As the global population ages, the increasing prevalence of age-related conditions like depression and cognitive decline has become a major public health challenge. Although these disorders do not affect all individuals equally, they can substantially compromise quality of life for those affected. At the societal level, these conditions contribute to rising healthcare costs, diminished workforce productivity, and an escalating demand for long-term care services. Collectively, these factors threaten to undermine economic stability and overwhelm social welfare systems in aging societies [1]. Depression accounts for a substantial portion of the global burden of disease, potentially leading to reduced physical activity, diminished social interaction, and lower quality of life [2, 3]. The World Health Organization reported approximately 970 million people worldwide were affected by mental disorders in 2019, and the COVID-19 pandemic dramatically worsened this situation [4]. Studies estimate that depression affects 28.4%–35.1% of people aged over 60 years, with higher prevalence in low- and middle-income countries [5]. Depressive symptoms are also prevalent in China, a middle-income country with a rapidly aging population [6]. Cognitive decline, including mild cognitive impairment, also climbs with age [7]. Additionally, depressive symptoms and cognitive decline frequently co-occur in older adults [8], with depression often present in those with cognitive impairment [9]. Elucidating the intricate relationship between cognitive decline and depressive symptoms, as well as any mediating factors, is an urgent priority in geriatric research.

Several studies have explored the association between depression and cognition, with findings generally falling into three categories. First, they appear to be bidirectionally linked. Meta-analyse show a significant association between depression severity and cognitive impairments (including executive function, memory, and processing speed) [10]. Bennett and Thomas [11] posited that early-life depression may increase later dementia risk, and vice versa [11]. A recent 16-year study in the UK demonstrated that severe depressive symptoms predict accelerated memory decline [12]. From the scarring hypothesis perspective, depression triggers chronic physiological and neurochemical changes that impair cognitive function, resulting in long-term cognitive deficits [13]. Conversely, cognitive impairment can worsen depression by hindering self-regulation, communication, and social engagement [14]. Second, cognitive function can predict depression, but the reverse may not hold. Aichele and Ghisletta [15] found that delayed memory recall predicted future depression, but this effect lessened with age [15]. However, their study did not show that prior depression caused a decline in memory recall. Likewise, a study of older Americans revealed that higher baseline depression levels predicted future declines in episodic memory, but not the reverse [16]. Third, prior research, such as the study by Gale et al. [17], has indicated that depression does not predict cognitive impairment [17], suggesting that the two conditions are not necessarily causally linked. Previous studies have identified mechanisms linking depression and cognitive decline in older adults, including behaviors like physical inactivity, smoking, and social withdrawal, as well as mediators such as emotion regulation and sleep quality [15, 18, 19].

The rapid advancement of artificial intelligence and digital technologies has significantly increased internet adoption among Chinese middle-aged and older adults. According to official statistics, national internet penetration rates rose from 50.3% to 70.4% between 2015 and 2020, with the 50-and-older demographic exhibiting particularly dramatic growth (from 9.2% to 26.3%) [20]. The expansion of internet applications has facilitated online health services, such as medication ordering and online medical consultations, potentially enabling better self-management of health in middle-aged and older adults. In fact, previous multinational studies have demonstrated internet use’s protective effects on cognitive function [21], general healthy status [22], and psychological health [23]. While online activity necessitates some level of cognitive ability, cross-cultural evidence from North America, Europe, and East Asia suggests internet use may also enhance cognitive capacity [24] and mitigating overall cognitive decline [21, 22]. For instance, internet use may offer non-pharmacological relief from cognitive decline through cognitive stimulation and social engagement [21]. Online social activities (e.g., chats, social networking) may enhance cognitive function in seniors by promoting neuroplasticity and social engagement [25], similar to the cognitive stimulation that builds resistance to age-related brain changes [26]. Additionally, active internet users can effectively stay in touch with friends and expand their social networks, and this could reduce loneliness and depression, especially for older adults [27]. A longitudinal study of Americans over 50 found that internet use reduced depression risk by one-third [28], while Chinese cross-sectional studies have linked moderate internet use to positive outcomes for older adults, including reduced depressive symptoms and improved cognition [29]. Given that many digital tools, such as online chats and games, can affect both cognitive function and depression in this demographic, and although preferred platforms differ regionally [23, 24, 30], there may be a potential relationship between cognition, depression, and internet use. Additionally, lack of social contact and recreation increases the risk of both cognitive impairment and depression, with social activities acting as a crucial link [15, 27]. Importantly, online interaction and entertainment can provide this social engagement, improving psychological wellbeing through social activities and community connection [21, 22]. Considering internet use as an alternative form of social engagement or leisure, we hypothesize that access to the internet may mediate the relationship between cognitive function and depression.

However, current research lacks clarity on how variations in internet use impact the relationship between cognitive function and depression across different age groups [31]. The digital divide perspective underscores this uncertainty, highlighting significant disparities in internet access and its subsequent cognitive and mental health benefits between middle-aged and older adults, stemming from the latter’s limited resources and social capital [22, 32]. Furthermore, the reliance on cross-sectional and bilateral studies limits our ability to establish causality or examine trilateral associations [10, 29], and widely used longitudinal designs struggle to separate interpersonal and individual factors, hindering our understanding of the complex relationships and potentially causing biased estimates [33–35]. Therefore, we utilized the Random Intercept Cross-Lagged Panel Model (RI-CLPM), a type of structural equation modeling (SEM) [36], to disentangle within-person changes from between-person differences, capture dynamic reciprocal relationships, and enhance the rigor of causal inference [35].

This study aimed to address three key research questions: (1) What is the longitudinal association between cognitive function and depressive symptoms in middle-aged and older Chinese adults? (2) To what extent does internet use mediate this relationship? (3) Are there differences in these mediation effects between middle-aged and older populations? Using CHARLS 5-year longitudinal data and controlling for demographics [37], socioeconomic status [38], lifestyle behaviors [19], daily activities [39], and health conditions [14, 33], we systematically examined these relationships while also exploring differences in the mediating effect of internet use across different age groups.

This study provides novel evidence from the first large-scale, nationally representative longitudinal investigation examining the dynamic interrelationships among cognitive function, depressive symptoms, and internet use in Chinese adults aged 45 and older over a 5-year period. Our RI-CLPM analyses demonstrated a robust bidirectional association between depression and cognition (all p < 0.05), supporting a potentially cyclical relationship between these domains. Importantly, we identified internet use as a significant mediator in this association. Age-stratified analyses revealed that the mediation effects were generally stronger in middle-aged adults (45–64 years) than in older adults (≥65 years), although this difference was statistically significant only for the cognition-depression link. These findings advance our understanding of the complex psychocognitive mechanisms in middle-aged and older adults, while highlighting the need for further research to elucidate the observed age-dependent variations in mediation effects.

Firstly, to comprehensively assess cognitive function, we employed a three-dimensional scale encompassing orientation/calculation, episodic memory, and visual construction. Our analysis of a nationally representative longitudinal dataset revealed a bidirectional causal relationship between cognitive function and depressive symptoms, consistent with prior research conducted in the USA and Europe [34, 41]. Notably, our results demonstrated a stronger negative association between prior cognitive function and subsequent depressive symptoms than vice versa. After accounting for between-person differences, our model revealed a connection between changes in depression and cognition at the within-person level. This association might stem from severe depression impairing self-regulation, which could lead to unhealthy behaviors that increase the risk of cognitive decline [41]. Furthermore, evidence from cognitive task studies, biological mechanism investigations, and randomized controlled trials supports the link between depressive symptoms and cognitive impairment [7, 18, 47]. While acknowledging this potential bidirectional relationship, our study further examined how other factors influence their trajectories over time. Importantly, the significant reciprocal relationship between cognition and depression persisted even after controlling for covariates.

Secondly, prior research has indicated that the reciprocal relationship between cognition and depression likely involves mediating pathways encompassing multiple mechanisms, including social isolation [48], social participation [49], physical activity [50], and sleep quality [19]. Our findings provide valuable insights into the mechanisms driving the reciprocal relationship between cognition and depression. While internet use showed statistically significant but modest mediation effects (indirect effect: β = −0.003), these findings still hold importance. The effect sizes are comparable to other known modifiable risk factors in aging research. Importantly, internet users demonstrated better subsequent cognitive function and fewer depressive symptoms, suggesting meaningful clinical implications. These results align with the cognitive reserve theory, adults with better baseline cognition may benefit more from internet use through greater digital skill acquisition and selective engagement in stimulating online activities [21]. The relatively small effects may reflect our binary internet use measure, which likely underestimates the benefits of more intensive or high-quality digital engagement. Specifically, internet use offers access to social and emotional support, which may reduce social isolation and enhance overall wellbeing, thereby alleviating depressive symptoms [23]. Moreover, individuals with lower baseline cognitive function may be less likely to adopt internet use or engage in online social interactions, limiting their access to potential benefits and potentially exacerbating depressive symptoms. Conversely, we also observed internet use mediating the pathway between depression and cognitive function. This might occur because individuals experiencing more severe depression tend to withdraw socially, resulting in reduced internet use for social engagement [28]. Consequently, they may miss out on the cognitive benefits associated with internet use, including cognitive stimulation, enhanced social interaction, and access to health-related information, when compared to their more digitally engaged peers [22]. By contrast, individuals with milder depressive symptoms may be more likely to leverage digital tools to increase their social engagement, which may provide beneficial cognitive stimulation and potentially offer protection against subsequent cognitive decline [21, 51].

Furthermore, similar to established treatments for depression and cognitive impairment (e.g., pharmacotherapy, psychotherapy, neurostimulation, and exercise interventions), research on the beneficial effects of internet use faces several limitations. These include substantial heterogeneity in digital intervention quality and insufficient evidence from diverse socioeconomic populations [44]. For instance, several studies have raised concerns that inappropriate internet use may worsen depressive symptoms. Therefore, while promoting internet use among middle-aged and older adults, health professionals should guide them toward responsible usage to prevent over-dependence and potential negative health consequences [28]. Additionally, internet experience varies significantly across populations due to factors such as age, gender, education, geographic location, and socioeconomic status [23]. For example, For example, compared to younger adults, older adults may derive fewer benefits in terms of social network expansion and interpersonal relationship improvement through internet use, which are crucial for mental health resilience [24]. In light of these findings, recent studies suggest that implementing timely interventions when early changes are observed can help mitigate dementia-related cognitive decline and improve mental health outcomes [12].

Although the observed mediation effects of internet use were modest in magnitude, this likely reflects the complex, multifactorial etiology linking cognitive decline and depression in older adults. At the biological level, shared neuropathological processes (e.g., vascular changes and neuroinflammation) may drive both cognitive impairment and depressive symptoms regardless of internet use [11]. Socially, factors like loneliness, reduced social networks, and socioeconomic disadvantages that are often less modifiable through digital means may exert stronger direct effects on mental health [22]. Behaviorally, physical inactivity and sensory impairments could mediate the relationship while remaining relatively unaffected by internet engagement [50]. These competing pathways help contextualize why digital interventions alone capture only a portion of the observed association. Nevertheless, the population-level impact of scalable digital interventions may offset their individually modest effects, particularly given the high prevalence of subthreshold cognitive and depressive symptoms in aging populations, the progressive nature of cognitive decline where early small benefits may yield disproportionate long-term advantages [34], and the unique accessibility benefits for underserved groups (e.g., rural or mobility-impaired elders) who face systemic barriers to traditional care. Future interventions should adopt multimodal approaches that combine internet-based tools with established protective factors (e.g., physical activity programs and hearing correction), develop personalization algorithms targeting users’ specific risk profiles (e.g., those with vascular risks versus social isolation), and prioritize implementation research to optimize real-world effectiveness beyond efficacy studies [52].

Finally, our age-stratified analyses demonstrated that while middle-aged adults (45–64 years) generally exhibited stronger mediation effects of internet use in the cognition-depression pathway compared to older adults (≥65 years), with most comparisons reaching statistical significance (p < 0.05), some pathways showed marginally significant differences (0.05 < p < 0.1) that warrant cautious interpretation. These findings preliminarily suggest middle-aged adults may derive greater cognitive-mental health benefits from internet use, potentially making them more responsive to technology-based interventions. These exploratory findings align with existing research on age-related digital engagement [23] and underscore the importance of developing digital literacy [32]. The observed age-group differences highlight the need for age-appropriate digital support strategies, particularly for older adults who may benefit from tailored approaches to optimize internet use benefits. Further research should prioritize replication in larger samples to confirm these age-dependent effects, investigate platform-specific benefits, and examine how digital access disparities may moderate the mediation effects, in order to inform more equitable digital health policies that account for age-related variations in technology’s psychological impacts [32, 51].

This study demonstrates several important strengths. First, this nationally representative, longitudinal study confirms a potential vicious cycle between cognition and depression and identifies internet use as a mediating factor. These findings fill a gap in the literature, enriching the theoretical understanding of this link. By exploring internet use in healthy aging, our study demonstrates the potential of digital tools to alleviate cognitive decline and depressive symptoms. Second, by employing the RI-CLPM, we differentiate within-person from between-person associations, strengthening the reliability of our results. Third, our subgroup analyses identify distinct mediation effects of internet use between middle-aged (45–64 years) and older adults (≥65 years), providing empirical support for developing age-specific interventions. Lastly, our findings contribute to advancing healthy aging practices in middle-income countries, particularly through technology interventions aimed at improving the mental health and cognitive abilities of middle-aged and older adults.