Developments and Trends in Flow Research Over 40 Years: A Bibliometric Analysis

Flow is a fundamental research topic in positive psychology (Nakamura & Csikszentmihalyi, 2002; Seligman & Csikszentmihalyi, 2000). The study of flow was pioneered by Mihaly Csikszentmihalyi (1975), who initially intended to investigate people’s creativity but eventually discovered flow, a positive experience or state that occurs when one is fully absorbed in the task at hand (Beard, 2015; Csikszentmihalyi, 1990). To elucidate flow more clearly, Csikszentmihalyi introduced the renowned nine-dimensional framework, encompassing three antecedents and six characteristics of flow (Nakamura & Csikszentmihalyi, 2002). The antecedents include (1) challenge–skill balance, (2) clear goals, and (3) immediate feedback. The characteristics comprise (4) intense concentration, (5) action–awareness merging, (6) loss of self-consciousness, (7) sense of control, (8) transformation of time perception, and (9) intrinsic rewarding experiences.

As an optimal experience in daily life, flow benefits people in many ways. For instance, flow experience is inherently self-rewarding and tends to enhance individuals’ positive affects and motivation immediately (H. Chen, 2006; Keller & Bless, 2008; Rogatko, 2009). Csikszentmihalyi (1997) believed that the secret to a high-quality life is to engage oneself more in the flow state. Such an argument is supported by copious evidence found for a robust relationship between flow proneness and different aspects of well-being such as autonomy, self-esteem, flourishing, and satisfaction with life (Asakawa, 2010; Koehler et al., 2021; Tse et al., 2021). The positivity and ubiquity of flow make it widely studied in various contexts, such as sports (Jackson et al., 1998; Jackson & Csikszentmihalyi, 1999; Swann et al., 2018), video games (Hsu & Lu, 2004; Sweetser & Wyeth, 2005), education (Hamari et al., 2016; Shernoff et al., 2014), and human-computer interaction (HCI; Ghani & Deshpande, 1994; Webster et al., 1993).

After decades of research, a large body of studies has accumulated in the flow field. Such an ocean of literature is undoubtedly daunting for any researcher, either newcomer or veteran, to construct knowledge architecture of the field. Therefore, literature reviews become essential. To date, several systematic reviews have been conducted to summarize the application of the flow concept across various domains (Perttula et al., 2017; Swann et al., 2018; Tan & Sin, 2021). In addition, there have been a few meta-analytic reviews quantifying the relationship between flow and challenge–skill balance (Fong et al., 2015), time perception (Hancock et al., 2019), and performance (Harris et al., 2021). While these reviews provide valuable insights, their scopes are still relatively narrow, and they cannot answer broad questions in the field, such as ‘What are the main themes of flow research?’ and ‘What are the general trends in the field?’

When facing a research area as expansive as flow, bibliometric analysis is a helpful tool to capture the big picture of the field. Through quantitative and qualitative techniques, bibliometric analysis can extract developmental trends and hotspots in the field, thereby providing a macroscopic overview for researchers (e.g., Dominko & Verbič, 2019; Yan et al., 2020). Compared to systematic review and meta-analysis, bibliometric analysis is especially useful when the scope of the field is broad and the set of studies is enormous (for an overview, see Donthu et al., 2021). The data-driven nature of bibliometric analysis promotes unbiased document selection, and results can be efficiently visualized, allowing readers to grasp the state of the field quickly (Nakagawa et al., 2019). In positive psychology, bibliometric techniques have been successfully used to analyze the entire field (Rusk & Waters, 2013) and specific fields such as subjective well-being (Dominko & Verbič, 2019). Nevertheless, a depiction of the complete picture of the field of flow research is still missing.

The current study aimed to use large-scale bibliometric analysis to provide a holistic overview of flow research. Specifically, according to traditional bibliometrics (Donthu et al., 2021; Ruiz-Parrado et al., 2021), we sought to answer the following five broad research questions (RQs) regarding the field: (1) What is flow research’s quantitative status and growth trends? Answering this question helps describe the development of the field. (2) What is the publication pattern of flow research? Unveiling the distribution of literature by journal or country can offer benchmarks for research decision-making processes, such as journal selection in reading and submission. (3) What are the cornerstone publications and pivotal theories in this field? This question will be answered through analyses of citations, the result of which can facilitate researchers’ identification of key literature in the area. (4) What are the major research themes in the large body of flow-related studies? Identifying the knowledge structure of this field makes it easier to understand the relationship among topics and find research gaps. (5) What are the promising directions of flow research? Researchers have suggested that there still are some theoretical issues to be clarified in the field of flow, such as its components, antecedents, measurement methods, and even association with positive consequences (Engeser et al., 2021; Fong et al., 2015; Moneta, 2012; Norsworthy et al., 2021). A quantitative analysis of the published literature on flow over the past decades will shed more light on these issues and provide guidance for future research.

At the bibliometric analysis data acquisition stage, two problems must be solved: which database to use and how to retrieve desired literature. Two databases, Web of Science (WoS) and Scopus, offer the capability to export detailed data suitable for in-depth bibliometric analysis (Donthu et al., 2021). Notably, the journal coverage in WoS is predominantly a subset of that in Scopus (Singh et al., 2021). This distinction is particularly relevant given the interdisciplinary nature of flow studies, which are frequently published in journals with diverse impact factors. Consequently, for our analysis, Scopus was selected to search for flow-related studies, ensuring a comprehensive inclusion of literature from various disciplines.

To retrieve the desired documents, a search strategy should be appropriately defined, which is challenging for terms with ambiguous meanings such as ‘flow.’ Therefore, we adopted a cautious two-step retrieval protocol. First, to thoroughly summarize the search strategies previously employed in this field, we used the following search string to collect flow-related systematic review or meta-analysis:

TITLE-ABS-KEY(“flow experience” OR “flow state” OR “flow theory”) AND TITLE(“review” OR “meta-analysis”).

Among the 145 documents identified, 20 are review articles focused on flow, offering the search strategies essential for our research. A summary of the search strings used in these studies is presented in Table A1 of the Online Appendix. These studies mainly employed two strategies for including relevant literature and two for excluding irrelevant literature. For literature inclusion, one strategy was to search ‘flow’ directly, which would undoubtedly yield numerous irrelevant results; another way was to search for flow-related phrases (e.g., ‘flow experience,’ ‘flow state’), which got a subset of the results of the former strategy. For literature exclusion, one strategy was to eliminate documents that contain flow experience-irrelevant phrases (e.g., ‘water flow,’ ‘optic flow’) in the title, abstract, or keywords; another efficient way was to eliminate the documents in which ‘Csikszentmihalyi’ did not appear in the text. The consideration behind using ‘Csikszentmihalyi’ to filter real flow-related literature was that Mihaly Csikszentmihalyi was the father of flow, and his works have been extensively cited in this field.

The second step aims to integrate these retrieved search strategies to build a flow literature database with sensitivity and specificity. This database incorporates as much relevant literature as possible by directly searching using ‘flow’ (and ‘optimal experience’) instead of flow-related phrases (e.g., ‘flow state’). To guarantee specificity, we excluded documents that did not cite any of Csikszentmihalyi’s flow-related literature, in addition to those that contained flow experience-irrelevant flow phrases. Thus, the following search string was used to retrieve literature:

TITLE-ABS-KEY(“flow” OR “optimal experience”) AND ((REF(“Csikszentmihalyi M*”) AND REF(“flow” OR “optimal experience” OR “beyond boredom and anxiety”)) OR AUTHOR-NAME(“Csikszentmihalyi M*”)) AND PUBYEAR < 2022 AND NOT TITLE-ABS-KEY(“optic* flow” OR “*water flow” OR “expirat* flow” OR “exrat* flow” OR “cash flow” OR “cereb* flow” OR “venous flow” OR “ventil* flow”) AND (LIMIT-TO(DOCTYPE , “ar”) OR LIMIT-TO(DOCTYPE , “re”)) AND (LIMIT-TO(LANGUAGE , “English”)).

Specifically, the retrieval included the following steps: (1) Search for documents with ‘flow’ or ‘optimal experience’ in the title, abstract, or keyword. (2) Search for documents citing Csikszentmihalyi’s flow-related works or having him in the author list. (3) Merge the documents from steps one and two, and only keep documents that appear in the results of both steps. (4) Exclude documents that contain flow experience-irrelevant flow phrases in the title, abstract, or keywords. (5) Only keep documents published before 2022. (6) Only keep documents categorized as Article or Review. (7) Only keep documents that are indexed in English.

The search was conducted on September 29, 2022. A total of 2,622 records were found. To ascertain their relevance to flow experience, we selected all documents published in 2020 and conducted a comprehensive review. The results confirmed their significant association with flow (see Online Appendix B).

Bibliometric analyses mainly employ two categories of techniques: performance analysis and science mapping (for an overview, see Donthu et al., 2021). The former includes statistics on publication-related indicators (e.g., number of publications by year, journal, or country) and citation-related metrics (e.g., citation count per document or journal). Science mapping typically employs network analyses to explore the interconnections among research constituents such as authors, authors’ affiliations, keywords, or cited references. The concurrent presence of two research constituents within a singular document (e.g., two terms concurrently listed as keywords) indicates a link between them. The frequency of such co-occurrences directly determines the strength of links, where a higher frequency signifies a stronger connection. With research constituents serving as nodes and the links between them established, various networks reflective of the academic landscape can be formed. Here, we used Microsoft Excel, VOSviewer, Gephi, and CiteSpace to perform these bibliometric analyses.

Microsoft Excel 2019 was used to conduct descriptive analyses (i.e., performance analyses) and determine statistics about the distribution of publication year, subject area, and countries. These analyses were used to answer RQ 1 (What is flow research’s quantitative status and growth trends?) and RQ 2 (What is the publication pattern of flow research?).

We used VOSviewer version 1.6.18 to conduct international collaboration, keyword co-occurrence, and co-citation network analysis. VOSviewer is a freely available software for constructing and visualizing bibliometric maps (van Eck & Waltman, 2010). It is widely used to extract networks in bibliometric analysis fields due to its ease of use and reliable algorithm (van Eck et al., 2010). As mentioned, the networks (also called maps) created in VOSviewer consist of many nodes and links between nodes. The size of a node generally represents the frequency of its occurrence, while the thickness of links represents strength. Each network node is then classified into a cluster by the visualization of similarities (VOS) algorithm (van Eck et al., 2010), represented by different colors. These network analysis results were used to answer RQ 3 (What are the cornerstone publications and pivotal theories in this field?), RQ 4 (What are the major research themes in the large body of flow-related studies?), and RQ 5 (What are the promising directions of flow research?).

For aesthetics, all networks analyzed using VOSviewer were visualized using Gephi (Bastian et al., 2009). Gephi is an open-source software excelling in network analysis and visualization. It features a highly interactive graphical interface and ensures compatibility with the network data output of VOSviewer.

Additionally, CiteSpace version 6.1.R3 (C. Chen, 2006) was used to perform citation burst analysis. CiteSpace is another popular bibliometric analysis tool based on Java and is freely available. Compared to VOSviewer, it exhibits a superior capability in capturing the evolving trends of research foci over time within a given field. Specifically, the citation burst analysis employs Kleinberg’s (2002) burst-detection algorithm to detect the most active research areas. When a document has a citation burst within a certain period, it indicates that it has attracted extensive attention from researchers in the field (C. Chen, 2014). The results of the citation burst analysis can reveal research hotspots and tendencies, which were used to answer RQ 4 and RQ 5.

Notably, the raw records downloaded from the database must be preprocessed before conducting the above analyses. First, many synonymous keywords in the documents needed to be standardized. Since previous research has suggested that keywords that occur only once can be regarded as ‘noise’ (Hofer et al., 2010; Leydesdorff, 2001), we manually checked keywords that occurred in more than one document and then unified the synonyms (for a similar approach, see Ruiz-Parrado et al., 2021). For example, the keywords ‘flow,’ ‘flow experience,’ and ‘flow state’ were unified and renamed as ‘flow;’ the keywords ‘games’ and ‘game’ were unified and renamed as ‘games.’ The detailed materials can be found in our online repository at Open Science Framework (https://osf.io/m9p46/).

Second, references in different formats (e.g., APA 6^th and 7^th editions) are recognized as distinct documents by the software, which has been a longstanding source of challenge for researchers (e.g., A. Kaur et al., 2022; Leung et al., 2017). To address this issue, we developed a Python script to identify the literature title within each reference. References with identical titles were then uniformly reformatted for consistency.

This section is organized into three parts: the first part is about the basic status of this field, aiming to answer RQ 1 (What is flow research’s quantitative status and growth trends?); the second part is about the publication patterns and distribution of flow-related studies, aiming to answer RQ 2 (what is the publication pattern of flow research?); the third part is about the analyses of the references and keywords, aiming to answer RQ 3 (What are the cornerstone publications and pivotal theories in this field?), RQ 4 (What are the major research themes in the large body of flow-related studies?), and RQ 5 (What are the promising directions of flow research?).

The number of peer-reviewed studies reflects how much scholarly attention a field or subject has received. We identified a total of 2,622 flow-related studies that were published in the past four decades, and Figure 1 shows the number of flow-related publications by year. A detailed sub-disciplinary analysis has also been provided in the Online Appendix C, broadly in line with overall trends. The first peer-reviewed journal article on flow was from Csikszentmihalyi (1982), who attempted to introduce flow into higher education as a mechanism for the enjoyability and intrinsic motivation of teaching.

After that, the field of flow research has exhibited a sustained scarcity over an extended period. Despite the lack of academic attention, Mihaly Csikszentmihalyi made efforts to complete the main work of flow theory construction and published foundational books in 1975, 1990, and 1997 (see references; books were not included in the analyses due to not being in the database). In addition, some easy-to-use flow measurement tools were developed, including the multidimensional flow state scale in sports (Jackson & Marsh, 1996) and the unidimensional flow scale in computer-mediated environments (Novak et al., 2000).

The landscape has changed after the new millennium, and an exponential growth trend can be observed since then. The thriving flow field in the new millennium might benefit from the initiation of the positive psychology movement (Nakamura & Csikszentmihalyi, 2002) and its increased media attention. Another potential catalyst might lie in the rapid evolution of information technology, given the leading momentum of computer science in the growth of flow literature (see Online Appendix C). Overall, based on the current quantitative trends, we speculate that the number of flow-related studies will continue to proliferate in the coming years.

The 2,622 flow-related documents were published in 1,139 different journals. However, 61.3% of the journals published only one document, and 18.9% published only two. There were only 38 journals that published 10 or more flow-related studies. Table 1 lists the top 10 most productive journals, which can be divided into several categories: (1) general psychology journals, such as Frontiers in Psychology, which published the most documents (n = 84); (2) journals related to computer science, such as Computers in Human Behavior, Computers and Education; (3) journals focusing on other branches of psychology, such as the Journal of Happiness Studies, Psychology of Sport and Exercise, and Psychology of Music; (4) journal related to business, such as the Journal of Business Research; and (5) multi-disciplinary journals, such as PLoS ONE and Sustainability.

The total number of publications in the top 10 journals was 345 (accounting for 13.2% of the total sample). Notably, these publications received a combined total of 23,008 citations, with an average of 66.7 citations per article. These statistics suggest that flow studies are fragmented into publication outlets but generally have received considerable attention. The Journal of Business Research has the highest average number of citations per article (total citations [TC]/total publications [TP] = 166.1), followed by Computers and Education (TC/TP = 121.1) and Computers in Human Behavior (TC/TP = 105.3).

The 2,622 retrieved studies on flow were contributed by authors affiliated with institutions across 84 countries or regions. Among them, 44 countries or regions (52.4%) have published less than 10 documents, 32 countries or regions have published 10–100 documents, and only eight countries or regions have published more than 100 documents. Figure 2 visualizes the international cooperation network among the top 30 most prolific countries or regions. The size of the nodes represents the number of documents published in that country or region, while the thickness of the links represents the strength of cooperation between the two nodes. The cooperation strength of two countries or regions is determined by the co-author frequency of researchers affiliated with the corresponding countries or regions.

The USA is the most prolific country or region in the field of flow (n = 751), followed by the UK (n = 303), Taiwan (n = 231), Australia (n = 201), and the Chinese mainland (n = 164). This rather distributed pattern reflected a global research interest in flow.

Based on author collaboration data from 2622 documents analyzed with the VOS algorithm (van Eck et al., 2010), VOSviewer grouped these countries and regions into 5 clusters (indicated by color): cluster 1 (green) includes the USA, Canada, Germany, Sweden, Switzerland, and Norway; cluster 2 (orange) includes the UK, Australia, India, and Ireland; cluster 3 (pink) includes several Asian countries or regions such as Chinese mainland, Taiwan, and Japan; the remaining two clusters mainly include European countries. Among these countries and regions, the USA and the UK have the most collaborators (both n = 28), followed by Australia (n = 23). The most frequent cooperation is between the UK and Australia (link strength [i.e., frequency] = 52), followed by the collaboration between the USA and other countries or regions such as the Chinese mainland, Canada, Taiwan, South Korea, and the UK.

Table 2 lists the top 10 most cited documents in the flow-related studies, all published before 2003. The top three are Csikszentmihalyi’s fundamental books on flow theory, and three of his other works are in the top 10. Despite the potential influence of our literature search methodology, the results confirm Csikszentmihalyi’s widely recognized status as the field’s founding father. With him in the top 10 are the works of Hoffman and Novak (1996). They proposed a conceptual model of flow in the computer-mediated environment to explain user interaction behavior. They then further empirically tested this model and proposed a unidimensional method to measure flow (Novak et al., 2000). Besides, there are two other frequently cited documents: in one, Jackson and Marsh (1996) published a multidimensional flow measurement scale; in the other, Koufaris (2002) combined the technology acceptance model with flow theory for the first time. Notably, ranking based on total citation count naturally favors older publications. To counter this bias, a supplementary ranking using average annual citations is available in the online appendix D. However, a more comprehensive approach involves evaluating high-impact publications across various periods. Such a result is detailed in the citation burst analysis presented in Section 3.3.3 of this paper.

To further reveal the relationship between references, we performed a co-citation network analysis on the 2,622 documents (see Figure 3). For clarity, only documents with a frequency of more than 60 were included in the network (for a similar approach, see van Nunen et al., 2018). As illustrated in the figure, the algorithm classified the cited documents into three clusters. Containing the most documents, cluster 1 (cyan) represents a large body of research based on classical flow theory. Representatives of this cluster are Csikszentmihalyi and Jackson. Cluster 2 (orange) mainly includes fundamental studies on applying flow theory to the field of HCI. Representatives of this cluster are Hoffman, Novak, and Webster. Cluster 3 (purple) has only three documents, two of which are pioneers in applying flow theory to explain the game experience (Sherry, 2004; Sweetser & Wyeth, 2005); another successfully demonstrated the effect of flow on learning in serious games (Hamari et al., 2016). The presence of a small cluster may be explained by an overlap between the gaming literature and the other two categories.

Keywords are the condensed form of essential content researchers provide in a document (Rose et al., 2010). Analysis of keyword frequency helps reveal research topics, while analysis of keyword co-occurrence (i.e., two keywords appearing in one document) can further show the architecture of research themes in a field.

A total of 5,994 unique author’s keywords were extracted from the 2,622 documents. After completing synonym merging (as described in Methods), 903 standardized keywords were included in the follow-up analyses. Table 3 lists the top 20 high-frequency keywords in flow research. ‘Flow’ was the most frequent keyword (n = 1,161). Among the synonyms of flow, ‘flow’ had the highest frequency (accounting for 72.3%), followed by ‘flow experience’ (20.9%) and ‘flow state’ (3.2%). Other high-frequency keywords were mainly related to the positive effect of flow (e.g., motivation, well-being, and creativity), computer-mediated environments, and education.

In addition, two theories showed a close relationship with flow: self-determination theory (SDT) and the technology acceptance model (TAM). Rooted in the works of Deci and Ryan (1985), SDT posits that human motivation is guided by three innate psychological needs—autonomy, competence, and relatedness—with intrinsic motivation arising when activities align with these core needs, fostering engagement and satisfaction. The TAM, developed by Davis (1989), suggests that technology adoption is primarily influenced by perceived usefulness and ease of use, offering a streamlined framework for understanding user motivation and behavior.

In light of flow’s multifaceted nature, a more intriguing analysis would involve comparing the frequency with which each of its nine dimensions has been the focal point of research. Detailed methodologies and results are presented in the online Appendix E. Our findings indicate that among Csikszentmihalyi’s nine proposed dimensions of flow (Nakamura & Csikszentmihalyi, 2002), the challenge–skill balance has received the most attention, followed by intrinsic rewards, concentration, and time perception. The dimensions explored less frequently include clear goals, action–awareness merging, loss of self-consciousness, and a sense of control, with immediate feedback being the least examined aspect.

To further reveal the relationship between the keywords and extract themes in this field, we performed a keyword co-occurrence network analysis (see Figure 4). Each node represents a keyword that appeared at least 10 times (for a similar threshold, see van Nunen et al., 2018), with its size indicating the frequency of occurrence. As can be seen from Figure 4, the algorithm classified the keywords into eight clusters, representing the eight main research themes in this field. Three of them are theoretical themes, including mechanism, positivity, and health. The other five themes are the application directions of flow, including technology, gaming, sports, creativity, and education. Technology was the most dominant theme, with a relatively even distribution of nodes in other themes. When considering the link strength of each theme after excluding the keyword ‘flow,’ gaming showed the strongest relationship to other themes (for details, see Online Appendix F).

To reveal the active research hotspots over the past few decades, we used CiteSpace to perform a citation burst analysis. A citation burst indicates whether a document has received a surge of citations over a period (C. Chen, 2014). The top 30 citation bursts in the flow area are illustrated in Figure 5. The earliest three citation bursts observed were the exposition of the flow theory (Csikszentmihalyi, 1990), the development of multidimensional flow scales (Jackson & Marsh, 1996), and the introduction of positive psychology (Seligman & Csikszentmihalyi, 2000).

Notably, four documents showed currently active citation bursts (until the end of 2021), all related to flow in various computer-mediated environments. Specifically, Guo et al. (2016) proposed an integrated flow model to explain the antecedents and outcomes of flow in online learning. Hamari et al. (2016) found that flow (operationalized as enhanced challenge and skill) in game-based learning helps students learn. Kaur et al. (2016) developed a new scale to measure users’ flow experience in social network environments. Finally, Liu et al. (2016) demonstrated that interpersonal interaction in social commerce facilitated users’ flow and subsequently influenced purchase intention.

As a pioneering topic of the positive psychology movement, flow has attracted the attention of many researchers in various fields. The flourishing literature in this field necessitates a comprehensive literature review, so this study employed bibliometric methods to provide novel macro-level insights into published documents. The following few paragraphs answer the five RQs proposed above.

RQ 1: What is flow research’s quantitative status and growth trends? After Csikszentmihalyi (1975) systematically introduced the concept of flow, it received little attention in academia for a long time. This changed with the launch of the positive psychology movement around the new millennium. Seligman and Csikszentmihalyi (2000) stressed that psychology should aim to build a positive quality of life for mentally healthy individuals, and one of the experiences that constitutes a good life is flow (Nakamura & Csikszentmihalyi, 2002). As the positivity of flow has been widely demonstrated (e.g., Asakawa, 2010; Koufaris, 2002) and measurement tools have been well developed (e.g., Jackson & Marsh, 1996; Novak et al., 2000), the body of flow-related literature has grown exponentially in the past two decades. As of 2021, this field had accumulated 2,622 peer-reviewed documents.

RQ 2: What is the publication pattern of flow research? The flow-related studies were published in 1139 journals and written by authors in 84 countries or regions. Those researchers come from various disciplines, including sports, education, arts, and computer science. The body of literature on flow within these disciplines has seen a notable increase in volume, along with considerable impacts. Another striking finding was that the countries that published the flow studies were spread across continents, providing evidence of the global research interest in flow.

The breadth of flow research revealed in this study carries profound implications. Historically, many psychological theories have been embroiled in debates over cultural universality (Henrich et al., 2010; Pepitone & Triandis, 1987). More recently, behavioral science has faced what is termed a replicability crisis and generalizability crisis, where many psychological phenomena have failed to be validated across different contexts (Open Science Collaboration, 2015; Yarkoni, 2022). Our findings, however, indicate that flow has been extensively studied across various domains and cultures with highly heterogeneous paradigms and samples. Given such heterogeneity (Bryan et al., 2021), we might have greater confidence in the robustness of the existing of flow.

RQ 3: What are the cornerstone publications and pivotal theories in this field? Keyword-based analyses revealed that flow theory was frequently related to the SDT and TAM. Flow theory has a deep theoretical connection to the SDT (Deci & Ryan, 1985), as they both emphasize individuals’ intrinsic motivation. Satisfaction of the three fundamental psychological needs (i.e., autonomy, competence, and relatedness) proposed by the SDT has been reported to be associated with flow (Kowal & Fortier, 1999). The relationship between flow theory and TAM is mainly reflected in the field of HCI. Numerous studies have found that perceived usefulness and ease of use can predict users’ flow experience (e.g., Y.-C. Huang et al., 2013).

Some literature forms the cornerstone of the field of flow. According to the reference-based analyses, Csikszentmihalyi’s three books are the most cited theoretical foundations in this field (Csikszentmihalyi, 1975, 1990, 1997), and the scales developed by Jackson and Marsh (1996) and Novak et al. (2000) have been most widely used. It should be noted that the results also revealed that there are two major and one minor literature groups in this area. Most of the works in the first major group were written based on the classical flow theory that Csikszentmihalyi proposed, and they mainly apply flow to traditional domains such as music and sports (Bakker, 2005; Jackson et al., 1998). The second major group essentially contains studies of flow in HCI. In this field, many researchers have reconceptualized flow. For example, Trevino and Webster (1992) defined four dimensions of flow: control, attention focus, curiosity, and intrinsic interest. Ghani and Deshpande (1994) argued that flow has two characteristics: total concentration and enjoyment in an activity, and Hoffman and Novak (1996) proposed their four antecedents of flow in computer-mediated environments: high level of skill and control, high level of challenge and arousal, focused attention, and interactivity and telepresence. While these conceptualizations provide theoretical foundations for the flourishing of flow in the field of HCI, there is no doubt that future research needs to probe the nature of flow further to integrate these diverse interpretations. The third minor group typically contains literature about flow in video games. Representatively, Sweetser and Wyeth (2005) proposed the GameFlow model to explain players’ enjoyment of the games.

RQ 4: What are the major research themes in the large body of flow-related studies? Keyword co-occurrence analysis revealed eight major research themes in the flow field (theoretical themes: mechanism, positivity, and health; application scenarios: technology, gaming, sports, creativity, and education). The most prominent theme in this field was the application of flow in technologies, which is somewhat unexpected but reasonable. As stated in his books (Csikszentmihalyi, 1975, 1990), Csikszentmihalyi first conceived the flow theory through interviews with athletes and artists. While flow continues to hold significance in these traditional professional domains, the focal points and driving forces behind flow research in the post-millennium era have primarily emanated from the convergence of computer science (see Appendices C and F). The co-citation analysis also showcased a distinct divergence between studies based on classical flow theory and adapted theories in the context of HCI. In our view, the prominent momentum of the theme “technology” may be due to the flow theory’s effectiveness in explaining user experiences (Novak et al., 2000; Sherry, 2004) and the practicality of conducting flow experiments in computer environments, especially compared to the difficulties of recruiting participants for sports or workplace settings.

More interestingly, of the eight themes, gaming showed the strongest connection to the others. Video games have been considered ideal tools for creating and studying flow because of their inherent concrete goal, immediate feedback, and flexible challenges (Sherry, 2004; Zhang et al., 2022). Games are also increasingly being combined with other activities, resulting in exergames (H.-C. Huang et al., 2018), educational games (Hamari et al., 2016), and music games (L.-X. Chen & Sun, 2016). Since there is compelling evidence that flow facilitates gameplay satisfaction and loyalty (Chang, 2013; Sepehr & Head, 2018; Su et al., 2016), we can expect more applications of flow theory in games.

RQ 5: What are the promising directions of flow research? The results suggest several possible directions for future research. Firstly, we identified five key application themes of flow in the keyword co-occurrence network analysis. Early research under these themes typically combined unique domain features with flow theory, leading to novel conceptualizations (e.g., education: Heutte et al., 2016; technology: Hoffman & Novak, 1996; sports: Jackson & Marsh, 1996; gaming: Sweetser & Wyeth, 2005). Although these adaptations faced criticism for inconsistent interpretations of flow (Norsworthy et al., 2021), they significantly contributed to the understanding and application of flow in respective fields. However, the field of creativity, where flow theory originated, lacks such adapted conceptualizations (Cseh, 2016). Such absence is critical due to both theoretical and practical reasons. Theoretically, flow necessitates clear feedback and loss of self-awareness (Nakamura & Csikszentmihalyi, 2002), yet the creative process often lacks objective performance feedback and requires artists’ continuous conscious participation (Cseh, 2016). Unlike sports or gaming, which have clear benchmarks, artistic creation is inherently subjective. This subjectivity presents challenges for artists in assessing their success instantly, leading to a divergence from traditional flow conceptualizations. Practically, empirical studies on creative flow have shown obscure results regarding flow’s impact on creative performance (Cseh et al., 2015; Thomson & Jaque, 2023). Hence, more extensive theoretical and empirical research on creative flow is warranted.

Secondly, health and positivity have also been identified as key themes in flow research. As an optimal experience advocated by positive psychology (Seligman & Csikszentmihalyi, 2000), flow’s contribution to the positive development of happiness and well-being is widely acknowledged (Asakawa, 2010; Tse et al., 2021). However, its relationship with health factors is less definitive (Schüler, 2012). Take psychological stress as an example: its impact on flow, and vice versa, remains ambiguous. While most evidence suggests that stress impedes flow (e.g., Kim & Park, 2018), some studies report a positive or an inverted U-shaped association (Mesurado et al., 2016; Peifer et al., 2014). The effectiveness of flow in stress relief also presents mixed findings (Hirao et al., 2012; Zhao & Zhou, 2021). Even if flow experience can alleviate stress, it also carries a potential risk of adverse addiction, particularly within digital environments (Schüler, 2012; Zhao & Zhou, 2021). These necessitate comprehensive empirical research prior to the application of flow interventions.

Thirdly, the psychological and neural mechanisms of flow demand further investigation. Despite the debate surrounding its practical application (Moneta, 2012), Csikszentmihalyi’s nine ‘dimensions’ of flow remain vital for theoretical understanding (Nakamura & Csikszentmihalyi, 2002). Our keyword frequency analysis indicates that, among the three antecedents of flow—clear goals, immediate feedback, and challenge–skill balance—the investigation of goal and feedback is notably less extensive. The exploration into flow characteristics such as a sense of control, loss of self-consciousness, and action–awareness merging also needs intensification. Additionally, despite early recognition of the role of personality factors, especially autotelic personality, in influencing flow experiences (Csikszentmihalyi, 1975), this topic remains underexplored. A more crucial concern lies in the prevalence of associating trait characteristics with flow experience frequency (e.g., Jackson & Eklund, 2002). We believe the development of new research tools heralds the potential for more rigorous investigations in this domain (Tse et al., 2018). Neurologically, the specific mechanisms underlying flow occurrence still remain largely elusive (van der Linden et al., 2021). Current studies on its neural basis are limited and advancing slowly (see Online Appendix C), highlighting the need for increased research efforts.

Finally, and most importantly, our citation burst analysis from the past decade underscores a seminal theoretical work: the chapter by Moneta (2012) in Engeser’s (2012) book, which emphasized the methodological issues of flow measurement. Historically, research has predominantly employed multidimensional scales, particularly those grounded in the nine-dimensional framework (Jackson et al., 2008; Jackson & Marsh, 1996). Such componential approaches inadvertently conflate the antecedents of flow with its characteristics (Moneta, 2012). Moreover, the aggregated multidimensional scale scores can be misleading (Barthelmäs & Keller, 2021). For instance, an individual experiencing heightened attention due to external pressures might score highly on certain sub-dimensions, yet such scores do not conclusively indicate a genuine flow state. Subsequent studies have echoed these concerns, calling for more integrated and standardized measurement approaches (Barthelmäs & Keller, 2021; Jackman et al., 2019; Norsworthy et al., 2021). However, a decade later, our citation burst analysis reveals a lack of significant new integrative tools in flow measurement, with many researchers, especially newcomers, persisting in using traditional scales (Zhang et al., 2022). Swann et al. (2018) have even suggested that the field may be “reaching a crisis point.” Hence, the flow research community needs to undertake more revolutionary work and promote widespread changes.

In short, the field of flow is still in a period of rapid growth, and it can be expected that there will be increasing interdisciplinary research. However, further rigorous theoretical work is essential for this research field’s robust and enduring advancement. The current study provides a macroscopic overview of the field and guidance for future studies.

Some limitations of this bibliometric study should be noted. First, we selected Scopus, considering the balance of quality and quantity of literature. However, retrieving data from other databases (e.g., WoS, Google Scholar) may give different results. Second, considering the ambiguous meanings of ‘flow,’ we summarized the search strategies of previous flow-related review articles and distinguished the desired documents mainly by detecting whether they contained Csikszentmihalyi’s works in the reference list. While Csikszentmihalyi is widely regarded as the founding father of the field, it cannot be guaranteed that all the flow-related studies cited his work. Therefore, it should be noted that his influence may have been amplified in this study. Third, the bibliometric analyses screened high-frequency information based on previous criteria to balance the efficiency and comprehensiveness of information transmission. While this practice generally has minimal impact on findings, we have included the unscreened descriptive results in our Open Science Framework repository (https://osf.io/m9p46/) for potentially interested readers. Finally, since this study used data-driven techniques to extract a holistic, unbiased overview of flow-related literature, we did not peruse each article with flow as a core research focus. The inclusion of low-relevance literature is a common limitation of bibliometric analysis (e.g., Baminiwatta & Solangaarachchi, 2021; van Nunen et al., 2018), but we instead argue that only in this way can we fully portray the achievements of flow theory over the past decades. Future review studies can further distill the literature in our database to investigate theoretical flow issues.

Yongfa Zhang: conceptualization, data curation, formal analysis, visualization, writing – original draft, writing – review & editing.

Fei Wang: conceptualization, visualization, writing – review & editing, supervision.

The authors declare they have no conflict of interest.

Study data and materials are provided on the Open Science Framework and can be retrieved from https://osf.io/m9p46/.