Introduction

Lifestyle-related diseases, such as diabetes and hypertension, have resulted in a substantial increase in healthcare costs in recent years. These conditions are largely influenced by daily lifestyle habits, making it crucial to promote a healthy lifestyle from an early age. Adolescence, an important phase for establishing independent lifestyle choices, presents a critical opportunity to cultivate healthy behaviors, including diet and exercise, to prevent lifestyle-related diseases. However, studies have shown an elevated risk of such diseases among university students during this period, often due to factors such as unbalanced diets and skipping breakfast1,2.

Initiating a change in eating behavior may be relatively straightforward, but sustaining it poses a significant challenge. Long-term maintenance is hindered by the difficulty in perceiving improvements in physiological measures, such as body weight, leading to a reduced sense of self-efficacy. High dropout rates from improvement programs, often driven by increased fatigue and psychological distress, have been recognized as significant predictors of unsuccessful obesity treatment outcomes3,4.

Several studies have demonstrated the effectiveness of feedback in facilitating health behavior changes across domains such as dietary habits, physical activity, and smoking cessation5. Biological feedback, in particular, has shown promise in motivating behavioral adjustments and improving health-related outcomes, supported by robust evidence from prior research6,7. Additionally, personalized feedback has been highlighted as a key driver of behavioral change, offering tailored information that enhances awareness of current behaviors and fosters motivation8. These findings support the effectiveness of feedback in promoting health behavior change.

Building upon the established evidence for feedback’s role in driving health behavior change, the frontal pole cortex (FPC), located in the prefrontal region, has emerged as a key neural substrate underpinning complex cognitive processes such as decision-making, motivation, problem-solving, planning, attention, and goal-directed persistence9,10,11,12. Its involvement extends to future-oriented cognition, including prospection13 and outcome evaluation14,15,16,17, as well as the motivation required to sustain cognitive and physical effort18. Importantly, the FPC operates beyond the reward systems, directly contributing to persistence19,20. Boorman et al. demonstrated that the FPC could track the advantage of persisting current option versus alternative (counter-factual) option14. Another study also suggested that the FPC could subserve foraging (i.e., persisting) behavior in a changing environment and the direct current stimulation over the FPC enhances the motivation and persistence for their goals 21,22. On the other hand, He et al. showed that in subjects on a restricted diet, the frontal pole was more activated during NoGo trials with high-calorie foods than during Go trials, suggesting the importance of the frontal pole for self-control in eating behavior23.

The present study aimed to ascertain the involvement of the FPC in sustaining dietary improvements for long-term health, particularly among young individuals. Previous research has demonstrated that the pre-measurement of FP structure discriminates persistence for the near future—ranging from a few hours to several months—across various task domains19. Furthermore, personalized feedback has been shown to enhance the persistence of learning behaviors and induce plastic changes in the FPC19. Based on these findings, the present study tests the hypothesis that personalized dietary feedback—designed to reflect individual health outcomes and tailored to specific nutritional behaviors—can significantly influence the structural characteristics of FPC19, thereby enhancing the capacity for sustained health behavior change and reducing future health risks. To evaluate this hypothesis, we established two groups: the Personalized Feedback (PF) group and the Control Feedback (CF) group, matched for brain structure, gender, body mass index (BMI), and age. Participants in the PF group received individualized feedback based on a detailed analysis of their dietary intake, which included projections of future health risks associated with their current eating patterns. In contrast, the CF group was provided with generalized nutritional advice without specific reference to individual dietary behaviors or health risks. Our investigation focuses on the relationship between pre-measured structural characteristics of the FPC and persistence in pursuing long-term health goals. We hypothesize that in the PF group, the personalized feedback will enhance motivation to maintain dietary logs, altering the pre-existing correlation between FPC characteristics and persistence. Conversely, in the CF group, where feedback is limited to general nutritional advice, this correlation is expected to remain stable, reflecting the absence of behavioral adaptation. This approach will help us understand the impact of personalized feedback on health behavior, dependent on the FPC structure.

Finally, we explore improvements in dietary nutrients and anxiety reduction associated with different types of feedback. This focus is motivated by evidence that young adults with suboptimal dietary habits often experience poorer psychological well-being2, including elevated anxiety levels associated with nutritional inadequacies24. Dietary improvements can yield significant benefits for mental health, as seen in reduced anxiety symptoms following healthier eating interventions24,25. For instance, improved eating habits have been shown to attenuate anxiety in young adults26. Trait anxiety represents an individual’s enduring propensity to feel anxious, making it a more stable and informative measure of chronic mental health changes27. We thus hypothesized that participants receiving personalized dietary feedback would not only improve their nutritional intake but also exhibit a greater reduction in trait anxiety, reflecting enhanced overall mental well-being in the intervention group.

Results

Improved persistence, nutrition, and mental health through personalized feedback

A cohort study of 59 healthy college students was conducted, with participants randomly divided into two groups, and it included a 27-day dietary diary logging centered around the diligent maintenance of dietary diaries. The Personalized Feedback (PF) group received feedback on their nutrient intake, calculated from the submitted food records, and insights into potential future health risks associated with their current dietary habits. Conversely, the Control Feedback group (CF) received only generic nutritional information. All participants received feedback based on dietary analysis of 9 meals every 3 days at night (days 3, 6, 9, 12, 15, 18, 21, 24) during the 27-day intervention.

Firstly, to analyze whether there was a difference in the number of diary loggings between the CF and PF groups over the remaining 24 days, we performed an ANCOVA using the number of diary loggings (dietary logging persistence) during the baseline period (days 1–3) as a covariate to adjust for initial differences. The participants in the PF group submitted a notably higher number of food diaries compared to those in the CF group (F(1, 47) = 8.11, p = 0.007, η2 = 0.15) (Fig. 1a, Supplementary Table 1).

Fig. 1
figure 1

Effect of personalized and control feedback on food diary behavior and nutritional intake. (a) Bar plot showing the number of food diary loggings during the dietary intervention. Yellow bar represents the Personalized Feedback (PF) group, and blue bar represents the Control Feedback (CF) group. (b) A bar plot of the nutrient intake ratio compares the PF group (yellow bars) to the CF group (blue bars). A star symbol (*) denotes statistical significance (t-test with multiple comparison, p < 0.05). (c) The plots of trait anxiety scores pre and post the diary intervention for both the PF and CF groups. An asterisk (*) denote the significant difference with gray dotted lines illustrating individual participant changes (repeated measure ANOVA, p < 0.05).

Full size image

Next, we investigated whether differences in dietary composition between the PF and CF groups emerged over the 24-day intervention period, excluding the initial 3-day baseline period prior to the first feedback. Using ANCOVA, we analyzed improvements in nutrient intake across the 24 days, with baseline values (days 1–3) included as covariates to adjust for initial differences. The results revealed that the PF group showed significantly greater intakes of calcium (F(1, 46) = 14.9, p < 0.001, partial \(\:{\eta\:}_{p}^{2}\)= 0.10)= 0.25), vitamin A (F(1, 46) = 5.01, p = 0.03, partial\(\:{\eta\:}_{p}^{2}\)= 0.10, vitamin C (F(1, 46) = 5.77, p = 0.02, partial \(\:{\eta\:}_{p}^{2}\)=0.12), and protein(F(1, 46) = 8.17, p = 0.06, partial \(\:{\eta\:}_{p}^{2}\)= 0.15) compared to the CF group during the 24-day period (Fig. 1b, Supplementary Table 2).

Furthermore, we assessed the reduction of anxiety using the State trait anxiety (STAI) as described by Spielberger et al.28 A two-way repeated measure ANOVA on the trait anxiety (STAI-T) scores revealed a significant main effect of time　(F(1, 46) = 17.99, p < 0.001, η2 = 0.02) and an interaction effect between time and group (F(1, 46) = 8.87, p = 0.005, η2 = 0.01). Subsequent post-hoc analyses demonstrated that the PF group exhibited a notable reduction in trait anxiety levels post compared to pre (p = 0.001), suggesting that personalized feedback may contribute to sustained health behaviors and associated reductions in trait anxiety (Fig. 1c, Supplementary Tables 3,4).

FPC structural correlates of dietary-logging persistence differ by feedback condition

We hypothesized that in the absence of personalized feedback ( CF group), an individual’s persistence in dietary logging would reflect the prosperity of the FPC structure according to our previous study the prosperity of the FPC is associated with behavioral continuity19. In contrast, when personalized feedback was provided (PF group), this external motivation was expected to modulate behavior, diminishing any direct relationship between FPC structural features and logging persistence. To test this hypothesis, we performed correlation analyses between pre-intervention FPC structural measures and the number of weeks of sustained dietary logging in each group. T1-weighted and T2-weighted magnetic resonance imaging (MRI) and diffusion-weighted MRI were obtained before the diary submission. We defined the number of food diary loggings during the diary submission as an indicator of health behavior persistence for the distant future health because of the health behavior based on analysis of submitted food diaries. For each of the images capturing fractural anisotropy (FA), thickness, and myelin (T1/T2 ratio), regions of interest (ROIs) were defined within the FPC. The correlation analyses were conducted within the ROI, specifically examining the relationship between these neuroimaging parameters and the frequency of 24-day food diary loggings. In the CF group, persistence in dietary diary submission was positively correlated with multiple structural features of the FPC. In particular, participants’ diary submission frequency showed significant positive correlations with left FPC cortical thickness (r = 0.570, p = 0.005), left FPC myelin (T1w/T2w ratio; r = 0.559, p = 0.006), and fractional anisotropy (FA) in the FPC bilaterally (left: r = 0.707, p < 0.001; right: r = 0.547, p = 0.010), as shown in Fig. 2. In contrast, the PF group showed no significant correlations between diary submission frequency and any FPC feature (Supplementary Table 5). No significant negative correlations were observed in either group.

Fig. 2
figure 2

Correlation between the Structure of Frontal Pole Cortex and Food Diary Logging Frequency in the Control Group. Regression analysis reveals significant correlations within the frontal pole area, highlighted in red, indicating a family-wise error-corrected p-value of less than 0.05 (p < 0.05, FWE). (a) Left: The correlation between FPC cortical thickness and the number of food diary loggings. Right: Scatter plots depict the relationship between the mean values of cortical thickness in the FPC and the frequency of food diary loggings. Data points are color-coded by group, with the PF group in yellow and the CF group in blue (b) Left: An association between the T1-weighted to T2-weighted (T1w/T2w) signal ratio and the numbers of food diary loggings. Right: Scatter plots depict the relationship between the mean values of the T1w/T2w ratio in the FPC and the number of food diary loggings. (c) Left: The relationship between fractional anisotropy (FA) beneath the FPC and the number of food diary loggings. Right: Scatter plots depict the relationship between the mean values of FA beneath the FPC and the number of food diary loggings.

Full size image

Discussion

The effectiveness of personalized nutrition strategies is a burgeoning area of research with substantial implications for public health. This study investigated the role of a personalized feedback mechanism, leveraging daily food diaries, to influence healthier eating behaviors, with a particular focus on the frontal pole cortex’s (FPC) involvement in long-term dietary change. The research employed a randomized control design, with participants assigned to either a personalized feedback group (PF group) or a control feedback group (CF). Those in the PF group received individualized dietary feedback, which led to statistically significant improvements in dietary patterns when compared to the CF. Notably, in the PF group, a significant reduction in anxiety was shown, which may have also contributed to an improvement in well-being.

In the CF group that did not receive personalized feedback, we found significant positive correlations between FPC structural features – specifically cortical thickness, myelin content (T1w/T2w ratio), fractional anisotropy (FA), and behavioral persistence in food diary logging. This finding aligns with the FPC’s role in goal maintenance, self-regulation, and future-oriented behavior19,28,29,30,31,32,33. Without personalized feedback, individuals might rely on their intrinsic cognitive control capacities; consequently, those with a greater FPC structure could sustain the logging behavior more effectively. By contrast, in the PF group, personalized feedback was anticipated to externally reinforce behavioral persistence. Consequently, such tailored external reinforcement would reduce reliance on the pre-interventional prosperity of FPC. Indeed, our analyses revealed no significant correlation between the number of dietary logs and the structure of FPC in the PF group. This absence of correlations suggests that personalized feedback may serve as a potent external moderator of behavior, reducing the direct impact of baseline variability in FPC structure. Although the present study did not directly measure neuroplastic changes in the FPC, the observed lack of correlation within the PF group might imply that personalized feedback effectively modulates or overrides intrinsic neural predispositions underlying behavioral persistence.

Our findings delineate a significant correlation between the structural characteristics of the frontal pole cortex (FPC) and the consistency of daily dietary record-keeping. This relationship resonates with prior studies that have implicated the FPC in maintaining various behaviors, including language acquisition and motor learning19. Complementary evidence linking FPC activation with future-oriented thinking and motivational enhancement bolsters this association29,30. The role of FPC in complex decision-making processes, especially in dietary choices, emerges from its contribution to cognitive control, notably self-regulation, and strategic adaptability in the face of evolving contexts32,33. Such cognitive flexibility is crucial for the continuous appraisal and modification of eating patterns and maintenance of dietary modifications34. The FPC integrates multiple information streams to facilitate self-regulation in the diet. The observed consistency in dietary record-keeping underscores the role of FPC in long-term planning and goal-directed behavior.

Furthermore, the perseverance observed in participants maintaining a dietary diary for long-term health objectives underscores the FPC function in pursuing and realizing extended goals. Moreover, within the domain of behavioral psychology, the strategy of “pre-commitment” has been identified as an effective countermeasure to procrastination. This approach involves the deliberate limitation of future options to ensure prioritization of long-term objectives over immediate gratifications35,36. By establishing such constraints, pre-commitment serves as a safeguard against the allure of tempting alternatives that may derail goal-directed behavior. Empirical evidence suggests that the functionality of the FPC is intricately involved in the facilitation of pre-commitment strategies, as opposed to behaviors characterized by procrastination37,38. Further research utilizing transcranial direct current stimulation (tDCS) targeting the frontal pole has demonstrated an increase in the propensity for individuals to engage in pre-commitment choices, substantiating the FPC’s role in the neurocognitive processes underlying such decision-making strategy29,39.

Personalized feedback, by tailoring advice to individual dietary behaviors, has demonstrated efficacy in aligning with nutritionally advised healthy intake, underscoring the superiority of customized guidance over generic information in improving health outcomes40. In personalized learning contexts, feedback is a critical factor for promoting self-reflection, self-regulation, and self-assessment, which are all essential for active learning and lifelong success41. These metacognitive abilities, which encompass modifying behaviors, thoughts, emotions, and environmental interactions, are pivotal in achieving personal health goals. Previous studies have supported that metacognition is associated with various healthy lifestyle behaviors, such as regular physical activity, screen time management, and adherence to a Mediterranean diet42,43,44. Studies with fMRI have demonstrated that the FPC, specifically Brodmann area 10, was also activated during metacognitive tasks that involve self-evaluation and decision-making45. Moreover, the critical role of the frontal pole in orchestrating complex cognitive functions such as problem-solving and planning has been emphasized, functions that are integral to metacognitive operations46. It is postulated that personalized feedback may potentiate metacognitive processes involving the FPC, hence providing a particularly beneficial strategy for individuals dedicated to establishing healthy eating habits with a future-oriented perspective. The motivational potency of such feedback is presumably heightened when it is meticulously attuned to the recipient’s unique needs and objectives.

Moreover, our study reinforces the notion that personalized nutritional feedback aimed at promoting a balanced diet significantly enhances daily nutrient intake—particularly protein, calcium, and vitamins A and C—and contributes to reduced anxiety levels. This aligns with existing research which suggests that targeted dietary interventions can have beneficial effects on mental health by reducing anxiety47,48. In our study, individuals in the personalized feedback group showed significantly higher intakes of calcium, vitamin A, vitamin C, and protein compared to those in the control group. These nutrients are well-documented for their roles in mood enhancement and resilience building, consistent with previous findings49,50,51,52. Additionally, protein, vital for neurotransmitter synthesis, has been linked to improved stress management and mood regulation53. The individualized feedback increased the regularity of food diary documentation, implying enhancements in both nutritional intake and well-being with reduction in anxiety50. On the other hand, significant changes were not observed in nutrients other than vitamin A, C, calcium, and protein. Past studies highlight that long-term and targeted approaches are often necessary to achieve meaningful dietary changes54. In addition, the interventions with behavior-change techniques such as self-monitoring and goal-setting should be employed to enhance effectiveness by addressing underlying dietary habits55. Our results underscore the challenges of inducing nutrient intake changes within a limited timeframe and emphasize the importance of incorporating sustained, behaviorally-informed, and individualized strategies.

The present study provides important insights into the role of the FPC in sustaining long-term, goal-directed health behaviors, but several limitations should be addressed. First, brain imaging data were collected only at baseline, limiting the analysis to correlations between the persistence of health behaviors and the structural characteristics of the FPC. This approach precludes the investigation of potential neuroplastic changes induced by personalized feedback. Future studies should adopt longitudinal neuroimaging methods to capture dynamic changes in FPC structure and function before and after interventions. Second, the measurement of persistence was based solely on the frequency of dietary loggings. While informative, this metric does not fully capture the complexity of sustained health behaviors. Expanding future research to include more comprehensive measures, such as the quality and consistency of dietary records, would provide deeper insights into the factors driving behavioral persistence.

Despite these limitations, the present study offers the first evidence, to our knowledge, of the involvement of the FPC in maintaining long-term health behaviors. Furthermore, personalized feedback, by engaging FPC-associated mechanisms, was shown to significantly enhance dietary adherence and reduce the anxiety. We propose that the structural characteristics of the FPC may serve as a valuable marker for identifying individuals who are more likely to sustain health-promoting behaviors, paving the way for the development of more targeted and effective interventions.

Methods

We recruited 59 participants for this experiment. To ensure sufficient statistical power to detect significant differences in logging rates as previously observed, we conducted a power analysis informed by submission rates from our prior study19 (Hosoda et al., 2020). In that study, the personalized feedback group achieved a submission rate of 92.7%, significantly higher than the 47.5% observed in the general feedback group, with a moderate effect size (φ = 0.49). Based on these parameters, the analysis indicated that a minimum of 50 participants was necessary to achieve 95% power at a 5% alpha level. The final analysis included 50 participants, with 25 in the personalized feedback (PF) group (female: 11; mean age ± SD = 21.44 ± 1.81, 19 -24y.o) and 25 in the control feedback (CF) group (female: 9; mean age ± SD = 21.68 ± 2.68, 19–23 y.o). The mean (± SD) weight, height, and BMI for the PF group were 57.8 ± 9.2 kg, 165.7 ± 7.6 cm, and 21.0 ± 3.2, respectively. For the CF group, the corresponding values were 59.8 ± 6.9 kg, 168.8 ± 7.1 cm, and 21.0 ± 2.5, respectively. No significant differences were observed between the two groups for these measures (weight p = 0.13, height p = 0.86, BMI p = 0.30), independent samples t-test). Additionally, basic brain structure, including thickness, FA, and myelin (T1/T2 ratio) did not show the significant differences between the groups (t-test). Nine participants were excluded from the analysis for various reasons: three because they completely stopped submitting their food log altogether during the diary submission, three due to the trouble during brain imaging and all brain images could not be taken, two for not receiving the allocated food diary, and one for the days and nights of his life were reversed. All participants self-reported as right-handed using the Edinburgh Handedness Test. None of the participants had a history of neuropsychiatric disorders, psychotropic medication use, head injuries, or previous participation in psychotropic medication studies. Written informed consent was obtained from all participants following the study protocol approved by the institutional review board at the University of Tokyo (348-7) and the Human Ethics Committee of Kao Corporation (Sumida-ku, Tokyo, Japan). This study adhered to the principles of the Declaration of Helsinki.

Food diary logging and behavioral data

Before the start of the main test, the participants practiced recording their meals and photos on the Internet or their smartphones for three days and took the main test over 27 days. To record their data, the participants accessed the URL described in the email sent from the research staff every morning, inputting their meal content and uploading their meal photos. The meal details and meal photos recorded on the Web server were downloaded by the researcher, and the Asken application (Asken Inc., Shinjuku, Tokyo, Japan)56 was used to analyze the amounts of 13 nutrients (protein, fat, carbohydrates, calcium, iron, vitamin A, vitamin B1, vitamin B2, vitamin C, vitamin E, dietary fiber, cholesterol, and sodium) in the food.

Participants were required to maintain and submit detailed food diaries along with photographs documenting their food intake over a 27-day period. Every third day during the intervention—specifically on days 3, 6, 9, 12, 15, 18, 21, and 24—participants received personalized or general feedback based on a dietary analysis of the nine meals consumed over the preceding days. This feedback was designed to enhance dietary awareness and facilitate healthy eating behaviors by providing insights tailored to their dietary habits (Personalized Feedback, PF) according to the Standard Tables of Food Composition in Japan—2015, which were calculated according to the 3-day food records every third night, or general dietary guidelines (Control Feedback, CF). Feedback by email was sent every three days during the 27th day period, for a total of nine times.

In addition, the participants were evaluated using the State-Trait Anxiety Inventory (STAI)28. It comprises 40 items, with 20 describing state anxiety (STAI-S) and 20 describing trait anxiety (STAI-T). Participants rated their emotions on a scale of 1–4, with a minimum score of 20 and a maximum score of 80. The trait anxiety was assessed using the officially licensed Japanese version of the State-Trait Anxiety Inventory (2000). The STAI was legally obtained by purchasing the required number of copies from the authorized distributor, SACCESS.BELL Inc. (https://www.saccess55.co.jp/kobetu/detail/stai_s.html), in full compliance with Japanese copyright regulations.

Statistical analysis of the behavioral data and nutrition intake

To quantify and compare the persistence in dietary logging between the PF and CF groups, an Analysis of Covariance (ANCOVA) was performed. The dependent variable was the number of food diaries submitted over the 24-day intervention period, excluding the initial three days to control for initial adaptation effects. The independent variable was the group (PF/CF), with the baseline submission number (first three days), included as a covariate to control for initial compliance levels.

Furthermore, to assess and compare the nutritional quality of the diets between the two groups, an additional ANCOVA was utilized. This analysis focused on thirteen key nutrients: protein, fat, carbohydrates, calcium, iron, vitamin A, vitamins B1 and B2, vitamin C, vitamin E, dietary fiber, cholesterol, and sodium. The intake of these nutrients was analyzed for the 24-day period, with baseline nutrient values (from the initial three days) serving as covariates to account for individual dietary habits at the start of the intervention.

The effects of personalized and general feedback on anxiety levels, as measured by the State-Trait Anxiety Inventory (STAI)28, were assessed using a two-way repeated measures ANOVA. STAI comprises 40 items, with 20 describing state anxiety (STAI-S) and 20 describing trait anxiety (STAI-T). Participants rated their emotions on a scale of 1–4, with a minimum score of 20 and a maximum score of 80. The analysis considered time (pre vs. post-intervention) and group (PF vs.CF) as factors to identify significant changes in anxiety over the intervention.　Bonferroni-corrected post hoc tests were employed to dissect these interactions and ascertain the specific differences between groups at each time point.

Image acquisition

All images were scanned on a 3.0 T MR scanner (PRISMA trio, Simence, Munich, Germany) using a 64-channel phased-array head coil. T1-weighted images were acquired by a magnetization-prepared rapid gradient-echo (MPRAGE) sequence: repletion time (TR) = 1900 ms, echo time (TE) = 3 ms, flip angle: = 9°, slice thickness = 1 mm, field of view (FOV) = 256 × 256 mm2, voxel size = 1 × 1 × 1 mm3. T2-weighted images were acquired by a turbo spin-echo sequence: TR = 3200 ms, TE = 410 ms, flip angle: = 120°, slice thickness = 1 mm, FOV = 256 × 256 mm2, voxel size = 1 × 1 × 1 mm3. Additionally, whole-brain diffusion-weighted images (DWI) were acquired with TE of 62 ms, TR of 6500 ms, FOV = 276 × 276 mm, matrix size of 128 × 128, flip angle of 90°, 68 slices, and 1 × 1 × 1 mm3 isotropic voxels.

Image preprocessing

T1w and T2w images were preprocessed with the PreFreeSurfer, FreeSurfer and PostFreeSurfer using the HCP pipeline (ver.4.3.0, https://github.com/Washington-University/HCPpipelines/tree/master), Freesurfer FMRIB Software Library (FSL, ver.6.0.0), and Connectome Workbench (ver.1.5.0), to generate cortical thickness and myelin maps. Further details of these pipelines can be found in the reference57. PreFreeSurfer carried out bias field correction and registration of the T2w image to the T1w image, FreeSurfer generated pial-gray matter (GM) and GM-white matter surfaces using additional information from the T2w image. Post freesurfer used a Multimodal Surface Matching algorithm (MSM)58 based on cortical curvature and myelination to space (164k fs LR mesh). For diffusion image preprocessing, FSL (ver. 6.0.0) was used to perform voxel-wise statistical analysis of FA data via TBSS. The data was corrected for eddy currents and head motion using FMRIB’s Diffusion Toolbox (FDT, part of FSL), and brain extraction was undertaken using the Brain Extraction Tool (BET, part of FSL). Following preprocessing, the FA data was aligned into 1 mm × 1 mm × 1 mm Montreal Neurological Institute (MNI) 152 Space using FMRIB’s Nonlinear Image Registration Tool (FNIRT). The mean FA image (threshold 0.2) was created and thinned to make a mean FA skeleton representing the centers of all tracts common to each group. Each subject’s aligned FA data was then projected onto this skeleton, and the resulting data were fed into voxel-wise cross-subject statistics.

Image data analysis

We conducted a general linear model (GLM) using the Permutation Analysis of Linear Models (PALM) toolbox59 to analyze thickness and T1w/T2w ratio. We used statistical parametric regression analysis to detect the correlation between cortical thickness and Myelin and the number of food diary loggings within 24-day in FG and CF, correcting for age and sex as nuisance covariates. Statistical significance was evaluated with 5000 permutations and family-wise error (FWE) correction with threshold-free cluster enhancement (TFCE) corrected60, and we used a region of interest (ROI) mask of frontal pole cortex created from each participant’s surface file (*aparc.164k_fs_LR.label.gii). After correcting for both hemispheres, the significance threshold was set at p < 0.05. For FA data, we used FSL/Randomise59 to generate a test statistic image and p-value image and performed voxel-weighted statistical analysis to detect a significant correlation between FA data and the numbers of food diary within 24-day both in CF and FG using the FSL Randomize program. We used permutation-based statistical analysis with 500 permutations, corrected for multiple comparisons using the TFCE method with FWE correction, and applied a significance threshold of p < 0.05 within an ROI mask of the frontal pole.