A systematic review of metacognitions in Internet Gaming Disorder and problematic Internet, smartphone and social networking sites use

Abstract Background The use of new technologies is growing, and some authors have suggested that frequent use might hide a non‐chemical addiction (i.e., technological addiction). Over the last 5 years, several studies investigating the role of metacognitions in technological addictions have been published. We aim to provide the first systematic review focused on this topic, by updating the initial evidence highlighted by a previous systematic review on metacognitions across addictive behaviours (Hamonniere & Varescon, 2018). Methods Electronic literature databases (Pubmed, PsychINFO, SCOPUS and Web of Science) were searched to identify studies that examined the relationship between metacognitions and four different technological addictions (Internet Gaming Disorder, IGD; problematic Internet use, PIU; problematic smartphone use, PSU; and problematic social networking sites use, PSNSU). Results We found 13 empirical studies published between 2018 and 2021. Positive low to moderate cross‐sectional associations between the four technological addictions and both generic and specific metacognitions were found, in accordance with the metacognitive model of addictive behaviours. Positive beliefs about worry, negative beliefs about thoughts concerning uncontrollability and danger, beliefs about the need to control thoughts and a lack of cognitive confidence were associated with IGD, PIU, PSU and PSNSU. Conclusions The absence of longitudinal studies prevents us from providing definitive answers about the role of metacognitions in technological addictions. Despite this limitation, interventions that target metacognitions could be beneficial for people presenting with technological addictions.


| INTRODUCTION
Technology-related impacts are an area of particular interest for the speed with which new technology is implemented. The use of new technologies is growing, especially, but not only, among young people. Fifty percent of the world population use the Internet, and more than 3.8 billion people (i.e., almost 49% of the world's population) own a smartphone (Statista, 2021). The daily time mobile phone users spent using their devices rose from 152 min in 2014 to 215 min in 2018 and is expected to grow to 234 min by 2021.
The number of mobile devices operating worldwide was 14.02 billion in 2020, and it is expected to reach 17.72 billion by 2024, an increase of 3.7 billion devices compared to 2020 levels (Statista, 2021). In 2020, over 3.6 billion people were using social media worldwide, a number projected to increase to almost 4.41 billion in 2025.
This rapid growth in popularity of new technologies has led to various theoretical discussions and empirical investigations on the potential benefits of their use. Online social media and gaming, for instance, represent an important developmental context for the handling of certain issues, which are characteristic of adolescence, such as gender and identity exploration, self-expression and the need for peer acceptance (Gerwin et al., 2018). Online games seem to have great positive therapeutic potential in addition to their entertainment value (Griffiths et al., 2017), whilst smartphone apps provide a non-invasive, inexpensive and easy to use solution for clinicians. Despite various advantages, the lack of self-regulation in the use of the various new technologies has been well documented, and overuse of digital technologies has been recognized as a public health concern (World Health Organization, 2015). Some authors (e.g., Griffiths, 1995) have raised the possibility that frequent use might hide a non-chemical addiction, which involves humanmachine interaction (i.e., technological addiction). From this viewpoint, technological addictions could be considered a subset of behavioural addictions that are characterized by the six core dimensions of the components model of addiction (Griffiths, 2005): salience, tolerance, conflict, mood modification, withdrawal and relapse. Given the current debate on the topic (see, for a discussion, Montag et al., 2021;Starcevic et al., 2020), a further description of what we meant by technological addiction is needed. We argue that the concept of technological addiction makes sense only if the use of technology is essential for the addiction development. In other words, to define a behavioural addiction as a technological addiction, technology should not be a mere vehicle or a means to access the object of the addiction. For instance, the use of technology is not an essential feature of gambling-similarly, it is not an essential feature for shopping addiction or pornography addiction, because it is plausible to suppose that these dependencies would exist in the absence of technology and/or the Internet (Caplan, 2002;Casale et al., 2014;Davis, 2001). Conversely, in Internet Gaming Disorder (IGD), problematic smartphone use (PSU) and problematic social networking sites use (PSNSU), the utilization of technology is a necessary component of the addiction itself.
That said, some authors have suggested that excessive technology use might reflect a temporary compensatory strategy to cope with transient negative states rather than a pervasive stable pattern of behaviour (see Billieux et al., 2015;Carbonell & Panova, 2017;Kardefelt-Winther, 2014). Despite these conflicting positions, the empirical literature has shown that IGD, PSU and PSNSU share some core features with established addictions, suggesting that excessive technology use deserves scientific attention. Existing empirical evidence has thus far highlighted the commonality between the neural mechanisms underlying substance use disorder and IGD (e.g., Fauth-Bühler & Mann, 2017), PSU (e.g., Horvath et al., 2020) and PSNSU (e.g., Aydın, Obu ca, et al., 2020;Lee et al., 2021). In keeping with these results, craving symptoms have been reported among IGD subjects (e.g., King et al., 2016), social media users (e.g., Stieger & Lewetz, 2018) and smartphone users (e.g., Wilcockson et al., 2019) under abstinence conditions. Withdrawal effects across technological addictions have also been highlighted through various experimental studies (see, for a review, Fernandez et al., 2020). Similar results have been observed by those empirical studies that have used the umbrella category of problematic Internet use (PIU; Niu et al., 2016;Wang et al., 2017).
Over the past two decades, the concept of metacognition and its link with psychological problems has received increasing attention in clinical psychology. Flavell (1978Flavell ( , 1979 introduced the term 'metacognitive knowledge' defining it as knowledge about one's own (or someone else's) cognitions, motivations and emotions. Fifteen years later, Wells and Matthews (1994) Sun et al., 2017;Wells & Matthews, 1996).
In the field of addictive behaviours, metacognitions are conceptualized across three temporal phases: pre-engagement, engagement and post-engagement (Spada et al., 2013;Spada & Wells, 2009). Depending on their content, they can be separated into two factors: positive and negative metacognitions. Positive metacognitions refer to the benefits of engaging in a specific behaviour as a cognitive and affective self-regulation strategy (e.g., 'using my Smartphone will help me relax'). These metacognitions have been found to play a central role in the pre-engagement phase because they motivate individuals to engage in addictive behaviour. Negative metacognitions concern the uncontrollability and dangers of thoughts and outcomes relating to the addictive behaviour employed (e.g., 'thoughts about using my Smartphone interfere with my functioning') and are activated in the engagement and post-engagement phases. Because they trigger negative emotional states (Caselli et al., 2018;, these metacognitions are thought to be involved in the perpetuation of addictive behaviours as a means of regulating these internal states. Empirical evidence supports the role of generic metacognitions and positive and negative metacognitions specific to addictive behaviours, in substance-based addictive behaviours (Spada, Nikčevi c, et al., 2007;Spada & Wells, 2005;Spada, Zandvoort, & Wells, 2007) and recognized behavioural addictions, that is, gambling (e.g., Caselli et al., 2018;Jauregui et al., 2016;Spada, Giustina, et al., 2015). Positive and negative metacognitions have also been found to be a stronger predictor than outcome expectancies (i.e., the anticipated reinforcing and punishing consequences related to engaging in a specific behaviour) in cigarette use and nicotine dependence (Nikčevi c et al., 2017) and drinking behaviour .  to assess the eligibility of the studies: title screening, abstract screening and full papers screening. The titles and abstracts of each article identified were screened by two researchers (S. C. and A. M.), and articles that according to both reviewers did not appear to meet the inclusion criteria were excluded. A total of 13 articles met these inclusion criteria. As all the 13 studies used a cross-sectional design, we conducted the quality assessment of the 13 studies using the AXIS tool, a quality assessment tool for observational cross-sectional studies (Downes et al., 2016). The tool comprises 21 items for which there are three response options ('yes', 'no' or 'do not know') to assess study quality and reporting transparency (with 'yes' scored as 1 and 'no' or 'do not know' scored as 0). As the interpretation of the scores is subjective, we used the following guidelines (Moor & Anderson, 2019): scores indicating low quality = 1-7; scores indicating medium quality = 8-14; and scores indicating high quality = 15-20). A quality score out of 20 is then generated. Table 1 shows the quality score for each study identified by this systematic review, and additional comments have been provided in Section 3.

| Demographics of the included studies
Thirteen studies focusing on metacognitions in technological addictions were published between 2018 and 2021 (Table 1) All the studies adopted a cross-sectional design. Samples were predominantly mixed, consisting of both men and women, with the exception of two studies, which mainly involved men . Only one study (Hamidi & Ghasedi, 2020) recruited a clinical sample. The majority of the included studies consisted of community adult samples (e.g., Akbari et al., 2020;Casale et al., 2018), whilst two studies focused on high school students (Aydın, Güçlü, et al., 2020;Marino et al., 2019), and one study involved early adolescents from middle schools (Marci et al., 2021). In the review, samples of adults, adolescents and children are addressed together because the types of problems associated with technological addictions do not differ between age groups (Kuss et al., 2021).  -Significant bivariate correlations between IGD total score and positive meta-worry (r = 0.22**), negative meta-worry (r = 0.23**), beliefs about the need to control thoughts (r = 0.23**) and cognitive monitoring (r = 0.16**).

| Measures of metacognitions
After controlling for daily Internet use and negative emotion recognition: -Positive meta-worry independently predict salience, tolerance, conflict, relapse and the IGD-T total score -Negative meta-worry independently predict withdrawal conflict and the IGD-T total score -Cognitive monitoring predicted mood modification 16 Marino et al. Social Skills Inventory (Riggio, 1986). Positive meta-worry, negative metaworry, beliefs about the need to control thoughts, and cognitive monitoring independently predicted the BSMAS total score controlling for daily SNS use.

17
Marino et al.  Beyond the internal consistency, no information on the psychometric characteristics of this brief self-report measure was given.

| Generic metacognitions and technological addictions
Overall, bivariate correlations across studies showed a consistent pattern of results across PIU, IGD and PSNSU. Positive beliefs about or worry (also named positive meta-worry) and negative beliefs about thoughts concerning uncontrollability and danger (also named negative meta-worry) were consistently found to be positively associated with IGD (Aydın, Güçlü, et al., 2020), PSNSU (Balıkçı et al., 2020;Ünal-Aydin et al., 2021) and PIU (Hashemi et al., 2020;Marci et al., 2021). The correlations were small across the studies, with the exception for a higher (i.e., moderate) associations found with PIU (Hashemi et al., 2020). Noteworthy, the sample recruited in this study had a mean age higher than the other samples.
Lack of confidence in memory and attention, and beliefs about the need to control thoughts, were found to be associated with higher scores on IGD measures among both adolescents (Aydın, Güçlü, et al., 2020) and university students (Zhang et al., 2020). Similarly, social networking site (SNS) problematic users showed significantly lower cognitive confidence and higher beliefs about the need to control thoughts than non-SNS problematic users. Cognitive monitoring was the metacognitions with lower-albeit significant-associations with PIU and IGD (0.08 ≤ r ≤ 0.16), whilst SNS problematic users did not significantly differ with SNS non-problematic users in this dimension.
Although the cross-sectional design used by the studies does not allow for causal inferences, findings from studies positing a mediating role for generic metacognitions were consistent. It has been highlighted that generic metacognitions had a mediating role in the association between well-known risk factors (e.g., negative affect) and PSNSU (Casale et al., 2018) and that they predict IGD over and beyond these factors (e.g., Zhang et al., 2020). None of the studies investigated generic metacognitions in PSU.

| Specific metacognitions about technological addictions
A consistent pattern of results also emerged regarding the link between specific metacognitions and PIU, IGD, PSNSU and PSU. The association was stronger with negative metacognitions relative to positive metacognitions. For instance, bivariate coefficients of.051 and 0.64 were found with PSNSU  and IGD , respectively. Taken as a whole, the results highlight that specific metacognitions (i) had a mediating role in the association between negative affect and PIU (e.g.,  and IGD  and (ii) predict PSU beyond depressive and anxiety symptoms (Akbari et al., 2020; and time spent using online gaming (e.g., Zhang et al., 2020).

| DISCUSSION
The aim of the present study was to systematically review the current state of knowledge regarding metacognitions in technological addictions and to interpret these findings in relation to the metacognitive model of addictive behaviours (Spada et al., 2013;. We reviewed 13 cross-sectional studies examining metacognitions from four different technological addictions. In general, there is a paucity of studies examining metacognitions across these behaviours among treatment-seeking samples. Consequently, conclusions drawn from the findings of the current review are necessarily tentative. As a whole, the empirical evidence shows that people who engage in unregulated use of new technologies hold dysfunctional metacognitions, thus confirming the initial results based on a few studies mainly conducted on PIU highlighted by Hamonniere and Varescon (2018). Correlations between problematic use of new technologies and all the metacognitions, be they generic or specific, were significant and low to moderate. These results seem to further support previous arguments that metacognitions have a transdiagnostic nature . However, as already suggested (Wells & Matthews, 1996) and empirically highlighted (Sun et al., 2017), it is plausible that the type of metacognitions differ in the extent to which they are prominent in specific disorders.
Hamonniere and Varescon (2018)  in accordance with this previous systematic review, we found that cognitive monitoring was the metacognition with the lowest-albeit significant-associations with PIU and IGD, and no significant differences were found on this dimension between SNSs problematic users and non-problematic users. The belief that one's own thoughts need to be controlled, which is typical of OCD, and somewhat prevalent also in eating disorders and generalized anxiety disorders (Sun et al., 2017), might be less prominent in addictive behaviours, including technological addictions.
When it comes to metacognitions specific to addictive behaviours, we also found positive low to moderate associations between specific positive metacognitions and the four different technological addictions. The stronger the beliefs about the positive effects on emotions and cognitions of engaging in Internet, online games, social media and smartphone use, the higher the tendency to engage in these behaviours. In particular, a recent study  found that positive metacognitions predict weekly online gaming hours, which, in turn, predict negative metacognitions. A strong indi- Intriguing results also come from studies that have controlled for negative affect. Prior research in this field informs us that depression and anxiety levels need to be taken into account in research profiling metacognitions . In fact, previous findings have consistently shown that negative affect predicts PIU (Pettorruso et al., 2020), IGD (e.g., Lin et al., 2020), PSNSU (Hou et al., 2019) and PSU (e.g., Vahedi & Saiphoo, 2018). On the one hand, various included studies revealed a mediating role of metacognitions in the association between negative affect and technological addictions . On the other hand, the present review highlights that metacognitions predict IGD and PSU beyond the negative affect (Akbari et al., 2020;Zhang et al., 2020). Overall, these findings show that metacognitions affecting technological addictions cannot be entirely traced back to anxiety and depressive symptoms, which is consistent with evidence of the mediating role of metacognitions when other psychosocial vulnerabilities were considered (see, e.g., Casale et al., 2018).
Unlike what has been done with drinking behaviour and nicotine We also want to reiterate the encouragement stated repeatedly (Hamonniere & Varescon, 2018;Sun et al., 2017) to adopt longitudinal designs in this field, in order to verify whether metacognitions play a role in the initiation and maintenance of the addictive behaviours, as suggested by metacognitive model of addictive behaviours. We also need to consider that when it comes to problematic technology use a spiral effect has also been hypothesized (Slater, 2007). It is fundamental to consider what the person is actually doing on social media or through his/her smartphone, because metacognitions that had led to technology use in the first place might be reinforced by the use of a particular type of media content.
Furthermore, no research has examined metacognitions about craving in the present field, although levels of craving appear to increase following smartphone and social media abstinence (e.g., Stieger & Lewetz, 2018;Wilcockson et al., 2019) and adults with IGD report boredom and the need for stimulation as consequences of an 84-h Internet gaming abstinence (King et al., 2016). As previous research on smoking cessation has shown that people who tend to have a negative appraisal of their craving-related thoughts present a greater risk of relapse after cessation (Nosen & Woody, 2014), it might be useful for future research to focus its attention on metacognitions about craving for Internet, social media, online games and smartphone use. We were not able to explore differences and similarities in metacognitions across different technological addictions given the few studies conducted on each phenomenon. We encourage future research to make comparisons between the four phenomena considered in the current review, to determine whether some metacognitions that may be more typical of a specific technological addiction might be useful from a clinical perspective as well.
Moreover, future studies might want to include addictions in which over-use of technology might be present without being a necessary component-that is, in some cases, the use of technology might simply be a vehicle or a means to access the object of the addiction. In fact, even if it seems reasonable to assume that online gambling, compulsive shopping and sex addiction would exist in the absence of technology and/or the Internet (Davis, 2001), the very distinction may not always be clear-cut (Montag et al., 2021;Starcevic et al., 2020).
Despite the highlighted limitations, the current evidence gives initial support to the generalizability of the metacognitive model of addictive behaviours to technological addictions. Interventions that target metacognitions, like Metacognitive Therapy (Wells, 2013), could be beneficial for people showing problematic technology use, akin to what has been done for other addictive behaviours .

DATA AVAILABILITY STATEMENT
Data sharing is not applicable to this article as no new data were created or analysed in this study.