Solitary eating is linked to reduced brain volume in cognitively unimpaired individuals and possible future cognitive decline

Abstract

Social isolation and poor nutrition are recognized risk factors for cognitive decline in older adults. However, the relationship between solitary eating, brain structure, and dietary patterns remains unclear. This study investigated the relationship between solitary eating, brain structure, and diet in healthy older Japanese adults. A total of 727 cognitively normal participants (mean age 70.32 years) underwent MRI and dietary assessments. Results showed that individuals who frequently eat alone have reduced brain volumes in regions critical for cognitive function, particularly the medial temporal lobe and hippocampus. These individuals also demonstrated less healthy dietary patterns, characterized by higher sugar and alcohol consumption, with lower intake of proteins, vitamins, and minerals. Analysis revealed that the difference in hippocampal volume between solitary and social eaters diminished when accounting for dietary factors, suggesting that nutritional factors primarily drive this relationship. However, the difference in medial temporal lobe volume remained significant after dietary adjustment, indicating that other factors, such as reduced social interaction, may affect this region. These findings emphasize the importance of social eating for maintaining brain health in older adults, suggesting that both social and nutritional aspects of eating are crucial. This research provides insights for developing healthy aging strategies that address both social and nutritional dimensions of eating behavior.

Introduction

The social context of eating has received increasing attention in recent years as a potential factor influencing cognitive health and brain structure in older adults. As populations globally continue to age, understanding the mechanisms that promote healthy cognitive aging has become a critical public health priority^1,2. Solitary eating, defined as the habitual consumption of meals alone, has been linked to various health outcomes, including nutritional status and psychological well-being³. Recent research has begun to explore the possible connections between social isolation and brain health. For example, studies have reported associations between lower frequencies of social contact and greater brain atrophy in community-dwelling older adults^4,5. Additionally, dietary patterns are recognized as key contributors to cognitive function and brain structure, with certain nutrients and dietary components being associated with better cognitive outcomes and preserved brain volumes^6,7. However, the complex interplay between solitary eating, dietary patterns, and brain structure remains poorly understood, particularly in the context of healthy cognitive aging⁸.

Although the associations between social isolation, diet, and brain health have been studied individually, several key gaps remain in understanding how these factors interact, particularly in relation to solitary eating. First, most studies have focused on general social isolation rather than specifically examining the impact of solitary eating on brain structure^5,9. This leaves an important gap in our knowledge about whether solitary eating, independent of overall social engagement, has distinct effects on brain health. Second, the mechanisms through which solitary eating may affect brain structure are not well established. Some studies have proposed potential pathways involving stress, inflammation, and changes in the gut microbiome^10–12, but the specific role of dietary patterns in mediating these effects remains unclear. Furthermore, there is a shortage of comprehensive studies that simultaneously assess solitary eating, detailed dietary intake, and regional brain volumes, making it difficult to disentangle the relative contributions of social and nutritional factors to brain health^13,14. Additionally, much of the research in this area has been conducted in Western populations, leaving a significant gap in understanding how these relationships may manifest in different cultural settings, particularly in Asian populations where dining customs and social norms around eating may differ^15,16. Addressing these gaps is crucial for developing targeted interventions to promote healthy cognitive aging and to fully understand the range of factors that contribute to brain health in older adults¹⁷.

The present study aims to address these critical gaps by investigating the relationships between solitary eating, regional brain volumes, and dietary patterns in cognitively healthy older adults in Japan. Specifically, our objectives are threefold: (1) to examine the relationship between solitary eating and regional brain volumes, with a particular focus on regions important for cognitive function, such as the hippocampus and medial temporal lobe¹⁸; (2) to analyze the dietary patterns associated with solitary eating and their potential mediating role in the relationship between solitary eating and brain structure¹⁹; and (3) to explore how these associations persist or change when adjusting for both nutrient intake and specific food and beverage consumption. By examining these factors in a non-Western population, our study seeks to provide a more comprehensive understanding of the complex interplay between social eating habits, nutrition, and brain health. This research has the potential to inform culturally appropriate interventions aimed at promoting healthy cognitive aging and to contribute to the growing body of evidence on the importance of social factors in maintaining brain health in older adults^20,21.

Results

Participant characteristics

This study included 727 participants, with 598 (82.3%) engaging in social eating and 129 (17.7%) in solitary eating. In Table 1, the mean age of the total sample was 70.32 years, with the solitary eating group (71.17 ± 7.09 years) being slightly older than the social eating group (70.07 ± 6.30 years), though this difference was not statistically significant. Women comprised 56.8% of the total sample, with similar proportions in both eating style groups (social: 57.1%, solitary: 55.8%). Health indicators showed no significant differences between the two groups. Hypertension prevalence was comparable (social: 52.1%, solitary: 53.5%). Mean serum LDL cholesterol levels (social: 116.21 ± 32.48 mg/dL; solitary: 113.58 ± 33.26 mg/dL) and HDL cholesterol levels (social: 61.47 ± 15.06 mg/dL; solitary: 59.68 ± 15.56 mg/dL) were also similar. The prevalence of diabetes mellitus did not differ significantly between social (15.2%) and solitary (16.3%) eaters.

Table 1.

Characteristics of the study population based on eating style

		Eating style
Variables	Total	Median 1	Median 2
		Social	Solitary
N	727	598	129
Age, mean (SD), y	70.32	70.07 (6.30)	71.17 (7.09)
Women, %	56.8	57.1	55.8
Hypertension, %	52.3	52.1	53.5
Serum LDL-chol, mean (SD), mg/dL	115.67	116.21 (32.48)	113.58 (33.26)
Serum HDL-chol, mean (SD), mg/dL	61.05	61.47 (15.06)	59.68 (15.56)
Diabetes mellitus, %	15.4	15.2	16.3
Formal education, mean (SD), y	11.34	11.35 (2.23)	11.41 (2.28)
BMI, mean (SD)	23.85	23.80 (3.24)	23.87 (3.71)
Smoking, %	11.1	10.2	15.5
Exercise habits, %	44.4	45.4	39.8

Values are expressed as the mean (SD) unless otherwise stated. *p < 0.05 versus social eating group.

Participants in both groups had comparable educational backgrounds, with mean years of formal education being 11.35 ± 2.23 for social eaters and 11.41 ± 2.28 for solitary eaters. Body Mass Index (BMI) was also similar between the groups (social: 23.80 ± 3.24; solitary: 23.87 ± 3.71). Lifestyle factors showed apparent differences, but these were not statistically significant. The prevalence of smoking was higher in the solitary eating group (15.5%) compared to the social eating group (10.2%). Similarly, exercise habits were more common among social eaters (45.4%) than solitary eaters (39.8%).

Brain volume differences between social and solitary eating styles

All brain region volumes were normalized to the estimated total intracranial volume (eTIV) and expressed as a percentage to account for individual differences in head size. In Table 2, we observed differences in the regional brain between social and solitary eating styles, after adjusting for various confounding factors and applying Bonferroni correction for multiple comparisons across the combined hemisphere brain regions. Total brain volume (TBV) was lower in the solitary eating group (58.28%, 95% CI: 57.76–58.80) compared to the social eating group (59.02%, 95% CI: 58.77–59.26), with an F-value of 6.31 and -log10(p) of 1.91. Several specific brain regions also showed smaller volumes in the solitary eating group compared to the social eating group. These regions included the medial temporal lobe (1.812% vs 1.852%, F = 5.93, -log10(p) = 1.82), parietal lobe (5.918% vs 6.019%, F = 5.44, -log10(p) = 1.70), occipital lobe (2.375% vs 2.437%, F = 8.98, -log10(p) = 2.55), insula (0.817% vs 0.833%, F = 4.49, -log10(p) = 1.46), and hippocampus (0.473% vs 0.484%, F = 4.69, -log10(p) = 1.51). The occipital lobe demonstrated the most pronounced difference between groups, with the highest F-value (8.98) and the lowest p-value (-log10(p) = 2.55) among all examined brain regions. Other brain regions, including white matter hyperintensities (WMH), frontal lobe, lateral temporal lobe, cingulum, and amygdala, did not show statistically significant differences between the two groups, with -log10(p) values ranging from 0.24 to 1.11.

Table 2.

Differences in regional brain volumes among eating styles

	Eating Style
Variables	Median 1	Median 2	F-value	-Log10(p-value)
	Social	Solitary
Brain volume to eTiv, %
TBV	59.02 (58.77–59.26)	58.28 (57.76–58.80)*	6.31	1.91
WMH	0.322 (0.297–0.346)	0.339 (0.286–0.392)	0.31	0.24
Frontal lobe	9.128 (9.075–9.181)	9.042 (8.927–9.157)	1.78	0.74
Temporal lobe	6.008 (5.971–6.045)	5.929 (5.850–6.008)	3.14	1.11
Temporal lobe, medial	1.852 (1.838–1.865)	1.812 (1.784–1.841)*	5.93	1.82
Temporal lobe, lateral	4.156 (4.128–4.184)	4.117 (4.057–4.177)	1.38	0.62
Parietal lobe	6.019 (5.983–6.055)	5.918 (5.841–5.995)*	5.44	1.70
Occipital lobe	2.437 (2.420–2.455)	2.375 (2.338–2.412)*	8.98	2.55
Insula	0.833 (0.827–0.839)	0.817 (0.805–0.830)*	4.49	1.46
Cingulum	1.133 (1.123–1.142)	1.114 (1.093–1.135)	2.58	0.96
Hippocampus	0.484 (0.480–0.489)	0.473 (0.463–0.482)*	4.69	1.51
Amygdala	0.165 (0.163–0.167)	0.164 (0.160–0.167)	0.61	0.36

Values are presented as mean values (95% confidence interval). In the multivariable-adjusted model, the values were adjusted for age, sex, educational level, hypertension, diabetes mellitus, BMI, smoking, exercise habits, and LDL and HDL cholesterol levels. ^*Bonferroni-adjusted p-values, p < 0.05 versus social eating group. A complete list of analyzed regional brain volume is provided in Supplementary Table S1.

Table S1 provides a comprehensive analysis of regional brain volume differences between social and solitary eating styles. Several cortical regions showed significantly smaller volumes in the solitary eating group compared to the social eating group. In the left hemisphere, these included the cuneus (F = 5.86, -log10(p) = 1.80), isthmuscingulate (F = 8.65, -log10(p) = 2.47), lingual (F = 4.14, -log10(p) = 1.38), middle temporal (F = 4.51, -log10(p) = 1.47), pericalcarine (F = 13.94, -log10(p) = 3.69), and transverse temporal (F = 3.95, -log10(p) = 1.33) areas. In the right hemisphere, significant differences were observed in the entorhinal (F = 6.48, -log10(p) = 1.95), fusiform (F = 4.49, -log10(p) = 1.46), isthmuscingulate (F = 11.67, -log10(p) = 3.17), lateral occipital (F = 6.52, -log10(p) = 1.96), pericalcarine (F = 6.64, -log10(p) = 1.99), and precuneus (F = 8.66, -log10(p) = 2.47) regions. The pericalcarine cortex in the left hemisphere showed the most pronounced difference among all cortical regions (F = 13.94, -log10(p) = 3.69). Among subcortical structures, the left putamen (F = 5.36, -log10(p) = 1.68) and left hippocampus (F = 4.86, -log10(p) = 1.56) demonstrated significantly smaller volumes in the solitary eating group. Broader brain measures also revealed significant differences. The solitary eating group showed reduced volumes in total cortical gray matter (F = 4.57, -log10(p) = 1.48), subcortical gray matter (F = 5.42, -log10(p) = 1.69), and total gray matter (F = 5.30, -log10(p) = 1.67). Both left (F = 4.17, -log10(p) = 1.38) and right (F = 4.85, -log10(p) = 1.55) hemisphere cortical gray matter volumes were significantly smaller in the solitary eating group.

Differences in nutrient intake between social and solitary eating styles

Table 3 presents a comprehensive analysis of nutrient intake differences between social and solitary eating styles, after adjusting for various confounding factors and applying Bonferroni correction for multiple comparisons. Significant differences were observed in macronutrient intake between the two groups. The solitary eating group showed lower protein intake (F = 8.46, -log10(p) = 2.43) compared to the social eating group. Conversely, the solitary group had a higher carbohydrate energy ratio (F = 4.85, -log10(p) = 1.55) and alcohol consumption (F = 4.94, -log10(p) = 1.58). Polyunsaturated fatty acid intake was lower in the solitary eating group (F = 4.52, -log10(p) = 1.47).

Table 3.

Differences in nutrient intake among eating styles

Variables	Eating Style
	Median 1	Median 2	F-value	-Log10(p-value)
	Social	Solitary
Surplus ammonia, mg	18.75 (17.94–19.55)	15.56 (13.83–17.29)*	10.75	2.96
Beta-carotene equivalent, µg	3782.35 (3611.16–3953.55)	3137.95 (2769.62–3506.29)*	9.68	2.71
Alpha-carotene, µg	614.01 (583.19–644.84)	498.69 (432.37–565.02)*	9.56	2.68
Vitamin B6, mg	4.08 (3.91–4.24)	3.46 (3.11–3.82)*	9.37	2.64
N-3 polyunsaturated fatty acids, g	6.54 (6.28–6.80)	5.59 (5.03–6.14)*	9.3	2.62
Beta-carotene, µg	3067.48 (2921.17–3213.78)	2531.85 (2217.08–2846.62)*	9.16	2.59
Nitrate ion, g	0.22 (0.21–0.23)	0.18 (0.16–0.21)*	8.58	2.45
Glycine, mg	8212.63 (7878.90–8546.35)	7030.26 (6312.24–7748.28)*	8.58	2.45
Folic acid, µg	366.65 (354.43–378.87)	323.54 (297.24–349.84)*	8.5	2.44
22:0 behenic acid, mg	1841.79 (1737.25–1946.33)	1472.83 (1247.91–1697.75)*	8.51	2.44
Iron, mg	8.39 (8.17–8.60)	7.63 (7.17–8.10)*	8.47	2.43
Protein, g	550.64 (523.66–577.61)	455.69 (397.65–513.74)*	8.46	2.43
Mannitol, g	250.78 (237.36–264.19)	203.68 (174.81–232.55)*	8.42	2.42
18:2 octadecadienoic acid, mg	5.32 (5.00–5.63)	4.21 (3.54–4.89)*	8.41	2.42
24:0 lignoselinic acid, mg	875.63 (821.34–929.92)	690.84 (574.03–807.64)*	7.92	2.3
18:3 octadecatrienoic acid, mg	13.65 (12.93–14.37)	11.20 (9.65–12.75)*	7.9	2.29
8:0 octanoic acid, mg	987.09 (933.78–1040.41)	810.90 (696.19–925.62)*	7.46	2.19
Pantothenic acid, mg	6.60 (6.42–6.78)	6.02 (5.64–6.40)*	7.26	2.14
Potassium, mg	2834.32 (2758.71–2909.93)	2590.05 (2427.38–2752.72)*	7.13	2.11
Copper, mg	1.18 (1.15–1.20)	1.09 (1.03–1.15)*	7.06	2.09
Vitamin K, µg	279.87 (269.07–290.68)	246.80 (223.56–270.05)*	6.4	1.94
17:2 heptadecadienoic acid, mg	8.23 (7.79–8.68)	6.89 (5.93–7.85)*	6.19	1.88
Total dietary fiber, g	13.45 (13.04–13.85)	12.24 (11.36–13.11)*	6.08	1.86
Zinc, mg	8.31 (8.13–8.50)	7.76 (7.36–8.16)*	6.03	1.84
Trehalose, g	0.17 (0.17–0.18)	0.15 (0.13–0.17)*	5.81	1.79
Insoluble dietary fiber, g	9.21 (8.93–9.49)	8.40 (7.80–9.00)*	5.78	1.78
Tryptophan, mg	1096.36 (1064.51–1128.20)	1004.05 (935.53–1072.56)*	5.74	1.77
16:2 hexadecadienoic acid, mg	28.56 (27.25–29.86)	24.87 (22.05–27.68)*	5.43	1.7
Sucrose, g	22.49 (21.71–23.27)	24.68 (23.00–26.35)*	5.41	1.69
Sulfur-containing amino acids; cystine, mg	1370.64 (1333.42–1407.87)	1266.12 (1186.02–1346.21)*	5.39	1.69
Maltose, g	0.79 (0.73–0.85)	0.95 (0.82–1.08)*	5.19	1.64
Histidine, mg	2888.11 (2793.69–2982.54)	2628.78 (2425.63–2831.94)*	5.15	1.63
Alpha-tocopherol, mg	6.28 (6.10–6.46)	5.78 (5.38–6.18)*	5.02	1.6
Alcohol, g	10.69 (9.26–12.12)	14.53 (11.46–17.60)*	4.94	1.58
10:1 decenoic acid, mg	64.99 (61.65–68.32)	56.00 (48.82–63.17)*	4.97	1.58
Carbohydrate energy ratio 2, %	56.78 (56.20–57.36)	58.32 (57.07–59.57)*	4.85	1.55
18:3 n-6 gamma-linolenic acid, mg	7.19 (6.77–7.60)	6.09 (5.19–6.99)*	4.7	1.52
Polyunsaturated fatty acid, g	15.52 (15.08–15.96)	14.39 (13.44–15.33)*	4.52	1.47
22:1 docosenoic acid, mg	623.93 (594.67–653.20)	549.80 (486.84–612.77)*	4.38	1.44
Magnesium, mg	376.00 (365.67–386.33)	350.51 (328.28–372.74)*	4.16	1.38
Tridecanoic acid, mg	2.77 (2.54–2.99)	3.32 (2.84–3.81)*	4.13	1.37
Vitamin c, mg	110.95 (106.74–115.16)	100.75 (91.69–109.80)*	4.02	1.34
Organic acid, g	0.97 (0.93–1.02)	0.87 (0.77–0.96)*	3.97	1.33
Arginine, mg	4178.26 (4068.99–4287.52)	3915.04 (3679.95–4150.13)*	3.97	1.33
Threonine, mg	2865.66 (2785.27–2946.05)	2672.86 (2499.90–2845.82)*	3.93	1.32
Heptanoic acid, mg	0.87 (0.80–0.95)	1.05 (0.89–1.21)*	3.94	1.32
18:3 n-3 alpha-linolenic acid, mg	1452.10 (1403.00–1501.20)	1334.73 (1229.09–1440.37)*	3.9	1.31

Values are presented as mean values (95% confidence interval). In the multivariable-adjusted model, the values were adjusted for age, sex, educational level, hypertension, diabetes mellitus, BMI, smoking, exercise habits, and LDL and HDL cholesterol levels. ^*Bonferroni-adjusted p-values, p < 0.05 versus social eating group. Omitted in the figure are nutrients intake with -log10(p-value) <1.301. A complete list of analyzed nutrients is provided in Supplementary Table S2.

Several micronutrients showed significantly lower intake in the solitary eating group. These included vitamins such as beta-carotene equivalent (F = 9.68, -log10(p) = 2.71), vitamin B6 (F = 9.37, -log10(p) = 2.64), folic acid (F = 8.50, -log10(p) = 2.44), pantothenic acid (F = 7.26, -log10(p) = 2.14), vitamin K (F = 6.40, -log10(p) = 1.94), alpha-tocopherol (F = 5.02, -log10(p) = 1.60), and vitamin C (F = 4.02, -log10(p) = 1.34). Minerals such as iron (F = 8.47, -log10(p) = 2.43), potassium (F = 7.13, -log10(p) = 2.11), copper (F = 7.06, -log10(p) = 2.09), zinc (F = 6.03, -log10(p) = 1.84), and magnesium (F = 4.16, -log10(p) = 1.38) also showed lower intake in the solitary eating group.

The solitary eating group exhibited lower intake of several fatty acids, including N-3 polyunsaturated fatty acids (F = 9.30, -log10(p) = 2.62), behenic acid (F = 8.51, -log10(p) = 2.44), octadecadienoic acid (F = 8.41, -log10(p) = 2.42), and alpha-linolenic acid (F = 3.90, -log10(p) = 1.31). Additionally, lower intake of various amino acids was observed in the solitary eating group, including glycine (F = 8.58, -log10(p) = 2.45), tryptophan (F = 5.74, -log10(p) = 1.77), cystine (F = 5.39, -log10(p) = 1.69), histidine (F = 5.15, -log10(p) = 1.63), and arginine (F = 3.97, -log10(p) = 1.33).

Total dietary fiber intake was lower in the solitary eating group (F = 6.08, -log10(p) = 1.86), as was insoluble dietary fiber (F = 5.78, -log10(p) = 1.78). Notably, the solitary eating group showed lower intake of surplus ammonia (F = 10.75, -log10(p) = 2.96), nitrate ion (F = 8.58, -log10(p) = 2.45), and mannitol (F = 8.42, -log10(p) = 2.42). However, they had higher intake of sucrose (F = 5.41, -log10(p) = 1.69) and maltose (F = 5.19, -log10(p) = 1.64).

Differences in food and beverage intake between social and solitary eating styles

Table 4 presents a detailed analysis of food and beverage intake differences between social and solitary eating styles, after adjusting for various confounding factors and applying Bonferroni correction for multiple comparisons. Significant differences were observed in the consumption of various food and beverage items between the two groups. The solitary eating group showed higher intake of several items compared to the social eating group. Most notably, whisky consumption was markedly higher in the solitary eating group (F = 16.41, -log10(p) = 4.25). Other items with significantly higher intake in the solitary group included white sugar (F = 14.89, -log10(p) = 3.90), table salt (F = 13.51, -log10(p) = 3.59), and saccharides (F = 12.81, -log10(p) = 3.43).

Table 4.

Differences in food and beverage intake among eating styles

Variables	Eating Style
	Median 1	Median 2	F-value	-Log10(p-value)
	Social	Solitary
Whisky, g	0.35 (−0.33 to 1.02)	3.64 (2.20–5.09)*	16.41	4.25
White sugar, g	7.81 (7.51–8.12)	9.24 (8.58–9.90)*	14.89	3.9
Table salt, g	5.16 (5.01–5.31)	5.83 (5.51–6.15)*	13.51	3.59
Potato, g	24.09 (22.87–25.31)	18.68 (16.06–21.31)*	13.42	3.57
Saccharides, g	10.23 (9.65–10.81)	12.74 (11.50-13.99)*	12.81	3.43
Green juice kale, g	0.05 (−0.02 to 0.11)	0.30 (0.16–0.44)*	10.99	3.02
Mushrooms, g	14.84 (13.94–15.74)	11.26 (9.33–13.19)*	10.89	2.99
Soybean miso, g	8.90 (7.98–9.82)	5.27 (3.30–7.25)*	10.68	2.95
Shellfish, g	8.47 (7.77–9.17)	5.73 (4.21–7.24)*	10.41	2.88
Other vegetables, g	147.30 (140.40–154.20)	124.18 (109.33–139.03)*	7.66	2.24
Vegetables not high in beta-carotene, g	147.12 (140.24–153.99)	124.20 (109.41–138.99)*	7.59	2.22
Pickles, g	23.38 (21.12–25.65)	15.94 (11.06–20.82)*	7.37	2.17
Instant noodles, g	1.73 (0.96–2.50)	4.19 (2.54–5.84)*	7.03	2.09
Sweet sake, g	0.14 (−0.36 to 0.63)	1.69 (0.63–2.76)*	6.81	2.03
Mayonnaise egg yolk mold, g	2.37 (2.16–2.58)	1.71 (1.26–2.16)*	6.78	2.03
Mottled kidney bean, g	−0.01 (−0.32 to 0.30)	0.96 (0.30–1.62)*	6.72	2.01
Beef, g	12.90 (11.88–13.92)	9.76 (7.57–11.95)*	6.49	1.96
Noodles, g	47.10 (42.62–51.58)	60.82 (51.17–70.46)*	6.4	1.93
Vegetables high in beta-carotene, g	143.93 (136.12–151.73)	120.69 (103.90–137.49)*	6.05	1.85
Pork, g	23.23 (21.89–24.57)	19.70 (16.83–22.58)*	4.75	1.53
Grains, g	201.68 (196.76–206.60)	214.67 (204.09–225.26)*	4.77	1.53
Regular cow’s milk, g	67.37 (60.63–74.12)	84.93 (70.41–99.45)*	4.62	1.5
Boiled macaroni and spaghetti, g	0.37 (−0.39 to 1.12)	2.26 (0.64–3.89)*	4.32	1.42
Canned tomato juice, no salt added, g	1.12 (−0.54 to 2.77)	5.27 (1.71–8.83)*	4.31	1.42
Granulated sugar, g	2.03 (1.59–2.48)	3.15 (2.19–4.12)*	4.27	1.41
Regular soy sauce, g	1.96 (1.58–2.33)	1.02 (0.21–1.83)*	4.27	1.41
Orange concentrated juice, g	7.32 (4.71–9.93)	13.77 (8.15–19.38)*	4.17	1.38
Breads, g	23.82 (21.13–26.50)	30.35 (24.57–36.12)*	4.05	1.35
Sake lees, g	0.03 (−0.18 to 0.23)	0.52 (0.08–0.96)*	3.92	1.32

Values are presented as mean values (95% confidence interval). In the multivariable-adjusted model, the values were adjusted for age, sex, educational level, hypertension, diabetes mellitus, BMI, smoking, exercise habits, and LDL and HDL cholesterol levels. ^*Bonferroni-adjusted p-values, p < 0.05 versus social eating group. Omitted in the figure are food and beverage intake with -log10(p-value) <1.301. A complete list of analyzed food items is provided in Supplementary Table S3.

Interestingly, the solitary eating group also showed higher consumption of certain beverages and processed foods. These included green juice kale (F = 10.99, -log10(p) = 3.02), instant noodles (F = 7.03, -log10(p) = 2.09), sweet sake (F = 6.81, -log10(p) = 2.03), noodles in general (F = 6.40, -log10(p) = 1.93), and regular cow’s milk (F = 4.62, -log10(p) = 1.50). Conversely, the solitary eating group demonstrated lower intake of several food items compared to the social eating group. Notably, there was a significant reduction in the consumption of vegetables and plant-based foods. Potato intake was lower (F = 13.42, -log10(p) = 3.57), as was the consumption of mushrooms (F = 10.89, -log10(p) = 2.99), soybean miso (F = 10.68, -log10(p) = 2.95), other vegetables (F = 7.66, -log10(p) = 2.24), vegetables not high in beta-carotene (F = 7.59, -log10(p) = 2.22), and vegetables high in beta-carotene (F = 6.05, -log10(p) = 1.85). The solitary eating group also showed lower intake of certain animal-based foods, including shellfish (F = 10.41, -log10(p) = 2.88), beef (F = 6.49, -log10(p) = 1.96), and pork (F = 4.75, -log10(p) = 1.53). While the solitary group consumed more noodles and breads, they had a lower intake of regular soy sauce (F = 4.27, -log10(p) = 1.41), which is commonly used as a condiment in many Asian cuisines.

Regional brain volume differences between social and solitary eating styles after adjusting for nutrient intake

Table 5 presents the differences in regional brain volumes between social and solitary eating styles after adjusting for multiple confounding factors, including age, sex, educational level, hypertension, diabetes mellitus, BMI, smoking, exercise habits, LDL and HDL cholesterol levels, and 5 nutrient intake variables with significant differences in Table 3 and acceptable multicollinearity (VIF < 10).

Table 5.

Differences in regional brain volumes among eating styles

Variables	Eating Style
	Median 1	Median 2	F-value	-Log10(p-value)
	Social	Solitary
Brain volume to eTiv, %
TBV	59.0 (58.7–59.2)	58.3 (57.8–58.9)	4.66	1.51
WMH	0.32 (0.30–0.35)	0.33 (0.28–0.38)	0.04	0.08
Frontal lobe	9.13 (9.08–9.18)	9.05 (8.94–9.17)	1.31	0.6
Temporal lobe	6.01 (5.97–6.04)	5.95 (5.87–6.03)	1.62	0.69
Temporal lobe, medial	1.85 (1.84–1.86)	1.82 (1.79–1.85)*	3.87	1.31
Temporal lobe, lateral	4.16 (4.13–4.18)	4.13 (4.07–4.19)	0.55	0.34
Parietal lobe	6.02 (5.99–6.06)	5.92 (5.84–6.00)*	5.25	1.65
Occipital lobe	2.44 (2.42–2.45)	2.38 (2.34–2.42)*	6.69	2
Insula	0.83 (0.83–0.84)	0.82 (0.80–0.83)*	5.18	1.64
Cingulum	1.13 (1.12–1.14)	1.12 (1.10–1.14)	1.73	0.72
Hippocampus	0.48 (0.48–0.49)	0.48 (0.47–0.49)	2.43	0.92
Amygdala	0.17 (0.16–0.17)	0.16 (0.16–0.17)	0.01	0.04

Values are presented as mean values (95% confidence interval). In the multivariable-adjusted model, the values were adjusted for age, sex, educational level, hypertension, diabetes mellitus, BMI, smoking, exercise habits, LDL and HDL cholesterol levels, and 5 nutrient intake variables with -log10(p-value) > 1.301 in Table 3, and variance inflation factors (VIFs) values were below 10. *Bonferroni-adjusted p-values, p < 0.05 versus social eating group.

After this comprehensive adjustment and applying Bonferroni correction for multiple comparisons, several brain regions still showed significant differences between the social and solitary eating groups. The medial temporal lobe volume was significantly smaller in the solitary eating group (1.826%, 95% CI: 1.796–1.857) compared to the social eating group (1.852%, 95% CI: 1.848–1.865), with F = 3.87 and -log10(p) = 1.31. The parietal lobe also demonstrated a significantly smaller volume in the solitary eating group (5.920%, 95% CI: 5.847–6.004) compared to the social eating group (6.023%, 95% CI: 5.997–6.060), with F = 5.25 and -log10(p) = 1.65. The occipital lobe showed the most pronounced difference, with a significantly smaller volume in the solitary eating group (2.384%, 95% CI: 2.345–2.424) compared to the social eating group (2.447%, 95% CI: 2.429–2.454). This difference had the highest F-value (6.69) and p-value (-log10(p) = 2.00) among all examined brain regions. The insula demonstrated a significantly smaller volume in the solitary eating group (0.820%, 95% CI: 0.807–0.834) compared to the social eating group (0.833%, 95% CI: 0.837–0.840), with F = 5.18 and -log10(p) = 1.64.

Interestingly, the difference in hippocampal volume between the two groups was no longer statistically significant after adjusting for nutrient intake (F = 2.43, -log10(p) = 0.92). Similarly, other brain regions, such as the total brain volume (TBV), did not show statistically significant differences between the social and solitary eating groups after this comprehensive adjustment.

Regional brain volume differences between social and solitary eating styles after adjusting for food and beverage intake

Table 6 presents the differences in regional brain volumes between social and solitary eating styles after adjusting for multiple confounding factors, including age, sex, educational level, hypertension, diabetes mellitus, BMI, smoking, exercise habits, LDL and HDL cholesterol levels, and 28 food and beverage intake variables with significant differences in Table 4 and acceptable multicollinearity (VIF < 10).

Table 6.

Differences in regional brain volumes among eating styles

Variables	Eating Style
	Median 1	Median 2	F-value	-Log10(p-value)
	Social	Solitary
Brain volume to eTiv, %
TBV	58.9 (58.7–59.2)	58.5 (57.9–59.1)	1.75	0.73
WMH	0.32 (0.30–0.35)	0.33 (0.27–0.38)	0	0.02
Frontal lobe	9.13 (9.07–9.18)	9.06 (8.94–9.19)	0.84	0.44
Temporal lobe	6.01 (5.97–6.05)	5.95 (5.86–6.04)	1.35	0.61
Temporal lobe, medial	1.85 (1.84–1.87)	1.81 (1.78–1.84)*	4.61	1.49
Temporal lobe, lateral	4.16 (4.13–4.18)	4.14 (4.07–4.20)	0.26	0.22
Parietal lobe	6.02 (5.98–6.06)	5.93 (5.85–6.02)	3.23	1.14
Occipital lobe	2.44 (2.42–2.45)	2.38 (2.34–2.43)*	5.11	1.62
Insula	0.83 (0.83–0.84)	0.82 (0.80–0.83)	3.13	1.11
Cingulum	1.13 (1.12–1.14)	1.12 (1.09–1.14)	1.73	0.72
Hippocampus	0.48 (0.48–0.49)	0.48 (0.47–0.49)	1.26	0.58
Amygdala	0.16 (0.16–0.17)	0.17 (0.16–0.17)	0.06	0.1

Values are presented as mean values (95% confidence interval). In the multivariable-adjusted model, the values were adjusted for age, sex, educational level, hypertension, diabetes mellitus, BMI, smoking, exercise habits, LDL and HDL cholesterol levels, and 28 food and beverage intake variables with -log10(p-value) > 1.301 in Table 4 and variance inflation factors (VIFs) values were below 10. *Bonferroni-adjusted p-values, p < 0.05 versus social eating group.

After this comprehensive adjustment for food and beverage intake and applying Bonferroni correction for multiple comparisons, two brain regions maintained significant differences between the social and solitary eating groups. The medial temporal lobe volume remained significantly smaller in the solitary eating group (1.814%, 95% CI: 1.783–1.845) compared to the social eating group (1.852%, 95% CI: 1.848–1.876), with F = 4.61 and -log10(p) = 1.49. The occipital lobe continued to show the most pronounced difference, with a significantly smaller volume in the solitary eating group (2.385%, 95% CI: 2.345–2.435) compared to the social eating group (2.447%, 95% CI: 2.429–2.454). This difference had the highest F-value (5.11) and p-value (-log10(p) = 1.62) among all examined brain regions.

Notably, the hippocampus, which showed significant differences in previous analyses, no longer demonstrated a statistically significant difference between the two groups after adjusting for food and beverage intake (F = 1.26, -log10(p) = 0.58). Similarly, other brain regions, such as the total brain volume (TBV), parietal lobe, and insula, did not show statistically significant differences between the social and solitary eating groups after this comprehensive adjustment.

Discussion

This cross-sectional study investigated the associations between solitary eating, regional brain volumes, and dietary intake in cognitively healthy older adults. Our findings revealed two key results: first, solitary eating was associated with reduced brain volumes in regions critical for cognitive function, particularly the medial temporal lobe and hippocampus; second, individuals who frequently engage in solitary eating exhibited less healthy dietary patterns, characterized by higher intakes of sucrose, maltose, alcohol, and a higher carbohydrate energy ratio.

Our finding that solitary eating was associated with reduced brain volumes aligns with previous research on social isolation and brain health^5,9. This relationship persisted in several brain regions even after adjusting for various confounding factors, including nutrient intake and food consumption patterns. The observed association between solitary eating and reduced hippocampal volume is particularly noteworthy, given the hippocampus’s crucial role in memory processing and its susceptibility to age-related cognitive decline¹⁸. These results are consistent with previous studies linking social isolation to brain atrophy. For example, Hirabayashi et al. reported that lower frequencies of social contact were associated with greater brain atrophy, particularly the temporal lobe and hippocampus, in community-dwelling older people without dementia⁴. Similarly, Walter and Sandsmark emphasized the importance of social contact in preventing brain atrophy in older adults²².

Interestingly, our analysis revealed a differential effect of dietary adjustments on specific brain regions. The significant difference in hippocampal volume between solitary and social eaters disappeared after adjusting for both nutrient and food intake. However, the difference in the medial temporal lobe volume persisted even after these adjustments. This suggests that the relationship between eating habits and brain structure may be region-specific and mediated through different pathways. The hippocampus appears to be more directly influenced by nutritional factors related to solitary eating, while the medial temporal lobe may be more sensitive to the social aspects of eating or other factors not fully captured by our dietary measures^23,24.

Our second key finding revealed that individuals who frequently engage in solitary eating had significantly higher intakes of sucrose, maltose, alcohol, and a higher carbohydrate energy ratio compared to those who ate with others. This dietary pattern associated with solitary eating may have important implications for brain health. High sugar intake, particularly from added sugars, has been linked to cognitive decline and an increased risk of dementia^7,19. The observed higher alcohol intake among those who frequently engage in solitary eating is also concerning, given the well-documented negative effects of excessive alcohol consumption on brain health²⁵. These dietary differences may be explained by several factors, including reduced motivation to prepare nutritious meals when eating alone, lack of positive social influence on food choices, and potential emotional factors leading to comfort eating or increased alcohol consumption^3,14,20,21.

The complex relationships observed between solitary eating, brain volumes, and dietary patterns suggest multiple potential mechanisms at play. While psychological mechanisms, such as stress from social isolation and reduced cognitive stimulation, likely contribute^26,27, our findings indicate that the pathways linking solitary eating to brain health may be more intricate. The dietary differences we observed could impact brain health through pathways not directly measured in this study, such as inflammation or alterations in the gut-brain axis^10–12,28. The differential effects on the hippocampus and the medial temporal lobe further support the idea of multiple mechanisms, potentially involving both direct nutritional effects and broader social or lifestyle factors associated with solitary eating.

It is important to acknowledge the limitations of our study. The cross-sectional design precludes causal inferences about the relationships between solitary eating, brain volume, and dietary patterns. Longitudinal studies are needed to establish temporal relationships and potential causal pathways²⁹. Additionally, while we controlled for several confounding factors, other unmeasured variables may have influenced the observed associations. Future studies should consider a broader range of potential confounders, including detailed measures of overall social engagement, depression, and anxiety. The use of a single question to assess eating habits is another limitation; more comprehensive measures of social dining patterns, including frequency and context of shared meals, would provide a more nuanced understanding of the relationship between social eating and brain health¹⁷. Our dietary assessment, while comprehensive^30,31, may not capture all aspects of nutritional intake that could influence brain health. Future studies could benefit from more detailed dietary assessments, including biomarkers of nutritional status³². Another important limitation of our study is the absence of direct measures of broader social interaction or network size. Solitary eating may reflect broader social isolation rather than being an independent risk factor. Previous studies have shown that larger social networks are associated with larger hippocampal and amygdala volumes³³. Therefore, future research should incorporate direct assessments of social interaction to better distinguish these effects. Additionally, given evidence that social network size effects on brain structure may be sex-dependent, future studies should examine potential sex differences in the relationship between solitary eating and brain volumes. Finally, although our study focused on older adults in Japan^15,16, cultural differences in eating habits and social norms around dining may limit the generalizability of our findings to other populations. Cross-cultural studies are needed to explore how these associations may vary across different sociocultural contexts^6,13.

In conclusion, our study provides evidence of associations between solitary eating, reduced brain volumes in regions critical for cognitive function, and less healthy dietary patterns in cognitively healthy older adults. These findings underscore the potential importance of social dining in maintaining brain health and highlight the need for further research to elucidate the mechanisms underlying these associations. As populations age worldwide^1,2,34, understanding and addressing the impact of social factors on brain health and nutrition are crucial for promoting healthy cognitive aging⁸.

Methods

Study population

This report describes an ongoing study aimed at investigating cognitive deterioration among older Japanese adults. Dementia screening was conducted in a population-based, longitudinal cohort of individuals within a specific age range. A detailed description of the data has been previously published¹⁵. This study was conducted in Nakajima, which is situated in Nanao City in Ishikawa Prefecture, Japan. The research methodology has been previously described^15,16,29. From 2016 to 2018, 2454 inhabitants aged ≥60 (representing 92.9% of the total demographics within this age bracket) participated in the study. All participants in this study were older adults living in the Nakajima area. The majority of participants were community-dwelling individuals living independently in their own homes or with family members. However, a small subset of participants were residing in institutional care facilities (such as nursing homes or assisted living facilities) within the same geographical area at the time of the study. The inclusion of both community-dwelling and institutionalized individuals allows for a more comprehensive representation of the older adult population in the studied area. To determine whether solitary eating is a risk factor for cognitive impairment in cognitively unimpaired older adults, we excluded patients with MCI (n = 427) and dementia (n = 364). We also excluded individuals who had no brain MRI examinations (n = 632), hemorrhagic and/or ischemic stroke lesions on MRI regardless of the presence or absence of neurological symptoms (n = 59), no dental examinations (n = 240), and no dietary history evaluations (n = 5). Stroke lesions (hemorrhagic and/or ischemic) were identified by two trained neuroradiologists who were blinded to the clinical information. Finally, data from 727 individuals were included in the analysis (Fig. 1).

Fig. 1. Flow chart of participant enrollment.

MCI mild cognitive impairment, MRI magnetic resonance imaging.

Cognitive status

Diagnosis of dementia was based on the guidelines of the Diagnostic and Statistical Manual of Mental Disorders, third edition, revised (DSM-III-R)³⁵, whereas diagnosis of MCI was established according to the International Working Group on general criteria for MCI³⁶, which stated that (1) persons should be judged as abnormal using other modalities besides not fulfilling the DSM-III-R dementia criteria, (2) functional activities of the person are mainly preserved or at least impairment is minimal, and (3) the person should have evidence of cognitive decline, either by self-assessment and/or the use of an informative report in conjunction with deficits on objective cognitive tasks. Among participants without dementia, a Clinical Dementia Rating score of 0.5 was used as the objective cognitive impairment value to denote cognitive and functional impairment consistent with MCI. A detailed description of the data was previously published¹⁵.

Magnetic Resonance Imaging (MRI) analysis

Structural MRI was conducted using a 1.5-T system (ECHELON RX; Hitachi, Japan). A detailed description of brain MRI data acquisition and processing has been previously published¹⁵. Three-dimensional volumetric acquisition of T1-weighted turbo field echo images was conducted according to the brain MRI protocol for the Alzheimer’s Disease Neuroimaging Initiative study²³ (Echo time/Repetition time, 4.0/9.2 ms; flip angle, 8°; Field of View, 240 mm; acquisition matrix, 192 × 192; number of slices, 170; voxel size, 0.9375 × 0.9375 mm; slice thickness, 1.2 mm). All T1-structural images were analyzed using FreeSurfer version 5.3 (http://surfer.nmr.mgh.harvard.edu)³⁷ at Tohoku University and preprocessed according to the standard method. Volumes of the areas of interest were created using FreeSurfer²⁴. The medial temporal lobe and hippocampus were selected as primary regions of interest because they are highly involved in memory processing and are particularly vulnerable to age-related cognitive decline. Other lobar regions (parietal, occipital, frontal) and subcortical structures (insula, cingulum, amygdala) were analyzed based on previous studies linking social isolation to widespread brain atrophy patterns⁴. The following 37 brain regions were analyzed: frontal lobe, parietal lobe, occipital lobe, lateral temporal lobe, medial temporal lobe, cingulum, insula, hippocampus, amygdala, total cortical gray matter, total subcortical gray matter, and volumes of white matter hyperintensities. Cortical regions were further subdivided into specific gyri and sulci according to FreeSurfer’s standard parcellation (e.g., cuneus, precuneus, fusiform, entorhinal cortex, etc.). A complete list is provided in Supplementary Table S1. The 37 areas were calculated as the sum of the volumes of the right and left sides of each area. We conducted additional analyses to examine the left and right hemispheres separately. We also performed separate analyses of gray and white matter volumes in each brain region (results presented in Supplementary Table S1). The TBV was calculated by summing the white and gray matter volumes. WMHs are lesions that appear as regions of decreased signal intensity on T1-weighted MRI. These are considered manifestations of small vessel ischemic disease. Among the volumes of parcels, the volumes of the following brain regions were used as outcomes: cortical volumes in the lobes (frontal, temporal, parietal, and occipital), insula, cingulum, hippocampus, and amygdala. All brain region volumes were normalized to estimated total intracranial volume (eTIV) and expressed as percentages to account for individual head size differences.

Assessment of eating style

Solitary eating was assessed through a self-administered questionnaire. Participants were asked, “How many people do you eat with?” and chose from four options: four or more, three people, two people, or alone. We corroborated these answers with information from caregivers or family members to ensure accuracy.

For analysis, responses were categorized into two groups: 1. Solitary eating: defined as usually eating meals alone; 2, Social eating: defined as usually eating with one or more other people.

Nutritional status

Dietary intake was assessed using a self-administered FFQ (Supplementary Material). This questionnaire has been validated and reported^15,30. Each participant completed the questionnaire in advance, and trained dieticians checked it during the screening test. The average food intake per day was calculated using the frequency of meals per week and the amount of each food portion. Nutritional intake was determined using the Standard Tables of Food Composition in Japan (5th Revised Edition)³¹. The FFQ assessed the intake of major macronutrients (carbohydrates, proteins, fats) and micronutrients (vitamins, minerals), as well as specific food groups, including fruits, vegetables, grains, dairy, meat, fish, and beverages. A complete list of the analyzed nutrients and food items is provided in Supplementary Tables S2, and S3. All dietary nutrients were adjusted for total energy using the density method³².

Statistical analysis

To compare participant characteristics, we used one-way analysis of variance for mean calculations of continuous variables and chi-square test for categorical variables. A two-tailed p-value < 0.05 indicates statistical significance. Analysis of covariance (ANCOVA) was employed to estimate and compare multivariable-adjusted values and their 95% confidence intervals for eating style. Homogeneity of variance assumptions for ANCOVA were verified using Levene’s Test before group comparisons. Multivariable models were adjusted for the following covariates: age, sex, years of education, hypertension status, diabetes mellitus status, body mass index (BMI), smoking status, exercise habits, serum LDL cholesterol, and serum HDL cholesterol levels. To address the potential for multicollinearity among confounding variables in Tables 5 and 6, we calculated Variance Inflation Factors (VIF) for each covariate. Only variables with VIF values less than 10 were included in the final multivariable-adjusted models. For nutrient intake (Table 5), five variables with significant differences in Table 3 and acceptable multicollinearity (VIF < 10) were retained: carbohydrate energy ratio, 18:3 n-6 gamma-linolenic acid, alcohol, organic acid, and sucrose. For food and beverage intake (Table 6), we excluded saccharides, which showed VIF > 10, and retained 28 other variables in Table 4 with acceptable collinearity. Bonferroni adjustment was applied to multiple comparisons, with Bonferroni-corrected p-values < 0.05 considered statistically significant. The SPSS software suite (version 26; SPSS Inc., Chicago, IL, USA) was used for all statistical analyses.

Ethics statements

This study complied with the principles outlined in the Declaration of Helsinki, and all methodologies were approved by the Kanazawa University Medical Ethics Review Committee (Approval Number 2186-15) and the Ryukyu University Medical Ethics Review Committee (Approval Number 2153). Written informed consent was obtained from all participants.

Author contributions

M.I., H.N.: Study concept, design, Data acquisition, analysis, interpretation, Writing—original draft, Writing—review & editing, Project administration, Supervision; M.N.S., M.I.J., K.I., T.K., S.T., R.S., M.M., and E.H.N.: Data acquisition, interpretation, Writing—review & editing; Y.S., S.K., and K.O: Writing—review & editing, Project administration. All authors have read and agreed to the submission and publication of this manuscript.

Data availability

Because of the inclusion of sensitive medical information that could compromise participant privacy, the data analyzed in this study cannot be made publicly available. However, anonymized data can be accessed upon reasonable request and with permission from the corresponding author.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41514-025-00284-3.