Skip to main content
NCBI home page
Wiley Open Access Collection logoLink to Wiley Open Access Collection
. 2026 Oct 11;18(5):e70226. doi: 10.1111/aphw.70226

The weight of junk food: Smaller hippocampal, amygdala, and prefrontal volumes in adolescents—The Cogni‐Action Project

Carlos Cristi‐Montero 1,2, Humberto Peña‐Jorquera 1,3, Javier Sanchez‐Martinez 4, Gastón Pavez‐San Martin 1, Patricio Solis‐Urra 5,6, Gerson Ferrari 7,8, José Francisco López‐Gil 9,10,✉
PMCID: PMC13655188  PMID: 42860545

Abstract

Most studies investigating diet–brain relationships in youth focus on single foods. Whether the joint consumption of multiple food groups is associated with neuroanatomy during adolescence remains largely unexplored, particularly outside high‐income countries. In this cross‐sectional study, 61 Chilean adolescents (11.6 ± 1.07 years; 42.6% girls) underwent T1‐weighted MRI. Volumes of total gray matter, hippocampus, amygdala, and selected prefrontal regions were extracted. Dietary intake was assessed using the Global Food Quality Index questionnaire. Confounder‐adjusted linear regressions examined individual food group–brain associations, with correction for multiple comparisons. Canonical correlation analysis (CCA) was used to identify multivariate food group–brain covariation with permutation‐based inference. Univariate analyses showed a significant inverse association between pastry consumption and hippocampal volume (β = −0.402, 95% CI: −0.571 to −0.228, adjusted R 2 = 0.34). CCA identified one significant multivariate mode linking higher unhealthy food consumption to brain volumes (r = −0.70, p = .008). Pastries, fried foods, and soft drinks showed the strongest contributions, whereas the corresponding brain variate was characterized by smaller hippocampal, amygdala, and anterior cingulate/prefrontal volumes. In conclusion, higher consumption of unhealthy food groups is associated with smaller limbic–prefrontal volumes in adolescents. Although causality cannot be inferred from these cross‐sectional data, these findings highlight adolescence as a critical window for diet–brain health research.

Keywords: brain health, eating habits, neurodevelopment, nutrition, ultra‐processed food

INTRODUCTION

Adolescence is a critical neurodevelopmental period characterized by extensive cortical reorganization and synaptic pruning, during which the consolidation of dietary habits may significantly influence ongoing brain development, including cognitive performance and brain structure (Heslin & McNulty, 2023; Juraska, 2024). Particularly, the prefrontal cortex, which remains underdeveloped during this stage, plays a key central role in self‐regulation and decision‐making (Lowe et al., 2020). Its prolonged maturation trajectory may transiently limit inhibitory control, increasing vulnerability to reward‐driven behaviors such as the overconsumption of energy‐dense, nutrient‐poor foods (Lowe et al., 2020). Given this stage is often referred to as the “second developmental period” due to the rapid pace of physical and cognitive changes, the role of nutrition must be appropriately addressed to maximize benefits and mitigate potential health risks (Heslin & McNulty, 2023).

Globally, evidence indicates that adolescent diets are characterized by high intakes of ultra‐processed foods, ready meals, sugars, and fats, alongside low consumption of healthy foods such as fruits, vegetables, legumes, and lean meats, with a progressive shift in food group consumption from childhood to adolescence (Heslin & McNulty, 2023). This pattern is particularly concerning, as dietary components and food group consumption may influence key neurodevelopmental processes, including neurogenesis, axonal and dendritic growth, synaptogenesis, and axonal myelination (Naveed et al., 2020). Importantly, these processes follow region‐specific and non‐linear trajectories across adolescence, suggesting that certain brain structures may be differentially sensitive to dietary exposures during this period (Mou et al., 2023). Moreover, early‐life food consumption profiles may exert lasting influences on brain structure and function, even when assessed at later developmental stages (Mou et al., 2023).

On the one hand, previous evidence using magnetic resonance imaging (MRI) suggests that unfavorable food consumption profiles are associated with smaller brain volume in younger and older populations (Mou et al., 2023; Zhang et al., 2024). For example, Mou et al. (2023), in a prospective study, found that children with higher adherence to “snack, processed foods, and sugar” food consumption profile at ages 1 and 8 showed smaller global brain volume at age 10. Along with this, Stadterman et al. (2020) reported that although a Western diet was not directly associated with brain volume, higher fat consumption was significantly related to a reduction in left hippocampal volume. On the other hand, greater adherence to a “whole grains, soft fats, and dairy” food consumption profile at age 8 was significantly associated with larger total and cerebral gray matter volumes at age 10 (Mou et al., 2023).

Because prior studies have linked individual food groups to specific brain regions, univariate regressions remain essential for verifying these one‐to‐one associations while adjusting for confounders in understudied age groups (Cadenas‐Sanchez et al., 2020; Kennedy et al., 2016). Yet the intrinsically correlated nature of both diet and neuroanatomy demands complementary multivariate techniques, particularly canonical correlation analysis (CCA), to model multiple food groups and brain regions simultaneously, curb multiple‐testing inflation, and unveil modes of diet–brain covariation that single‐variable models cannot detect. Finally, because most MRI research originates in high‐income countries, its generalizability to low‐ and middle‐income contexts is uncertain (Bethlehem et al., 2022; Cristi‐Montero et al., 2025), leaving Latin American adolescents markedly under‐represented. To the best of our knowledge, no study using MRI analysis has explored the association between food group consumption and multiple brain regions in South America's developing countries. Collectively, these gaps underscore the need for the present study.

Therefore, this study will examine food group–brain relationships in Chilean adolescents. Confounder‐adjusted univariate regressions will first be conducted to verify previously reported associations between individual food groups and specific brain regions. Subsequently, CCA will assess multivariate diet–brain covariance and quantify the contribution (cross‐loadings) of each food group. By adopting a system‐level analytical approach in an under‐represented adolescent population, this exploratory study aims to generate hypotheses on how dietary food group composition relates to brain structure, while providing preliminary insights relevant to regional public health strategies and the broader understanding of adolescent neurodevelopment.

METHODS

This subsample was drawn from the Cogni‐Action Project, a cross‐sectional study conducted between March 2017 and October 2019 (Solis‐Urra et al., 2019). The present analysis should be considered exploratory, given the subsample size and the use of multivariate modeling techniques. The project originally included 1296 adolescents (50% girls) aged 10–14 years and was conducted in accordance with the principles of the Declaration of Helsinki (World Medical Association., 2001). It received approval from the Ethic Committee of the Pontificia Universidad Catolica de Valparaiso (BIOEPUCV‐H103‐2016) and was retrospectively registered in the Research Registry on July 8, 2020 (ID: researchregistry5791). Written informed consent was obtained from school principals, parents, and participants before their inclusion in the study.

Participants

For the main Cogni‐Action Project, all children enrolled in fifth to eighth grade were eligible to participate. However, due to resource constraints (specifically cost limitations), it was not possible to collect neuroimaging data for the full cohort. Thus, the present subsample for neuroimaging analysis was constituted by adolescents from the main project who, along with their parents and guardians, voluntarily provided informed consent for the MRI assessment. Therefore, for this subsample analysis, baseline data were collected from 61 adolescents (mean age: 11.6 ± 1.07 years; 42.6% girls) from 10 schools. Following ethical standards, children with physical, psychiatric, or psychological disabilities were allowed to participate in the cross‐sectional phase, contingent upon informed consent from both the child and their parent or guardian. However, to avoid potential confounding influences on brain structure and function, specific exclusion criteria were applied for the neuroimaging subsample. Specifically, no children with diagnosed psychiatric, psychological, or neurodevelopmental disorders (e.g., attention deficit hyperactivity disorder [ADHD], depression) or neurological conditions were included in the final neuroimaging analysis cohort.

Additional inclusion criteria for the neuroimaging analysis required adolescents to be right‐handed due to structural differences in brain organization between left‐ and right‐handed individuals (Sha et al., 2021). Further exclusion criteria for MRI analysis were (a) absence of a valid T1‐weighted image (n = 14), (b) incidental tumor findings evaluated by a trained health professional (n = 1), and (c) concerns about image quality after visual inspection and preprocessing steps (n = 3; specifically, presence of severe motion artifacts such as ringing, ghosting, or poor gray‐white matter contrast, as well as uncorrectable surface segmentation errors).

Food variables

Food group consumption was assessed using the Global Food Quality Index questionnaire (Ratner et al., 2017), which allows us to determine the daily or weekly frequency of consumption of different food groups. This instrument consists of nine food group items, five healthy (fruits, vegetables, dairy, legumes, and fish), and four unhealthy (sugar‐sweetened beverages (soft drinks), pastries, fried foods, and added table sugar), along with three main meal items. Each food group was rated on a scale from 1 (less nutritious) to 10 (more nutritious, as recommended by the Chilean Ministry of Health). A composite score derived from the sum of individual food group ratings was used to classify diets as healthy (90–120), in need of changes (60–89), or unhealthy (<60) (Ratner et al., 2017). Each food group item had six possible response options categorized as follows: 1: never or rarely; 2: once a week; 3: two/three times a week; 4: four/six times a week; 5: once a day; and 6: ≥twice a day. Further, we dichotomized these variables into two groups: lower consumption (Responses 1–3) and higher consumption (Responses 4–6) for each respective food group.

Brain structure: Magnetic resonance imaging (MRI)

MRI acquisition

Structural brain images were acquired using a 1.5 Tesla MRI scanner (AVANTO, Siemens Medical Systems, Erlangen, Germany). High‐resolution T1‐weighted images were obtained using a magnetization‐prepared rapid gradient echo (MPRAGE) sequence with the following acquisition parameters: repetition time (TR) = 2200 ms, echo time (TE) = 2.6 ms, and an isotropic voxel size of 1.0 × 1.0 × 1.0 mm3. The total acquisition time for the structural sequence was approximately 4 min and 32 s. Only structural T1‐weighted images were used for the present analyses.

Structural image preprocessing

T1‐weighted magnetization‐prepared rapid gradient echo (MPRAGE) images were processed using FreeSurfer 7.4.1 (https://surfer.nmr.mgh.harvard.edu) within the Neurodesk platform (https://www.neurodesk.org/) employing the cross‐sectional “recon‐all” pipeline (“recon‐all − s subject_id − i T1image − all”; https://surfer.nmr.mgh.harvard.edu/fswiki/recon-all). From the FreeSurfer output (aseg.stats file), several brain and intracranial volume metrics were extracted, including total gray matter volume (defined as the sum of cortical, subcortical, and cerebellar gray matter from both hemispheres), cortical volume (measured as the volume between the pial and white matter surfaces, excluding non‐cortical tissue such as the hippocampus), subcortical gray matter volume (comprising structures such as the thalamus, caudate, putamen, pallidum, hippocampus, amygdala, accumbens, ventral diencephalon, and, when present, the substantia nigra). The volume of the prefrontal cortex was estimated by summing the volumes of specific regions extracted from the left and right hemisphere parcellation files (lh.aparc.stats and rh.aparc.stats, respectively), including the caudal anterior cingulate, lateral and medial orbitofrontal cortices, pars opercularis, pars orbitalis, pars triangularis, rostral anterior cingulate, rostral middle frontal, superior frontal, and frontal pole (Desikan et al., 2006). These parcellations were based on the Desikan‐Killiany atlas (Desikan et al., 2006), which was used to define the cortical regions of interest. Further, the total hippocampal and amygdala volumes were calculated by summing the left and right hemisphere volumes, which were derived using the cross‐sectional hippocampal subfield segmentation module (“segmentHA_T1.sh subject_id SUBJECT_DIR”) (Iglesias et al., 2015). JS‐M conducted quality control procedures following the ENIGMA Consortium Cortical Quality Control (QC) Protocol 2.0 (https://enigma.ini.usc.edu).

Covariates

Seven covariates (sex, peak height velocity, body mass index, school vulnerability index, intracranial volume, global fitness score, and global food group–based dietary score) were included in the first approach to the analysis, owing to their relevance to the outcome based on the existing literature. However, in the sensitivity analysis, global fitness score and body mass index were excluded due to the non‐significant association in the model.

Sex and maturation are key factors influencing cognitive and brain development (Stillman et al., 2020). Maturity was estimated using the peak height velocity (PHV), calculated as the difference between chronological age and predicted age of PHV attainment using the validated somatic equation (Mirwald et al., 2002). Furthermore, physical measurements were conducted by trained evaluators following standardized procedures: Standing and sitting height were measured without shoes using a portable stadiometer (SECA, model 213, GmbH, Germany), and body weight was assessed with light clothing using an electronic scale (OMRON, HN‐289‐LA, Kyoto, Japan). Socioeconomic context was assessed using the School Vulnerability Index (SVI), a measure developed by the Chilean Government to quantify the socioeconomic vulnerability of students attending partially (subsidized) or fully funded schools (López et al., 2017). The SVI integrates personal and family variables such as socioeconomic status, parental education, student health, emotional well‐being, and the school's geographic context, generating a single score per institution. Ranging from 0 to 100, the SVI is updated annually. Private schools, excluded from this measure as they do not receive state funding, are assigned a default score of 0. Based on their SVI, schools are categorized into low (<10), middle (≥10 to <60), and high (≥60) vulnerability (López et al., 2017). Intracranial volume estimated with FreeSurfer was used to adjust brain metrics, controlling for inter‐individual differences in brain size and reducing structural confounding (Cristi‐Montero et al., 2025). The global food group‐based dietary score was determined using a pre‐established questionnaire including nine‐item referring foods and three‐item regarding main meals (Ratner et al., 2017). This score was calculated to isolate the link of each food group by accounting for and subtracting the combined influence of all other food groups. Thus, we separately calculated healthy and unhealthy food group‐based composite scores. More details on the Cogni‐Action Project measures have been published elsewhere (Solis‐Urra et al., 2019).

Statistical analysis

First, missing data were imputed using the nonparametric random forest method through the “missForest” R package (Stekhoven & Bühlmann, 2012). This approach effectively imputes large and complex mixed‐type datasets (quantitative and/or categorical variables), accounting for complex interactions and nonlinear relationships by training a random forest model on observed values to predict missing ones. Little's MCAR test was conducted to assess the missing data mechanism (Little, 1988). The results yielded a chi‐square statistic of χ2(130) = 148 and a p‐value of .131, indicating insufficient evidence to reject the null hypothesis that the missing data are completely at random. The average percentage of missing data across the variables used in this study was low = 4.2%. Outliers were identified as values exceeding four standard deviations from the mean and were winsorized by replacing them with the nearest non‐outlier value. Data normality was assessed using the Shapiro–Wilk test and QQ‐plots. Non‐normally distributed variables were transformed using the orderNorm procedure (R Studio) to meet statistical assumptions (e.g., hippocampal volume).

Second, to avoid overadjustment in the regression models, given the reduced sample size, a principal component analysis (PCA) was conducted to generate a single confounding factor. This PCA‐derived factor explained 58.8% of the variance and included key variables related to the primary outcomes: sex, peak height velocity, school vulnerability index, and intracranial volume. Furthermore, given the multiple comparisons that increase the risk of Type I error and the study's exploratory nature, p values were adjusted using the Benjamini–Hochberg False Discovery Rate (FDR) procedure (Benjamini & Hochberg, 1995), including bootstrapping with 5000 replications to generate coefficient estimates and confidence intervals (Hochberg & Benjamini, 1990; Yekutieli & Benjamini, 1999). This approach controls the probability of false positives while maintaining the robustness and statistical power of the results.

Third, a correlation matrix was initially computed to explore the relationships among all variables included in the study. Further, to enhance the robustness of our analysis, two statistical approaches were performed. First, a univariate analysis was conducted using linear regression to determine the isolated associations between each dichotomized food group and specific brain volumes. Standardized beta coefficients were estimated, and p values were used to assess significant associations while accounting for confounding variables. Furthermore, we report the adjusted R 2, a commonly used measure of goodness‐of‐fit in linear regression analysis, which indicates the proportion of variance explained by the model, adjusted for the number of independent variables.

Further, a CCA was employed to perform a complementary multivariate approach to explore covariation patterns between two sets of variables: (a) two latent food group factors (comprising five healthy and four unhealthy food variables, as separately analyzed) and (b) a latent brain volume factor (10 variables, including hippocampus, amygdala, caudal anterior cortex, orbitofrontal cortex, pars opercularis, pars orbitalis, pars triangularis, rostral anterior cortex, dorsolateral, and frontal pole regions). This method identifies modes of covariation between two sets of variables without assuming predefined relationships, considering the canonical cross‐loading or strength of correlation (CCA rho) of each variable on its opposite canonical variate (Figure 1). Each mode is defined by a pair of CCA canonical variates or subject vectors that exhibit the maximum correlation. The total number of generated modes is always equal to the number of variables in the smaller dataset (here, four and five). To perform CCA, we used the script “permcca” (Winkler et al., 2020), available at (https://github.com/andersonwinkler/PermCCA), while controlling for confounding variables previously described. It is important to note that the identified canonical mode should be interpreted as a system‐level covariation pattern rather than evidence of region‐specific neuroanatomical effects.

FIGURE 1.

FIGURE 1

Open in a new tab

Canonical food group–brain mode of covariation derived from CCA, illustrating latent factors, loadings, and cross‐loadings.

All variables were standardized using z‐scores prior to analysis. The significance of the CCA modes was assessed via nonparametric inference testing with 1000 permutations among subjects within schools. Family‐wise error correction (FWE‐corr) was applied across all CCA modes to adjust for multipl e comparisons. A CCA (covariation, loading, and cross‐loading) was deemed significant at FWE corr p < .05 and a minimal association of r values ≥0.10 (small effect size). The magnitudes of the correlations were interpreted as negligible (0.00–0.10), weak (>0.10–0.39), moderate (0.40–0.69), strong (0.70–0.89), and very strong (0.90–1.00) (Akoglu, 2018; Schober et al., 2018). All analyses were adjusted for sex, peak height velocity, school vulnerability index, and intracranial volume as principal component analysis. Moreover, the global food group‐based composite score was also included as a covariate to reduce the combined influence of the remaining food groups on the outcome (Ratner et al., 2017).

The statistical significance of the canonical functions was assessed through a comparative analysis using Wilks' Lambda for parametric testing, whereas Jackknife validation was used to estimate the stability of the canonical correlations due to the small number of subjects for the classic approach, as was recommended previously (Lee, 2007). Additionally, canonical correlations and permutation p values were obtained through non‐parametric testing (1000 permutations) as a more robust analytical approach. Statistical significance was set at p < .05.

RESULTS

We aimed to describe differences in brain volume and food group intake among adolescents. Descriptive and comparative analyses are presented as means ± standard deviations for continuous variables, whereas categorical variables are reported as frequencies and percentages. Regarding overall diet quality assessed via the Global Food Quality Index, 0% of adolescents were classified as having a healthy diet, 13.1% were in need of changes, and 86.9% presented an unhealthy diet. Table 1 summarizes the main characteristics of the adolescents.

TABLE 1.

Descriptive characteristics of participants and key study variables.

Variables All (n = 61)
Age (y) 11.6 ± 1.07
Weight (kg) 50.8 ± 12.2
Height (cm) 152 ± 8.96
BMIz (SD) 1.13 ± 1.05
PHV (y) −0.67 ± 1.11
GFS 0.35 ± 2.64
SVI 2.07 ± 0.79
Food groups (those with higher consumption)
Fruits 15 (24.6%)
Vegetables 10 (16.4%)
Dairy 13 (21.3%)
Legumes 42 (68.9%)
Fish 48 (78.7%)
Pastries 35 (57.4%)
Soft drinks 42 (68.9%)
Sugar 30 (49.2%)
Fried foods 45 (73.8%)
Global diet quality classification
Healthy diet (90–120 points) 0 (0%)
In need of changes (60–89 points) 8 (13.1%)
Unhealthy diet (<60 points) 53 (86.9%)
Open in a new tab

Note: BMIz, body mass index z‐score, according to the WHO growth reference for children and adolescents aged 5–19 years; BMI‐for‐age is classified as severe thinness (< −3 SD), thinness (< −2 SD), normal weight (−2 to +1 SD), overweight (> +1 SD), and obesity (> +2 SD). Data are present as mean and standard deviation for continuous variables, whereas counts and percentages were used for categorical or dichotomous variables.

Abbreviations: GFS, global fitness score; PHV, peak height velocity; SVI, school vulnerability index; y, years.

Figure 2 illustrates the correlation matrix highlighting the relationships between brain volume, individual food group intake variables, and the PCA‐derived factor. Notably, strong and moderate positive correlations were observed among several brain regions, reflecting their interconnected nature. In contrast, the inverse associations between brain volumes and food groups, as well as with the PCA‐derived factor, ranged from weak to moderate (r = −0.42 to −0.01). Particularly, higher moderate associations were observed between sex and amygdala (r = −0.50) and school vulnerability index and whole gray matter (r = −0.49). Moreover, stronger negative correlations were observed for sex and whole gray matter (r = −0.67).

FIGURE 2.

FIGURE 2

Open in a new tab

The correlation matrix of whole brain volume measures, individual food group variables, and the PCA‐derived factor. The black square highlights the specific correlation between brain and food variables. Abbreviations: BMIz, body mass index (standard deviation); CAC, caudal anterior cingulate cortex; GM, gray matter; PHV, peak height velocity; RAC, rostral anterior cingulate; SVI, school vulnerability index.

Figure 3 presents the results of the univariate linear regression analyses examining the associations between brain volumes and healthy and unhealthy food groups. Among the dichotomized food groups, neither healthy nor unhealthy foods showed significant associations with any brain volume after false discovery rate correction (p‐FDR ≥ .05), with the exception of pastry intake, which was inversely associated with total hippocampal volume (β = −0.402, 95%CI [−0.571; −0.228], p‐FDR = .007; adjusted R 2 = 0.34). Additional details are provided in Tables S1 and S2.

FIGURE 3.

FIGURE 3

Open in a new tab

Dot‐and‐whisker plot associations between healthy and unhealthy food groups and whole brain volumes. Analysis was adjusted for basic confounders (sex, peak height velocity, school vulnerability index, and intracranial volume as principal component analysis) and global food group–based composite score. Significance threshold set at p < .05. * = significant association (larger dot size). Abbreviations: CAC, caudal anterior cingulate; RAC, rostral anterior cingulate.

Canonical correlation analysis (CCA)

Table 2 illustrates the specific significance analysis of the CCA. In detail, CCA results showed that the relationship between the two latent factors was significant only for the unhealthy group, defined by the joint variation of multiple unhealthy food groups (r = −0.70; permutation p value = .008), indicating a stronger association between the predictor and outcome variables. In contrast, for the healthy group, none of the five canonical functions reached statistical significance (p ≥ .05), suggesting a weaker relationship between the food variables. Additionally, Jackknife validation indicated a mean canonical correlation of 0.597 for the healthy group and 0.701 for the unhealthy group, suggesting that the coefficients were more stable and robust in the unhealthy group.

TABLE 2.

Results for significance analysis.

Canonical model Wilks' Lambda X 2 df p‐value Jackknife correlation (95%CI) Canonical correlation Permutation p‐value
Healthy foods
1 0.370 51.8 50 .405 −0.597 (−0.552, −0.608) −0.575 .537
2 0.552 32.1 36 .656 – −0.491 .356
3 0.728 18.2 24 .794 – −0.455 .069
4 0.917 5.3 14 .982 – −0.226 .881
5 0.967 2.7 6 .945 – −0.183 .489
Unhealthy foods
1 0.313 61.3 40 .017 −0.701 (−0.682, −0.721) −0.699 .008
2 0.609 26.5 27 .489 – −0.461 .323
3 0.773 14.8 16 .543 – −0.406 .083
4 0.926 4.8 7 .683 – −0.273 .182
Open in a new tab

Note: “Canonical model” column, each number represents a distinct mode of covariation identified in the analysis. Canonical correlations and permutation p‐values were obtained using nonparametric testing (1000 permutations). Significance was set at p < .05; significant values are shown in bold.

This CCA did not reveal any significant mode of covariation for healthy foods. In contrast, for unhealthy food, CCA revealed one statistically significant mode of covariation (Figure 4). Specifically, pastries, soft drinks, and fried foods exhibited a moderate contribution to the unhealthy food factor (0.38 to 0.62), whereas sugar intake displayed a negligible contribution. Regarding brain volume, the hippocampus, amygdala, caudal anterior cingulate, pars opercularis, and pars triangularis showed a moderate contribution to the latent brain volume factor (−0.43 to −0.15), whereas the rest of the brain regions displayed a negligible contribution.

FIGURE 4.

FIGURE 4

Open in a new tab

Mode of covariation link sets of food variables with sets of brain volumes in adolescents. Pink: cross‐loadings for unhealthy food groups; gray: cross‐loadings for brain volumes (n = 61). Canonical correlations and permutation p‐values were obtained through non‐parametric testing (1000 permutations). Analysis was adjusted for basic confounders (sex, peak height velocity, school vulnerability index, and intracranial volume as principal component analysis), and healthy food group–based composite score. Significance threshold set at p < .05.

DISCUSSION

This study aimed to examine the relationship between food group consumption and brain volumes in a subsample of Chilean schoolchildren. Univariate analyses revealed a significant association between pastry consumption and hippocampal volume. In contrast, the multivariate analysis indicated that greater intake of unhealthy foods, particularly pastries, soft drinks, and fried foods, was significantly associated with smaller brain volumes in key regions such as the hippocampus, amygdala, caudal anterior cingulate cortex, pars opercularis, and pars triangularis. By integrating region‐specific analyses with a system‐level framework, our findings provide converging but distinct insights into how dietary food group composition relates to adolescent brain structure.

In the univariate analyses, only pastry consumption emerged as significantly associated with hippocampal volume, whereas no other isolated food group–brain associations survived correction for multiple comparisons. This limited number of significant one‐to‐one associations suggests that isolated food group effects may be modest during early adolescence, a developmental stage characterized by substantial inter‐individual variability and ongoing neuroanatomical remodeling. These findings are broadly consistent with previous evidence suggesting that ultra‐processed foods and diets high in sugar and fat may have a detrimental impact on brain health. For instance, Clark et al. (2020) found that higher intake of monosaccharides (particularly fructose and glucose) was associated with increased volume in specific hippocampal subfields. However, this increase was attributed to inflammation, not structural improvement. Their post hoc analysis revealed increased mean diffusivity in the right cingulate‐prefrontal cortex, a marker of myelin loss or delayed axonal pruning, highlighting the relevance of neurodevelopmental processes during adolescence. These findings are further supported by Mou et al. (2023), who found that at age 10, children with higher consumption of snacks and processed foods exhibited an inverse association with overall brain volume.

In contrast, CCA identified a robust system‐level covariation between higher consumption of unhealthy food groups and smaller limbic–prefrontal volumes. Pastries, fried foods, and soft drinks primarily drove this latent dietary profile, whereas no comparable multivariate pattern emerged for healthy food groups, indicating an asymmetric association during adolescence. These findings are particularly relevant, as the affected regions, including the hippocampus and the prefrontal cortex, exhibit non‐linear developmental trajectories during adolescence (Selmeczy et al., 2019) and play a critical role in brain activation, executive functions, self‐regulation, and overall cognitive performance (Lowe et al., 2020; Selmeczy et al., 2019). Importantly, the structural integrity of these regions is fundamental for cognitive functioning, as cognitive maturity in adulthood is supported by the integrity of hippocampal and prefrontal cortical structures during adolescence (Calabro et al., 2020). Conversely, smaller prefrontal cortex volumes have been associated with poorer inhibitory control, potentially increasing susceptibility to impulsive behaviors such as the overconsumption of ultra‐processed foods (Lowe et al., 2020). Such neurodevelopmental dynamics may partially explain why diet‐related associations are more readily detectable at a multivariate, system‐level than through isolated volumetric measures.

It is worth noting that the modest associations observed in our study may be partly explained by well‐established neurodevelopmental differences between adolescents and adults, particularly those related to synaptic pruning (Lowe et al., 2020). During adolescence, synaptic pruning and myelination occur concurrently, a developmental duality that enhances neural efficiency and functional specialization. This process may help explain why diet‐related macroscopic volumetric differences are less readily detectable in adolescents than in adult populations (Lowe et al., 2020). Whereas in adults, the effects of poor diet may be more visibly reflected in brain structure due to mature regulatory mechanisms, in adolescents, dietary influences may operate at a deeper neurobiological level that is not fully captured by gross volumetric measures and may be masked by ongoing somatic and neural development (Lowe et al., 2020). Supporting this notion, a 10% increase in ultra‐processed food consumption during adolescence has been associated with a 0.11 standard deviation reduction in composite executive function scores, independent of adiposity and socioeconomic status (Mottis et al., 2025).

On the contrary, our findings regarding healthy food options did not align with previous evidence. For example, most studies assessing specific, isolated food groups have been conducted in adults, consistently showing favorable results (Croll et al., 2018; Gaudio et al., 2023; Kokubun & Yamakawa, 2019; Zhang et al., 2024). In contrast, in younger populations, healthy foods are less frequently evaluated as individual categories and are more commonly examined collectively as part of broader dietary patterns (Kokubun & Yamakawa, 2019). In comparison, in the present study, neither individual healthy food groups nor their aggregation into food group‐based composite scores showed consistent favorable associations with brain volumes. This discrepancy suggests that such associations may be less evident during early adolescence or may vary across sociocultural contexts, including Latin American populations.

One possible explanation is that the negative association of unhealthy foods may outweigh the potential benefits of healthier alternatives in this specific adolescent population. Adolescents' developing brains may be particularly vulnerable to the detrimental effects of energy‐dense, nutrient‐poor foods (Reichelt & Rank, 2017). Evidence suggests that such diets can alter the brain's reward systems and impact frontostriatal and frontotemporal neurocircuitry, which are still maturing during this developmental stage (Reichelt & Rank, 2017). Furthermore, the persistence of negative effects even after returning to a healthier diet (Kullmann et al., 2025) may help explain why no beneficial associations were observed for healthy food groups in our study. These findings emphasize the importance to consider both individual and aggregated food analyses when examining dietary influences on brain structure.

Lastly, the present analyses do not aim to link dietary exposure to structural brain changes over time but rather to characterize patterns of covariation between food group consumption and neuroanatomical measures during a sensitive developmental period. Whereas single‐point dietary assessment and MRI acquisition may appear simplistic, CCA is specifically designed to capture multivariate relationships between two high‐dimensional datasets, allowing the identification of coherent food group–brain covariation profiles rather than isolated region–food associations. Therefore, considering these methodological constraints, we suggest that both the sociocultural context and the developmental stage of the studied population may play a critical role in shaping the observed associations, which should be interpreted as hypothesis‐generating and warrant confirmation in longitudinal studies with repeated dietary and neuroimaging assessments.

Perspectives

From a public health perspective, the present findings provide preliminary neuroanatomical evidence that is highly relevant to the Chilean context, where childhood and adolescent obesity rates are among the highest worldwide (von Hippel & Bogolasky Fliman, 2024). In this setting, national policies such as the Chilean Food Labeling and Advertising Law (Law 20.606), which regulates foods high in sugars, saturated fats, sodium, and energy density, represent a cornerstone strategy to improve dietary behaviors in youth. Beyond global diet quality scores, our results highlight the importance of food group composition and consumption patterns, suggesting that higher intake of specific unhealthy food groups may be particularly relevant for brain health during adolescence (Kullmann et al., 2025).

Importantly, these findings should be interpreted within the broader socioeconomic context of the studied population, as most neuroimaging studies examining diet–brain relationships have been conducted in high‐income countries. In our sample, approximately 66% of participants attended schools characterized by low‐to‐middle socioeconomic vulnerability, which may influence dietary behaviors, food availability, and exposure to ultra‐processed foods in ways that differ from those observed in high‐income country settings (Lommi et al., 2021). Although causal inferences cannot be drawn from this cross‐sectional design, the observed multivariate associations between higher consumption of unhealthy food groups and smaller volumes in brain regions suggest that the potential benefits of such policies may extend beyond cardiometabolic health. Specifically, these results suggest that reducing access to such foods in school and community environments may contribute to protecting not only physical health but also aspects of brain structure and function. Future longitudinal and intervention studies are warranted to evaluate whether sustained reductions in the consumption of foods subject to warning labels are associated with more favorable brain developmental trajectories in Chilean adolescents.

Strengths and limitations

Our study has several strengths worth highlighting. First, our analysis is based on a sample from a less‐studied age group, helping to fill gaps in the literature on this population. This is particularly relevant given that most studies examining this association using MRI analysis have been conducted in developed countries. Additionally, a key strength of our study is the incorporation of relevant confounding variables, such as peak height velocity and school vulnerability index, that could strongly influence the outcomes, thereby enhancing the robustness of our analytical framework. Finally, the complement analysis including both univariate and multivariate analysis, with the inclusion of bootstrapping (5000 replications) and Jackknife validation, ensures the stability and robustness of our findings.

Nonetheless, some limitations should also be acknowledged. The cross‐sectional nature of our study prevents us from establishing causality, even with careful adjustments for relevant variables. In addition, both dietary intake and brain MRI measures were assessed at a single time point, which limits the ability to capture developmental changes or temporal relationships between diet and neuroanatomical outcomes. Although CCA considers multiple food groups simultaneously, it does not derive dietary patterns per se. Therefore, findings should be interpreted as multivariate associations between food group consumption and brain structural measures, rather than as indicators of overall diet quality. Regarding dietary data, we were unable to differentiate specific food types, quantify exact amounts in grams, or adjust for total energy intake, which limits the depth of our analysis due to the nature of the questionnaire. Finally, the small number of subjects is necessary to keep in mind, positioning this work as exploratory and hypothesis‐generating. Therefore, future studies should evaluate whether the associations observed here are replicable in larger cohorts, ideally incorporating sex‐specific analyses.

CONCLUSION

Our findings suggest that the consumption of specific food groups during adolescence is associated with brain morphology. Multivariate analyses identified a canonical unhealthy food group profile associated with smaller volumes in the amygdala, hippocampus, and prefrontal cortex regions. Importantly, we observed an asymmetric pattern: Only unhealthy consumption patterns (such as pastries, soft drinks, and fried foods) showed significant associations with reduced brain volumes, whereas healthy food consumption showed no measurable benefits for brain structure at this developmental stage. This suggests that the negative association of unhealthy foods may outweigh the potential structural benefits of healthy eating during adolescence. These results highlight the importance of promoting healthier eating habits during this critical period and provide empirical support for targeted nutritional policies. Longitudinal and intervention studies are needed to confirm these associations and elucidate their underlying causal mechanisms.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

The study was approved by the Bioethics and Biosafety Committee of the Ethics Committee of the Pontificia Universidad Catolica de Valparaiso (BIOEPUCV‐H103‐2016) and conducted in accordance with the guidelines of the Declaration of Helsinki.

Supporting information

Table S1. Associations between whole brain volumes and dichotomized intake of healthy food groups.

Table S2. Associations between whole brain volumes and dichotomized intake of unhealthy food groups.

APHW-18-0-s001.docx (28KB, docx)

ACKNOWLEDGMENTS

We want to acknowledge Jorge Olivares‐Arancibia and Tamara Huber‐Perez and all university students who gathered the data at schools. Furthermore, we thank school principals, parents, and adolescents who supported this project. We also want to acknowledge the Independent Imagenology Center Quintaimagen, Viña del Mar, Chile.

Cristi‐Montero, C. , Peña‐Jorquera, H. , Sanchez‐Martinez, J. , Martin, G.‐S. , Solis‐Urra, P. , Ferrari, G. , & López‐Gil, J. F. (2026). The weight of junk food: Smaller hippocampal, amygdala, and prefrontal volumes in adolescents—The Cogni‐Action Project. Applied Psychology: Health and Well‐Being, 18(5), e70226. 10.1111/aphw.70226

Funding information Carlos Cristi‐Montero received funding for the Cogni‐Action Project from the National Commission for Scientific and Technological Research CONICYT/FONDECYT INICIACION 2016 Grant No. 11160703 (Chile) and the National Research and Development Agency (ANID) from Chile‐2019, Postdoctoral Grant No. 74200071. H.P.‐J is supported by the National Agency for Research and Development (ANID)/SUBDIRECCIÓN DE CAPITAL HUMANO/DOCTORADO BECAS CHILE/2025 (Grant No. 72250115). J.S.‐M is supported by the National Agency for Research and Development (ANID)/Scholarship Program/DOCTORADO BECAS CHILE/2022 (Grant No. 72220164).

DATA AVAILABILITY STATEMENT

The data generated during and/or analyzed during the current study could be made available from the corresponding author upon reasonable request.

REFERENCES

  1. Akoglu, H. (2018). User's guide to correlation coefficients. Turkish Journal of Emergency Medicine, 18(3), 91–93. 10.1016/j.tjem.2018.08.001 [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Benjamini, Y. , & Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society. Series B, Methodological, 57(1), 289–300. 10.1111/j.2517-6161.1995.tb02031.x [DOI] [Google Scholar]
  3. Bethlehem, R. a. I. , Seidlitz, J. , White, S. R. , Vogel, J. W. , Anderson, K. M. , Adamson, C. , Adler, S. , Alexopoulos, G. S. , Anagnostou, E. , Areces‐Gonzalez, A. , Astle, D. E. , Auyeung, B. , Ayub, M. , Bae, J. , Ball, G. , Baron‐Cohen, S. , Beare, R. , Bedford, S. A. , Benegal, V. , … Alexander‐Bloch, A. F. (2022). Brain charts for the human lifespan. Nature, 604(7906), 525–533. 10.1038/s41586-022-04554-y [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Cadenas‐Sanchez, C. , Migueles, J. H. , Erickson, K. I. , Esteban‐Cornejo, I. , Catena, A. , & Ortega, F. B. (2020). Do fitter kids have bigger brains? Scandinavian Journal of Medicine & Science in Sports, 30(12), 2498–2502. 10.1111/sms.13824 [DOI] [PubMed] [Google Scholar]
  5. Calabro, F. J. , Murty, V. P. , Jalbrzikowski, M. , Tervo‐Clemmens, B. , & Luna, B. (2020). Development of hippocampal–prefrontal cortex interactions through adolescence. Cerebral Cortex, 30(3), 1548–1558. 10.1093/cercor/bhz186 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Clark, K. A. , Alves, J. M. , Jones, S. , Yunker, A. G. , Luo, S. , Cabeen, R. P. , Angelo, B. , Xiang, A. H. , & Page, K. A. (2020). Dietary fructose intake and hippocampal structure and connectivity during childhood. Nutrients, 12(4), 4. 10.3390/nu12040909 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cristi‐Montero, C. , Sanchez‐Martinez, J. , Espinoza‐Puelles, J. P. , Martinez‐Flores, R. , & Hernandez‐Jaña, S. (2025). Adherence to 24‐hour activity cycle and a whole brain volumetric approach in adolescence: A cross‐sectional study—The Cogni‐Action project. Brain and Cognition, 185, 106282. 10.1016/j.bandc.2025.106282 [DOI] [PubMed] [Google Scholar]
  8. Croll, P. H. , Voortman, T. , Ikram, M. A. , Franco, O. H. , Schoufour, J. D. , Bos, D. , & Vernooij, M. W. (2018). Better diet quality relates to larger brain tissue volumes: The Rotterdam study. Neurology, 90(24), e2166–e2173. 10.1212/WNL.0000000000005691 [DOI] [PubMed] [Google Scholar]
  9. Desikan, R. S. , Ségonne, F. , Fischl, B. , Quinn, B. T. , Dickerson, B. C. , Blacker, D. , Buckner, R. L. , Dale, A. M. , Maguire, R. P. , Hyman, B. T. , Albert, M. S. , & Killiany, R. J. (2006). An automated labeling system for subdividing the human cerebral cortex on MRI scans into gyral based regions of interest. NeuroImage, 31(3), 968–980. 10.1016/j.neuroimage.2006.01.021 [DOI] [PubMed] [Google Scholar]
  10. Gaudio, S. , Rukh, G. , di Ciommo, V. , Berkins, S. , Wiemerslage, L. , & Schiöth, H. B. (2023). Higher fresh fruit intake relates to larger grey matter volumes in areas involved in dementia and depression: A UK Biobank study. NeuroImage, 283, 120438. 10.1016/j.neuroimage.2023.120438 [DOI] [PubMed] [Google Scholar]
  11. Heslin, A. M. , & McNulty, B. (2023). Adolescent nutrition and health: Characteristics, risk factors and opportunities of an overlooked life stage. Proceedings of the Nutrition Society, 82(2), 142–156. 10.1017/S0029665123002689 [DOI] [PubMed] [Google Scholar]
  12. Hochberg, Y. , & Benjamini, Y. (1990). More powerful procedures for multiple significance testing. Statistics in Medicine, 9(7), 811–818. 10.1002/sim.4780090710 [DOI] [PubMed] [Google Scholar]
  13. Iglesias, J. E. , Augustinack, J. C. , Nguyen, K. , Player, C. M. , Player, A. , Wright, M. , Roy, N. , Frosch, M. P. , McKee, A. C. , Wald, L. L. , Fischl, B. , van Leemput, K. , & Alzheimer's Disease Neuroimaging Initiative . (2015). A computational atlas of the hippocampal formation using ex vivo, ultra‐high resolution MRI: Application to adaptive segmentation of in vivo MRI. NeuroImage, 115, 117–137. 10.1016/j.neuroimage.2015.04.042 [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Juraska, J. M. (2024). The last stage of development: The restructuring and plasticity of the cortex during adolescence especially at puberty. Developmental Psychobiology, 66(2), e22468. 10.1002/dev.22468 [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Kennedy, J. T. , Collins, P. F. , & Luciana, M. (2016). Higher adolescent body mass index is associated with lower regional gray and white matter volumes and lower levels of positive emotionality. Frontiers in Neuroscience, 10, 413. 10.3389/fnins.2016.00413 [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Kokubun, K. , & Yamakawa, Y. (2019). Association between food patterns and gray matter volume. Frontiers in Human Neuroscience, 13, 384. 10.3389/fnhum.2019.00384 [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Kullmann, S. , Wagner, L. , Hauffe, R. , Kühnel, A. , Sandforth, L. , Veit, R. , Dannecker, C. , Machann, J. , Fritsche, A. , Stefan, N. , Preissl, H. , Kroemer, N. B. , Heni, M. , Kleinridders, A. , & Birkenfeld, A. L. (2025). A short‐term, high‐caloric diet has prolonged effects on brain insulin action in men. Nature Metabolism, 1–9. 10.1038/s42255-025-01226-9 [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Lee, H.‐S. (2007). Canonical correlation analysis using small number of samples. Communications in Statistics ‐ Simulation and Computation, 36(5), 973–985. 10.1080/03610910701539443 [DOI] [Google Scholar]
  19. Little, R. J. A. (1988). A test of missing completely at random for multivariate data with missing values. Journal of the American Statistical Association, 83(404), 1198–1202. 10.1080/01621459.1988.10478722 [DOI] [Google Scholar]
  20. Lommi, S. , Engberg, E. , Tuorila, H. , Kolho, K.‐L. , & Viljakainen, H. (2021). Sex‐ and weight‐specific changes in the frequency of sweet treat consumption during early adolescence: A longitudinal study. British Journal of Nutrition, 126(10), 1592–1600. 10.1017/S0007114521001112 [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. López, V. , Oyanedel, J. C. , Bilbao, M. , Torres, J. , Oyarzún, D. , Morales, M. , Ascorra, P. , & Carrasco, C. (2017). School achievement and performance in Chilean high schools: The mediating role of subjective wellbeing in school‐related evaluations. Frontiers in Psychology, 8, 1189. 10.3389/fpsyg.2017.01189 [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Lowe, C. J. , Morton, J. B. , & Reichelt, A. C. (2020). Adolescent obesity and dietary decision making—A brain‐health perspective. The Lancet Child & Adolescent Health, 4(5), 388–396. 10.1016/S2352-4642(19)30404-3 [DOI] [PubMed] [Google Scholar]
  23. Mirwald, R. L. , Baxter‐Jones, A. D. G. , Bailey, D. A. , & Beunen, G. P. (2002). An assessment of maturity from anthropometric measurements. Medicine and Science in Sports and Exercise, 34(4), 689–694. 10.1097/00005768-200204000-00020 [DOI] [PubMed] [Google Scholar]
  24. Mottis, G. , Kandasamey, P. , & Peleg‐Raibstein, D. (2025). The consequences of ultra‐processed foods on brain development during prenatal, adolescent and adult stages. Frontiers in Public Health, 13, 1590083. 10.3389/fpubh.2025.1590083 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Mou, Y. , Blok, E. , Barroso, M. , Jansen, P. W. , White, T. , & Voortman, T. (2023). Dietary patterns, brain morphology and cognitive performance in children: Results from a prospective population‐based study. European Journal of Epidemiology, 38(6), 669–687. 10.1007/s10654-023-01012-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Naveed, S. , Lakka, T. , & Haapala, E. A. (2020). An overview on the associations between health behaviors and brain health in children and adolescents with special reference to diet quality. International Journal of Environmental Research and Public Health, 17(3), 953. 10.3390/ijerph17030953 [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Ratner, R. , Hernández, P. , Martel, J. , & Atalah, E. (2017). Propuesta de un nuevo índice de calidad global de la alimentación. Revista Chilena de Nutricion, 44(1), 33–38. 10.4067/S0717-75182017000100005 [DOI] [Google Scholar]
  28. Reichelt, A. C. , & Rank, M. M. (2017). The impact of junk foods on the adolescent brain. Birth Defects Research, 109(20). 10.1002/bdr2.1173 [DOI] [PubMed] [Google Scholar]
  29. Schober, P. , Boer, C. , & Schwarte, L. A. (2018). Correlation coefficients: Appropriate use and interpretation. Anesthesia and Analgesia, 126(5), 1763–1768. 10.1213/ANE.0000000000002864 [DOI] [PubMed] [Google Scholar]
  30. Selmeczy, D. , Fandakova, Y. , Grimm, K. J. , Bunge, S. A. , & Ghetti, S. (2019). Longitudinal trajectories of hippocampal and prefrontal contributions to episodic retrieval: Effects of age and puberty. Developmental Cognitive Neuroscience, 36, 100599. 10.1016/j.dcn.2018.10.003 [DOI] [PMC free article] [PubMed] [Google Scholar]
  31. Sha, Z. , Pepe, A. , Schijven, D. , Carrión‐Castillo, A. , Roe, J. M. , Westerhausen, R. , Joliot, M. , Fisher, S. E. , Crivello, F. , & Francks, C. (2021). Handedness and its genetic influences are associated with structural asymmetries of the cerebral cortex in 31,864 individuals. Proceedings of the National Academy of Sciences of the United States of America, 118(47), e2113095118. 10.1073/pnas.2113095118 [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Solis‐Urra, P. , Olivares‐Arancibia, J. , Suarez‐Cadenas, E. , Sanchez‐Martinez, J. , Rodríguez‐Rodríguez, F. , Ortega, F. B. , Esteban‐Cornejo, I. , Cadenas‐Sanchez, C. , Castro‐Piñero, J. , Veloz, A. , Chabert, S. , Sadarangani, K. P. , Zavala‐Crichton, J. P. , Migueles, J. H. , Mora‐Gonzalez, J. , Quiroz‐Escobar, M. , Almonte‐Espinoza, D. , Urzúa, A. , Dragicevic, C. D. , … Cristi‐Montero, C. (2019). Study protocol and rationale of the “Cogni‐action project” a cross‐sectional and randomized controlled trial about physical activity, brain health, cognition, and educational achievement in schoolchildren. BMC Pediatrics, 19(1), 260. 10.1186/s12887-019-1639-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
  33. Stadterman, J. , Belthoff, K. , Han, Y. , Kadesh, A. D. , Yoncheva, Y. , & Roy, A. K. (2020). A preliminary investigation of the effects of a Western diet on hippocampal volume in children. Frontiers in Pediatrics, 8, 58. 10.3389/fped.2020.00058 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Stekhoven, D. J. , & Bühlmann, P. (2012). MissForest—Non‐parametric missing value imputation for mixed‐type data. Bioinformatics, 28(1). 10.1093/bioinformatics/btr597 [DOI] [PubMed] [Google Scholar]
  35. Stillman, C. M. , Esteban‐Cornejo, I. , Brown, B. , Bender, C. M. , & Erickson, K. I. (2020). Effects of exercise on brain and cognition across age groups and health states. Trends in Neurosciences, 43(7), 533–543. 10.1016/j.tins.2020.04.010 [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. von Hippel, P. T. , & Bogolasky Fliman, F. (2024). Did child obesity decline after 2016 food regulations in Chile? Revista Panamericana de Salud Pública, 48, e16. 10.26633/RPSP.2024.16 [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Winkler, A. M. , Renaud, O. , Smith, S. M. , & Nichols, T. E. (2020). Permutation inference for canonical correlation analysis. NeuroImage, 220, 117065. 10.1016/j.neuroimage.2020.117065 [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. World Medical Association . (2001). World Medical Association Declaration of Helsinki. Ethical principles for medical research involving human subjects. Bulletin of the World Health Organization, 79(4), 373–374. [PMC free article] [PubMed] [Google Scholar]
  39. Yekutieli, D. , & Benjamini, Y. (1999). Resampling‐based false discovery rate controlling multiple test procedures for correlated test statistics. Journal of Statistical Planning and Inference, 82(1), 171–196. 10.1016/S0378-3758(99)00041-5 [DOI] [Google Scholar]
  40. Zhang, R. , Zhang, B. , Shen, C. , Sahakian, B. J. , Li, Z. , Zhang, W. , Zhao, Y. , Li, Y. , Feng, J. , & Cheng, W. (2024). Associations of dietary patterns with brain health from behavioral, neuroimaging, biochemical and genetic analyses. Nature Mental Health, 2(5), 535–552. 10.1038/s44220-024-00226-0 [DOI] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Table S1. Associations between whole brain volumes and dichotomized intake of healthy food groups.

Table S2. Associations between whole brain volumes and dichotomized intake of unhealthy food groups.

APHW-18-0-s001.docx (28KB, docx)

Data Availability Statement

The data generated during and/or analyzed during the current study could be made available from the corresponding author upon reasonable request.

Articles from Applied Psychology. Health and Well-Being are provided here courtesy of Wiley

RESOURCES