Human survival and thriving depend heavily on social interaction. However, this social dependency carries a significant cost: perceived social isolation, or loneliness, which negatively affects physical and mental health, cognitive performance, and lifespan. Loneliness increases vulnerability to dementia, yet its neural underpinnings remain unclear. This study directly investigates the neural correlates of trait loneliness (persistent feelings of unmet social needs) in a large population sample, distinguishing it from simply spending time alone or having infrequent social contact. While prior research has linked social connectedness to brain structure and function, focusing primarily on lower-level circuits (visual, affective, attentional), this study examines higher-order social brain regions, particularly those associated with mentalizing and self-referential thought, hypothesizing that these regions would show significant involvement in the experience of trait loneliness.

Previous research has investigated the impact of loneliness on various aspects of brain function and structure. Animal studies have highlighted differences in subcortical reward systems related to social isolation. Human studies have observed dampened reward signaling to social cues in mesolimbic systems and increased vigilance for negative social information. Differences have also been reported in visual cortices, attention networks, limbic structures, and prefrontal cortex, highlighting the involvement of perceptual, attentional, and affective processing in loneliness. However, fewer studies have explored the role of higher-order brain regions involved in complex social cognition, such as those comprising the 'social brain' (medial prefrontal and medial temporal lobes, temporoparietal junction, and posteromedial parietal cortex). This study aimed to fill this gap by conducting a comprehensive multimodal investigation, considering structural, functional, and diffusion MRI data.

This study utilized data from the UK Biobank imaging-genetics cohort (approximately 40,000 participants aged 40-69), assessing trait loneliness using a single binary question: "Do you often feel lonely?" Three neuroimaging modalities were employed: structural MRI (gray matter volume), resting-state functional MRI (intrinsic functional connectivity), and diffusion MRI (white matter tract microstructure). For structural MRI, a Bayesian hierarchical framework was used to analyze gray matter volume variation across 100 brain regions, categorized into seven large-scale brain networks (default, limbic, dorsal attention, somatomotor, visual, control, and salience). Resting-state fMRI data were analyzed using partial least squares to identify patterns of functional connectivity associated with loneliness. Diffusion MRI data were analyzed to assess white matter integrity of 48 major tracts. All analyses controlled for potential confounders (e.g., body mass index, head size, head motion). Sex-specific analyses were also conducted to examine potential gender differences in brain-loneliness associations.

The analysis revealed a prominent association between loneliness and the default network. In structural MRI, the default network showed the strongest relationship with loneliness, with volume variations in multiple regions within the network correlating with loneliness scores. Specifically, positive volume associations were seen in the posterior superior temporal sulcus, temporoparietal junction, and fusiform gyrus, while negative associations were found in the dorsolateral prefrontal cortex, retrosplenial cortex, and inferior visual cortex. Functional MRI analysis showed that lonely individuals exhibited stronger within-network coupling within the default network and increased between-network coupling between the default network and limbic, dorsal attention, and somatomotor networks. Diffusion MRI revealed increased microstructural integrity in three fornix tracts, which connect the hippocampus to the default network, further supporting the default network's prominent role in loneliness. Sex-specific analyses indicated more pronounced brain-loneliness associations in men compared to women, particularly in functional connectivity and white matter microstructure.

The findings highlight the significant role of the default network in the neural expression of loneliness. This contrasts with previous studies that emphasized lower-level brain regions. The observed increased activity and structural integrity within the default network in lonely individuals may reflect a compensatory mechanism, with individuals engaging in more self-referential thought, reminiscing, and mental simulations of social interactions to cope with their social isolation. The stronger association of these effects in men compared to women warrants further investigation, considering potential factors such as social stigma or reporting biases. The study's focus on a mid-to-late adulthood cohort raises questions about the dynamic interplay between loneliness, brain structure, and aging, particularly given previously reported decreased default network connectivity in younger adults with loneliness. Future research should investigate these longitudinal relationships.

This multimodal neuroimaging study using the UK Biobank demonstrates a strong association between loneliness and alterations in the default network, a brain region crucial for self-referential processing and social cognition. These findings suggest that the brain's response to loneliness might involve an increased reliance on internally directed cognitive processes, including mental simulations of social experiences. Further research should explore these findings in longitudinal studies, examining the development of loneliness-related brain changes across the lifespan and across genders, and investigating the clinical implications for intervention strategies.

The study's cross-sectional design prevents causal inferences regarding the relationship between loneliness and brain structure/function. The use of a single item to measure loneliness, while validated, might not fully capture the complexity of this subjective experience. Future studies could benefit from utilizing more comprehensive measures of loneliness and social isolation, as well as incorporating longitudinal data to track changes over time. The potential influence of other co-occurring conditions was minimized through statistical control, however, these factors might have played a role.