Mid Brain Activation Course For Adults
Geppe Warton
We exploit novel data from brain-training games to examine the impacts of air pollution on a comprehensive set of cognitive skills in adults. We find that exposure to particulate matter (PM2.5) impairs adult cognitive function, and that these effects are largest for those in prime working age. These results confirm a hypothesized mechanism for the impacts of air pollution on workforce productivity. We also find that the cognitive effects are largest for new tasks and for those with low ability, suggesting that air pollution increases inequality in productivity.
We thank our editor (Daniel Millimet), anonymous reviewers, Karen Clay, Jonathan Colmer, Tatyana Deryugina, Nick Kuminoff, Paulina Oliva, Nick Sanders, seminar participants at the Australian National University, Carnegie Mellon University, Cornell University, Lumos Labs, Inc., and the University of Adelaide, and conference participants at the 8th IZA Workshop: Environment, Health and Labor Markets for invaluable comments and suggestions. The authors gratefully acknowledge financial support from the University of Pittsburgh and the Heinz College at Carnegie Mellon University. The views expressed herein are those of the authors and do not necessarily reflect the views of the National Bureau of Economic Research.
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Ageing-related cognitive decline is highly prevalent with one in four community-dwelling adults over the age of 60 experiencing noticeable decline in cognitive functions [1]. Cognitive decline is a key dimension of ageing; it heralds dementia, illness, and death [2, 3]. Brain structural measures are a well-established set of cognitive ageing biomarkers [4]. It is critical to understand risk factors of structural brain differences, potential mechanisms, and biomarkers to inform rational bases for early interventions and to identify those at greatest risk of cognitive decline and dementia. Brain structure and functioning in later life has been linked to more proximal social determinants of health across the life course, including adversities [5, 6] and individual socioeconomic position [7,8,9]; however, the long-term impact of distal factors in the life course, such as neighbourhood deprivation, are less understood. Neighbourhood disadvantage is an important modifiable predictor of old age health [10] and cognition [11] with effects distinct from individual socioeconomic factors.
Residing in disadvantaged areas might influence brain morphology through multiple interrelated ways and the underlying processes may differ across the life course. Socioeconomically disadvantaged areas are likely to suffer from poorer housing conditions [12], higher crime and violence [13], have lower provision of high quality green space [14], and residents are more likely to be exposed to higher levels of contaminants (e.g., air pollution [15]), and to health-damaging commodities and services, such as alcohol, fast food and tobacco outlets [16]; these environmental features, in turn, can associate directly with brain morphology and function (e.g., green space [17], air pollution [18]). In childhood, the quality of local schools may be pertinent, as might the availability of neighbourhood resources and amenities (e.g., community and cultural centres) in late adulthood. Research shows that children exposed to stressors in their social environment experience activated hypothalamic-pituitary-adrenal axis which leads to long-term dysregulation and changes in the brain [19, 20]; investigated pathways in adulthood include inflammatory [21, 22], neuroendocrine [23] and cardiovascular [24] mechanisms.
Existing research on neighbourhood deprivation and brain health in late adulthood is limited by cross-sectional or short-term longitudinal study designs that offer only a snapshot of current environmental conditions [23,24,25,26]. Given that neighbourhood exposures, especially during sensitive periods of brain development, might have very long-term impacts on brain health, it is important to account for exposures across the whole life, though such data are rare. Applying the life-course approach (i.e., examining the life-course impact of social and physical exposures on later health and diseases risk [27]) has the potential to partially overcome methodological biases [11], and it has been applied to explore individual-level risk factors of brain health among older adults [7,8,9]. Still, reconstructing objectively measured historical neighbourhood exposures over several decades remains a significant challenge due to the lack of consistently measured neighbourhood-level data and residential history covering the entire life course [28, 29].
Diffusion MRI (dMRI) can quantify water molecule diffusion in white matter microstructure [36]. All raw dMRI data were converted from DICOM to NIfTI-1 format using TractoR v2.6.2 [37]. Using tools freely available in the FSL toolkit v4.1.9 (FMRIB, Oxford University: ) [38], data underwent brain extraction [39] performed on the T2-weighted EP volumes acquired along with the dMRI data. The brain mask was applied to all volumes after correcting for systematic eddy-current induced imaging distortions and bulk patient motion using affine registration to the first T2-weighted EP volume of each participant [40]. For all dMRI volumes, diffusion tensors were fitted at each voxel and water diffusion measures were estimated for mean diffusivity and fractional anisotropy at each voxel. Tractography was performed using an established probabilistic algorithm with a two-fibre model per voxel (BEDPOSTX/ProbtrackX) [41, 42]. Analysis of twelve major white matter tracts was performed using probabilistic neighbourhood tractography [37]. These tracts were the genu and splenium of the corpus callosum, left and right arcuate fasciculus, left and right anterior thalamic radiation, left and right rostral cingulum, left and right inferior longitudinal fasciculus, and left and right uncinate fasciculus (see Supplementary Fig. 2 for their locations). All tracts were visually quality checked, and exclusions were made on a tract basis. Tract-averaged diffusion parameters (i.e., fractional anisotropy, mean diffusivity) weighted by the streamline visitation count were then calculated from all voxels by tract [35, 43].
Confounders included in the main analyses are coloured grey: dark grey are considered as confounders for all life course models, medium grey for young adulthood and mid- to late adulthood exposures, light grey for mid- to late adulthood exposures only. Sensitivity analyses addressing selective mobility (S1) and potential confounding (S2, S3) are blue, red, and green, respectively. Links between confounders are not shown for simplicity. BMI = body mass index; IQ = intelligence quotient; ND = neighbourhood deprivation; OSC = occupational social class.
Models were fitted with full information maximum likelihood (FIML) estimation within structural equation modelling (SEM) using the lavaan package v0.6-12 [45] in R v4.2.1 [46]; codes used in this study are available from the corresponding authors upon request. FIML regression has the advantage of estimating model parameters based on all available information, including participants with missing variables, increasing power, and thus lowering type II error. Importantly, fitting models in the context of all available data for confounders enables to calculate model residuals in a larger and more comprehensive sample, thus estimating the impact of exposure more accurately. FIML regression produces equivalent results to models handling missing data with multiple imputation [47]. In addition to standardised regression coefficients (β), standard errors (SE), and two-sided p-values (p) adjusted for false discovery rate (pFDR), we reported total sample size for each analysis (N) and the number of pairwise complete observations (n) for associations of interest.
Global brain measures included six macro-structural outcomes (total brain, grey matter, normal-appearing white matter, and white matter hyperintensity [log-transformed to approximate normal distribution [43]] volumes, cortical surface area, and mean cortical thickness) and the two markers of white matter microstructure. Consistent with prior work [48], general factors of fractional anisotropy and mean diffusivity across the twelve white matter tracts were estimated as latent factors; we included residual correlations between the splenium and the genu of corpus callosum, and between right and left sides of the bilateral tracts (see Supplementary Table 1 for fit indices and factor loadings). We corrected for multiple comparison between the eight global brain outcomes using FDR adjustment [49]. Goodness of fit indices were provided for the general factors of fractional anisotropy and mean diffusivity, for all other outcomes models were fully saturated (i.e., Comparative Fit Index=1, Tucker-Lewis Index=1, Root Mean Square Error of Approximation=0, Standardised Root Mean Square Residual=0).
Local brain associations were explored in further analyses. First, we estimated associations with life-course neighbourhood deprivation scores across the entire cortical surface. Vertex-wise analysis was performed in a common space (the FreeSurfer average template) for 327,684 cortical vertices using all MRI participants with cortical surface data. Three vertex-wise brain measures were assessed: cortical volume, surface area, and thickness. Vertex measures were exported from SurfStat and then SEM with FIML was used to iteratively estimate standardised coefficients and the corresponding p-values by vertex for each neighbourhood deprivation exposure. Correction for multiple comparison was performed by FDR and the findings were presented in cortical surface maps. The spatial overlap between significant cortical regions was assessed by the Dice coefficient [50]. Second, in addition to the general factors, we presented associations in each of the twelve white matter tracts after FDR correction.
c80f0f1006