Digital solastalgia: exploring user attachment and perceived degradation in social media environments

Abstract

Social media platforms are increasingly perceived as real-world environments, developing attachments like those for physical places. When these platforms undergo significant changes or degradation, users may experience solastalgia: grief and loss linked to environmental degradation. This exploratory study investigates whether solastalgia can be felt for social media platforms. Through an online survey (n = 200), participants provided insights into social media usage, User Interface (UI) preferences, and perceived platform degradation, alongside psychometric assessments of solastalgia, social media addiction, technostress, and technology readiness. Findings support the hypothesis that solastalgia can occur for social media platforms. Key predictors include online interaction style, perceived platform mismanagement, aggressive monetization, technostress, low technology readiness, and identifying as male. Additionally, a preference for past UI designs correlates with higher solastalgia scores. We discuss implications for User Interface (UI) and User Experience (UX) design and propose directions for future research.

Introduction

Ever since internet usage started spreading, websites became spaces for aggregation, social interaction, and community building to the point of evoking parallels to physical world places such as “parks”, “forums”, “town squares”, or “galleries” (Halegoua and Polson, 2021. This became even more relevant with the advent of social media and the increasing interactivity and immersion brought in by new technologies, such as affordable virtual and augmented reality devices (Hespanhol, 2023; Oleksy et al., 2023).

As digital places become more widespread, new research is necessary to investigate whether existing psychological place-related constructs still hold up when applied to digital environments. One case is constituted by the phenomenon of “Place Attachment” (Relph, 1976; Scannell and Gifford, 2010): the bonding that occurs between individuals and their meaningful environments (Hernández et al., 2020; Lewicka, 2011). Place attachment is a frequent phenomenon and studies have already demonstrated its appearance in digital and virtual places (Oleksy et al., 2023; Plunkett, 2011). The development of place attachment can be supported by a multitude of factors. Scannell and Gifford (2010) formulated a tripartite organizing framework (the “PPP framework) that divides these in factors related to the Person, those relating to the psychological Process, and factors relating to the Place itself. Of the latter type, individuals can develop attachment to a place thanks to its capacity to support their goals, either directly (Kyle et al., 2004), or indirectly by affective self-regulatory processes (Korpela, 1989; Korpela et al., 2008). The intensity of place attachment can vary and, in some cases, can reach maladaptive levels, turning into genuine “place dependence” (Jorgensen and Stedman, 2001).

Displacement, the experience of being forcibly removed from a place of attachment, may interrupt this regulatory function of place, leading to emotional distress, grief, and longing (Fried, 2017; Fullilove, 1996). This negative affective experience is commonly defined as Nostalgia (Albrecht et al., 2007), and it is frequently reported by refugees, migrants, and individuals forcibly removed by urbanization processes. Displacement is not the only way an individual can experience the loss of a place: as they undergo changes over time, such as those due to environmental degradation, the place might end up irremediably altered. As a result, individuals that developed place attachment towards them may experience intense feelings of loss or grief. This emotional experience became particularly relevant in recent years in the research field of environmental psychology, focusing on the emotional consequences of global environmental change, especially that brought on by climate change. This experience of loss and grief due to environmental degradation was named “Solastalgia” (Albrecht et al., 2007; Galway et al., 2019). Albrecht et al. (2007) coined this term as a portmanteau of the English words “solace” and “nostalgia”. Solastalgia describes the loss of the restorative qualities and positive affectivity (solace) of a place of attachment. Besides being an essential addition to the taxonomy of “climate emotions” (Pihkala, 2022), solastalgia constitutes a source of emotional distress, both for the general population and clinical settings (Cáceres et al., 2022; Vela Sandquist et al., 2025). While similar, solastalgia and nostalgia differ in one key aspect: nostalgia refers to a place that is no longer accessible, while solastalgia refers to present environments. In this sense (Albrecht et al., 2007). Nostalgia and its self-regulatory aspects have already been described in the context of media and online spaces (Menke, 2017), however, to our knowledge, no literature exists on solastalgia in the context of digital environments.

Online places share remarkable properties with some physical ones, as they can be communal and are subject to multiple individuals’ simultaneous interaction (Malecki, 2017). Consequently, they might be construed as “common goods” (Hofmokl, 2010). These properties may lead to issues arising from human inhabitation in physical places. One such example, which parallels environmental degradation in the physical world, is the phenomenon of “data pollution” (Ben-Shahar, 2019), for example: the exponential proliferation of low-quality data and contents as a byproduct of the recent widespread adoption of artificial intelligence models (Pan et al., 2023; Xing et al., 2024). One type of online place that plays an essential role in the everyday life of sizable segments of the population is social media (Erfani and Abedin, 2018). Through social media environments, individuals share experiences of aggregation, socialization, entertainment, commerce, and work (Bayer et al., 2020). More than two decades have passed since the deployment of widely used social networks (for example, Friendster was launched in 2002, and Myspace was launched in 2003). Through the years, many existing social media environments have undergone considerable changes. While many of these changes are sometimes intentional or a reflection of the growth and popularity of a particular social media platform, sometimes they are unintentional or symptomatic effects of degradation and loss of quality in the user experience. Frequently cited forms of degradation of virtual environments are management issues, reduction in the general content quality, increases in instances of disruptive user behaviors, perceived reductions in user freedom, widespread technical issues, increases in monetization practices, flight of users from the platform, and proliferation of spam, bots, or low-quality AI-generated content.

Historically, there have been multiple instances of events in social media platforms that evoked in users a feeling of longing for the past status quo of the platform. These events may range from relatively smaller changes in the mechanics or the user interface of the platform, such as the removal of the “dislike” button on YouTube (YouTube Removing Dislike“discourages Trolls” but “Unhelpful for Users”, 2021), to large sweeping changes in platform policy, culture, and style, such as the recent acquisition of Twitter, and its rebranding into X (Conger and Hirsch, 2022). In consideration to these social media platform dynamics and parallels with real-world places, we hypothesize that individuals may perceive feelings of grief and loss, that is, solastalgia, due to social media environmental change, similarly to solastalgia for physical places.

Research questions and hypotheses

This study is exploratory and aims to investigate whether changes and degradation in social media environments may lead to feelings of Solastalgia, what moderates this relationship, and if it is reflected in preferences for physical qualities of social media, such as user interfaces (UIs). This led us adapt a psychometric scale for the measurement of solastalgia (the Brief Solastalgia Scale; Christensen et al., 2024) for the social media context and to formulate a series of research questions (RQs), and to develop a series of pre-registered hypotheses:

RQ1: Do individuals perceive solastalgia for online environments?

Our first research question is centered on the very possibility that solastalgia for social media environments exists. We hypothesize that one possible predictive factor of Solastalgia is simply time spent on one’s favorite social media platform. Time spent in contact with a place is an important factor in the development of place attachment, and the more time is spent in a particular place, the more an individual is likely to notice and experience changes with it (Lomas et al., 2024; Smaldone, 2006). We operationalized this with the following hypothesis:

H1.1: Years spent of favorite social media platform positively predict Brief Solastalgia Scale scores.

Since solastalgia stems from the degradation of an environment, a possible predictor of solastalgia in social media platform is the subjective feeling that an environmental degradation is currently happening. We operationalized this with the following hypothesis:

H1.2: Feelings of degradation of favorite social media platform positively predict Brief Solastalgia Scale scores.

Solastalgia is a complex affective state, and it may be influenced by mediators such as sense of presence (Lee, 2004; Turner and Turner, 2006), technology readiness (Blut and Wang, 2020), nostalgic feelings towards the imagined past place (Lomas et al., 2024), dependency to social media (Duradoni et al., 2020; Hou et al., 2019), and stress towards current technologies (Ayyagari et al., 2011). To research the complexities behind this possible phenomenon, we formulated the following research question:

RQ2: How do sense of presence in online environments, attitudes towards new technologies, social media addiction, affective states, and technostress factor in this relationship?

The rationale behind our investigation of sense of presence is derived from the question whether it is necessary to perceive a place as tangible and navigable to experience effects related to it (Turner and Turner, 2006). We expect that an increased sense of presence for social media platforms might increase the possible feelings of Solastalgia in individuals. To explore these questions, we formulated the following hypothesis:

H2.1: Sense of presence in social media platforms positively predicts solastalgia scores.

We expect technology readiness, that is, the inclination to adopt and pursue new technologies (Blut and Wang, 2020), to reduce the possibility of experiencing feelings of solastalgia. On the one hand, since technologically ready individuals are quick to embrace novel trends, they might not be as likely to develop attachment to social media environments. On the other hand, less technologically ready individuals might develop negative experiences to changes in social media environments. We operationalized this investigation as follows:

H2.2: Technology Readiness Scale scores negatively predict solastalgia scores.

The presence of place dependence may be an indicator of place attachment (Jorgensen and Stedman, 2001), and thus to susceptibility towards experiencing solastalgia. To quantify place dependence in the context of social media platforms, we will use the construct of “Addiction to social media” (Hou et al., 2019). This phenomenon concurs with resistance to change, and emotional attachment to the object of addiction, that is, the social media platform. For these reasons, we expect social media dependency to be a significant predictor of solastalgia. Wo operationalized this hypothesis as follows:

H2.3: Bergen Social Media Addiction Scale score positively predicts solastalgia scores.

Another possible indicator of emotional attachment might be derived from differences between emotional reactions for present social media use and recalled emotionality for past use: reporting more positive feelings towards a remembered past might be an indication of attachment. We operationalized this like so:

H2.4: Higher differences in PANAS scale scores between those referring to the present state of the social media platform and past recollections of it significantly predicts solastalgia scores.

Beside recalled emotionality, we speculate that current feelings of stress attributed to technological devices might contribute to a sense of loss and longing towards an online place of attachment (Ayyagari et al., 2011). For this reason, we formulated the following hypothesis:

H2.5: Technostress positively predicts solastalgia scores.

One aspect we also wanted to investigate was how solastalgia might be reflected in the more spatially/visually coded features of social media environments, that is, their user interfaces (UIs). A possible relationship between UI perception and solastalgia might inform future developments in UI/UX (i.e. User experience) design. For this investigation, we formulated the following research question:

RQ3: Is preference for older styles of user interfaces associated with solastalgia?

Specifically, we speculate that factors usually associated with qualitative perception of physical spaces might be reflected in solastalgia. We focus on ratings of perceived safety, comfort, and esthetic appeal, as these were identified as key predictors of restorative effects in favorite places by Korpela and colleagues (Korpela, 1989; Korpela et al., 2008). We operationalized this with the following hypothesis:

H3.1: Higher difference in ratings of safety, comfort, and esthetic appeal for user interfaces between 2024 and 2010 significantly predicts solastalgia scores.

We chose 2010 as a time point of comparison because that year marked a point of mass adoption of social media platforms and there are several archival screenshots available from that year.

Methods

Preregistration

This study methodology, together with its hypotheses and data analysis plan, was pre-registered on OSF, and it is available for consultation at the following link: https://doi.org/10.17605/OSF.IO/ZYG5J.

Sampling

Our target sample size was 200 participants. This number was established following a power analysis using the software G*Power (ver. 3.1.9.6). The analysis showed that a sample size of at least 193 is required to detect a low-moderate correlation coefficient (r = 0.20) with an 0.80 statistical power. The full G*power analysis output is displayed in the Supplementary materials (S1.0).

Participants were recruited using the services provided by the online portal Prolific.com, among residents of the United States of America, selecting a gender-balanced sample. Besides gender, we also restricted recruitment to individuals aged between 18 and 45 years old. This was done to limit the possible confounding effects of the limited social media use in older adults (Bell et al., 2013).

Materials

In this section, we will describe the scales and measurement instruments utilized as part of the present study. The formulated ad hoc items are fully displayed in the Supplementary Materials (S4.0).

Demographics

In the first part of the survey, participants were asked their age in years and their gender identification, including male, female, and others.

Social media usage questions

Participants were asked a series of questions concerning their social media usage. These questions surveyed recent usage of specific platforms and which one of them is their favorite, perceived frequency of social media use, and favorite access modality (e.g., whether the participants prefer to access their favorite social media through desktop computer, mobile phones, or others). These questions were formulated ad hoc for the purpose of this study. The list of social media platforms that could be selected as part of questions of “which of these platforms have you used in the last month” or “which of these platforms is your favorite” were sourced from a list of social platforms by the number of reported users available on Wikipedia (“List of Social Platforms with at Least 100 Million Active Users”, 2024). The original list included both “traditional” social media platforms (e.g. Facebook) and other websites or applications with social media features which do not have a focus on large-scale social interaction (e.g. WhatsApp). We decided to remove these from our list to not introduce possible confounders due to the more one-on-one interaction focus of instant messaging services rather than the broader, content-focused mode of interaction of “traditional” social media. The decision on which social platforms to retain was made independently by two of the authors (EC & SG). Analysis of interrater agreement through Cohen’s k suggests substantial agreement between the two raters (κ = 0.747, z = 4.46, p ≤ 0.001, N = 35). Disagreements were resolved through discussion. The final list of social media platforms is composed of the following 25 items: Facebook, Youtube, Instagram, TikTok, LinkedIn, Douyin, Kuaishou, Weibo, X (formerly known as Twitter), Qzone, Reddit, Pinterest, Quora, JOSH, Tieba, Threads, Xiaohongshu, Twitch, Discord, Likee, Picsart, Vevo, Tumblr, VK, and None of these. Participants who selected “None of these” were allowed to give a personalized answer using a text box. The full items are displayed in the Supplementary Materials (S4.1).

Online social interaction questions

Participants were asked to rate their habitual mode of interaction with family members, friends, and acquaintances. The rating was done through “sliders” (i.e., semantic/visual analog scales) ranging from 0 (“Mostly in real life”) to 10 (“Mostly online”). These items were formulated ad hoc for the purpose of this study. The full items are displayed in the Supplementary Materials (S4.2).

Social media years

To obtain a quantitative measure of the total usage of social media, we adapted the “Joystick years” index used by Kühn and Gallinat (2014). The participant was asked to respond numerically to the following questions: “How many days per week do you use social media platforms?”, “How many hours do you use social media platforms on these days on average?”, and “How many years have you been using social media platforms on a regular basis?”. The index was then obtained by multiplication of the numerical answers (${Social}{Media}{Years}={Days}{Per}{Week}\times {Hours}{Per}{Day}\times {Years}$). The full items are displayed in the Supplementary Materials (S4.3).

Social media degradation items

Participants were required to rate whether they experienced issues related to social media environment degradation through sliding scales ranging from 0 (“Not at all”) to 10 (“Very much”). The full items are displayed in the Supplementary Materials (S4.4).

Brief solastalgia scale (BSS)

The Brief Solastalgia Scale (BSS) is a novel psychometric tool to assess feelings of Solastalgia (Christensen et al., 2024). It is composed of 5 items on a Likert scale ranging from “Strongly disagree” to “Strongly agree”. The original wording of the scale items refers to physical environments. For this reason, we adapted the items to refer specifically to social media environments. Our version with altered wording is displayed in the Supplementary Materials (S2.0).

Bergen Social Media Addiction Scale (BSMAS)

The Bergen Social Media Addiction Scale (BSMAS) is a 6-item psychometric instrument for the measurement of symptoms of behavioral addiction towards social media (Andreassen et al., 2012, 2017; Duradoni et al., 2020). The instrument is rated on a 5-point Likert scale ranging from “Very rarely” to “Very often”.

Tech readiness scale (TRI 2.0)

To measure individual differences in the tendency to adopt new technologies, we used the second iteration of the Tech Readiness Scale (TRI 2.0), a trait-like measurement scale to assess individual propensity to embrace and use new technologies for accomplishing goals (Parasuraman and Colby, 2015). This instrument is composed of 16 items on a 5-point Likert scale ranging from “Strongly Disagree” to “Strongly Agree”. The TRI possesses four subscales, each underpinned by four items: Optimism, Innovativeness, Discomfort, and Insecurity. The first two represent motivators of tech readiness, while the other two represent inhibitors. The use of this scale was formally authorized by the original authors for the purposes of this study.

Multimodal Presence Scale (MPS)

The Multimodal Presence Scale (MPS) is a psychometric tool conceived to measure the sense of presence in mediated environments through three modalities: Physical presence, Social presence, and Self-presence (Makransky et al., 2017). The scale is composed of 15 items on a 5-point Likert scale, ranging from “Completely disagree” to “Strongly agree”. Each modality is measured by one of three subscales. The original scale was developed specifically concerning virtual reality environments. For our study purposes, we changed the wording of the items to reflect social media environments. The reworded items are displayed in the Supplementary Materials (S2.0).

Technostress scale

The technostress scale (Nimrod, 2018) measures stress induced by Information and Communication Technology (ICT) use. The original scale is composed of four subscales (“Overload”, “Complexity”, “Privacy”, and “Exclusion”), and it incorporates a total of 14 items on a 5-point Likert scale ranging from “Strongly disagree” to “Strongly agree”. We opted to implement this measurement tool instead of others, as most available scales for the measurement of technostress were specifically conceived for a workplace context, and thus, the items do not apply to the general population (e.g. Tarafdar et al., 2010). However, while the items of Nimrod’s scale can easily apply to the general population, the scale was originally focused on technostress in senior citizens. To adapt it for our sample of individuals under 45 years of age, we decided not to implement the “Exclusion” subscale since it measures feelings of “[…] inferiority compared with younger users and consequent pressure to make an effort to be included in the contemporary technological environment.” (Nimrod, 2018).

International Positive and Negative Affect Schedule—Short Form (I-PANAS–SF)

The international positive and negative affect schedule—short form (I-PANAS–SF) is a brief psychometric tool widely used to gauge the affective state (either positive or negative) of individuals. It is composed of 10 items, each a word representing either a positive or a negative emotion. Each item is rated on a 5-point Likert scale ranging from “Not at all” to “Extremely”. This tool has been developed from the original PANAS (Watson et al., 1988) into a version specifically designed to be brief and suitable for administration to international cross-cultural samples (Thompson, 2007). In this study, we will administer the scale both about the present moment and retrospectively to survey affectivity related to past social media use.

UI ratings

In addition to the psychometric scales above, participants were asked to rate six screenshots of user interfaces (UIs) on three analogic scales ranging from 0 to 10: “Safety”, “Comfort”, and “Esthetic Appeal”. The UIs belonged to three different Social Media Platforms (Facebook, YouTube, and Reddit) at two different moments in time: 2010 and 2024. The screenshots were sourced from the Web Design Museum (Web Design Museum - Discover Old Websites, Apps and Software, n.d.) and the personal accounts of the authors. A blurring effect was applied to the images to hide textual information and other potentially confounding elements, and the presentation order of the items was randomized. The six screenshots are displayed in the Supplementary Materials (S3.0).

Attention checks

Three attention check items were placed in the survey. These items prompted the participant to select a specific answer (e.g. “Please select ‘absolutely agree’ for this question”) to ensure they were not giving random or low-effort answers. Participants failing two or more attention checks were excluded, in conformity with Prolific.com site policy.

Procedure

The study followed a cross-sectional online survey design. The survey was hosted and run on Psytoolkit.org (Stoet, 2010, 2017). We estimated that the entire protocol would require no more than 10 min to be completed. A flowchart of the study processes is displayed in Fig. 1.

Fig. 1: Flowchart of the study process.

Images: Freepik, Eucalyp, Iconjam, Smashicons, Parzival’1997, Juicy_fish, Three musketeers, on Flaticon.com.

Data analysis

We performed a Confirmatory Factor Analysis (CFA) to verify whether the dimensional structure of the Brief Solastalgia Scale and the Multimodal Presence Scale remained unaltered after changing the wording of the items.

To test the relationships between variables, we used linear regression models. Below, we will outline the specific procedures we employed for each operationalized hypothesis.

• To test H1.1, we performed a linear regression, entering “years spent on favorite social media platform” as a predictor and BSS as a dependent.

• To test H1.2, we performed a multiple linear regression, entering “Feelings of degradation of favorite social media platform” as a predictor and BSS as a dependent.

• To test H2.1, we performed a linear regression, entering each of the “Multimodal Presence Scale” factors as a predictor and BSS as a dependent.

• To test H2.2, we performed a linear regression, entering the four subscales of TRI 2.0 as a predictor and BSS as a dependent.

• To test H2.3, we will perform a linear regression, entering BSMAS scores as predictor and BSS as dependent.

• To test H2.4: we will perform a linear regression entering each of the differences (Deltas, Δ) in ratings of “PANAS Positive” and “PANAS Negative” between recalled past social media experience and present social media experience as predictor and BSS as dependent.

• To test H2.5, we will perform a linear regression, entering “Technostress Scale” scores as a predictor and BSS as a dependent.

• To test H3.1: we will perform a linear regression entering each of the differences (Deltas, Δ) in ratings of “safety”, “comfort”, and “esthetic appeal” between user interfaces from 2010 and 2024 as a predictor, and BSS as dependent.

Our data analysis plan presented a single divergence from pre-registration: instead of using the total TRI 2.0 score to test H2.2, we entered all four of its subscales as predictors.

All data analyses and handling procedures were performed in R (ver. 4.2.0), running in R Studio. Confirmatory Factor Analysis (CFA) calculations were performed using the “lavaan” package (ver. 0.6-17).

Results

Participants

A total of 201 participants accessed the survey; only one failed two or more attention checks and was excluded from data analysis (attrition rate: 0.5%). The final sample consisted of 200 participants (mean age: 32.04; SD: 7.22; percentage of females: 50%). Participants took an average of 11’42” to perform the survey. The survey ran from 28/08/2024 at 18:30 GMT, to 29/08/2024 at 15:00 GMT.

Descriptive analysis

Means, standard deviations, and correlations between study variables are displayed in Table 1. Statistics about declared social media preferences and usage in the last month are reported separately in Table 2.

Table 1 Spearman’s correlation matrix of observed variables with means and standard deviations.

Table 2 Statistics of surveyed social media platforms.

Confirmatory factor analyses for modified scales

We performed CFA to evaluate whether the dimensionality of our re-worded psychometric scales remained unaltered. Computing a CFA using Diagonally-Weighted Least Squares (DWLS) estimator for our modified BSS scale shows that our version possesses good fit (χ²(5, N = 200) = 7.74, p = 0.171; baseline model χ²(10, N = 200) = 3387.08, p < 0.001; CFI = 0.999, TLI = 0.998; RMSEA = 0.052, 90% CI [0.000, 0.121], p = 0.404; SRMR = 0.028). Performing a CFA with DWLS estimator for our modified version of the MPS also shows that also in this case dimensionality was preserved (χ²(87, N = 200) = 133.55, p = 0.001; baseline model χ²(105, N = 200) = 13,498.95, p < 0.001; CFI = 0.997, TLI = 0.996; RMSEA = 0.052, 90% CI [0.033, 0.069], p = 0.413; SRMR =0.056). The reworded versions of the scales used in this study are displayed in the Supplementary Materials (S2.0).

Test of hypotheses

In this section, we will address each operationalized hypothesis on a point-by-point basis.

H1.1

Fitting a linear regression model with “Years spent on favorite social media platform” as a predictor and “Brief Solastalgia Scale” scores as outcomes yields a non-significant model (F(1, 198) = 1.303, p = 0.255, R² = 0.007).

H1.2

We performed a multiple linear regression model to assess how specific perceived social media degradation effects factor on BSS scores. The model was statistically significant (F(9, 190) = 15.78, p < 0.001), with an R² of 0.428, indicating that approximately 42.8% of the variance in total BSS scores can be explained by the independent variables. The significant predictors included “Presence of management issues” (b = 0.483, t(190) = 3.989, p < 0.001) and “Increased push in monetization” (b = 0.210, t(190) = 2.337, p = 0.020), while “Increase in low-quality content” was approaching significance (b = 0.242, t(190) = 1.951, p = 0.052). Other elements, such as “Increase in disruptive behaviors by users” and “Privacy concerns”, did not show significant effects. Performing a Shapiro-Wilk test of residuals showed that they follow a normal distribution (W = 0.994, p = 0.617), and computing a Breusch-Pagan test confirms that their distribution is homoscedastic (χ² = 0.403, Df = 1, p = 0.525). Durbin-Watson test is non-significant (D = 1.85, p = 0.262), indicating no evidence of autocorrelation in the residuals (lag one autocorrelation = 0.074).

H2.1

We fitted a multiple linear regression model to investigate the role of each presence modality in predicting BSS scores. The resulting model is not statistically significant (F(3, 196) = 1.127, p = 0.339, R² = 0.002).

H2.2

We performed a multiple regression, using all four TRI2.0 subscales as predictors (“Optimism”, “Innovativeness”, “Discomfort”, and “Insecurity”) and BSS scores as outcomes. The resulting model is statistically significant (F(4,193) = 5.158, p < 0.001, R² = 0.077). “Discomfort” (b = 1.029, t(193) = 2.824, p = 0.005) and “Insecurity” (b = 1.063, t(193) = 2.592, p = 0.010) were significant predictors, while “Optimism” and “Innovativeness” were non-significant. Performing a Shapiro-Wilk test on residuals suggests the normality assumption is violated (W = 0.985, p = 0.032); however, performing a Breusch-Pagan test shows that the assumption of homoscedasticity was not violated (χ² = 2.709, Df = 1, p = 0.100). Moreover, the Durbin-Watson test is non-significant (D = 1.90, p = 0.502), suggesting no autocorrelation of residuals (lag one autocorrelation = 0.039).

H2.3

Computing a linear regression model using Bergen Social Media Addiction Scale scores as a predictor and BSS scores as outcome results in a statistically significant model (F(1,198) = 4.755, p = 0.030). However, the percentage of variance explained by the model is modest (R² = 0.018), accounting for circa 1% of the variance. The Bergen Social Media Addiction Scale is a significant positive predictor (b = 0.140, t(196) = 2.181, p = 0.030). Shapiro-Wilk test of normality of residuals is significant (W = 0.985, p = 0.035), suggesting a violation of the normality assumption. However, both the Breusch–Pagan test (χ² = 0.356, Df = 1, p = 0.550) and the Durbin-Watson test (D = 1.777, p = 0.134, lag one autocorrelation = 0.100) are non-significant, suggesting that the other assumptions were not violated.

H2.4

To test the impact of perceived affectivity, we fitted a multiple regression model using differences in positive and negative affectivity measured through the I-PANAS–SF between the present and the start of social media usage as predictors (ΔPANAS) and BSS scores as output. The model produced by this calculation is non-significant (F(2,197) = 1.717, p = 0.182, R² = 0.007).

H2.5

We fitted a multiple regression model using as predictors the four dimensions of the Technostress scale we administered (“Overload”, “Complexity”, and “Privacy”) and the BSS scores as outcomes. The resulting model was statistically significant (F(3,196) = 5.486, p = 0.001, R² = 0.063): “Privacy” (b = 0.266, t(196) = 2.887, p = 0.004) and “Complexity” (b = 0.283, t(196) = 2.115, p = 0.036) were significant predictors, while “Overload” remained non-significant. Shapiro-Wilk test for normality of residuals was significant (W = 0.985, p = 0.030), while the Breusch-Pagan test (χ² = 2.802, Df = 1, p = 0.094) and the Durbin-Watson test (D = 1.857, p = 0.344, lag 1 autocorrelation = 0.063) were non-significant.

H3.1

To investigate whether social media UI preferences predict feelings of solastalgia, we computed a multiple regression model using BSS scores as outcomes and the differences in “Safety” ratings (ΔSafety), differences in “Comfort” ratings (ΔComfort), and differences in “Esthetic appeal” ratings (ΔAppeal) between UI screenshots from 2010 and 2024 as predictors. This yielded a statistically significant model (F(3,196) = 3.106, p = 0.028, R² = 0.031), with “ΔComfort” as a significant predictor (b = 0.251, t(196) = 2.096, p = 0.037). Instead, “ΔSafety” and “ΔComfort” were non-significant predictors. Shapiro-Wilk test of normality of residuals was statistically significant (W = 0.980, p = 0.468), suggesting a violation of the normality assumption, while the Breusch-Pagan test and the Durbin-Watson test remained non-significant (D = 1.770, p = 0.098, lag one autocorrelation = 0.107)

Exploratory analyses

In this section, we will outline additional exploratory analyses performed with our data.

Social Media Years as a predictor of BSS scores

As complementary to the pre-registered analysis to investigate H1.1, we also fitted a linear regression model using “Social Media Years” as predictor and BSS scores as outcomes. The resulting model is non-significant (F(1,198) = 0.988, p ≤ 0.001, R² = 0.321).

Degradation predictors of solastalgia controlling for Social Media Years

Fitting a multiple regression model by entering the perceived factors of degradation and their interaction with “Social Media Years” suggest a moderate relationship between the outcome variable (BSS) and the predictors. The model explained approximately 45.8% of the variance in the dependent variable (R² = 0.4577). The overall model was significant (F(19, 180) = 7.997, p < 0.001). Several individual predictors significantly contribute to the model. The main effects of management issues (β = 0.560, p = 0.002), low quality content (β = 0.477, p = 0.016) indicate significant differences in predictors when total time spent on social media is considered. However, interaction terms between Social Media Years and the degradation factors were largely nonsignificant, except for a borderline effect for User Flight (β = 0.001, p = 0.054). The model is robust according to diagnostic tests: Shapiro-Wilk for normality of residuals is non-significant (W = 0.992, p = 0.304), as well as the Breusch-Pagan test (χ² = 0.965, Df = 1, p = 0.326) and the Durbin-Watson test (D = 1.846, p = 0.29, lag one autocorrelation = 0.074).

Age and gender as predictors of BSS scores

We examined the influence of demographic variables on social media Solastalgia by fitting a multiple regression model including “Age” and “Gender” (as a dummy variable) as predictors and BSS scores as outcomes. This calculation resulted in a statistically significant model (F(2,197) = 3.223, p = 0.042, R² = 0.021). While “Gender” was a statistically significant predictor (b = 1.667, t(197) = 2.475, p = 0.014), “Age” remained non-significant. This result suggests that individuals identifying as male are more likely to experience Solastalgia for social media platforms. Shapiro-Wilk test of normality of model residuals was significant (W = 0.983, p = 0.015), while both the Breusch-Pagan test (χ² = 1.840, Df = 1, p = 0.175) and the Durbin-Watson test remained non-significant (D = 1.783, p = 0.138, lag one autocorrelation = 0.099).

Online social interactions as predictors of BSS scores

We investigated how habitual online (as opposed to in real-life) interaction with family, friends, and acquaintances might influence feelings of Solastalgia for social media spaces by entering each of the three items as predictors and BSS scores as outcome. The overall resulting model is borderline non-significant (F(3,196), p = 0.069, R² = 0.020). However, habitual online interaction with acquaintances significantly predicts BSS scores (b = −0.305, t(196) = −2.675, p = 0.008). The Shapiro-Wilk test (W = 0.987, p = 0.073), the Breusch-Pagan test (χ² = 1.699, Df = 1, p = 0.192) and the Durbin-Watson test (D = 1.783, p = 0.110, lag one autocorrelation = 0.101) were all non-significant.

X (formerly known as Twitter) as a case study

The acquisition and subsequent change in management of Twitter (currently known as X) introduced considerable changes in the mechanics, policies, and internal culture of the social media platform. We decided to exploratorily analyze this event as a “natural experiment” in which to observe the effects of rapid change in a social media platform in feelings of solastalgia. While the number of our participants who indicated X as their favorite platform was modest (n = 13, 7%) and thus unsuitable for a comparison, half of our total sample declared to have used X at least once in the past 30 days (n = 100, 50%). We compared the BSS scores of recent X users with X non-users using a Mann-Whitney U test. The scores were not significantly different between the two groups (W = 4732, p = 0.513).

Discussion

The present study investigated the evolving phenomenon of solastalgia within social media environments, explicitly examining whether changes and perceived degradation in social media platforms are associated with solastalgia toward social media environments among users. Additionally, we explored factors that might moderate this relationship, such as sense of presence, technology readiness, social media addiction, affective states, technostress, and user UIs. Our findings partially supported our hypotheses and contributed to our understanding of how individuals emotionally respond to the transformation of digital spaces. A summary of our results is displayed in Table 3.

Table 3 Summary of hypotheses results.

Coherently with our first research question, the results indicate that individuals experience solastalgia about online environments (RQ1). We found that feelings of degradation of one’s favorite social media platform significantly predicted higher scores on the BSS, and thus feelings of solastalgia (i.e. H1.2). Notably, perceptions of management issues and increased monetization efforts were significant predictors of BSS scores. This suggests that when users perceive negative changes in platform management or feel pressured by monetization strategies, they may experience emotional distress, which is common in solastalgia (Albrecht et al., 2007; Pihkala, 2022).

However, contrary to our hypothesis, the years spent on a favorite social media platform did not significantly predict solastalgia scores (i.e. H1.1). This finding implies that the duration of platform use alone is insufficient to elicit feelings of solastalgia; instead, specific negative experiences and perceptions of environmental degradation have a more direct impact on users’ emotional responses. This is also corroborated by our exploratory analyses, in which we found that absolute social media usage time (i.e. Social Media Years) does not predict solastalgia scores. However, when total time spent on social media and its interactions with the perceived degradation factors are controlled for, while management issues remain a significant predictor, monetization practices cease to be significant, and loss of content quality emerges as statistically significant. These results suggest that solastalgia does not directly depend on the duration of the exposure to the digital social environment but is subject to the quality and nature of the changes the users perceive. However, the duration of exposure does influence which degradation factors are seen as salient.

To answer the second research question (RQ2: How do sense of presence in online environments, attitudes towards new technologies, social media addiction, affective states, and technostress factor in this relationship?), we studied how various factors might influence the relationship between social media changes and solastalgia. As measured by the MPS, the sense of presence in social media platforms did not significantly predict solastalgia scores, failing to support our hypothesis, in which we expected a predictive relationship between sense of presence and solastalgia (i.e. H2.1.). This suggests that being immersed or feeling present in an online environment may not necessarily be connected to solastalgia: unlike physical environments, where a strong sense of presence can mediate emotional attachments, online environments may elicit solastalgia through different mechanisms. It is also worth noting that although its dimensional structure remains stable when adapted for social media contexts, the MPS scale is specifically designed to evaluate the sense of presence in immersive environments such as virtual and augmented reality. It may be inadequate to measure the sense of presence in the case of social media or other non-immersive computer-generated environments.

Hypothesis H2.2 (i.e. Technology Readiness Scale scores negatively predict solastalgia scores) was partially supported. The “Discomfort” and “Insecurity” subscales of the TRI (i.e. the “inhibitors” of technology readiness) were significant positive predictors of solastalgia, whereas “Optimism” and “Innovativeness” (i.e. the “promotors”) were not. This indicates that users who feel uneasy or lack confidence in new technologies are more likely to experience solastalgia in response to changes in social media platforms. These findings align with prior research suggesting that negative emotions toward technology can hinder technological acceptability.

Supporting Hypothesis H2.3 (i.e. Bergen Social Media Addiction Scale score positively predicts solastalgia scores), the BSMAS scores were a significant predictor of solastalgia, although the effect size was modest. Individuals with higher levels of social media addiction may experience more significant emotional distress when their favored platforms change, possibly due to a stronger emotional dependence and emotional attachment to the platform (Andreassen et al., 2012). This finding suggests that addictive behaviors toward social media may exacerbate feelings of loss and grief when the platform undertakes changes.

Hypothesis H2.4 (i.e. Higher differences in PANAS scale scores between those referring to the present state of the social media platform and past recollections of it significantly predict solastalgia scores) was not supported. Differences between past and present positive and negative affect (ΔPANAS) did not significantly predict solastalgia scores. This indicates that changes in the general affective state over time are not directly associated with solastalgia in the context of social media environments. Solastalgia may be related to specific emotional responses not captured by general measures of positive and negative effects measured by the PANAS.

Regarding technostress, the hypothesis that it positively predicts solastalgia scores (i.e. H2.5) was supported. The “Privacy” and “Complexity” dimensions of the technostress scale were significant predictors of solastalgia. Users who experience stress related to privacy concerns and the complexity of technology are more likely to feel solastalgia. This finding underscores the role of internal stress factors in contributing to solastalgia, highlighting that not only external changes but also users’ stress in navigating these changes play a significant role.

Our third research question examined whether a preference for older styles of user interfaces is associated with solastalgia (RQ3). Supporting our hypothesis, we found that more significant differences in comfort ratings between UIs from 2010 and 2024 significantly predicted higher solastalgia scores (H3.1). Users who perceive older interfaces as more comfortable are more likely to experience solastalgia when faced with newer, less familiar UIs. This suggests that UI design may play a role in users’ emotional attachment to platforms and that changes in design can contribute to feelings of loss.

Gender emerged as a significant predictor of solastalgia, with males reporting a higher likelihood of experiencing feelings of loss or distress related to social media platform changes compared to females. This disparity may reflect fundamental differences in how men and women interact with and form emotional attachments to online spaces. Men, for example, may be more engaged with specific platform features, forums, or specialized communities that gather information on subjects, that encourage a strong sense of belonging or identity (Krasnova et al., 2017). When these features are altered, removed, or degraded, men may feel a more acute sense of loss due to the deeper investment in these virtual environments.

Additionally, these findings may stress gender-specific emotional and cognitive responses to digital spaces. Men may be more inclined to externalize their emotional reactions to changes in social media environments, possibly by focusing on tangible losses, such as diminished functionality or the disruption of routines. In contrast, women may process such changes more internally or adopt alternative coping strategies that mitigate their emotional response to platform degradation. Varying social media usage patterns may also influence the observed gender differences; for instance, some research has suggested that women often engage in more social and relationship-focused interactions online (Muscanell and Guadagno, 2012), which could make them less sensitive to changes in the platform’s technical or visual aspects.

Moreover, the way men and women build attachments to online spaces might be related to differences in emotional investment or the perceived role of social media in their daily lives. Men may view specific platforms as an extension of their identity or social status, particularly in spaces where competition, achievement, or community reputation are essential. In addition, information seeking is a stronger motivation for social media usage in men than women (Krasnova et al., 2017). When these platform aspects are compromised, men may experience personal loss beyond dissatisfaction with the user interface. On the other hand, women may prioritize the relational and communicative aspects of social media, allowing them to adapt more quickly to changes in the platform as long as their core social connections remain intact. These gender-related patterns of solastalgia call for more research on how emotional attachment to digital environments is formed and disrupted. Future research should investigate the psychosocial mechanisms behind these differences, examining how men and women develop, maintain, and respond to the loss of digital spaces.

Our results suggest frequent online interaction with acquaintances is associated with lower solastalgia scores. This finding indicates that users who regularly engage with acquaintances online may experience less emotional distress when faced with changes or degradation in social media platforms. One possible explanation for this relationship is that these users have social networks spread across multiple platforms and communication tools. This diversity in their online interactions may act as a buffer, reducing their dependence on any single platform, and alterations or declines in the digital environment in one platform have a diminished emotional impact because these users can maintain their social connections through alternative channels.

Furthermore, users who mainly interact with acquaintances might perceive social media platforms as interchangeable tools for communication rather than unique spaces with which they have a personal connection. This practical view of the digital space may allow them to adapt more quickly to platform modifications or migrations, as their primary concern is maintaining their social networks rather than preserving specific platform features or esthetics. This flexibility reduces the likelihood of experiencing solastalgia because their emotional attachment lies with the social interactions themselves, not the medium through which they occur.

We also examined the case of X (formerly known as Twitter) as a natural experiment of rapid platform change. However, we did not find significant differences in solastalgia scores between recent X users and non-users. This null result might be due to the modest sample size we used and to the fact that we analyzed the subjects that reported having used X at least once in the past 30 days, instead of those reporting it as their favorite platform. As we have noted when discussing H1.1, time spent on a particular platform does not predict solastalgia. Thus, usage frequency might be an inadequate criterion to subdivide a sample to investigate this issue. Further research with larger samples focused on attachment to the platform, instead of time spent on it, may provide more insights.

The findings of this study have several important implications for the design and management of social media platforms. They support the idea that changes in digital environments can have tangible and significant emotional consequences for users, akin to the solastalgia experienced due to environmental alterations in physical spaces. This highlights online spaces’ important role in individuals’ emotional landscapes. As such, platform developers and managers must recognize that significant changes, especially those perceived negatively, may elicit feelings of loss and distress among users. It is important to remember though that even if strong, these affective reactions are often temporary.

In addition, the link between UI preferences and solastalgia points up the significance of user-centered design in UI/UX development. Designers should be mindful of users’ emotional attachment to interfaces that they perceive as familiar. Introducing new features or design elements gradually can help users adapt without feeling overwhelmed. Moreover, providing options for users to choose between new and classic interface styles can empower them to personalize their experience, thereby mitigating feelings of loss associated with abrupt UI changes. For example: as of September 2024, Reddit users can choose whether to use the new site UI, or to revert to the “old” Reddit interface.

Limitations and future research

Some limitations of this study should be acknowledged. First, the sample was restricted to U.S. residents aged 18–45, which may limit the generalizability of the findings to other demographic groups, such as older adults or international populations, who may experience solastalgia differently due to variations in cultural context, technological engagement, or generational perspectives. Future research could expand on these findings by including a broader age range and participants from different cultural contexts to explore social and generational differences in solastalgia related to social media. Further investigation into gender differences could further help to explain the reasons behind the higher levels of solastalgia among males found in this study. Exploring other potential moderators and mediators, such as personality traits, coping strategies, and community engagement, could provide a more comprehensive understanding of solastalgia in online environments.

Second, the study relied on self-report measures, which can introduce biases such as social desirability, where participants may respond in ways they believe are more socially acceptable, or recall bias, particularly when participants are asked to retrospectively assess their emotions or affect. These biases can potentially distort the accuracy of the data. While self-report is a commonly used method in psychological research, supplementing it with behavioral data or longitudinal tracking (usage logs) in future studies could provide a broader view of how users react emotionally to platform changes over time.

Third, we used questionnaires adapted from existing measures that were validated for different contexts. Although these adaptations were made with consideration of the original context, they may not fully capture all the aspects of emotions, perception, and solastalgia in the specific context of social media environments. Future research should develop and validate instruments tailored to measure these aspects in the context of social media environments to ensure more precise assessment of the emotional impact of platform changes.

Fourth, in our study procedure, we did not randomize the order of presentation of the psychometric scales, this may introduce several biases such as response consistency, satisficing, or common method bias (CMB). To check for CMB, we performed Harman’s single-factor test (Podsakoff et al., 2003), in which all items were entered into an exploratory factor analysis. The unrotated single factor solution accounted for only 14% of the total variance, far below the 50% threshold typically considered problematic. Fit indices (e.g., RMSEA = 0.19, TLI = 0.47) further indicated that a single-factor model did not adequately fit the data. Together, these results suggest that common method bias is unlikely to pose a serious threat to the validity of our findings (Podsakoff et al., 2003).

Finally, the cross-sectional design of this study limits our ability to infer causality. While we identified associations between various factors and solastalgia, we cannot determine whether these factors are connected to solastalgia or are simply correlated. Longitudinal studies and studies employing an experimental design are needed to track the development of solastalgia over time and determine causal relationships between the studied variables.

Overall, despite the exploratory nature of this study, given that solastalgia in social media environments has not been extensively examined in existing literature, our research benefits from being pre-registered, which adds robustness to our findings and mitigates potential confirmation biases. The results suggest that solastalgia in social media environments may indeed represent an emerging psychological phenomenon, justifying further future research. The potential impact of the development of this line of research extends to areas such as user experience design, platform sustainability, and mental health interventions within digital environments. The findings reported in this study should be validated by future research employing more informative and robust design, as well as more diverse samples to establish the generalizability of the results. Additionally, solastalgia as an environmentally linked emotional experience might be generalizable to other online spaces where user interaction is pivotal, such as Massively Multiplayer Online videogames (MMOs) or virtual reality platforms. Further studies could also investigate the specific mechanisms by which platform degradation leads to emotional distress, as well as explore potential coping strategies that users may develop in response to such changes.

Conclusions

In conclusion, this study explored the phenomenon of solastalgia within social media environments, revealing that perceived degradation of platforms, technological discomfort and insecurity, technostress related to privacy concerns and complexity, and a preference for older, more familiar user interfaces are all associated with feelings of solastalgia among users. These findings parallel the concept of solastalgia in physical spaces – where environmental changes lead to emotional distress – putting the emphasis on the emotional significance of digital environments in people’s lives. Recognizing solastalgia in digital contexts can enrich our understanding of human-computer interaction, emphasizing the need to balance innovation with familiarity to mitigate emotional distress among users. This underscores the importance of sensible platform evolution that considers user attachment and well-being, safeguarding that technological progress does not come at the expense of users’ emotional health.