Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Advertisement
Communications Psychology volume 4, Article number: 44 (2026)
3570
181
Metrics details
Loneliness is increasingly recognized not only as a stable trait but also as a dynamic affective process, marked by short-term fluctuations in mood, social perception, and behavior. This study examined how self-reported experiences of loneliness, perceived rejection, and social behavior unfold across time in daily life. A community sample of 157 midlife adults completed ecological momentary assessments five times daily for 20 days, reporting on feelings of loneliness, social threat, self-disclosure, and interaction frequency. Dynamic structural equation and multilevel models demonstrated reciprocal associations between momentary loneliness and perceived rejection. Greater variability in loneliness was associated with more unstable threat appraisals, and increases in loneliness predicted subsequent reductions in both social interaction and self-disclosure. These within-person dynamics were moderated by trait loneliness: individuals higher in trait loneliness exhibited more persistent loneliness, stronger coupling between loneliness and perceived rejection, and greater social withdrawal. Findings support a multi-timescale framework in which recursive patterns of emotion, perception, and behavior contribute to the maintenance of loneliness in everyday life.
Loneliness—a subjective sense of insufficient social connection—has emerged as a pressing public health issue, particularly among midlife and older adults1. Accumulating evidence links loneliness to increased risk for premature mortality, cardiovascular disease, depression, and cognitive decline2,3,4,5. These long-term consequences underscore the need to understand not only who is lonely but also how loneliness is maintained in daily life.
Traditionally viewed as a stable individual difference, loneliness is increasingly recognized as a dynamic affective process that varies in response to changing social contexts and internal psychological states6,7. Momentary experiences of loneliness ebb and flow throughout the day and co-occur with changes in emotion, cognition, and behavior. Studies using high-frequency sampling methods have found that daily fluctuations in loneliness are associated with elevated perceptions of social threat, reduced positive affect, and diminished social engagement8,9,10. These findings suggest that loneliness is not merely a static individual difference, but a self-reinforcing process in daily experience.
A growing literature has highlighted three temporal features that characterize loneliness as a dynamic process. Intra-individual variability refers to the extent to which a person’s loneliness levels fluctuate over time, capturing the overall range of experiences. Instability, often measured by changes between consecutive timepoints, reflects the unpredictability of these shifts. Loneliness inertia captures the short-term temporal persistence of loneliness, suggesting a diminished capacity to shift out of the state once it has been experienced (see Fig. 1). Each of these features captures distinct regulatory vulnerabilities: higher variability and instability have been associated with emotion regulation difficulties and diminished sensitivity to social cues11,12, whereas elevated inertia—particularly when coupled with social withdrawal—has been linked to reduced flexibility and impaired re-engagement with social environments13,14. Recent work has further demonstrated that patterns of daily loneliness variability differ systematically based on chronic loneliness levels, with individuals higher in trait loneliness exhibiting distinct patterns of within-person variation15. Together, these temporal indices offer a descriptive taxonomy for characterizing loneliness dynamics by distinguishing dispersion, successive change, and short-term persistence.
Each panel depicts simulated trajectories of loneliness over time for illustrative cases of low (top row) and high (bottom row) values of three key dynamic features: inertia, variability, and instability. Inertia reflects the degree to which loneliness persists from one time point to the next (left column). Variability captures the overall range of fluctuation in loneliness around an individual’s mean level (middle column). Instability, indexed by the magnitude of successive changes, reflects short-term lability in the experience (right column). These temporal signatures provide distinct yet complementary insights into how loneliness unfolds within individuals in daily life.
Importantly, individuals differ in how these dynamics unfold, with trait loneliness shaping both the intensity of momentary states and their impact on subsequent perception and behavior. Individuals higher in trait loneliness tend to exhibit heightened sensitivity to perceived social threat and slower emotional recovery following episodes of disconnection16. Neuroimaging and behavioral evidence further suggest that chronically lonely individuals exhibit reduced responsiveness to positive social cues, potentially constraining their ability to update social interpretations when interpersonal conditions improve17. As a result, trait loneliness may both amplify exposure to threatening social perceptions and constrain opportunities for emotional recalibration, reinforcing patterns of withdrawal and sustaining feelings of isolation over time.
These altered perception-action cycles may also manifest in social behavior. Loneliness is theorized to function as a motivational signal for reconnection, analogous to how hunger drives food-seeking18. However, among individuals with elevated trait loneliness, this signal may become dysregulated, amplifiying vigilance to rejection and undermining effective social responding19,20. These patterns align with sociometer theory, which conceptualizes self-esteem as an internal monitor of perceived relational value that guides social behavior21. When belonging needs are unmet, individuals exhibit heightened sensitivity to social cues, including facial expressions and vocal tone, reflecting an activated social monitoring system22. While adaptive in contexts where social belonging is uncertain, heightened attunement to social cues may overemphasize signs of perceived social rejection, undermining engagement and reinforcing cycles of avoidance and disconnection.
Extending from these perceptual–motivational processes, maladaptive regulation is also evident in interpersonal expression. One behavioral mechanism through which loneliness may perpetuate itself is self-disclosure, the voluntary sharing of personal or emotionally meaningful information. Disclosure supports both intimacy formation and emotion regulation23,24,25, yet among lonely individuals the process often becomes dysregulated, manifesting as excessive guardedness or poorly timed oversharing26,27. These disruptions may impair reciprocity and limit mutual understanding. Over time, these patterns can become self-reinforcing sequences in which emotional states (loneliness, perceived rejection) and behaviors (reduced disclosure, social avoidance) mutually sustain one another, creating feedback loops that perpetuate psychological vulnerability. These recursive patterns may erode the capacity of everyday interactions to restore a sense of connection, thereby maintaining loneliness across time.
Building on this prior work, the present study used ecological momentary assessment (EMA) to examine the dynamic interplay between loneliness, perceived social rejection, and social behavior as it unfolds in daily life. Across 20 days, 157 midlife adults completed five smartphone-based surveys per day, reporting on their experiences of loneliness, perceived rejection, self-disclosure, and interaction frequency. This high-temporal-resolution design enabled us to test how emotional, cognitive, and behavioral states co-vary within individuals and whether these associations differ by trait levels of loneliness.
We tested three preregistered hypotheses: First (H1), momentary loneliness and perceived rejection would exhibit bidirectional associations, reinforcing one another over short timescales. Second (H2), increases in momentary loneliness would predict reductions in subsequent self-disclosure and social interaction. Third (H3), trait loneliness would moderate these within-person associations, such that individuals higher in trait loneliness would show stronger and more persistent links between momentary loneliness, perceived rejection, and social disengagement.
Beyond these confirmatory hypotheses, we examined an exploratory aim motivated by recent evidence that temporal features of affective experience, including intra-individual variability (overall dispersion around one’s mean) and instability (magnitude of successive changes), index distinct regulatory vulnerabilities11,28,29. Specifically, we examined whether individuals exhibiting greater variability and instability in momentary loneliness also demonstrate corresponding patterns in perceived rejection, suggesting coupled dysregulation across emotional and cognitive systems. Figure 2 summarizes the primary conceptual and analytic framework.
The framework depicts three temporal pathways: (I) autoregressive effects, capturing the persistence of each construct across time points; (II) near-concurrent associations, reflecting the near co-occurrence of loneliness and PR within the same 3-h window; and (III) prospective effects, modeling how loneliness at t–1 predicts subsequent social behavior. Trait loneliness was examined as a moderator of all pathways. Consecutive assessments were spaced approximately 3 h apart.
Data were drawn from a community-based sample of 157 midlife adults (age range: 46–74; M = 55.9, SD = 8.4). Participants were recruited from three sources. First, eligible participants registered with UC Davis Sona (study pool) received an email inviting them to take part in the study. Second, the study was advertised on the home page of TestMyBrain.org. Finally, ads were placed on social media for adults in the US aged 45–74. All participants were prescreened for smartphone ownership and willingness to complete repeated daily assessments.
To ensure adequate representation across the loneliness spectrum, stratified sampling was implemented midway through recruitment. Individuals scoring at the extreme low end of the UCLA Loneliness Scale (scores <1.8, representing the bottom 20th percentile) were excluded, as their low loneliness levels provided insufficient within-person variability to capture dynamic processes. Retaining participants with at least moderate trait loneliness ensured that our sample captured meaningful fluctuations in momentary loneliness, thereby optimizing statistical power for within-person analyses while maintaining heterogeneity in loneliness experiences. The distribution of trait loneliness in the final sample (Fig. 3) demonstrated adequate variability (M = 2.39, SD = 0.71, range = 1.00–4.00), confirming that the enrichment strategy preserved substantial individual differences necessary for testing moderation hypotheses.
Trait loneliness was calculated as the mean of UCLA Loneliness Scale items at baseline. The histogram displays individual differences in trait loneliness (M = 2.39, SD = 0.71, range = 1.00–4.00).
Participants were required to meet two pre-registered criteria for inclusion in the final analytic sample. First, participants must have completed at least 50 of 100 total EMA prompts (50% overall compliance) across the 20-day study period. Second, participants must have completed at least 50% of prompts on two designated anchor days—Day 7 and Day 14—which served as critical mid-protocol checkpoints. These anchor days were selected a priori to ensure adequate mid-study engagement: Day 7 allowed early identification of technical issues or declining motivation before substantial attrition occurred, while Day 14 captured sustained engagement through the study midpoint without being confounded by end-of-study fatigue or completion effects that might emerge in the final week. No exclusions were applied for low compliance on individual days outside these anchor assessments, allowing participants to miss occasional prompts due to situational factors (e.g., travel, illness, work demands) while still contributing usable data.
Of the 218 participants initially enrolled, 61 (27.9%) were excluded for failing to meet one or both criteria: 43 participants (19.7%) completed fewer than 50 total prompts, 12 participants (5.5%) met overall compliance requirements but failed anchor day thresholds, and 6 participants (2.8%) failed both criteria. The resulting analytic sample (N = 157) demonstrated high protocol adherence, with a mean compliance rate of 87.2% (SD = 9.8%, range = 50–100%) across all prompts. When aggregated to the daily level, participants completed an average of 4.37 of 5 prompts per day (SD = 0.93), corresponding to 89.4% daily compliance (SD = 18.5%).
The final analytic sample comprised 157 participants (Mage = 56.0 years, SD = 8.39), including 117 women, 38 men, and 2 individuals identifying as another gender. Participants self-identified as White (n = 132), Black (n = 9), Asian (n = 8), or another race/ethnicity (n = 8). Educational attainment varied, with 48 participants holding bachelor’s degrees and 48 holding graduate degrees. Marital status included 73 married participants, 32 divorced participants, and 24 never married participants. Participants were geographically dispersed across the four U.S. Census regions (South: n = 45; West: n = 41; Midwest: n = 36; Northeast: n = 34).
The study included a baseline assessment followed by a 20-day EMA protocol. After providing informed consent, participants completed an online baseline questionnaire assessing sociodemographic characteristics and trait-level psychological measures. Participants received five smartphone-based surveys per day via text message (SMS), delivered through a custom platform developed using the jsPsych library30, hosted on a secure, HIPAA-compliant server, with message delivery coordinated through Twilio’s API.
At study onboarding, each participant selected a personalized daily start time aligned with their typical wake time (e.g., 7:00 a.m., 8:00 a.m.) in their local time zone. The EMA platform then generated five daily prompts distributed pseudo-randomly within five sequential 3-h windows anchored to the participant’s chosen start time. For example, a participant selecting an 8:00 a.m. start time received prompts randomly distributed within the following windows: Window 1 (08:00–10:59), Window 2 (11:00–13:59), Window 3 (14:00–16:59), Window 4 (17:00–19:59), and Window 5 (20:00–22:59). Within each window, the exact prompt time was drawn from a uniform distribution, with a constraint that no two consecutive prompts could occur within 10 min of each other to reduce response burden and preserve temporal independence.
This structured five-window design ensured (1) consistent within-person temporal spacing across the study period, (2) adequate coverage of waking hours while avoiding late-night or early-morning periods, and (3) sufficient temporal resolution (maximum inter-prompt interval ≈3 h within days) for capturing dynamic processes without excessive burden. Between-person variability in timing was limited to participant-selected start times; all participants experienced the same five-window structure relative to their anchor. Overnight periods (typically 11:00 p.m.–7:00 a.m.) were excluded from sampling, resulting in a maximum inter-prompt interval of approximately 9–11 h between the final evening prompt and the first morning prompt of the subsequent day.
Each text message included a unique survey link delivered at scheduled times. Links did not time out; however, participants were informed that responses submitted after the next prompt or fewer than 10 min apart would not receive credit. Surveys required approximately 2–3 min to complete. Compensation was $1 per completed prompt plus a $10 bonus for achieving ≥90% compliance (maximum compensation = $125: $15 baseline + $100 EMA + $10 bonus). Study staff monitored compliance in real time and contacted participants who experienced technical difficulties or showed declining engagement. All procedures were approved by the Institutional Review Board at the University of California, Davis (Protocol #2102760-4).
Participants self-reported age, gender, race/ethnicity, education, employment status, household income, and habitual sleep/wake times.
Trait loneliness was assessed using the 20-item UCLA Loneliness Scale31, a validated measure of perceived social disconnection. Items were rated on a 4-point scale (1 = “Never” to 4 = “Always”), with positively worded items reverse-coded. A composite score was computed by averaging across items, with higher scores reflecting greater trait loneliness. We present the distribution of trait loneliness in Fig. 3 (M = 2.39, SD = 0.71, range = 1.00–4.00), confirming adequate variability despite excluding extremely low scorers.
Momentary loneliness was assessed using a 3-item EMA-adapted version of the UCLA Loneliness Scale32. At each prompt, participants rated how often they had felt “isolated from others,” “left out,” and “lacking companionship” since the previous prompt. Items were rated on a 5-point scale (1 = “Often” to 5 = “Hardly ever or never”), reverse-coded, and summed. For time-lagged analyses, values were shifted forward to predict subsequent outcomes. Momentary variability was quantified using within-day standard deviation (SD), and temporal instability was indexed using the mean squared successive difference (MSSD). Momentary loneliness demonstrated excellent reliability at both the between- ((omega) = 0.99) and within-person ((omega) = 0.86) levels.
Perceived rejection was measured with a single item adapted from the Health and Retirement Study33: “To what extent did you feel rejected or criticized during your most recent social interaction?” Responses were rated on a 7-point scale (0 = “Not at all” to 6 = “Extremely”). SD and MSSD were computed at the person level to quantify variability and instability. Perceived social threat encompasses concerns such as exclusion, devaluation, and interpersonal risk. Because our measure captures one component of this broader construct, we use the term “perceived rejection” throughout. The item captured experiences from participants’ most recent interaction, which typically occurred between the prior and current prompt rather than at survey completion. Therefore, associations between perceived rejection and momentary loneliness reflect near-concurrent rather than simultaneous dynamics.
At each prompt, participants indicated whether they had engaged in any interaction defined as “talking or spending time with someone in person, via phone/video call, or by text.” They also reported (1) time since the most recent interaction, (2) partner type(s), and (3) communication modality. For each inter-prompt interval (e.g., gap₁₂ to gap₄₅), a binary indicator of social interaction (SI) was coded as “1” if the interaction occurred within the relevant time window (≤1 h for short gaps, ≤6 h for longer gaps), and “0” otherwise.
Self-disclosure was assessed with the item: “In your last social interaction, how much did you disclose or tell them about personal/private things about yourself?” Responses were rated on a 7-point scale (0 = “Not at all” to 6 = “Extremely”), indexing momentary interpersonal openness and depth.
The study protocol was preregistered prior to data collection on October 8, 2024 (https://osf.io/r79cd/overview?view_only=a9f61b13bad045d2a65b4bf187bf4198). We note three deviations from the pre-registered protocol. First, we implemented two-part DSEM models to address substantial floor effects (i.e., high frequencies of zero values in both momentary loneliness and perceived rejection) which became apparent only during preliminary data inspection. The pre-registration specified standard continuous models but did not anticipate the distributional properties that necessitated binary/continuous partitioning. Second, analyses examining temporal variability (SD) and instability (MSSD) were exploratory additions not specified in the original protocol; these were motivated by recent literature11,28,29 highlighting the importance of dynamic features beyond mean-level associations. Third, sensitivity analyses testing alternative thresholds for binary categorization (e.g., scores >8 or >10 for loneliness) and robustness checks for trait moderation across model specifications were conducted post-hoc to evaluate the stability of primary findings. All pre-registered hypotheses (H1–H3), measures, sampling procedures, and primary analytic approaches (DSEM, multilevel models) were implemented as planned.
All data manipulation and analyses were performed in R34. Data analysis proceeded in two main stages, each aligned with a corresponding hypothesis.
To examine the dynamic interplay between momentary loneliness and perceived rejection, we employed two-part Dynamic Structural Equation Modeling (DSEM) using Mplus v8.935. This approach was selected to account for substantial floor effects in both constructs (i.e., a high frequency of zero values) necessitating separate modeling of the presence versus absence of each experience (binary part), and the intensity among those endorsing the experience (continuous part). Substantively, this distinction allows us to test whether trait loneliness moderates the occurrence of momentary states (e.g., Do chronically lonely individuals more frequently experience acute loneliness or perceived rejection?) versus their intensity when present (e.g., When experiencing loneliness, is it more severe for chronically lonely individuals?).
Each construct (loneliness and perceived rejection) was modeled using (1) a binary component capturing whether the experience was present (coded as 1) or absent (coded as 0), and (2) a continuous measure of intensity among non-zero responses. Formally, for individual i at time t:
and
where ({{mathrm{Loneliness}}}_{{it}}^{{mathrm{Binary}}}) and ({{mathrm{PR}}}_{{it}}^{{mathrm{Binary}}}) are the binary indicators, while ({{mathrm{Loneliness}}}_{{it}}) and ({{mathrm{PR}}}_{{it}}) are the observed values. The probability models are:
and
The continuous component for loneliness can be specified as follows: for observations where(,{{mathrm{Loneliness}}}_{{it}} > ,0):
Similarly, for PR,
The autoregressive and concurrent paths for the binary components can be specified as:
The autoregressive and concurrent paths for the continuous part can be specified as:
Where ({eta }^{(L)}) and ({eta }^{(P)}) represents the latent variables for loneliness and PR binary components, ({eta }^{({LC})}) and ({eta }^{({LP})}) represent the latent variables for loneliness and PR continuous components. ({{{rm{varphi }}}}_{11}^{left({{rm{L}}}right)}), ({{{rm{varphi }}}}_{11}^{left({{rm{P}}}right)}), ({varphi }_{11}^{left({LC}right)}), and ({varphi }_{11}^{left({PC}right)}) represents the autoregressive parameters for the loneliness and PR binary parts and continuous parts, respectively, while ({{{rm{varphi }}}}_{12}^{left({{rm{L}}}right)}), ({{{rm{varphi }}}}_{12}^{left({{rm{P}}}right)}), ({varphi }_{12}^{left({LC}right)}), and ({varphi }_{12}^{left({PC}right)}) represents the concurrent parameters between these parts.
The 3-item momentary loneliness scale yields raw scores ranging from 3 to 15 after reverse-coding and summing. We linearly transformed these scores by subtracting 3, resulting in a 0–12 range to facilitate modeling. Binary loneliness was defined as scores exceeding the scale midpoint (>6 on the transformed scale, corresponding to >9 on the original scale), indicating the presence of loneliness at that assessment.
DSEM estimated both autoregressive (inertia) and near-concurrent effects at the within-person level, allowing for examination of short-term coupling between loneliness and threat. Trait loneliness was included as a moderator, with interaction terms tested in follow-up models to evaluate whether dynamic processes differed as a function of baseline loneliness severity. All models included demographic covariates (age, gender, education) at the between-person level. Bayesian estimation was used with four MCMC chains (5000 iterations; 2500 burn-in), and all parameters converged adequately (PSRF < 1.05).
To examine temporal variability and instability in loneliness and PR, we computed daily indices of intraindividual standard deviation (SD) and mean squared successive difference (MSSD). SD captured dispersion across five daily assessments; MSSD quantified the squared differences between successive time points, indexing instability. For participant i on day d:
and
Where ({L}_{t}) is the momentary loneliness at each assessment, with a maximum of 5 assessments each day, and d = 1,2,..,20; i represents the ith participant.
Parallel estimates (i.e., ({{mathrm{SD}}}_{{mathrm{PST}},i,d}) and ({{mathrm{MSSD}}}_{{mathrm{PST}},i,d})) were computed for perceived rejection. Multilevel models (MLMs) were performed using the brms package in R36 to examine the temporal variability with level 1 equations:
Where ({gamma }_{10}) and ({gamma }_{11}) are the average levels of the associations between loneliness and PR temporal variabilities.
To examine how fluctuations in momentary loneliness influence subsequent social behavior, we estimated Bayesian multilevel models (MLMs) using the brms package in R36. This modeling approach accounts for the nested data structure (observations within individuals), accommodates missingness under a Missing At Random (MAR) assumption, and supports random slopes to capture individual differences in within-person effects. Momentary loneliness scores were person-mean centered to disaggregate within-person deviations from between-person differences. This centering isolates the effect of transient increases or decreases in loneliness relative to an individual’s typical level, allowing for a focused examination of how short-term deviations in loneliness predict subsequent behavior. Separate models were fit for two behavioral outcomes: (1) social interaction (binary), and (2) self-disclosure (continuous). For example, the model for predicting the likelihood of a social interaction at time t as a function of loneliness at t–1 was specified as:
Level 1:
Level 2:
Where ({eta }_{{PC},{L}_{{it}},C}) is the person-centered loneliness for participant i at assessment t; ({gamma }_{10}) is the average within-person effect of daily loneliness on the likelihood of engaging in social interaction across participants.
Bayesian estimation was conducted using default weakly informative priors unless otherwise specified. Convergence was assessed using the potential scale reduction factor (R̂), with all estimates indicating acceptable convergence (R̂ < 1.1). Posterior predictive checks supported the adequacy of the model fits.
To evaluate whether trait loneliness moderates the dynamic processes identified in Hypotheses 1 and 2, we conducted a series of interaction models. These analyses tested whether individuals higher in trait loneliness exhibit stronger coupling between momentary loneliness, perceived rejection (PR), and social behavior. For Hypothesis 1, moderation analyses were implemented within the two-part DSEM using Mplus v8.935. Specifically, we tested trait loneliness × time interaction terms for both autoregressive (inertia) and concurrent paths involving loneliness and PR. Given estimation constraints, moderation models were conducted separately and did not include random slopes.
To assess moderation of temporal variability (i.e., standard deviation [SD] and mean squared successive difference [MSSD]), trait loneliness × SD and trait loneliness × MSSD interaction terms were included in multilevel models (MLMs) using the brms package36 in R. These models evaluated whether individuals higher in trait loneliness show stronger associations between variability in loneliness and instability in perceived rejection.
For Hypothesis 2, we tested whether trait loneliness moderated the within-person effects of momentary loneliness on subsequent social behavior (i.e., social interaction and self-disclosure). Interaction terms between trait loneliness and person-centered momentary loneliness were included in the respective MLMs.
To test the hypothesis that loneliness and perceived rejection are dynamically coupled over time, we implemented two-part dynamic structural equation modeling (DSEM). Each construct was partitioned into a binary component (presence vs. absence) and a continuous intensity measure for nonzero reports, permitting the joint estimation of autoregressive and cross-lagged effects while accommodating floor effects37. Models included age, gender, and education as covariates.
Consistent with predictions, bidirectional associations emerged across both the binary (presence vs. absence) and continuous (intensity) components of loneliness and perceived rejection (Table 1). Specifically, perceiving rejection was associated with greater concurrent loneliness intensity (β = 0.261, 95% CrI [0.205, 0.322]), and reporting loneliness was associated with greater perceived rejection intensity (β = 0.110, 95% CrI [0.068, 0.155]). Reverse pathways also held: higher rejection intensity was associated with an increased likelihood of reporting loneliness (β = 0.154, 95% CrI [0.103, 0.207]), and elevated loneliness intensity was associated with a greater likelihood of perceiving rejection (β = 0.085, 95% CrI [0.066, 0.103]).
Autoregressive estimates revealed high temporal persistence in the presence of loneliness (β = 0.618, 95% CrI [0.578, 0.657]) and moderate persistence for the presence of perceived rejection (β = 0.404, 95% CrI [0.366, 0.442]). The intensity of loneliness (β = 0.321, 95% CrI [0.297, 0.346]) and perceived rejection (β = 0.169, 95% CrI [0.134, 0.203]) were less stable, suggesting greater fluctuation in severity than occurrence (Table 2).
To further test the robustness of these bidirectional associations, we estimated supplementary models treating both constructs as purely binary (presence vs. absence). These models revealed positive concurrent associations in both directions: perceived rejection was associated with increased likelihood of concurrent loneliness (β = 0.34, 95% CrI [0.29, 0.39]), and loneliness was associated with increased likelihood of concurrent perceived rejection (β = 0.31, 95% CrI [0.27, 0.36]). Moderation by trait loneliness showed a selective pattern: credible intervals excluded zero when loneliness predicted perceived rejection (binary→binary: β = 0.28, 95% CrI [0.15, 0.41]; binary→continuous: β = 0.31, 95% CrI [0.09, 0.48]), and when perceived rejection predicted loneliness in the binary→binary model (β = 0.22, 95% CrI [0.08, 0.35]). In contrast, there was little credible evidence for moderation in the binary perceived rejection→continuous loneliness (β = 0.19, 95% CrI [−0.03, 0.39]), the continuous perceived rejection→binary loneliness (β = 0.03, 95% CrI [−0.07, 0.13]), or the continuous loneliness→continuous perceived rejection models (β = −0.22, 95% CrI [−0.47, 0.06]). Together, these findings suggest that trait loneliness selectively moderates specific pathways within the loneliness-rejection dynamic, particularly amplifying how the presence of loneliness influences both the occurrence and intensity of perceived rejection, while showing less consistent moderation of intensity-based pathways.
To address our exploratory aim, we examined whether temporal variability (SD) and instability (MSSD) in loneliness corresponded to similar patterns in perceived rejection. We computed person-level indices of intra-individual standard deviation (SD, capturing overall dispersion) and mean squared successive difference (MSSD, indexing magnitude of successive changes) for both constructs. Multilevel models revealed that greater variability in loneliness was associated with greater variability in perceived rejection (β = 0.22, 95% CrI [0.14, 0.29]), and greater instability in loneliness was associated with greater instability in rejection appraisals (β = 0.17, 95% CrI [0.12, 0.23]). These findings suggest that individuals who experience more pronounced fluctuations in loneliness tend to interpret social experiences in a less stable manner, potentially undermining consistent appraisal of interpersonal context.
To test our second hypothesis, we examined whether elevations in loneliness predicted subsequent social behavior using Bayesian multilevel models estimated in R (brms package)36. Time points were nested within individuals, and momentary loneliness was person-mean centered to disaggregate within- from between-person effects. Bayesian estimation was used, and all models showed acceptable convergence (R̂ < 1.1) and fit.
Consistent with our hypothesis, higher momentary loneliness predicted a reduced likelihood of subsequent social interaction (β = –0.12, 95% CrI [–0.18, –0.06]) and lower levels of self-disclosure (β = –0.248, 95% CrI [–0.266, –0.231]) (see Table 3). These findings suggest that transient increases in loneliness are followed by a reduction in affiliative behavior, including fewer interpersonal interactions and reduced interpersonal openness.
To examine whether trait loneliness alters the magnitude of within-person associations, we tested interaction effects between trait and momentary states using two-part DSEM and multilevel models. Trait loneliness selectively moderated specific pathways between momentary loneliness and perceived rejection. Individuals higher in trait loneliness reported greater perceived rejection during moments of acute loneliness (binary loneliness → continuous rejection: β = 0.308, 95% CrI [0.090, 0.484]). The pattern of moderation was selective across model components. Credible evidence for moderation emerged when loneliness served as the predictor (binary loneliness → binary rejection: β = 0.113, 95% CrI [0.004, 0.231]; binary loneliness → continuous rejection: β = 0.308, 95% CrI [0.090, 0.484]), as well as when binary rejection predicted binary loneliness (β = 0.157, 95% CrI [0.026, 0.299]). In contrast, there was little credible evidence for moderation in other pathways, including binary rejection → continuous loneliness (β = 0.190, 95% CrI [−0.025, 0.394]), continuous rejection → binary loneliness (β = 0.030, 95% CrI [−0.074, 0.128]), and continuous loneliness → continuous rejection (β = −0.063, 95% CrI [−0.128, 0.004]).
This pattern of selective moderation suggests that trait loneliness intensifies bidirectional associations between loneliness and rejection, particularly for the presence/absence components. The significant binary-to-binary moderation (β = 0.113 for loneliness → rejection; β = 0.157 for rejection → loneliness) indicates that chronically lonely individuals are more likely to perceive rejection when experiencing even minimal loneliness, and are more likely to experience loneliness when perceiving rejection. The stronger binary-to-continuous moderation further suggests that chronically lonely individuals perceive more intense rejection during lonely states compared to their less lonely counterparts. Conversely, there was no credible evidence that trait loneliness moderated the rejection-to-loneliness pathway (β = 0.190, 95% CrI [−0.025, 0.394]), suggesting that the association between perceived rejection presence and loneliness intensity does not meaningfully differ across levels of trait loneliness. Similarly, there was little credible evidence for moderation in the reverse pathways from continuous loneliness to binary rejection (β = −0.220, 95% CrI [−0.466, 0.057]).
Finaly, trait loneliness moderated behavioral responses (Fig. 4), though there was little credible evidence for moderation of the momentary loneliness–social interaction association (β = −0.09, 95% CrI [−0.19, 0.00]). By contrast, moderation was significant for self-disclosure: individuals higher in trait loneliness showed larger reductions during lonely moments (β = −0.044, 95% CrI [−0.072, −0.011]).
Plots depict predicted values for the analytic sample (n = 157) across the observed range of person centered momentary loneliness (x-axis) at low (−1 SD; teal), mean (pink), and high (+1 SD; purple) levels of trait loneliness. In the left panel, individuals high in trait loneliness exhibit decreased likelihood of social interaction as momentary loneliness increases, whereas those low in trait loneliness maintain relatively stable interaction patterns. In the right panel, higher momentary loneliness is associated with reduced self-disclosure, with steeper declines observed among individuals high in trait loneliness. Shaded bands represent 95% credible intervals.
This study provides evidence for the dynamic processes that sustain loneliness in everyday life. Our findings support conceptualizing loneliness not only as a stable trait but also as a fluctuating affective state shaped by moment-to-moment experiences, individual vulnerability, and social behavior6,7. Integrating EMA with trait-level measures, our analyses indicated recursive affective–cognitive–behavioral patterns, such as affective persistence38, heightened sensitivity to perceived rejection20, and reduced affiliative behavior19 that maintain loneliness over time. These dynamics operate across multiple timescales—micro (momentary), meso (daily), and macro (trait-level)—demonstrating the value of high-resolution temporal designs for elucidating the mechanisms that sustain psychological risk39,40.
Consistent with our preregistered hypotheses, momentary loneliness and perceived rejection were reciprocally associated across time. Increases in loneliness predicted heightened threat perception at the subsequent assessment, and vice versa. These bidirectional associations were evident for both binary indicators (presence vs. absence) and continuous measures (intensity) and were accompanied by robust autoregressive effects, particularly for loneliness, which showed strong temporal persistence. These findings support a short-term feedback loop in which loneliness and threat appraisals mutually reinforce one another, consistent with dynamic models of emotion regulation that attribute affective persistence to attentional narrowing and reduced flexibility in interpretive processing41,42. Conceptually, this reciprocal coupling can be understood as analogous to a flywheel: once loneliness-related threat appraisal is set in motion, successive moments of perceived social danger can add “momentum,” making the affective state more likely to carry forward unless sufficient countervailing social information or regulatory flexibility introduces “friction” that slows the cycle.
Trait loneliness moderated these dynamics in complex ways. Individuals higher in trait loneliness reported greater perceived rejection intensity when experiencing momentary loneliness and exhibited stronger temporal persistence in loneliness across the 20-day study period. Interestingly, trait loneliness also moderated the binary-to-binary pathways bidirectionally: chronically lonely individuals showed stronger coupling between the occurrence of loneliness and rejection states in both directions. This bidirectional moderation suggests that trait loneliness creates a lower threshold for both perceiving rejection when lonely and experiencing loneliness when perceiving rejection. However, the absence of credible evidence for moderation in the binary rejection → continuous loneliness pathway indicates that while trait loneliness affects whether rejection triggers loneliness, it does not amplify the intensity of loneliness once triggered. These findings align with evidence that chronic loneliness predicts distinct patterns of daily loneliness fluctuation15, suggesting that trait loneliness shapes how loneliness unfolds across momentary experience. These selective moderation patterns accord with neuroimaging evidence of blunted responsiveness to affiliative cues17 and behavioral work indicating impaired recalibration to improved social conditions16. Together, these moderation effects indicate that trait loneliness amplifies the emotional and behavioral impact of momentary disconnection, reinforcing self-perpetuating cycles of affective withdrawal and helping to explain the persistence of chronic loneliness even when objective social circumstances improve.
Analyses of intra-individual variability and instability further refined this account. Participants who exhibited greater fluctuations in momentary loneliness throughout the study period also showed less consistent patterns of perceived rejection. Although exploratory, this finding is consistent with prior research linking affective instability to emotion regulation deficits, social unpredictability, and increased vulnerability to psychological distress28,43. Incorporating dynamic features such as variability and instability into models of loneliness may enhance the detection of dysregulated emotional trajectories that are not captured by mean-level approaches.
At the behavioral level, these emotional signatures had interpersonal consequences. When individuals reported higher-than-usual loneliness, they were subsequently less likely to engage in social interaction and reported lower self-disclosure. These associations suggest that loneliness coincides with behaviors that limit opportunities for social reconnection7,16. These effects were moderated by trait loneliness: among individuals low in trait loneliness, higher-than-usual loneliness was associated with stable or increased social engagement, consistent with compensatory affiliation-seeking22. In contrast, those higher in trait loneliness were more likely to reduce social interaction and self-disclosure when experiencing elevated loneliness, a pattern indicative of behavioral inhibition under conditions of perceived social threat18. These divergences highlight the role of trait loneliness in shaping whether momentary disconnection prompts compensatory social engagement or reinforces withdrawal.
Taken together, these findings support a multi-timescale framework for understanding the maintenance of loneliness. At the micro level, momentary loneliness and perceived rejection exhibited reciprocal coupling across short lags. At the meso level, greater daily variability in loneliness was linked to instability in rejection perception. At the macro level, trait loneliness intensified emotional persistence and inhibited affiliative behavior. This multilevel perspective aligns with dynamic systems models of affect regulation41, highlighting the utility of high-temporal-resolution methods in identifying recursive patterns of emotion and behavior that sustain psychological vulnerability over time44.
Several limitations warrant consideration. The sample consisted of middle-aged adults (ages 46–74) residing in the United States, which may limit generalizability to younger adults, older-old adults (75+), and populations in non-Western cultural contexts where loneliness may manifest and function differently45,46. Additionally, our exclusion of individuals with extremely low trait loneliness indicates that the findings apply primarily to those experiencing at least moderate loneliness levels. While this sampling strategy optimized our ability to detect within-person dynamics (i.e., by ensuring sufficient momentary variability and reducing floor effects) and enhanced statistical power for moderation analyses, it limits generalizability to individuals with minimal loneliness who may exhibit different temporal patterns. Future research should examine whether similar recursive dynamics emerge among individuals low in trait loneliness, or whether these patterns are specific to moderate-to-high loneliness contexts.
Beyond these sampling considerations, unmeasured socio-structural factors may also shape the observed dynamics. Employment status shapes daily social contact patterns and temporal structure, potentially moderating the persistence of lonely states through workplace interactions or their absence following retirement47. Relationship status and household composition alter the social context in which loneliness operates. Married individuals, for example, have proximal support that may buffer against loneliness-driven withdrawal, while those living alone may have fewer opportunities for spontaneous social contact48. Future research with larger, more diverse samples should examine whether these socio-structural factors moderate the strength and persistence of within-person loneliness-rejection dynamics.
Although EMA enhances ecological validity and reduces recall bias, our design involved several measurement constraints. Our binary operationalization of social interaction (presence/absence) captured only whether participants engaged socially between assessments, not the frequency, duration, or subjective quality of interactions—dimensions that may show graded responses to loneliness and carry distinct implications for well-being. For instance, loneliness may differentially affect the quantity versus quality of social contact, or individuals may maintain interaction frequency while reducing emotional depth. Event-contingent sampling designs, in which participants complete brief surveys immediately following each social interaction, could capture these nuanced aspects with greater temporal precision and reduced aggregation bias. Similarly, passive sensing technologies—such as GPS-based location tracking or acoustic monitoring via the Electronically-Activated Recorder (EAR)49—could provide objective indicators of interaction frequency, duration, and ambient social exposure without relying on self-report. Integration of multi-method approaches would enable more comprehensive characterization of how momentary loneliness shapes multiple dimensions of social engagement.
Additionally, while our core hypotheses were preregistered, the two-part modeling approach was selected to accommodate distributional properties identified during data inspection, and variability analyses were exploratory. Although this enhances methodological rigor by addressing floor effects, it introduces analytic flexibility that warrants independent replication in preregistered confirmatory studies. Future work should preregister confirmatory tests and examine alternative distributional approaches (e.g., zero-inflated models, log-transformations) as robustness checks, acknowledging the trade-offs each entails for assumptions and interpretability.
Finally, a methodological consideration concerns the temporal misalignment between loneliness and threat assessments. Loneliness ratings reflected the interval between prompts, whereas threat appraisals were anchored to the most recent social interaction. Although this structure mirrors the ecological timing of each construct, more temporally aligned or event-contingent designs may be necessary to disentangle their ordering more precisely. In addition, person-specific modeling approaches, such as idiographic vector autoregressive modeling or dynamic network analysis, may identify subgroups characterized by distinct patterns of emotional persistence, affective reactivity, or behavioral response50,51,52.
This study demonstrates that loneliness is not merely a static individual difference but a dynamic affective process that unfolds in daily life. Momentary loneliness and perceived rejection were reciprocally linked, shaped by trait loneliness, and predictive of downstream social behavior (e.g., reduced interaction and self-disclosure). These findings support a temporal process model in which persistent loneliness, heightened sensitivity to rejection, and diminished social engagement may contribute to sustaining loneliness over time. By identifying when and for whom these patterns are most likely to emerge, this work provides a foundation for temporally informed and personalized strategies to disrupt maladaptive cycles and restore social connection.
De-identified ecological momentary assessment (EMA) and baseline data used in the present study are publicly available at https://osf.io/pj95s. The shared dataset includes only the variables necessary to reproduce the analyses reported in this manuscript, and all participant identifiers have been replaced with anonymized IDs. Although these data are provided to support transparency, reproducibility, and teaching, researchers interested in using the data for publication should contact the data maintainer (E.D.B., edbeck@ucdavis.edu) for permission and to request access to additional variables not included in the public release.
The R and Mplus code used for data preprocessing, model estimation, and figure generation is publicly available on GitHub and archived on Zenodo at https://doi.org/10.5281/zenodo.17934530.
National Academies of Sciences. Social Isolation and Loneliness in Older Adults: Opportunities for the Health Care System (National Academies Press, 2020).
Cacioppo, J. T. & Cacioppo, S. The growing problem of loneliness. Lancet 391, 426 (2018).
Article PubMed PubMed Central Google Scholar
Holt-Lunstad, J. The potential public health relevance of social isolation and loneliness: prevalence, epidemiology, and risk factors. Public Policy Aging Rep. 27, 127–130 (2017).
Article Google Scholar
Holt-Lunstad, J., Smith, T. B., Baker, M., Harris, T. & Stephenson, D. Loneliness and social isolation as risk factors for mortality: a meta-analytic review. Perspect. Psychol. Sci. 10, 227–237 (2015).
Article PubMed Google Scholar
Leigh-Hunt, N. et al. An overview of systematic reviews on the public health consequences of social isolation and loneliness. Public Health 152, 157–171 (2017).
Article PubMed Google Scholar
Buecker, S., Horstmann, K. T. & Luhmann, M. Lonely today, lonely tomorrow: temporal dynamics of loneliness in everyday life and its associations with psychopathological symptoms. Soc. Psychol. Personal. Sci. 15, 170–181 (2024).
Article Google Scholar
Hawkley, L. C., Preacher, K. J. & Cacioppo, J. T. Multilevel modeling of social interactions and mood in lonely and socially connected individuals: the MacArthur social neuroscience studies. In Oxford Handbook of Methods in Positive Psychology (eds Ong, A. D. & van Dulmen, M. H. M.) 559–575 (Oxford University Press, 2007).
Compernolle, E. L., Finch, L. E., Hawkley, L. C. & Cagney, K. A. Momentary loneliness among older adults: contextual differences and their moderation by gender and race/ethnicity. Soc. Sci. Med. 285, 114307 (2021).
Article PubMed PubMed Central Google Scholar
Van Bogart, K. et al. Examining the bidirectional nature of loneliness and anxiety among older adults in daily life. J. Gerontol. Ser. B 78, 1676–1685 (2023).
Article Google Scholar
Zhaoyang, R., Harrington, K. D., Scott, S. B., Graham-Engeland, J. E. & Sliwinski, M. J. Daily social interactions and momentary loneliness: the role of trait loneliness and neuroticism. J. Gerontol. B Psychol. Sci. Soc. Sci. 77, 1791–1802 (2022).
Article PubMed PubMed Central Google Scholar
Meng, J., Wang, X., Wei, D. & Qiu, J. State loneliness is associated with emotional hypervigilance in daily life: a network analysis. Pers. Individ. Differ. 165, 1–7 (2020).
Witzel, D. D., Van Bogart, K., Harrington, E. E., Turner, S. G. & Almeida, D. M. Loneliness dynamics and physical health symptomology among midlife adults in daily life. Health Psychol. 43, 528–538 (2024).
Article PubMed PubMed Central Google Scholar
Elmer, T., Geschwind, N., Peeters, F., Wichers, M. & Bringmann, L. Getting stuck in social isolation: solitude inertia and depressive symptoms. J. Abnorm. Psychol. 129, 713–723 (2020).
Article PubMed Google Scholar
Koval, P., Kuppens, P., Allen, N. B. & Sheeber, L. Getting stuck in depression: the roles of rumination and emotional inertia. Cogn. Emot. 26, 1412–1427 (2012).
Article PubMed Google Scholar
Van Bogart, K., Smyth, J. M. & Graham-Engeland, J. E. Daily loneliness deconstructed: examining patterns of within-and between-person variation. Emotion https://doi.org/10.1037/emo0001606 (2025). Epub ahead of print.
van Roekel, E. et al. Trait and state levels of loneliness in early and late adolescents: examining the differential reactivity hypothesis. J. Clin. Child Adolesc. Psychol. 47, 888–899 (2018).
Article PubMed Google Scholar
Cacioppo, S., Capitanio, J. P. & Cacioppo, J. T. Toward a neurology of loneliness. Psychol. Bull. 140, 1464–1504 (2014).
Article PubMed PubMed Central Google Scholar
Cacioppo, J. T. & Hawkley, L. C. Perceived social isolation and cognition. Trends Cogn. Sci. 13, 447–454 (2009).
Article PubMed PubMed Central Google Scholar
Qualter, P. et al. Loneliness across the life span. Perspect. Psychol. Sci. 10, 250–264 (2015).
Article PubMed Google Scholar
Spithoven, A. W., Bijttebier, P. & Goossens, L. It is all in their mind: a review on information processing bias in lonely individuals. Clin. Psychol. Rev. 58, 97–114 (2017).
Article PubMed Google Scholar
Baumeister, R. F. & Leary, M. R. The need to belong: desire for interpersonal attachments as a fundamental human motivation. Psychol. Bull. 117, 497–529 (1995).
Article PubMed Google Scholar
Pickett, C. L., Gardner, W. L. & Knowles, M. Getting a cue: the need to belong and enhanced sensitivity to social cues. Personal. Soc. Psychol. Bull. 30, 1095–1107 (2004).
Article Google Scholar
Collins, N. L. & Miller, L. C. Self-disclosure and liking: a meta-analytic review. Psychol. Bull. 116, 457–475 (1994).
Article PubMed Google Scholar
Greene, K., Derlega, V. J. & Matthews, A. Self-disclosure in personal relationships. in Cambridge Handbook of Personal Relationships. (eds Vangelisti, A. L. & Perlman, D.) 409–427 (Cambridge University Press, Cambridge, 2006).
Irwin, A. & Taylor, D. Social Penetration: The Development of Interpersonal Relationships (Holt, Rinehart, and Winston, 1973).
Solano, C. H., Batten, P. G. & Parish, E. A. Loneliness and patterns of self-disclosure. J. Personal. Soc. Psychol. 43, 524–531 (1982).
Article Google Scholar
Wei, M., Russell, D. W. & Zakalik, R. A. Adult attachment, social self-efficacy, self-disclosure, loneliness, and subsequent depression for freshman college students: a longitudinal study. J. Couns. Psychol. 52, 602–614 (2005).
Article Google Scholar
Houben, M., Van Den Noortgate, W. & Kuppens, P. The relation between short-term emotion dynamics and psychological well-being: a meta-analysis. Psychol. Bull. 141, 901 (2015).
Article PubMed Google Scholar
Koval, P., Sutterlin, S. & Kuppens, P. Emotional inertia is associated with lower well-being when controlling for differences in emotional context. Front. Psychol. 6, 1997 (2016).
de Leeuw, J. R. jsPsych: a JavaScript library for creating behavioral experiments in a Web browser. Behav. Res. Methods 47, 1–12 (2015).
Article PubMed Google Scholar
Russell, D. W. UCLA Loneliness Scale (Version 3): reliability, validity, and factor structure. J. Personal. Assess. 66, 20–40 (1996).
Article Google Scholar
Hughes, M. E., Waite, L. J., Hawkley, L. C. & Cacioppo, J. T. A short scale for measuring loneliness in large surveys. Res. Aging 26, 655–672 (2004).
Article PubMed PubMed Central Google Scholar
Juster, F. T. & Suzman, R. An overview of the health and retirement study. J. Hum. Resour. 30, S7–S56 (1995).
Article Google Scholar
R Core Team. R: A Language and Environment for Statistical Computing (R Foundation for Statistical Computing, 2024).
Muthén, L. K. & Muthén, B. O. Mplus: Statistical Analysis with Latent Variables: User’s Guide (Version 8) (Muthen & Muthen, 2017).
Bürkner, P.-C. brms: an R Package for Bayesian multilevel models using Stan. J. Stat. Softw. 80, 1–28 (2017).
Article Google Scholar
Hamaker, E. L., Asparouhov, T., Brose, A., Schmiedek, F. & Muthen, B. At the frontiers of modeling intensive longitudinal data: dynamic structural equation models for the affective measurements from the COGITO study. Multivar. Behav. Res. 53, 820–841 (2018).
Article Google Scholar
Koval, P., Pe, M. L., Meers, K. & Kuppens, P. Affect dynamics in relation to depressive symptoms: variable, unstable or inert? Emotion 13, 1132–1141 (2013).
Article PubMed Google Scholar
Hamaker, E. L., Ceulemans, E., Grasman, R. P. P. P. & Tuerlinckx, F. Modeling affect dynamics: state of the art and future challenges. Emot. Rev. 7, 316–322 (2015).
Article Google Scholar
Kuppens, P. & Verduyn, P. Emotion dynamics. Curr. Opin. Psychol. 17, 22–26 (2017).
Article PubMed Google Scholar
Hollenstein, T. This time, it’s real: affective flexibility, time scales, feedback loops, and the regulation of emotion. Emot. Rev. 7, 308–315 (2015).
Article Google Scholar
Kuppens, P., Allen, N. B. & Sheeber, L. B. Emotional inertia and psychological maladjustment. Psychol. Sci. 21, 984–991 (2010).
Article PubMed PubMed Central Google Scholar
Koval, P., Sütterlin, S. & Kuppens, P. Emotional inertia is associated with lower well-being when controlling for differences in emotional context. Front. Psychol. 6, 1997 (2016).
Article PubMed PubMed Central Google Scholar
Batra, R., Johal, S. K., Chen, M. & Ferrer, E. Consequences of sampling frequency on the estimated dynamics of AR processes using continuous-time models. Psychol. Methods https://doi.org/10.1037/met0000595 (2023).
Lykes, V. A. & Kemmelmeier, M. What predicts loneliness? Cultural difference between individualistic and collectivistic societies in Europe. J. Cross Cult. Psychol. 45, 468–490 (2014).
Article Google Scholar
Rokach, A. The Psychological Journey to and from Loneliness: Development, Causes, and Effects of Social and Emotional Isolation (Academic Press, 2019).
Wang, M. & Shi, J. Psychological research on retirement. Annu. Rev. Psychol. 65, 209–233 (2014).
Article PubMed Google Scholar
Hawkley, L. C. & Kocherginsky, M. Transitions in loneliness among older adults: a 5-year follow-up in the national social life, health, and aging project. Res. Aging 40, 365–387 (2018).
Article PubMed Google Scholar
Mehl, M. R. The electronically activated recorder (EAR): a method for the naturalistic observation of daily social behavior. Curr. Dir. Psychol. Sci. 26, 184–190 (2017).
Article PubMed PubMed Central Google Scholar
Beck, E. D. & Jackson, J. J. Personalized prediction of behaviors and experiences: an idiographic person–situation test. Psychol. Sci. 33, 1767–1782 (2022).
Article PubMed PubMed Central Google Scholar
Bringmann, L. F. et al. A network approach to psychopathology: new insights into clinical longitudinal data. PLoS ONE 8, e60188 (2013).
Article PubMed PubMed Central Google Scholar
Fisher, A. J., Reeves, J. W., Lawyer, G., Medaglia, J. D. & Rubel, J. A. Exploring the idiographic dynamics of mood and anxiety via network analysis. J. Abnorm. Psychol. 126, 1044 (2017).
Article PubMed Google Scholar
Download references
This research was supported in part by the National Institute on Aging (R01-AG082954; PI: E.K.G.; RF1-AG088206; PI: E.D.B.) and the National Science Foundation (NSF-2336406; PI: E.D.B.). The funders had no role in study design, data collection and analysis, decision to publish or preparation of the manuscript. The authors thank the study participants and research staff for their contributions to data collection and study coordination. This work was completed while A.D.O. was on sabbatical leave at Harvard University.
Department of Psychology, Cornell University, Ithaca, NY, USA
Sijing Shao & Anthony D. Ong
Department of Psychology, University of California, Davis, Davis, CA, USA
Emorie D. Beck
Brooklyn Health, Brooklyn, NY, USA
Zoe Hawks
Feinberg School of Medicine, Northwestern University, Chicago, IL, USA
Karina Van Bogart & Eileen K. Graham
Department of Psychology, Northwestern University, Evanston, IL, USA
Eileen K. Graham
Department of Medicine, Weill Cornell Medicine, New York, NY, USA
Anthony D. Ong
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
PubMed Google Scholar
S.S. and A.D.O. conceptualized the study. S.S. and A.D.O. developed the analytic approach. S.S. conducted the statistical analyses. E.D.B. secured funding for the project. S.S. and A.D.O. drafted the initial manuscript. All authors (S.S., E.D.B., Z.H., K.V.B., E.K.G., and A.D.O.) contributed to data interpretation, critically revised the manuscript, and approved the final version.
Correspondence to Anthony D. Ong.
The authors declare no competing interests.
Communications Psychology thanks Max A. Halvorson and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editor: Troby Ka-Yan Lui. A peer review file is available.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.
Reprints and permissions
Shao, S., Beck, E.D., Hawks, Z. et al. Loneliness modulates social threat detection in daily life. Commun Psychol 4, 44 (2026). https://doi.org/10.1038/s44271-026-00410-1
Download citation
Received:
Accepted:
Published:
Version of record:
DOI: https://doi.org/10.1038/s44271-026-00410-1
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
Collection
Advertisement
Communications Psychology (Commun Psychol)
ISSN 2731-9121 (online)
© 2026 Springer Nature Limited
Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.