Psychological triggers and behavioral mechanisms of relapse in social media addiction

Abstract

Social media addiction relapse, a growing concern within the realm of social psychology, remains an underexplored topic despite its increasing prevalence in modern society. This study delves into the emotional and cognitive mechanisms underpinning relapse, utilizing the Stimulus-Organism-Response (S-O-R) framework to investigate how instant gratification, perceived irreplaceability, separation anxiety, and fear of missing out (FOMO) act as triggers for relapse through both approach and avoidance urges. Drawing on theories of self-regulation failure and compulsive behavior, an online survey conducted with 273 individuals who experienced relapse shows that these factors significantly explain variation in relapse behavior, with strong statistical support for the proposed model. By integrating media effects, emotional responses, and behavioral relapse within the framework of applied social psychology, this research contributes to the understanding of technology addiction in the context of a hyperconnected society. The findings offer valuable insights for developing interventions aimed at fostering healthier digital behaviors and mitigating the adverse effects of excessive media consumption, particularly in relation to relapse.

Introduction

With the development and transformation of information technology, social media has emerged as an integral component of people’s daily lives^1,2, which significantly affects users’ personal and professional lives^3,4. However, the overuse of social media may contribute to the emergence of addictive behaviors^5,6, which may result in a range of adverse consequences for individuals, including diminished efficiency, increased fatigue, and depression^7,8. Prior research indicated that when users consciously attempt to reduce the time and frequency in order to break a habit, they often swiftly revert to their original addictive state or may even experience exacerbated effects^9–11. Unlike substance-based addictions, where relapse is typically linked to physical withdrawal symptoms and well-defined cycles of clinical treatment, relapse in social media addiction often occurs in informal, unregulated contexts. Users may lack clear support mechanisms or structured intervention plans, making their attempts to disengage more vulnerable to failure. The return to compulsive usage often signals both the strength of platform-induced reinforcement and the inadequacy of personal or behavioral self-regulation strategies.

Nevertheless, an increasing number of scholars have shifted their attention from the beneficial effects of social media to exploring the dark side of IT use, highlighting and emphasizing the detrimental consequences of social media addiction¹². To understand the recurrence of problematic social media use, this study draws on the Stimulus-Organism-Response (S-O-R) framework, a model well-suited for analyzing how external and internal cues interact to shape behavioral outcomes. The S-O-R model enables the identification of emotionally charged stimuli (e.g., gratification or anxiety) that elicit internal psychological responses (approach or avoidance urges), which in turn drive behavioral relapse. This framework helps link affective experiences with action tendencies in a structured way, offering a coherent basis for exploring relapse as a psychological-behavioral cycle. Turel et al.¹³ proposed that individuals often turn to social media to manage negative emotions, finding temporary emotional relief. This short-term effectiveness reinforces habitual engagement, gradually entrenching reliance on digital coping. Over time, this pattern increases the likelihood of relapse into excessive use, particularly when individuals face new emotional stressors. Cao et al.¹⁴ suggested that social media serves as a novel medium for social interaction, and individuals may fall back into the use of social media due to their desire for social interaction, while the absence of social support or heightened social pressure may contribute to the relapse of addictive behaviors. The multidimensional factors of social media addiction relapse require a comprehensive understanding and intervention. Thus, researching relapse is crucial for helping individuals recognize usage patterns and reduce negative impacts.

Although addiction to social media has attracted growing attention, focused investigations into its relapse patterns remain scarce. Much of the existing literature concentrates on the onset or escalation of compulsive use⁸, while few studies explore what causes individuals to return to problematic behaviors after a period of abstention. Turel et al.¹³ suggested that relapse can be a consequence of failed emotional regulation, where users revert to social media as a coping tool. Similarly, Chen et al.¹⁵ examined relapse through the lens of self-regulation failure, emphasizing the cyclical nature of reinforcement and withdrawal in digital habits. Importantly, the relapse process in social media use mirrors that of other behavioral addictions, where craving, habit loops, and psychological triggers play central roles^6,16. Despite these scattered insights, a systematic understanding of the relapse mechanism in social media contexts remains underdeveloped, underscoring the urgency and originality of the present study.

The S-O-R framework distinguishes between external stimuli, internal psychological states, and behavioral responses, making it a suitable lens to study relapse in social media addiction. Following prior studies^17–19, we apply this model to conceptualize emotionally charged experiences (such as gratification, separation anxiety, and FOMO) as stimuli that shape motivational urges and ultimately drive relapse. This framing provides a clear theoretical basis for linking environmental cues to psychological mechanisms of re-engagement.

This study constructs a model of influencing factors of social media addiction relapse, grounded in the S-O-R framework. It investigates how four emotional variables—instant gratification, perceived irreplaceability, separation anxiety, and FOMO—act as stimuli that make the approach urge and avoidance urge affect the relapse response of addicted users through these emotional experiences. The selection of these four emotional stimuli is grounded in their relevance to social media environments and their empirical linkage to relapse-prone behaviors. Instant gratification and perceived irreplaceability are reward-focused emotions that foster compulsive approach tendencies, while separation anxiety and FOMO reflect discomfort-avoidance responses that have been repeatedly linked to withdrawal distress and re-engagement^20,21. This constellation captures both hedonic and stress-relief pathways of relapse, offering a balanced view across the motivational spectrum. This study aims to offer a novel perspective for the research of the influential factors of addiction relapse by analyzing this model, thereby enriching the existing literature and facilitating the development of effective intervention strategies to mitigate the risk of addiction.

This study makes three significant contributions. Firstly, although there is a lot of literature on using the S-O-R model to study information systems^19,22,23, the literature on social media addiction relapse based on the S-O-R model is very rare. This offers an innovative perspective and thinking path for future research. Second, this paper identifies emotional factors as a source of stimuli, recognizing the significant role and impact of emotional experiences within information systems. Ultimately, this paper’s exploration of relapse of social media addiction enriches the existing literature on the adverse effects of information technology, thereby equipping individuals with a comprehensive understanding necessary to effectively mitigate impulsive behaviors stemming from these negative consequences.

Research model and hypothesis

Research model

Drawing upon prior research and the established S-O-R framework, we formulated a model to elucidate the determinants of relapse in social media addiction. The study examines how the four emotional factors act on approach urge and avoidance urge to ultimately influence social media addiction relapse. Figure 1 illustrates the conceptual framework of our research.

Fig. 1.

Research model.

Hypothesis

Instant gratification and approach urge

Social media enriches users’ lives by providing diverse functions that support relationship building and maintenance, while also enabling users to experience social support and a sense of fulfillment²⁴. Instant gratification offers online users a favorable emotional state, which can stimulate their subjective sense of happiness and amplify the rewarding sensations associated with positive emotions while eliminating negative feelings²⁵. Additionally, users are constantly pursuing instant gratification by using social media platforms; the user’s brain has established an association with this behavior, thereby continuously promoting and reinforcing the user’s usage behavior, and users derive satisfaction from social media following their individual needs²⁶. Therefore, during periods of withdrawal from social media, users experience a heightened cognitive and behavioral drive for instant gratification, coupled with an urge to mitigate the adverse effects associated with withdrawal. This culminates in a pronounced approach urge, which can precipitate an impulsive return to their prior usage patterns.

In this study, “approach urge” refers to the motivational drive to engage with social media to obtain positive rewards—such as enjoyment, social connection, or self-enhancement. In contrast, “avoidance urge” describes the impulse to return to social media to alleviate negative emotions, including anxiety, irritability, or a fear of exclusion. These urges reflect two distinct motivational orientations identified in behavioral addiction literature, with approach urges tied to reward-seeking behaviors and avoidance urges linked to coping-oriented behaviors. Recent studies on mobile social media use reinforce these emotional pathways, showing that users’ attachment, social connectivity, and benefit anticipation can significantly shape compulsive engagement and continuance behavior^27–29. These findings support the role of emotional salience as a precursor to motivational urges. Thus, it is predicted that,

H1

Instant gratification from social media is positively correlated with approach urge.

Perceived irreplaceability and approach urge

Users’ acceptance of new information technologies (such as social media) is significantly influenced by both perceived usability and perceived usefulness³⁰. The psychological needs of users have driven their engagement with information technology. Furthermore, the specific functionalities offered by information systems have subsequently facilitated the fulfillment of these psychological needs, reinforcing users’ engagement³¹. Perceived irreplaceability refers to the belief that social media platforms are uniquely capable of satisfying core emotional or informational needs²². This perception stems not from mere convenience but from a sense of functional exclusivity—users may believe there is no equivalent substitute for maintaining relationships, expressing identity, or accessing immediate information. This evaluation functions as a stimulus within the S-O-R model, reinforcing internal approach urges by signaling that value will be lost if one does not re-engage. Similar to other reward-based motivations, perceived irreplaceability can lead users to resume platform use even after attempts to abstain. Thus, we hypothesize:

H2

The perceived irreplaceability of social media is positively correlated with the approach urge.

Separation anxiety and avoidance urge

Individuals engage in self-motivation, cultivate and sustain stable relationships to satisfy their needs, thereby fulfilling the fundamental requirement for belonging³². Social media platforms provide users with a means of staying in touch and accessing real-time updates on demand. Users are reluctant to separate from their phones mainly because they have formed a strong emotional attachment to them, and they experience anxiety when they are separated from their phones³³. Separation anxiety in the context of digital platforms arises when individuals feel distress over the inability to remain connected to online social environments. This anxiety is not only about missing content but about losing a key emotional refuge. Prior research has demonstrated that individuals with high dependence on social technologies exhibit heightened distress when disconnected, often due to the psychological role these platforms play in emotional regulation^20,34. The avoidance urge here emerges not from positive anticipation, but from a desire to escape discomfort associated with disconnection—an urge to soothe anxiety rather than pursue pleasure. Therefore, it is predicted that,

H3

Separation anxiety is positively correlated with avoidance urge.

Fear of missing out (FOMO) and avoidance urge

FOMO leads users to maintain a continuous connection with social networks, and individuals with high levels of FOMO may even overuse their smartphones to satisfy their constant desire for connectivity²⁰. Fear of missing out intensifies users’ perception of potential loss when disengaged from online activity, often triggering compulsive checking behaviors. This response is frequently motivated by emotional discomfort—such as anxiety about exclusion or falling behind peers—which individuals seek to avoid^21,35. Rather than being driven by reward-seeking, the avoidance urge here reflects an internal drive to minimize anticipated negative emotional states. Users may thus re-engage with social media not for gratification, but to avoid the stress of missing out. Thus, it is predicted that,

H4

FOMO is positively correlated with avoidance urge.

Approach urge, avoidance urge, and relapse

McAuliffe³⁶ emphasized that individuals attempting to quit a drug addiction must possess the resilience to resist both temptation and craving, otherwise the addict cannot achieve the goal of quitting, and the role of desire or impulse has been shown to be particularly significant in relapse of addiction. The motivation and urges to smoke may stem from various factors, including an addiction to the perceived benefits of smoking or a reluctance to confront the discomfort associated with cessation, as well as being directly influenced by nicotine withdrawal³⁷. Similarly, smartphones provide users with a wealth of content and satisfaction, including online social interaction, entertaining games, and timely access to information, the desire for these contents will drive social media addicts to generate an urge again. Consequently, when individuals are exposed to relevant stimuli associated with social media, they are significantly influenced and exhibit a heightened approach urge, which may precipitate a relapse into addictive behaviors. In contrast to the approach-avoidance motivation, a study by Doherty et al.³⁸ found that greater negative emotions (such as anxiety, sadness, anger, and confusion) and social psychological stress predict stronger smoking urges, this urge is a strong predictor of relapse. Avoiding urge involves using avoidance to reduce negative emotions. Heavy social media users often turn to these platforms to alleviate real-life stress. However, addicted individuals experience withdrawal symptoms that intensify original stress and trigger additional negative emotions upon cessation. This tendency to evade discomfort significantly contributes to social media addiction relapse. So, the following hypotheses are proposed:

H5

Approach urge is positively correlated with relapse.

H6

Avoidance urge is positively correlated with relapse.

Method

Measurement

To ensure content validity, the scales were adapted from existing literature, with modifications tailored to fit this research context. Initially, this study employed the research scale established by Chen et al.³⁹ to measure instant gratification. The metric for assessing perceived irreplaceability was derived from Subramani⁴⁰. The Fear of Missing Out Scale was adapted from Przybylski et al.²¹. The scale for measuring separation anxiety was adapted from the NMP-Q scale developed by Yildirim and Correia³⁴. The scales measuring approach and avoidance urge were based on the research conducted by Tiffany and Drobes⁴¹. The relapse scale was established based on the research conducted by Lee et al.⁴². Although the original relapse subscale was developed in the context of gaming addiction, the items were adapted to focus specifically on the behavioral sequence of withdrawal and return in social media use. These items do not simply capture addiction severity, but rather the experiential pattern of reducing use followed by re-engagement, consistent with the operational definition of relapse in behavioral addiction literature¹⁵. Participants selected for the study reported personal attempts to reduce social media use followed by resumption of prior behaviors, providing a valid population for the constructs under study. Specifically, respondents were asked: “Have you ever intentionally reduced or stopped your use of social media for a period, and then later returned to your previous or increased usage level?” Only those who selected “Yes” could proceed to the main questionnaire. An additional logic validation item embedded later in the survey confirmed the presence of this behavioral pattern. Participants failing to confirm consistent relapse behavior across these two checks were excluded, ensuring that the final sample reflected genuine relapse experience in line with the study’s operational definition. Importantly, relapse in this context was operationalized behaviorally—not clinically—as a self-reported return to excessive use following an intentional reduction. This aligns with behavioral addiction models used in technology studies and ensures the construct reflects repeated loss of control rather than temporary disengagement or normative fluctuation. The aforementioned scale employs a five-point Likert scale for empirical measurement, with response options spanning from ‘strongly disagree’ to ‘strongly agree’. Survey items are shown in Table 1. The proposed survey model includes gender, age, occupation, years of social media use, and frequency as control variables to examine the potential influence of demographic characteristics.

Table 1.

Measurement items.

Variable	Item	Measurement
Instant gratification (Chen et al. 39)	IG1	I use social media because it responds quickly and meets my needs
	IG2	I use social media for instant gratification
	IG3	I often use social media on impulse because of the instant enjoyment it gives me
Perceived irreplaceability (Subramani 40)	PI1	I’d be hard-pressed to find another tool that provides as much value as social media
	PI2	I’d be hard-pressed to find another platform to replace the value of social media
	PI3	I can’t easily replace social media use with other activities
Separation anxiety (Yildirim and Correia 34)	SA1	If I can’t use social media, I feel anxious because I can’t maintain social relationships
	SA2	I feel anxious if I can’t use social media because I can’t keep up with what’s happening on social media
	SA3	If I can’t use social media, I will feel uneasy and even affect my sleep
Fear of missing out (Przybylski et al. 21)	FO1	I get anxious when I don’t know what my friends are doing
	FO2	Sometimes I wonder if I spend too much time trying to understanding what’s going on
	FO3	When I’m in a good mood, I think it’s important to share details online (like updates)
Approach urge (Tiffany and Drobes 41)	AU1	I have the urge to use social media to make myself feel good
	AU2	I have the urge to use social media to increase happiness
	AU3	I have an urge to use social media for satisfaction and rewards (such as social connection, achievements, or play)
Avoidance urge (Tiffany and Drobes 41)	AvU1	I have the urge to use social media to alleviate depression
	AvU2	I have the urge to use social media to reduce irritability
	AvU3	I have the urge to use social media to reduce boredom
Relapse (Lee et al. 42)	R1	After a period of withdrawal, I started using social media again
	R2	When I reduce my use of social media, I tend to revert to my previous usage pattern
	R3	After reducing my use of social media, I went back to it and used it more often

Sample and data collection

This study used established scales from international scholars. Data were collected online via Credamo software, targeting randomly selected individuals with diverse professional backgrounds. Participants were screened through self-report questions to ensure they had previously attempted to reduce their social media use and later returned to high-frequency use, thereby confirming their relevance to the relapse phenomenon under investigation. To align with conceptual clarity, only participants who reported having consciously attempted to reduce their social media use—and subsequently resumed excessive use—were included in the final sample. This operationalization is consistent with existing work in behavioral relapse^15,42 and reflects the practical realities of relapse in non-clinical populations. Although longitudinal data would provide stronger causal inference, our cross-sectional design with retrospective self-report offers valid initial insights into relapse dynamics in digital settings. Responses were anonymous and used for academic research only. To incentivize participation, respondents received one lottery entry for a material reward upon completing the survey.

In two weeks, we collected 280 complete responses. Questionnaires completed in less than 2 min were deemed invalid due to potential insufficient attention to the questions. 273 questionnaires were considered valid survey samples. To ensure that the sample size was adequate for our analysis, we conducted an a-priori power analysis using G*Power. Based on a medium effect size (f² = 0.15), a significance level of 0.05, and a desired power of 0.80, the required minimum sample size for six predictors was calculated as 138. With 273 valid responses, our sample exceeds this threshold and provides sufficient statistical power to support the conclusions drawn from the structural model. Table 2 shows the demographic characteristics of the participants.

Table 2.

Demographics.

Category	Items	Frequency	Percentage
Gender	Male	86	31.5
	Female	187	68.5
Age	< 18	0	0
	18–24	54	19.8
	25–30	107	39.2
	31–40	96	35.2
	> 40	16	5.8
Occupation	Students	72	26.4
	Civil administration	26	9.5
	Company staff	136	49.8
	Freelance work	18	6.6
	Others	21	7.7
Usage years	< 1 year	1	0.4
	1–3 years	13	4.8
	4–6 years	52	19
	7–9 years	88	32.2
	> 10 years	119	43.6
Frequency of everyday use of social media	1–10 times	39	14.3
	11–20 times	99	36.2
	21–30 times	69	25.3
	31 or above	66	24.2

Data analysis

We used SPSS 26.0 for descriptive statistics and reliability tests, and SmartPLS 4.0 for Partial Least Squares Structural Equation Modeling (PLS-SEM), including confirmatory factor analysis (CFA) and evaluation of both measurement and structural models. PLS is particularly well-suited for our research, especially in the context of theoretical development^18,43, and it effectively facilitates the evaluation of both measurement and structural models. The model was evaluated employing the two-step methodology established by Anderson and Gerbing⁴⁴.

Measurement model

Confirmatory factor analysis was employed to evaluate both convergent validity and discriminant validity, with three criteria for assessing convergent validity proposed: a composite reliability (CR) of 0.70 or higher and an average variance extracted (AVE) exceeding 0.50^45–47, and item loadings greater than 0.60¹⁷. Results of confirmatory factor analysis are shown in Table 3. Our data achieved the established criteria for optimal performance. In addition to composite reliability AVE, we assessed the internal consistency of each construct using Cronbach’s alpha coefficients. All alpha values exceeded the recommended threshold of 0.70, indicating strong internal reliability across constructs. Specifically, Cronbach’s alphas ranged from 0.83 to 0.92, confirming the robustness and consistency of the measurement scales used.

Table 3.

Result of confirmatory factor analysis.

Construct	Item	Mean	SD	loading	Cronbach’s alpha	CR	AVE
Instant gratification	IG1	4.29	0.70	0.86	0.83	0.87	0.70
	IG2	4.37	0.73	0.83
	IG3	4.18	0.74	0.81
Perceived irreplaceability	PI1	3.95	0.91	0.88	0.89	0.90	0.75
	PI2	4.05	0.88	0.85
	PI3	3.93	0.93	0.86
Separation anxiety	SA1	3.90	0.92	0.87	0.89	0.86	0.72
	SA2	3.94	0.95	0.82
	SA3	3.67	0.90	0.86
Fear of missing out	FO1	3.46	1.01	0.90	0.92	0.93	0.81
	FO2	3.57	1.07	0.89
	FO3	3.73	1.13	0.91
Approach urge	AU1	4.15	0.71	0.82	0.83	0.85	0.66
	AU2	4.21	0.77	0.78
	AU3	4.21	0.75	0.82
Avoidance urge	AvU1	3.90	0.93	0.88	0.90	0.92	0.79
	AvU2	3.92	0.88	0.88
	AvU3	4.25	0.91	0.91
Relapse	R1	4.15	0.77	0.80	0.83	0.84	0.64
	R2	4.11	0.79	0.80
	R3	4.03	0.81	0.81

Table 4 shows that all items loaded higher at their corresponding constructs than others. They confirm the discriminant validity of the measurement model.

Table 4.

Item factor loadings and cross-loadings.

Variable	IG	PI	SA	FO	AU	AvU	R
IG1	0.86	0.06	0.10	0.02	0.07	0.01	0.06
IG2	0.83	0.08	0.14	0.12	0.16	0.06	0.09
IG3	0.81	0.04	0.04	0.06	0.18	0.11	0.18
PI1	0.07	0.88	0.07	0.11	0.06	0.14	0.16
PI2	0.04	0.85	0.16	0.05	0.04	0.14	0.16
PI3	0.08	0.86	0.12	0.12	0.06	0.13	0.16
SA1	0.11	0.11	0.87	0.14	0.11	0.05	0.12
SA2	0.10	0.17	0.82	0.15	0.22	0.09	0.13
SA3	0.09	0.10	0.86	0.12	0.24	0.09	0.10
FO1	0.10	0.06	0.09	0.90	0.11	0.09	0.10
FO2	0.04	0.09	0.12	0.89	0.07	0.12	0.12
FO3	0.06	0.13	0.19	0.91	0.08	0.08	0.06
AU1	0.21	0.12	0.16	0.07	0.82	0.02	0.10
AU2	0.12	0.04	0.17	0.16	0.78	0.15	0.17
AU3	0.11	0.02	0.21	0.04	0.82	0.06	0.19
AvU1	0.05	0.15	0.06	0.11	0.07	0.88	0.11
AvU2	0.09	0.11	0.05	0.11	0.08	0.88	0.12
AvU3	0.03	0.13	0.10	0.07	0.06	0.91	0.07
R1	0.16	0.11	0.10	0.08	0.19	0.10	0.80
R2	0.08	0.17	0.09	0.14	0.15	0.12	0.80
R3	0.11	0.22	0.14	0.07	0.11	0.10	0.81

Instant gratification = IG, Perceived irreplaceability = PI, Separation anxiety = SA, Fear of missing out = FO, Approach urge = AU, Avoidance urge = AvU, Relapse = R.

Table 5 presents the correlation coefficients and the square roots of the AVE for each construct. For adequate discriminant validity, the square roots of the AVE for each construct must surpass the correlation coefficients with other constructs^48–50. Additionally, there are varying degrees of correlations between the constructs. In summary, all variables show significant correlations, and the correlation coefficients are smaller than the corresponding AVE square roots, indicating that the variables have a certain degree of discriminant validity.

Table 5.

Correlations and square roots of AVE.

Construct	IG	PI	SA	FO	AU	AvU	R
IG	0.84
PI	0.20**	0.87
SA	0.28***	0.32***	0.85
FO	0.20**	0.26***	0.34***	0.90
AU	0.38***	0.21***	0.46***	0.26***	0.81
AvU	0.18**	0.33***	0.23***	0.25***	0.22***	0.89
R	0.32***	0.41***	0.34***	0.28***	0.41***	0.30***	0.80

Structural model

The model was evaluated using PLS analysis, as illustrated in Fig. 2. The findings indicate that the research model received comprehensive support from the data, with instant gratification and perceived irreplaceability exerting a significant positive influence on approach urge, thereby confirming hypotheses H1 and H2. Furthermore, separation anxiety and fear of missing out significantly positively affect avoidance urge, thus validating hypotheses H3 and H4. Additionally, the results provide support for hypotheses H5 and H6: Instant Gratification significantly influences Approach Urge (β = 0.447, p < 0.001), thereby validating Hypothesis 1; Perceived Irreplaceability significantly affects Approach Urge (β = 0.182, p < 0.01), confirming Hypothesis 2; Separation Anxiety significantly impacts Avoidance Urge (β = 0.190, p < 0.01), supporting Hypothesis 3; Fear of Missing Out significantly influences Avoidance Urge (β = 0.201, p < 0.01), verifying Hypothesis 4; Approach Urge significantly affects Relapse (β = 0.469, p < 0.001), substantiating Hypothesis 5; and Avoidance Urge significantly influences Relapse (β = 0.242, p < 0.001), affirming Hypothesis 6. Thus, all hypotheses are supported by the empirical data. To assess the model’s explanatory power, R-squared values were calculated for all endogenous constructs. The results indicate that 41% of the variance in Approach Urge, 32% in Avoidance Urge, and 44% in Relapse were explained by the model, demonstrating moderate to strong predictive strength. Additionally, none of the control variables (gender, age, occupation, years of use, and usage frequency) showed statistically significant effects on the endogenous constructs, underscoring the centrality of emotional and motivational pathways in predicting relapse.

Fig. 2.

Structural equation model diagram.

Discussion and conclusion

Key Findings

The study’s findings underscore a dual-path mechanism through which emotional stimuli influence relapse in social media addiction. Instant gratification and perceived irreplaceability significantly predicted approach urge, suggesting that both hedonic and utilitarian attachments to social media platforms contribute to users’ compulsion to re-engage. This supports earlier assertions by Turel and Serenko¹³, who noted that users often develop strong emotional associations with the rewarding nature of digital engagement.

Notably, our findings highlight that separation anxiety and FOMO predict avoidance urge, reinforcing the view that digital platforms serve as a coping mechanism for negative affect^20,35. While prior literature often treats FOMO as a direct antecedent of excessive use, our study distinguishes its effect as mediated through an avoidance-based psychological pathway, emphasizing users’ need to alleviate emotional discomfort rather than pursue active rewards.

The differential impacts of approach and avoidance urges on relapse further enrich our understanding. Approach urges had a stronger effect size, which aligns with the stimulus–response reinforcement logic in compulsive behavior models. However, the significant effect of avoidance urges reveals an often-overlooked dimension: the drive to escape negative emotions can be as potent as the desire to gain positive stimulation.

These findings integrate well with the S-O-R framework, but they also extend its application by illuminating how emotional triggers initiate not just cognitive appraisal but distinct motivational orientations (approach vs. avoidance). This adds nuance to existing relapse models in behavioral addiction, which have historically emphasized reward-seeking over emotion avoidance.

Theoretical implications

This study has three important theoretical implications. Firstly, this study used the S-O-R model as a theoretical framework in exploring the factors that contribute to the relapse of social media addiction. Through empirical research, we not only validated the efficacy of the S-O-R model in analyzing the phenomenon of information system technology addiction but also further broadened its application scope by extending it to the emerging domain of social media. This application not only enriches the empirical basis of the S-O-R model in the study of information technology addiction, but also provides a new perspective for understanding and predicting the behavioral responses of social media users in the face of addiction challenges. In this manner, this study not only enhances our understanding of the relapse mechanisms associated with social media addiction but also offers a more comprehensive analytical framework for future research, facilitating the exploration of the intricate dynamics of social media usage and its long-term effects on individual behavior.

Secondly, the study demonstrates that relapse in social media addiction is not driven by a single emotional state but by a constellation of triggers—instant gratification, perceived irreplaceability, separation anxiety, and FOMO—each of which exerts distinct motivational pressure. By recognizing these varied emotional stimuli as integral inputs in the relapse cycle, this research expands the emotional scope traditionally considered in information systems addiction studies.

Thirdly, the research advances relapse theory in technology use by explicitly situating these emotional drivers within the S-O-R framework. Rather than treating emotions as background conditions, the model illustrates how they operate as structured stimuli feeding into approach and avoidance urges, providing a more nuanced, process-oriented lens for understanding relapse.

Practical implications aligned with emotional and motivational mechanisms

This study highlights that relapse in social media addiction is not merely a function of time spent online but of deeper emotional and motivational dynamics. Therefore, interventions should be tailored to the specific emotional triggers and the form of motivational urge (approach vs. avoidance) they activate. For individuals highly sensitive to instant gratification, platform features such as real-time notifications and rapid feedback cycles can reinforce compulsive engagement. Tools that introduce deliberate latency (e.g., delayed notifications or batch delivery systems) may help reduce the immediate reward loop and attenuate approach urges.

In users who exhibit strong perceived irreplaceability, relapse may stem from the belief that social media is the only viable channel for connection or self-expression. Here, interventions might focus on cognitive reframing techniques that highlight alternative sources of value—such as offline community involvement or creative offline projects—thereby weakening the belief in digital exclusivity. When it comes to separation anxiety, driven by emotional discomfort when disconnected, solutions like gradual withdrawal protocols or "digital exposure therapy" could help. These approaches slowly extend periods of disconnection while providing reassurance through offline support or mood regulation tools. Moreover, affective strain—including depressive mood and perceived technostress—has been linked to maladaptive coping and problematic app use, reinforcing the significance of emotional discomfort as a relapse precursor^51,52.

Mindfulness-Based Cognitive Therapy (MBCT) offers a particularly promising intervention because it cultivates present-moment awareness and tolerance for discomfort—skills that counter both approach-driven compulsions and avoidance-based tendencies like FOMO or separation anxiety. By training users to observe urges without immediate reaction, MBCT directly weakens the emotional reactivity that drives relapse. Consolidating MBCT as a standalone approach also allows for integration with broader digital well-being strategies, such as gradual disconnection or algorithmic feed curation, ensuring that emotional regulation skills are learned alongside behavioral adjustments.

Finally, while sleep quality is undoubtedly important, we now present it not as a standalone recommendation but as a contextual factor that amplifies vulnerability to emotional reactivity, especially anxiety and FOMO. Thus, sleep management is part of a broader emotional resilience strategy that complements interventions targeting specific emotional drivers of relapse. Many existing interventions—such as limiting screen time or enforcing digital curfews—tend to focus on behavioral regulation without addressing the emotional drivers of re-engagement^53,54. Our findings suggest that these measures may fall short if they ignore the internal motivational dynamics triggered by emotional stimuli like FOMO and perceived irreplaceability. Effective intervention must go beyond time management to include emotional literacy and motivational retraining, particularly for users whose avoidance urges are rooted in anxiety or perceived social loss.

Regarding MBCT, its value lies in cultivating present-moment awareness and tolerance for discomfort—skills that directly counter the compulsive avoidance patterns associated with FOMO and separation anxiety. Sleep hygiene strategies, similarly, are not simply wellness tips; they play a foundational role in emotional regulation and stress resilience, reducing the likelihood of relapse triggered by negative affect.

Limitations and future research

This study has some limitations. Firstly, our study did not differentiate between the types of social media; we only studied the cases of relapse in addiction caused by general social media platforms. Due to the different characteristics of different social media platforms, the process of preventing relapse may involve specific factors of different types of information technology. Future research should aim to differentiate among various types of social media platforms and examine the potential differences in relapse rates.

Second, this study identified four factors as the sources of stimulation; however, there may be additional factors that are pertinent to the model. For instance, this study does not investigate the impact of individual characteristics, such as loneliness and other related variables, on the relapse of social media addiction; furthermore, the identified stimulus sources are confined to emotional factors. Future research could enhance the S-O-R model by incorporating additional intervening variables, thereby constructing a more comprehensive framework for understanding social media addiction relapse, advancing theoretical inquiry in this domain, and providing improved guidance for practical applications.

Third, the data collected in this study were obtained through a questionnaire survey method, which offers notable advantages; however, as it relies on self-assessment by users, there is potential for self-report bias^55,56. Consequently, future research should consider employing a more diverse array of methodologies.

Finally, our research was conducted in China and involved social media addiction patients from culturally similar countries, without considering the impact of cultural differences. Prior research has demonstrated that culture can significantly influence the utilization of information technology^57–59; consequently, future studies should investigate samples from diverse countries to enhance the generalizability of the findings.

Conclusion

This study contributes to the growing literature on digital relapse by mapping emotional stimuli to relapse behavior through the lens of motivational urgency. Overall, this study contributes to the literature in three distinct ways. First, it extends the S-O-R framework by positioning emotionally charged experiences as functional stimuli that trigger relapse in digital addiction. Second, it advances research on social media addiction by showing that both approach and avoidance urges play critical roles in relapse, with avoidance urges being equally consequential as reward-driven urges. Third, it integrates recent insights on emotional strain and digital well-being to propose intervention pathways, thereby bridging theoretical development with practical relevance.

Author contributions

XC: Conceptualization, Methodology, Resources, Supervision, Writing—Review & Editing, Project administration. SG and MN: Conceptualization, Methodology, Formal analysis, Investigation, Data Curation, Writing—Original draft preparation, Visualization.

Funding

There was no funding taken for this study.

Data availability

Data for this study can be attained at the request from the corresponding author.

Declarations

Competing interests

The authors declare no competing interests.

Ethical approval

According to the Declaration of Helsinki, the research was approved by SEM Hefei University of Technology’s Ethical Committee, and all respondents were required to provide written informed consent. Participation was voluntary, and participants were told of the study’s goal. Privacy was maintained, and responses were submitted anonymously.

Human or animal rights

Not applicable.