions. This finding among older youth may indicate that older children who experienced greater radon exposure necessarily recruited regions we would typically expect to support attentional reorienting to a greater extent than their same-aged peers with less exposure in order to match their performance on the task. On the other hand, younger children exhibited a reverse validity effect within the orienting network. Given that their attentional networks are less matured, these younger children with greater radon exposure may have exhausted their resources early on and were unable to support the additional recruitment necessary for this type of reorienting due to their higher radon exposure.

Outside of the classically-defined orienting network, we also observed age-by-radon interaction effects in the left IFG and IPL, two areas commonly implicated in the executive control attention network (Petersen and Posner, 2012). Interestingly, the executive control and orienting networks are often difficult to disentangle in youth (Taylor et al., 2018) and are believed to diverge into functionally and anatomically distinct networks across later adolescent development (Petersen and Posner, 2012). In both impacted regions, we detected age-by-radon interactions that were largely driven by radon-related increases in neural recruitment among younger children. The data may suggest compensatory recruitment of the executive control attention network among youth exposed to greater degrees of environmental toxins. The heightened plasticity of the cross-coupled, undifferentiated executive control and orienting networks in this developmental range may be critically sensitive to environmental exposures, yielding these aberrant patterns of compensatory oscillatory activity.

Outside of these two attention networks, we found significant age-by-radon interactions in the right parahippocampal area and the left cuneus. Younger children exhibited robust recruitment of these regions during the task, with an increase in recruitment for valid trials as a function of increasing radon exposure. This reverse validity effect in younger children further supports a compensatory response, as mentioned above. Conversely, older youths seemed to show either consistent, as in the parahippocampal gyrus, or increasing, as in the cuneus, recruitment as a function of radon exposure. They did not show the same exhaustion of resources as their younger counterparts; rather, older youth showed a pattern of necessary and increasing engagement with more toxic exposure. Several previous studies of attentional reorientation have reported validity effects in the parahippocampal area (Gómez et al., 2008; Vossel et al., 2006). Research suggests that parahippocampal activity related to attentional reorientation during invalid trials may be connected to revision of the task model due to the critical role of the hippocampus and parahippocampal area in forming transient memory traces of the characteristics within an environment (Gómez et al., 2008). Similarly, this region has been implicated as part of a network of regions supporting contextual updating and spatial processing (Aminoff et al., 2013; Kveraga et al., 2011). Local and long-range aberrations in functional and structural connectivity involving the parahippocampal gyrus have been linked to impairments in efficient attentional allocation and ability to generate predictions and expectations of upcoming events within a person’s environment (Huang et al., 2018; Peterson et al., 2011; Xiao et al., 2016). In regard to the cuneus, many studies of spatial attention, especially in relation to validity and interference effects have implicated this region (Arif et al., 2020b; Picci et al., 2023; Taylor et al., 2021). In other studies of attentional reorientation, alpha/beta oscillations within the right cuneus varied based on the validity of the cue (Picci et al., 2023), as well as personal factors such as age and chronic disease (Arif et al., 2020b), suggesting that attention-related oscillatory dynamics in this neural substrate are highly sensitive to a multitude of internal and external factors. Our data add to this rich literature, implicating chronic radon exposure as another potential modulator of neural dynamics in the cuneus, though the exact mechanism by which radon impacts activity in this region is unknown.

Finally, we detected multiple regions which exhibited a main effect of radon exposure. In the beta band, a radon-related reverse validity effect was evident in the ACC, a region which plays a critical role in monitoring and managing cognitive conflict (Gómez et al., 2008) and is implicated in the executive control network (Konrad et al., 2005; Petersen and Posner, 2012). Prior studies on selective attention tasks involving interference have found the ACC to show interference and age effects throughout development (Corbetta and Shulman, 2011; Gómez et al., 2008; Petersen and Posner, 2012; Taylor et al., 2021). Stronger alpha and beta oscillations have been related to better task performance in visuospatial attention and spatial working memory tasks (Proskovec et al., 2019, 2018b), allowing for efficient processing of the relevant stimuli while inhibiting distractions, in order to flexibly orient attention where necessary (Diepen et al., 2016; Embury et al., 2019; Petro et al., 2019; Picci et al., 2023; Proskovec et al., 2019; Wilson et al., 2017). Thus, it is possible that the youths with greater exposure in our sample required stronger beta responses to sustain performance, though this was unique to activity during the easier trials, which could reflect exhaustion of resources in the more difficult invalid condition.

We also found decreasing or reverse validity effects in the theta band within the left dlPFC as a function of increasing home radon exposure. Theta oscillations, especially in frontal regions, play a key role in allocation of spatial attention across the lifespan (Conejero et al., 2018; Picci et al., 2023; Rajan et al., 2019; Spooner et al., 2020) and are critical in long-range neuronal communication and coordinating information processing and transfer (Colgin, 2013; Herrmann et al., 2016; Wiesman et al., 2017). Thus, radon-related disruptions to this region may explain a potential mechanism by which radon exposure impacts the development of attentional processes. The left dlPFC has been implicated in the dorsal attention network (Corbetta and Shulman, 2011; Vossel et al., 2014), and is important for processes of spatial attention and reorienting, additionally playing a multitude of roles in higher order cognition (Corbetta and Shulman, 2011; Picci et al., 2023; Taylor et al., 2020). MEG studies have found theta-specific effects within bilateral dlPFC that are related to attentional interference and show developmental sensitivity and sex-specific differences (Picci et al., 2023; Taylor et al., 2020, 2021). Further, effects in dorsal prefrontal regions were found in a study of childhood toxin exposure where white matter measures in the dlPFC were significantly related to polycyclic aromatic hydrocarbon exposure in children (Peterson et al., 2015). Considering research connecting white matter abnormalities with radon exposure and behavioral or structural deficits (Gómez-Anca and Barros-Dios, 2020; Groves-Kirkby et al., 2016; Zhang et al., 2022), this may be a potential mechanism by which radon impacts neural function within this region.

Before closing, we must note several limitations of the current study. Firstly, home radon concentrations were assessed through short-term radon test kits which measured for a three- to seven-day period. Generally, the kits utilized are reliable, but they are susceptible to inaccuracies due to factors such as damp conditions in the location of the test, open doors or windows during the exposure period, and adverse weather at the time of testing. Studies suggest longer-term kits which are exposed for 30 days to one year can more accurately measure radon concentrations within a home, and these tests tend to be more robust against weather-based or seasonal variations (Novilla et al., 2021). Secondly, the test kits used within this study only assess current home radon concentrations. Although we approximated cumulative exposure based on each person’s current home radon concentration and the amount of time they had lived in that home, there may be better ways to assess lifetime exposure. This is especially true for children who have lived in multiple homes for which the radon concentration is unknown. Future studies could approximate exposure prior to the current home by using average radon concentrations for the zip code areas where children previously resided. It is also possible to use glass-based measurements of historical radon exposure if the family can provide a sample of glass which has been in the home for the duration of the child’s lifetime (Mahaffey et al., 1993; Samuelsson, 1988), though these estimates would likely be far less specific than individual home measurements. Another limitation is that we only measured radon concentrations in homes. Even though the home is the greatest source of radon exposure for most people (Kendall et al., 2021), future works may consider accounting for other locations where children may be exposed to radon by testing other commonly occupied facilities (e.g., schools) or accounting for how much time the child usually spends in the home each day. Lastly, the present study employed a cross-sectional design in a relatively small sample, but a longitudinal design with a larger study population would allow for clearer linkages between chronicity of radon exposure and individual trajectories of change in the neural dynamics serving attentional reorienting, as well as exploration of potential sex effects.

5. Conclusion

To conclude, the findings of the present study contribute to a growing body of work investigating the consequences of toxin exposure on neurodevelopment, specifically investigating the impacts of home radon exposure on neural indices of attention in developing youths. We found significant relationships between radon exposure and neural oscillations serving attentional reorientation in nodes of multiple neurocognitive brain networks and substrates implicated in attentional orienting and executive control. Importantly, our findings point to unique patterns of compensatory neural oscillatory activity across this dynamic and sensitive developmental window of childhood into midadolescence that align with well-established theories of exhausted neural resources (i.e., CRUNCH). These data provide key insight on the impact that home radon exposure may have on the ability of youth to effectively attend to and adapt to changes within their environment, at least within this sample of typically-developing youth. As such, future works should explore the degree to which radon exposure is associated with clinically-relevant psychological symptoms including inattention and hyperactivity.

Data availability

Data will be made available on request.