Greater loneliness is associated with dense, less modular, brain connections; greater sense of life meaning is associated with increased, more modular, connectivity between default & limbic networks

Loneliness and meaning in life are reflected in the intrinsic network architecture of the brain. Mwilambwe-Tshilobo et al. Soc Cogn Affect Neurosci., Mar 29 2019, nsz021. https://doi.org/10.1093/scan/nsz021

Abstract: Social relationships imbue life with meaning, whereas loneliness diminishes the sense of meaning in life. Yet the extent of interdependence between these psychological constructs remains poorly understood. Loneliness and meaning are associated with different patterns of functional connectivity; however, no studies have investigated this directly. We took a multivariate network approach to examine resting-state fMRI functional connectivity's association with loneliness and meaning in a large cohort of adults (N=942). Loneliness and meaning in life were negatively correlated with one another. In their relationship with individually parcelled whole-brain measures of functional connectivity, a significant and reliable pattern was observed. Greater loneliness was associated with dense, and less modular, connections between default, frontoparietal and externally-directed attention and perceptual networks. A greater sense of life meaning was associated with increased, and more modular, connectivity between default and limbic networks. Low loneliness was associated with more modular brain connectivity, and lower life meaning was associated with higher between-network connectivity. These findings advance our understanding of loneliness and life meaning as distinct, yet interdependent, features of sociality. The results highlight a potential role of the default network as a central hub, providing a putative neural mechanism for shifting between feelings of isolation and purpose.

KEYWORDS: individual differences; partial least squares; personality; resting-state functional connectivity

---
Introduction
 Loneliness and life meaning are psychologically-bound constructs closely tied to sociality
(Lambert et al., 2013; Stillman et al., 2009; Twenge, Catanese, & Baumeister, 2003). As a social
species, humans typically seek out social bonds and search for meaning and purpose throughout
the life-course. Indeed, both loneliness and a reduced sense of meaning are closely associated
with declines in functional capacity (Perissinotto et al., 2012), dementia onset (Holwerda et al.,
2014; Boyle et al., 2012), and mortality in later life (Holt-Lunstad et al., 2015; Boyle et al., 2009;
Hill and Turiano, 2014). Despite these psychological and functional relationships, loneliness and
meaning in life (MIL) are considered to be distinct constructs and their degree of
interdependence remains poorly understood. Loneliness reduces the perception of a meaningful
existence (Stillman et al., 2009)—the sense that life has purpose, significance, and coherence
(Martela & Steger, 2016). This association appears to be reciprocal as MIL is strongly associated
with the presence of close relationships (Ebersole, 1998; Klinger, 1977), and previous reports
show that the subjective perception of a meaningful life promotes social engagement and helps
sustain close social bonds (Steptoe & Fancourt, 2019; Stillman & Lambert, 2013). Loneliness
arises due to deficiencies in the quality or quantity of social ties and the absence of social
connectedness, in turn, diminishes MIL, suggesting that this relationship may also be reinforcing
(Baumeister & Leary, 1995). But are these constructs opposite sides of the same coin, or are they
emergent from distinct mechanisms?
Loneliness is characterized by implicit hyper-vigilance for social threats (Cacioppo et al.,
2016). While this can facilitate the identification of viable social partners and prevent rejection,
prolonged loneliness shifts exogenous attentional processes towards perceived social threats
(Bangee et al. 2014; Cacioppo, Balogh, & Cacioppo, 2015). Altered attention to external stimuli
may affect how individuals internalize perceived information and make endogenous judgments
about MIL (Hicks, Schlegel, & King, 2010). Externally-and internally-guided cognitive
processes are mediated by different neural networks and their interactions (Corbetta & Shulman,
2002; Spreng et al., 2010). This raises the possibility that loneliness and MIL are dissociable at
the level of the brain, and subserved by distinct brain networks. Investigating how individual
differences in loneliness and MIL are reflected within these neurocognitive systems may advance
our understanding of their interdependence, and how they interact to guide adaptive and
maladaptive behaviors.
A growing body of neuroimaging studies have provided important insights into the neural
correlates of loneliness, reflecting changes in brain regions associated with processing of social
information. In a task-based functional magnetic resonance imaging (fMRI) study, lonely
individuals showed increased bilateral activation in the visual cortex in response to unpleasant
social images compared to unpleasant non-social images. Regions implicated in reward
processing (e.g. ventral striatum, amygdala) and perspective-taking (e.g. temporoparietal
junction) showed lower activation when positive social images were presented, suggesting that
lonely individuals may derive less pleasure from rewarding social stimuli (Cacioppo et al.,
2009). Furthermore, other studies have linked loneliness to changes in brain morphology within
the default network (DN), a neural system involved in social and self-related processes
(Andrews‐Hanna, Smallwood, & Spreng, 2014). Loneliness is negatively correlated with grey
matter volume (Kanai et al., 2012) and white matter density (Nakagawa et al., 2015) in the left
posterior superior temporal sulcus (pSTS). These findings indicate that loneliness may
compromise the structural and functional integrity of multiple brain regions.
Resting-state functional connectivity (RSFC) has been an invaluable analytic approach
for investigating the functional interactions between anatomically separate brain regions and
their relationship with behavior (Stevens & Spreng, 2014). Unlike task-based fMRI paradigms,
resting-state functional magnetic resonance imaging (rs–fMRI) is task-free and can be used to
simultaneously identify multiple functional networks correlated with behavior. Furthermore,
previous analyses of rs-fMRI data from healthy adult populations have consistently shown strong
congruence between brain networks derived from resting-state and those from task-based studies
(Cole et al., 2014; Stevens & Spreng, 2014; Tavor et al., 2016). In fact, task-based connectivity
estimates have been demonstrated to introduce potentially spurious functional relationships (Fox
& Raichle, 2007).
Prior studies have used rs-fMRI to characterize intrinsic functional brain networks related
to loneliness and MIL. Greater feelings of loneliness have been associated with less integrated
connectivity between attention networks (Tian et al., 2017), as well as increased RSFC within
the cingulo-opercular network, which is implicated in cognitive control (Layden et al., 2017).
These intrinsic changes are consistent with behavioral reports of associations between hyper
vigilance and loneliness (Cacioppo et al., 2016). An investigation of the neural basis of meaning
(Waytz et al., 2015) reported increased connectivity among regions of the medial temporal lobe
subsystem of the DN, implicated in autobiographical remembering and mental simulation (Andrews‐Hanna et al., 2014). While loneliness and MIL are correlated at the level of behavior, the analytical approaches used to characterize the neural representation of each construct have
focused on functional connectivity of select brain regions or networks of interest, thus precluding
inferences on a whole-brain level of integrated networks that can provide insight regarding the
relationship between loneliness and MIL. Here, we investigate individual differences in the
neural representation of loneliness and MIL within a single analytical framework.
The goal of the present study was to assess how whole-brain RSFC is associated with
individual differences in loneliness and MIL. We characterized the intrinsic architecture of brain
connectivity within a large population of healthy young adults using RSFC and individually
parcellated brain regions (Chong et al., 2017), respecting that the localized topology varies
across individuals in the cortex (e.g. Stevens et al., 2015) in order to identify the pattern of
functional connectivity within and between large-scale networks. Using multivariate partial least
squares (PLS), we characterized how patterns of RSFC relate to individual differences in
perceived loneliness and MIL. This approach permits both replication of previous RSFC
patterns, and exploratory examination of behavioral associations outside previously examined
networks.
By examining the intrinsic functional connectivity underlying individual differences in
loneliness and MIL, we test two hypotheses: First, loneliness would be associated with greater
connectivity between regions that support attention, including the FPN, dorsal attention (DAN),
and the ventral attention networks (VAN; Corbetta and Shulman, 2002). In contrast, MIL would
be associated with greater connectivity within the DN. Our second hypothesis was that these
patterns of RSFC would be inversely related (i.e. individuals with high levels of loneliness will
share the same pattern of brain connectivity as those with a low sense of MIL and vice-versa). If
confirmed, this would provide support for theoretical models of sociality suggesting that
loneliness and MIL are distinct yet interdependent constructs (Lambert et al., 2013).


Discussion

Loneliness and meaning in life are important for guiding everyday behavior and
sustaining mental health and well-being over the life course and into advanced age. Yet their
neural signatures remain poorly understood. Here we used a multivariate analytical model to
examine patterns of intrinsic functional connectivity associated with individual variability in
loneliness and MIL in a large sample of healthy adults. There were three primary findings. First,
we identified reliable patterns dissociating whole-brain RSFC related to individual differences in
loneliness and MIL. Second, we observed a core role for default network connectivity in
differentiating loneliness and meaning in life. While default and frontoparietal interactions,
among others, were associated with higher levels of loneliness, this pattern differed for MIL
where connectivity between default and limbic brain regions was associated with a greater sense
of meaning. Finally, greater feelings of loneliness were associated with lower modularity, or
increased integration, between the default and frontoparietal networks and more externally
oriented networks including somatosensory and visual brain regions. In contrast, a stronger sense
of life meaning was associated with greater modularity among more internally-oriented systems
including the limbic and default networks.
Current theoretical models of sociality suggest that loneliness and MIL are discrete yet
interdependent, and potentially reinforcing (Lambert et al., 2013; Stillman et al., 2009; Twenge
et al., 2003). However, only a few studies have investigated the relationship between the loss of
social functioning (i.e. loneliness) and MIL, and these have primarily employed behavioral
methods (Lambert et al., 2013; Stillman & Lambert, 2013). More recently, investigations into the
intrinsic functional architecture of the brain at rest (i.e. in the absence of explicit task demands)
have demonstrated that these durable features of brain organization can enhance our
understanding of enduring features of mental function (Smith et al., 2015; Stevens and Spreng,
2014). Here we leveraged this idea to explore patterns of functional connectivity associated with
individual differences in loneliness and MIL.
 We predicted that the DN, through its role in mediating internally directed cognition,
would be associated with MIL. A greater sense of life meaning has previously been associated
with increased connectivity within the medial temporal lobe subsystem of the DN (Waytz,
Hershfield, & Tamir, 2015). Our data complements this finding by showing increased
connectivity within nodes of the DN associated with higher MIL. Additionally, we observed a
robust, albeit unpredicted, pattern of connectivity within and between networks typically
implicated in internally-directed cognitive processes associated with higher MIL, including the
limbic and default networks, as well cognitive control regions of the FPN. The limbic network is
involved in emotional processing, which involves monitoring, evaluating, and adjusting
emotional reaction to align with current goals. Thus the ability to internally reflect upon one’s
affective state, may be important for a sense of meaning, particularly when experiencing
negative emotions (Kross & Ayduk, 2011). Consistent with this idea, individuals with a clear
sense of purpose in life report lower levels of negative affect and less emotional reactivity to
stressors in daily life (Hill, Sin, Turiano, Burrow, & Almeida, 2018).
The evolutionary theory of loneliness posits that feeling lonely is an aversive biological
signal that motivates the individual to repair or seek new social relationships, and leads to neural
changes that impact attention and processing of social information (Cacioppo & Cacioppo,
2018). While our findings are in accordance with previous studies linking loneliness with altered
RSFC in networks related to attention and executive control (Layden et al., 2017), the results
point to broader changes in brain connectivity across multiple networks. As with MIL, the most
robust associations were observed for between network interactions, and specifically between the
DN and FPN as well as networks implicated in more externally-directed cognition including
attentional (e.g. VAN) or perceptual (e.g. SOM and visual networks) processing. While the
breadth of these associations was not predicted, the VAN is associated with bottom-up or
externally monitoring for behaviorally salient features of the environment (Corbetta and
Shulman, 2002), presumably detected through connections with these perceptual systems. While
we are unable to directly confirm this with the current data, this is consistent with behavioral
accounts of hyper-vigilance for external social threat associated with loneliness. Further, the DN
has been implicated in low mood and ruminative thoughts (DuPre & Spreng, 2018), which may
be elevated by a sense of loneliness. However, several methodological considerations may
account for differences between Layden et al. (2017) and the current findings. While a whole
brain analytic approach was used in both, we examined connectivity strength using individually
parcellated neurocognitive networks—thereby accounting for inter-subject functional
connectivity variability—rather than focusing on standardized network parcellation schemes.
Further, we used multivariate, data-driven analytical methods and a single model approach,
including MIL whereas the earlier study focused on attention networks to test their hypotheses.
Further, PLS methods allow for identification of both within and between network connectivity
strengths in a single analytical model (McIntosh & Mišić, 2013; McIntosh & Lobaugh, 2004).
Here, the between network associations were among the most robust, and most discriminating,
patterns observed for loneliness and life meaning.
Our second hypothesis was based in part on recent findings that individual differences in
both positive and negative behavioral traits have been associated with a unique configuration of
intrinsic functional connectivity (Smith et al., 2015). Specifically, increased connectivity within
regions encompassing the DN was linked to positive behavioral traits such as life satisfaction,
and inversely related to negative behavioral traits such as perceived stress (Smith et al., 2015).
Similarly, by including both loneliness and MIL in a single model, here we were able to identify
a single pattern of functional connectivity implicating the DN that was associated with these
positive and negative constructs. Connectivity within the DN, and its connections to the limbic
network, were associated with a higher sense of life meaning and lower feelings of loneliness. In
contrast, DN connectivity to externally-oriented attentional systems and cognitive control
networks was associated with a higher sense of loneliness, and lower life meaning.
We further examined the features of whole-brain RSFC organization related to loneliness
and MIL by interrogating the modular intrinsic network architecture. Increased modularity has
been associated with more efficient processing operations and is generally considered to be a
marker of brain health (Bullmore and Sporns, 2009; Wig et al., 2017). The intrinsic network
organization of brain networks associated with loneliness was less modular as the DN and FPN
were less differentiated from externally-directed attention and perceptual networks. As suggested
above, this pattern of network dedifferentiation may reflect increased vigilance for social threat.
Consistent with this idea, less modular brain network architecture has been associated with
negative affect including depression, as well as normal and pathological aging (Andrews-Hanna
et al., 2014).
To our knowledge, this is the first study to investigate whole-brain patterns of RSFC
associated with loneliness and MIL. Both MIL and loneliness are predictors of successful aging
and an important future direction would be to examine how these patterns of intrinsic brain
networks change in normal and pathological aging. Future examinations will also be necessary to
explore how the connectivity patterns identified in the present study are dynamically shaped in
response to task demands that require judgments of belonging and/or existential meaning.
Further, MIL is distinct from meaning-seeking and meaning maintenance, and these differences
will need to be explored with respect to loneliness and patterns of RSFC. This question is
particularly relevant in light of past work demonstrating a distinction between the presence of
meaning and the search for meaning (Heine, Proulx, & Vohs, 2006; Steger, 2012), and may have
important implications for the interpretation of our results for loneliness given that the lack of
belonging could both motivate or discourage an individual’s search for meaning.
By investigating associations between brain function, loneliness and MIL within a
common analytical framework, we were able to identify a pattern of intrinsic functional
connectivity that differentiated brain networks associated with higher MIL and lower loneliness
from those associated with lower MIL and higher loneliness. Critically, between network
interactions, particularly those involving the DN, were among the most robust and discriminating
intrinsic network markers of loneliness and MIL. Behaviorally, these findings advance our
understanding of these two constructs as distinct, yet interdependent, features of sociality
(Lambert et al., 2013; Stillman et al., 2009; Stillman & Lambert, 2013). While speculative, the
data also implicate the DN as a candidate network hub, suggesting that these brain regions may
provide a neural conduit for shifting between feelings of isolation and purpose. If confirmed,
these findings may inform future research to design behavioral and neural intervention strategies
targeted at disrupting the reinforcing cycle of loneliness and life meaning.