Using Theoretical Models of Problematic Internet Use to Inform Psychological Formulation: A Systematic Scoping Review

Abstract

Background

Empirical research has been produced on the topic of ‘Internet Addiction’ or ‘Problematic Internet Use’ (PIU) for more than 20 years, with a variety of theoretical approaches suggested by scholars to account for the behaviour. However, the discourse has been fraught with debate around construct definition, measurement, and validity.

Aims

This review aimed to systematically review the extant literature on the topic of PIU, to identify the published psychological theories in the area, and to synthesise the findings to produce actionable information for practicing psychologists as well as academics.

Method

Given the breadth of the aims, a scoping review methodology was utilised. Four major reference libraries (Scopus, Proquest, Pubmed, Technology Research Database) were searched using a string of relevant terms.

Results

Of 1412 initial search results, eighteen theories were included in the study. Nine theories related to generalised PIU, seven related to specific Internet use issues, such as online gaming or social media, while two theories took account of both a generalised and specific view. Data were analysed using Formulation-Based Thematic Analysis (FBTA) to synthesise theory elements under the deductive headings of Predisposing, Precipitating, Maintaining, and Protective factors.

Discussion

The lack of protective factors against PIU was a prominent finding. The utility of the psychological formulation approach, particularly in an area fraught with conceptual debate and frustration with traditional medical classification systems, is emphasised.

Introduction

Academics from the fields of psychology, communication studies, and human-computer interaction have studied young people’s relationship with the Internet from many points of view for the past few decades. One important factor, discussed in a growing body of literature, is excessive use of the Internet. Researchers have largely conceptualised excessive use in one of two ways: as an addiction or specific pathology, or more generally as problematic use (Beard & Wolf, 2001; LaRose et al., 2003). However, as Beard and Wolf (2001) state, none of the literature up to the early 2000s had attempted to produce a testable theory, instead researchers simply disagreed about how to operationally define the construct.

Addiction versus Problematic Behaviour

While numerous theories have been developed at this point in time, the debate about construct definition continues. Internet Addiction (IA) was originally conceptualised as having four primary diagnostic elements: (1) an increasing level of investment of resources in online activities, (2) a negative change in emotional states when offline, (3) a tolerance to the positive effects of Internet usage, and (4) denial that Internet use is a problem (Kandell, 1998). While the Internet Addiction point of view was widely held by many researchers it suffered from numerous issues, including a lack of conceptual specificity or empirical backing, and primarily that it does not account for the very wide range of behaviours in which a person can engage while online (Davis, 2001; Shaffer et al., 2000). We will use the term Problematic Internet Use (PIU) to reflect the broader conceptualisation described by Beard and Wolf. This decision has two advantages: firstly, it maintains a broad focus for the range of potential Internet use behaviours, and 2) it avoids drawing a distinction about whether a particular online behaviour is or is not a disorder/disease/addiction.

Cognitive-Behavioural Models and Advancements

Working from the conceptualisation of excessive Internet use as a problematic behaviour, rather than an addiction/disease, Davis (2001) outlined a cognitive-behavioural model of excessive use. Rather than excessive Internet use causing psychosocial difficulties or leading to an addiction pathology, Davis proposed that pre-existing psychosocial difficulties in fact predisposed people to develop maladaptive cognitions and behaviours that in turn lead to excessive use. A key element to Davis’ theory was that he outlined an important difference between specific and generalised PIU. He argued that certain specific online behaviours, such as stock trading, accessing sexual material, or online gambling, were distinct from generalised online behaviours, whereby an individual is drawn by the experience of the Internet itself and particularly the unique communication platform that it offers. To measure PIU Caplan (2002) designed the Generalised Problematic Internet Usage Scale (GPIUS), for which he found a 7-factor structure that closely mapped onto Davis’ model. Caplan adding important elements such as a preference for online social interaction (POSI) and specific social skills deficits. He also found that scores on the GPIUS were correlated with the psychosocial factors of depression, self-esteem, loneliness, and social reticence (Caplan, 2003).

Another major theory is the Interaction of Person-Affect-Cognition-Execution (I-PACE) model (Brand et al., 2016). This model has a basis in cognitive behavioural theory, but also includes neuropsychological factors, and processes associated with substance use addictions. Other researchers have also proposed theories of PIU focusing on meeting basic needs, general proneness to problem behaviours, perceived importance of online behaviour, and emotional dysregulation (Casale et al., 2016; Kim & Davis, 2009; Ko et al., 2008; Wong et al., 2014). However, there is considerable variety in the literature, which is maintained in part by the lack of agreement on conceptual or even descriptive frameworks of PIU.

Controversies in PIU

Scholarly debate on the topic of generalised versus specific PIU is ongoing since Davis coined the terms in 2001. Research using generalised models has observed good fit and accounted for substantial variance in online behaviours (Haagsma et al., 2013; Moreau et al., 2015). Other studies have directly tested this issue and observed greater specificity and nuance by considering specific online behaviours individually (Van Rooij et al., 2017). More contentious however is the move to formalise diagnostic criteria for PIU. In the fields of medicine and psychology both classification systems and clinical practice often develop more quickly than academic disciplines can produce research findings. A prominent example is the inclusion of Internet Gaming Disorder (IGD) in Section 3 of the Diagnostic and Statistical Manual of Mental Disorders fifth Edition (American Psychiatric Association, 2013), which was met with considerable criticism (Kuss et al., 2017; Van Rooij & Kardefelt-Winther, 2017).

Formulation as an Alternative to Diagnosis

The current models of classification of mental illness, the DSM-V (American Psychiatric Association, 2013) and ICD-11 (WHO | ICD-11 Revision, 2018), view PIU through the lens of ‘disease’. An alternative to the disease model of mental health is a process called psychological formulation. This approach is how psychologists organise and synthesise information about a clinical case. In contrast to the disease model, formulation takes into account the broad bio-psycho-social view of an individual and accepts that their presenting difficulty will have idiosyncrasies, rather than fit into a diagnostic category (Johnstone, 2018). Thinking of an individual’s difficulty regulating their time spent on online activities in this way avoids the need to engage in the debate around diagnostic categories and instead offers the opportunity to work with individuals on a case-by-case basis. Though formulation is commonly used in clinical practice it is more difficult to bring into the research space, due to its flexibility, and the bespoke nature of every individual’s case formulation. In this paper we present one way to achieve this, by using a psychological formulation model as a deductive framework to analyse elements of published theories.

The present review

As discussed above, the scientific literature has not a reached consensus about a definitive theory of PIU. At the point of writing no systematic review has been published on this topic yet there is a clear need to identify all of the attempts to explain PIU and synthesise these diverse approaches to make the academic efforts actionable for practicing psychologists. To this end ‘Formulation-Based Thematic Analysis’ (see Supplemental Material) will be used to synthesise the information proposed by theories, categorising the findings in a useable structure.

Method

Study Design

Given the breadth of the above aim a scoping review methodology was utilised following Colquhoun et al. (2014) and guidelines published by Arksey and O’Malley (2005) and furthered by Levac et al. (2010). As such the current review progressed through the following steps:

• • Identifying the research question

• • Identifying relevant studies

• • Study selection

• • Charting the data

• • Collating, summarising, and reporting the results

Search Strategy

Three key components within the search strategy were the Internet as the medium/context, Addiction or Problematic use, and Theory or model. Each of these three domains was reflected in the scoping search with the following terms and variations: ((Internet OR Online OR Gaming OR Social Media OR Facebook) AND (Addiction OR Pathology OR Problematic OR PIU OR Excessive OR Compulsive) AND (Theor* OR Model)). The following databases: Proquest, Scopus, Google Scholar, and Technology Research Database. In the instance where full-text papers were not available through institutional subscriptions authors were e-mailed using the corresponding author address on the paper and the texts were requested through the ResearchGate Web site. Where no response was received, as occurred for two papers, the studies were excluded from the review.

Inclusion/Exclusion criteria

As this is a scoping review, strict methodological criteria were not applied to papers. Papers were included if they presented a model that attempted to explain problematic Internet use, Internet Addiction, or other conceptual variation, in a way that allows testable hypotheses about the relationships between theoretical factors to be generated. Papers were excluded if they: did not present an original model or theory or significant revision of a previous model, presented a model or theory that does not allow for testable hypotheses to be generated, or presented a model that takes no account of psychosocial factors, i.e. simply present a traditional substance use model without taking account of the specific psycho-social context of the Internet.

Screening procedures and validity

One reviewer ran the searches in each database, downloading the titles and abstract files. The analysis of duplicates and the formal Title and Abstract screening stages were conducted using the EPPI-Reviewer 4.0 web application (Thomas et al., 2011). Papers that were relevant to the aims of the review, or that could not be ruled out based on their Titles or Abstracts, were reviewed at the Full-Text level. The first author screened full-text records for inclusion in the study and the second author independently applied the coding frame to 50% of the papers with an agreement rate of 85.71%. In the 14.29% of papers where the first and second authors’ inclusion screening did not match the papers were brought to a discussion forum where the inclusion criteria were jointly applied. An example of the reasons for the difference in coding was where multiple papers were published based on the same theory, though applied to different populations or research topics. Another example was the extent to which the paper met the criteria of a theory that allowed for testable hypotheses to be created. Following the discussion and agreement between authors the remaining studies proceeded to data extraction. The reference sections of included studies were checked to search for any references that may be relevant, but not accounted for by the search strategy.

Data extraction

As per the Arksey and O’Malley (2005) guidelines an iterative approach to data extraction was taken, whereby the data extraction form was revised as needed to reflect the information reported in the papers. In general, the data extracted reflected: a description of the conceptual construct, a description of the relevant problematic Internet behaviours (generalised Internet use, social media, online gaming, etc.), a description of the proposed bio-psycho-social factors (predisposing, precipitating, maintaining, protective), and a description of how the proposed factors relate to one another to explain the PIU. One reviewer completed this process; however, two reviewers piloted the procedure on a small number of studies and difficulties or amendments to the procedure were discussed and resolved.

Data Analysis/Synthesis and Reporting

Data analysis will take the form of Formulation-Based Thematic Analysis (FBTA, see Supplemental Material). This approach uses the primary formulation categories described by (Carr, 2015) involving the organisation and synthesis of Predisposing, Precipitating, Maintaining, and Protective Factors pertinent to the presenting problem. The proposed bio-psycho-social factors from theories were sorted into one or more of these categories based on how these factors and their interaction were described by authors. This approach to data synthesis allowed for a meta-view of the theoretical landscape to be taken, through the lens of psychological formulation. It is hoped that this will meet the aim of the study and widen the potential space for hypothesis generation for future studies. Reporting of the review followed the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement (Moher, Liberati, Tetzlaff, & Altman, 2009).

Results

Overall search findings

Figure 1 presents the flow chart to depict an overview of the literature search. Searches returned 1412 records, after the removal of duplicates. Two additional studies were added to the study at the full-text review stage based on titles identified in reference sections. Given the broad nature of the search strategy many of the titles were clearly not relevant to the research question and a large proportion were removed at title and abstract screening level. Of the 35 articles that progressed to full-text review stage, two were not accessible to the authors, while 15 were either not theoretical papers or were not theories of problematic Internet use. Eighteen articles were included in the data analysis stage. Details of these papers are included in Table 1.

Figure 1.

Prisma flow chart. Overview & characteristics of models.

Table 1.

Table of included studies, theoretical details, and formulation elements reflected in theories.

Paper details		Model details				Formulation elements
Authors & study code number	Year	Model/Theory name	Theoretical background	Process of development	Internet use difficulty	Predisposing	Precipitating	Maintaining	Protective
Brand, young, & christian (1)	(2014)	Unnamed	Cognitive-behavioural + neuropsych	Systematic rev. + synthesis of literature	Generalised + specific	• Executive Function• Prefrontal control	• Daily hassles• Internet-based coping	• Reinforcement• Ignoring other coping mechanisms• Biased cognitions• Further reduced prefrontal processing	None described
Brand, young, laier, wolfling, & potenza (2)	(2016)	Interaction of person-affect-cognition-execution (I-PACE)	Cognitive-behavioural + neuropsych	Theoretical revision: Adding concepts of substance abuse	Specific internet use disorders	• Biopsychological constitution• Psychopathology• Personality traits• Loneliness/perceives social support• Specific motivations for internet use	• Internet-based coping• Conditioned cognitive and emotional responses to internet stimuli	• Internet-based coping + reduction in non-internet-based coping methods• Internet-related cognitive biases• Conditioned cognitive and emotional responses to internet stimuli• Conditioning-based reduction in EF	None described
Caplan (3)	(2003)	Unnamed	Cognitive-behavioural	Series of empirical studies	Generalised	• Psychosocial difficulties: Depression, loneliness, low sense of social competence	• Development of preference for online social interaction	• Perceived greater online social competence• Deterioration in offline relationships/Responsibilities leads to increased usage	None described
Caplan (4)	(2010)	Unnamed	Cognitive-behavioural	Modelling study	Generalised	• Social anxiety	• Preference for online social interaction• Using internet for mood regulation	• Preference for online social interaction• Using internet for mood regulation	None described
Cimino and Cerniglia (2018)	(2018)	Unnamed	Early emotional regulation + psychodynamic	12-year longitudinal study	Generalised	• Early childhood emotion regulation strategies: Self- or other-focused• Internalising or externalising mental health difficulties in teens• self-doubt• sensation-seeking tendencies	• Attempt to emotionally interact with others online• Attempt to strengthen identity online	• Attempt to emotionally interact with others online • Attempt to strengthen identity online	None described
Davis (6)	(2001)	Cognitive-behavioural model of pathological internet use	Cognitive-behavioural	Theoretical description based on extant literature	Generalised + specific	• Psychopathology: Depression, social anxiety, substance dependence	• Discovery of internet platforms• Reinforcement by positive response from engagement	• Reinforcement by positive response from engagement• Seeking out new platforms• Maladaptive cognitions re self and world.• Rumination/preoccupation	None described
Dong & potenza (7)	(2014)	Unnamed	Cognitive-behavioural + drug addiction models		Internet gaming disorder	• High reward sensitivity• Low loss sensitivity	None described	• High reward sensitivity• Cognitive bias• Altered set-shifting & behavioural inhibition for game specific stimuli• Diminished consideration of outcomes from usage	None described
Gao, zhaopeng, & jingyuan (8)	(2017)	Unnamed	Social belongingness theory	Conceptual model and empirical modelling study	Social media use	None described	• Sense of belonging online – social contact• Enjoyment• Escapism	• Escapism• Pleasure/enjoyment• Arousal	None described
Griffiths (2005)	(2005)	Components model	Substance addictions	Theoretical based on extant literature	Behavioural addictions, incl. Internet use	None described	• Mood/state modification• Habituation to stimulus	• Mood/state modification• Cognitive distortion• Cognitive preoccupation• Habituation to stimulus leading to need for more volume or intensity• Conflict: Internal & social	None described
Kaliszewska-Czeremska (2011)	(2011)	Unnamed	Cognitive-behavioural	Empirical modelling study	Generalised	• Temperament trait: Low persistence, high shyness, low sentimentality• Deficient self-regulation	None described	• Ineffective control over behaviour change• Deficient self-regulation	None described
Kardefelt-Winther (2014)	(2014)	Compensatory internet use	Behaviour stems from motivation to meet needs	Theoretical description based on extant literature	Generalised	• Psychosocial vulnerability: Social anxiety	• Affordances of internet to compensate for needs• Motivations for going online	• Affordances of internet to compensate for needs• Motivations for going online	None described
Kim & davis (12)	(2009)	Unnamed	Cognitive-behavioural	Series of modelling studies	Generalised	• Low self-esteem• Trait anxiety, worry• Sensation seeking tendencies	None described	• Value placed on online activities• Tendency to access flow state	None described
Kuss, shorter, van rooji, mheen, & griffiths (13)	(2014)	Unnamed	Components model + personality types in addiction	Series of modelling studies	Generalised	• Low agreeableness• High neuroticism• Low conscientiousness (adolescents only)• Low resourcefulness (adolescents only)	• Mood/state modification• Habituation to stimulus	• Mood/state modification• Cognitive distortion• Cognitive preoccupation• Habituation to stimulus leading to need for more volume or intensity	None described
LaRose, lin, & eastin (14)	(2003)	Unnamed	Self-regulation (bandura)	Modelling study	Generalised	• Depressive cognitive bias• Depression	• Depressive cognitive bias• Depression• Prior conscious decisions to use Internet• Repeated use to alleviate dysphoric moods• Lapses in effective self-regulation	• Habit formation• Automatic processing• Conditioning – usage paired with positive emotions	• Internet self-efficacy
Lee et al. (2017)	(2017)	Unnamed	Self-regulation (bandura, LaRose)	Modelling study	Social media use	• Poor relationship with father	• Seeking identity formation• Mood state alteration	• Reliance on social media for mood state alteration	None described
Snodgrass, bagwell, patry, dengah, smarr-foster, vanoostenburg, & lacy (16)	(2018)	Unnamed	Compensatory internet use	Modelling study	Online gaming	• Loneliness	• Failing to access online social support	• Intense gaming without social support	• Highly involved in online gaming• Online social support• Offline social support
Wang et al. (2015)	(2015)	Unnamed	Theory of rational addiction + cognitive-affective-behaviour paradigm	Modelling study	Social media - microblogs	None described	• Desire to increase utility of internet use• Habitual use	• Cognitive distortions about usage• Negative anticipation of non-usage impact on affect	None described
Wei et al. (2017)	(2017)	Tripartite neurocognitive model	Neurocognitive models of addiction	Theoretical conceptual model based on extant neuropsych literature	Internet gaming disorder	• Lowered efficiency in impulse control and self-reflection• Poor interoceptive awareness	• Increased automatic motivational responses	• Increased automatic motivational responses• Lowered efficiency in impulse control and self-reflection• Poor interoceptive awareness	None described

Overview & Characteristics of Models

Nine of the models conceptualised the presenting problem as generalised problematic Internet use, seven conceptualised specific Internet use disorders, while two took account of both generalised and specific PIU in their models. In terms of general theoretical underpinnings eight models cited cognitive-behavioural theory as the foundation for their models. These models favour the role thoughts, feelings, and behaviours, as per the cognitive model presented by Aaron Beck (Beck, 1997; Beck & Haigh, 2014). Two are based on substance addiction models. Two are based on the Social Cognitive Theory of Self-Regulation, which encompasses the monitoring of one’s own behaviour, judgements about this behaviour, and affective responses (Bandura, 1991). Three models are based on the concept that Internet use is a means to satisfy needs that are otherwise unmet: one based on The Theory of Rational Addiction and two based The Compensatory Internet Use Model. Finally, one model is based on neuropsychological principles, tracking the physical structures in neuroanatomy that are linked to addiction. Figure 2 presents the results of the Formulation-Based Thematic Analysis (FBTA) in a graphical model. Results are furtehr described below.

Figure 2.

Formulation-Based Thematic Analysis (FBTA) results model.

Predisposing factors for problematic Internet use

Analysis of the codes that were grouped under the over-arching deductive heading of Predisposing Elements led to the creation of five themes that theoretical models have indicated could predispose an individual towards problematic use of the Internet. These are titled: Mental Health Difficulties, Personality Traits, Temperament Traits, Neuropsychological Profile, and Attachment/Early Experiences. Mental Health Difficulties or psychosocial issues was described as a predisposing variable in nine of the eighteen theoretical models (9/18). Depression or mood difficulties, along with social anxiety were conceptualised as the primary mental health ‘risk-factors’ for over-use. Personality Traits were also described as predisposing factors (2/18). Lower levels of agreeableness and higher levels of neuroticism were described for individuals of any age, while lower levels conscientiousness and resourcefulness were indicated for adolescent populations only. Linked to personality, Temperament Traits (1/18) were described as low persistence and sentimentality, and high shyness. An individual’s Neuropsychological Profile was described as an important factor (4/18). Specifically, difficulties with executive functioning, self-regulation, the capacity to be reflective about one’s own behaviours, as well as high sensitivity to reward were described as risk factors. Finally, Early Life Experiences and Attachment, as well as parental relationships, was considered a predisposing factor (2/18). The impact of attachment on emotional regulation strategies leading to an infant becoming highly self-focused or highly-other focused, as opposed to having a healthy secure attachment was described as a risk factor for overuse of the Internet. Further, the role of the father-child relationship was conceptualised as an important factor, with a negative relationship being positively associated with overuse.

Precipitating factors for problematic Internet use

Precipitating factors, those variables that initially facilitate the usage of an online activity that eventually develops into a pattern of usage that causes problems were described in fourteen of the eighteen models (14/18). It is important to note that many of the factors described here also feature in the next section as the processes that initially lead to problematic usage were also presented in the models as processes that maintain problematic usage. Analysis of the codes that were under the over-arching heading of Precipitating Factors yielded six distinct themes: Internet/Program Features, Motivations for Using the Internet, Use of the Internet for Mood Regulation, Conditioning and Habit Formation, Life Stress, and Lapses in Self-Regulation. The Internet/Program Features theme (5/18), reflects the descriptions in theories of the things that the Internet affords users, namely an enjoyable experience, the opportunity to escape into a different media, as well as the opportunity to learn and discover new programs and experience the enjoyment and excitement of that. Motivation for Using the Internet however, describes the specific motivations that individuals have for going online that can lead to problematic use. This theme represents motivations such as the desire to connect with people, an attempt to build a greater sense of identity, and, either a perceived sense of belonging to an online community, or, a failed attempt to access social support online. Using the Internet for Mood Regulation (6/18) reflected the tendency of some users to engage with the Internet to modify their mood or emotional state. While this theme could be considered a part of the Motivation for Using the Internet theme it was described differently in the theories. Here, mood regulation was described as a by-product of use, and played a role in conditioning and habit formation that eventually leads to problematic use. Conditioning and Habit Formation (5/18) was also described in terms of precipitating problematic use. This theme refers to classical and operant conditioning processes that occur from the pairing of behaviours and positive responses. However, an important element related to ‘addictive’ processes is that habituation occurs, leading to the need for greater amounts of the conditioned stimulus to achieve the same level of conditioned response. Life stresses (1/18) reflect the daily hassles that could prompt an individual to use the Internet to access distraction or relief. A final precipitating factor was Lapses in Self-Regulation (1/18). This factor describes the initial automatization of use, whereby room is created for decreased vigilance over the length of time or the intensity of engagement with the Internet.

Maintaining factors for problematic Internet use

This section of the results reports on the analysis of codes that describe the factors that perpetuate unregulated Internet use, once an individual has already developed a problematic pattern of usage. Of note this over-arching deductive formulation category was the only one to contain an element of every theoretical model in the review. This greater representation is reflected in the greater number of themes that emerged from the analysis of this category: Features of the Internet, Cognitive Processes, Coping & Mood Regulation, Socialising, Conditioning and Habit Formation, Deficient Self-Regulation, and Conflict.

The Features of the Internet theme is much the same as in the Precipitating Factors section, in that it represents the enjoyable experience offered by the Internet, escapism, and excitement of discovering new uses. However, in terms of maintaining factors it also represents the value that users place on their online activities, whereby online relationships have been established and investments in programs or communities have reached a significant level. A multifaceted theme that emerged was Cognitive Processes. This theme represents the described impact that problematic Internet usage has on prefrontal control, set-shifting abilities for game/internet related content, increased reward sensitivity, and reduced interoceptive awareness and self-reflection. Other important maintaining cognitive processes described were preoccupation and rumination, both of which were conceptualised as triggers or cues to re-engage with the Internet. Finally, cognitive distortions or biased cognitions are suggested to become a maintaining factor to overcome the internal conflict associated with the problematic use. The theme of Coping and Mood Regulation is also much the same as in the Precipitating section, except that as a maintaining process it is described as an increasing focus on Internet use as the only coping mechanism, to the detriment of other previously utilised coping strategies. The Socialising theme reflects the proposed role that establishing a sense of greater social efficacy online on maintaining problematic usage. It is also described as accounting for the on-going preference for online, rather than face-to-face communication. This is then suggested to be reinforced by the Conflict in close relationships caused by the over-use, which makes face-to-face contact less attractive, and, using the Internet is the automatic coping strategy employed to then improve mood following conflict.

The theme of Conditioning and Habit Formation is also described as a maintaining process. Building on their role in precipitating the problematic use both the conditioned relationships between stimuli and the level of habituation to stimuli are presented as now being further embedded. Added to this are secondary reinforcement factors such as time of day, location, notification tones have become paired with conditioned stimuli and responses. Each of these factors are proposed by the authors as increasing the automaticity of the engagement with the Internet, maintaining the problem. Deficient Self-Regulation, in the context of each of the processes described up to now, lessens, and the more it lessens the more the usage can increase unchecked. Finally, Conflict was described as a maintaining factor in that interpersonal conflict could push someone towards more intense usage as a means to cope, but that intense usage could also fuel interpersonal conflict if an individual has a problematic pattern of use.

Protective factors against Internet use becoming problematic

Codes representing variables that could be preventative against developing, or lessening the impact of, problematic Internet use were only identified in two theoretical models (2/18). Within this formulation element three themes emerged: Social Support, Level of Involvement, and Internet Self-efficacy. Social Support (1/18) was conceptualised as a buffer against unregulated use and was described as comprising both online and offline social support. Offline social support was described as an individual who has good relationship with their social system and is not lonely. However, the access to online social support is linked to the second theme – Level of Involvement (1/18) – as without a high enough level of engagement with an online social group (the authors describe online gaming) the experience of being online can be further socially isolating. As described in the Precipitating factors section failure to access this online support can lead to unregulated use. Finally, Internet Self-Efficacy (1/18) was presented by the authors as a factor that would be associated with less dysregulated use.

Specific versus Generalised Problematic Internet Use

Thus far, all themes have represented the analysis of all the included theories, which is consistent with the aim of the study – to synthesis all of the theoretical attempts to explain the phenomenon of PIU. However, given the debate around classification of Internet use disorders, and the move for IGD to be included in the classification systems, a further analysis was carried out on six theories, the three that specifically focused on problematic internet gaming, and the three that focused on problematic social media use. While models such as that of Brand et al. (2016) are designed for specific Internet use disorders the model can still be applied to any specific Internet use disorder, so is not included in this section. All themes have been described in the preceding sections so will be reported on briefly below.

Looking first to problematic Social Media use, the analysis of three theories produced themes under the formulation categories of Predisposing, Precipitating, and Maintaining. Under Predisposing the only theme was Attachment/Parenting, with a particular focus on the paternal relationship. Under Precipitating the following themes emerged: Internet/Program Features – the enjoyment of using the Internet and opportunities if affords, Motivation for Using the Internet – the desire to connect with people and discover one’s own identity, Using the Internet for Mood Regulation, and Habit Formation. Under Maintaining factors Internet/Program Features, Using the Internet for Mood Regulation, and Habit Formation all emerged. However, the additional theme of Socialising, with the authors suggesting that a greater investment in online social presences leads to greater enjoyment of the social network, also emerged.

Looking next to problematic Internet Gaming, the analysis of these three models produced themes under each of the four formulation categories. Under Predisposing Cognitive Processes – high reward sensitivity, poor executive functioning in the form of impulse control, self-reflection, and interoceptive awareness, and Mental Health Difficulties – in particular loneliness, emerged as themes. Under Precipitating Motivations for using the Internet, specifically gaming as a result of failure to access offline social support, and Lapses in Self-Regulation – leading to automatic usage, were described in theories. Under Maintaining the themes of Cognitive Processes and Socialising emerged. Cognitive Processes represents factors such as set-shifting ability, biased cognitions regarding usage, high reward sensitivity, and poor executive functioning, while Socialising represents the assertion that intense gaming in the absence of social support is a maintaining factor. Finally, under the Protective category Social Support and Level of Involvement emerged as linked themes. Here the theory proposes that both offline and online social support is protective, but in order to access online support a gamer must be highly involved and invested to access social support online.

Discussion

Overview of results

This paper provides the first over-view of the theoretical attempts to explain the phenomenon of problematic Internet use (PIU). While this area is fraught with disagreements about the conceptualisation of the issues and the need to regard it as a diagnosable condition the approach taken to synthesising the results of this review is an attempt to transcend this issue by using a novel analytical framework, Formulation-Based Thematic Analysis, to categorise and synthesise the elements described in theories. In terms of contributions to the literature this review makes two distinct original contributions. Firstly, we offer a comprehensive systematic review identifying the diverse attempts to offer theoretical explanations of PIU. Secondly, we offer practicing psychologists a framework for a) understanding the issue in academic terms, and b) for building a formulation for client-work, which is individual to the personal and social circumstances of the client as well as their unique pattern of engagement with online media.

Our systematic review methodology identified eighteen theoretical models that met inclusion criteria for the study. Of these, eight models cited cognitive-behavioural theory as the foundation for their models. Two are based on the Social Cognitive Theory of Self-Regulation. Three models are based on the concept that Internet use is a means to satisfy needs that are otherwise unmet: one based on The Theory of Rational Addiction and two based The Compensatory Internet Use Model. Two are based on substance addiction models. Finally, one model is based on neuropsychological principles, tracking the physical structures in neuroanatomy that are linked to addiction. Taken together these models offer a view of PIU from a wide variety of perspectives. Another layer to their view is whether the theory was attempting to account for generalised or specific PIU. Our analysis of three models specifically looking at gaming and three at social media revealed some important differences between factors included in the theories. Most notable was the focus on cognitive processes, such as executive-functioning (particularly set-shifting), biased cognitions, and an innate high reward-sensitivity, in the gaming theories while these factors were absent from the social media theories. Importantly the theme of Socialising emerged from the analysis of both set of theories, a finding that highlights the social element of online gaming, and, in one theory a high level of involvement was considered a protective factor (Snodgrass et al., 2018).

Lack of protective factors

An interesting finding of our analysis was the dramatic under-representation of other protective factors in the included theoretical models. While all models included factors that were considered maintaining, and the vast majority included elements that were considered predisposing and precipitating only two theories specifically identified protective elements. It could be argued that the lack of, or opposite of, many of the precipitating and maintaining factors could in fact be protective. However, our analytic methodology set out to synthesise only what was explicitly described by theory authors, and not to make inferences beyond this. As such, the lack of identified protective factors – a key element in any therapeutic treatment plan – is an important result of our study. Of note, Dong and Potenza (2014) do suggest specific areas for therapeutic focus based on their model, but they do not describe protective factors per se.

Contextualisation of findings

An ominous absence from any of the theoretical models in the study is the role of parenting. Given that most of the societal-level concern surrounding PIU is linked to adolescents, there is no theory which address this factor. In a previous study on this topic we reported on practicing psychologists’ formulations of PIU. Results observed a high focus on the role of parenting – both in terms of monitoring/managing usage, and in terms of modelling healthy use of technology (Authors, In Preparation). In general, our review is best contextualised within the debate surrounding classification and diagnosis. While this paper attempts to avoid engaging head-on in the debate surrounding Internet Addiction and Internet Gaming Disorder we believe that our approach offers a means for advancing research in the area. Multiple studies have pointed to the issues with construct validity of IGD and PIU (Aarseth et al., 2016; Kardefelt-Winther, 2017; Ryding & Kaye, 2017; Van Rooij & Kardefelt-Winther, 2017). Further, in their study on the clinical relevance of IGD Przybylski et al. (2017) observed that of the 18,932 gamers in their study only between 0.3 and 1.0% reported symptoms consistent with the ‘disorder’. In this context research should consider whether classifying online behaviours as addictions is of any real benefit. Rather, we propose a move away from classifying common behaviours, like Internet use, as disorders and move in favour of helping those few individuals who do develop problematic usage patterns.

While the present review has breadth as a core strength of its approach this also brings an important weakness. Given that the scoping review aimed to capture the diversity of academic thinking on this issue and synthesis the findings in an actionable way for psychologists it lacks a critical review of the empirical literature, both used to justify the arguments in the theoretical models, and the published literature since models were published. We acknowledge that this was beyond the scope of this review but it is an important caveat to place on our findings. Nevertheless, we believe that the present review offers a foundation from which sound inferences could be drawn and tested, ideally taking into account mediating and moderating factors as outlined by Kardefelt-Winther (2017).

Practical applications

As a practical contribution, our results offer a framework to psychologists or psychiatrists conducting an assessment with a client who uses the Internet to a problematic extent. Specifically, the predisposing factors suggest important elements to consider when taking a clinical history, while the precipitating and maintaining factors indicate the processes which should be considered when assessing how a problem behaviour originated. Our findings complement the review of the existing psychometric measurement tools by Laconi et al. (2014). Finally, the processes identified in our review point to therapeutic approaches that could help a client to re-gain control over their Internet use.

Conclusions & Future Research

The present study offers an over-arching view of the topic of PIU, from both a theoretical and applied perspective. Future research should explore clinicians’ formulations of PIU further to assess their over-lap and complimentary relationship with scientific theory, particularly at this early stage of research on the topic. An important factor that future studies should explore is the protective factors, that were under-represented in the models included in our review. Finally, while producing the scientific basis to build a sound conceptual theory of these difficulties takes time, practicing clinicians are strongly influenced by the major classification systems. Our findings offer an alternative approach to these systems, while acknowledging the developing theoretical landscape.

ORCID iD

Brendan Rooney https://orcid.org/0000-0001-9842-1492