European education policies have increasingly committed to the development of ‘online learning resources’, an omnibus term for study materials, assessments, and/or entire courses gathered on online platforms and websites (e.g. Learning Corner, Online Linguistic Support) (European Commission, 2021). These online learning resources, and digital technologies that underpin them, are thereby positioned as ‘tools’ that grant new (virtual or blended) learning environments beyond institutional borders (European Commission, 2020). They are further considered as means to introduce new forms of personalized, flexible, and student-centered learning, especially by drawing on data-driven practices (e.g. learning analytics, dashboards, real-time feedback) (ibid). It can thus be said that European education policy stresses the ‘usefulness’ of online learning resources for learners who want to design their own, personal learning settings and practices. As the NW28 Special Call elicits studies on the prioritization of ‘learning’ in transnational policies, it invites us to query whether and how online learning resources establish the aforementioned or other forms of learning and, accordingly, to consider their possible problematics.

In doing so, this study draws on a line of literature that emphasizes how digital technologies, besides allowing ‘human learning’, are drawing on (data of) these latter learning practices to learn themselves (see ‘machine learning’) (Knox, Williamson, & Bayne, 2020). This is illustrated by the operations of dashboards and/or automated assessments, which continuously draw on and respond to people’s clicking data, while practices that produce these data are immediately and constantly steered by these same data too (ibid). This stresses the blurry border and continuous cycles of in-formation between people traditionally considered ‘learners’ and digital technologies. Moreover, these cyclical patterns generate ‘feedback loops’ that enclose particular ideas about what learning is, thus limit possibilities forthinking about learning anew and, ultimately, limit possibilities for learning to think (ibid; Thompson & Sellar, 2018). As this highlights problematic implications of online-mediated, data-driven learning practices, this study intends to scrutinize whether and how such feedback loops persist in European online learning resources and seeks possibilities for (non-)human learners to intervene in these loops.

 While building on this topical and conceptual literature, this study further draws on social topology as a theory that addresses how such human-technological entanglements install particular structures or, rather, forms of practices (cf. Marres, 2012).  For example, social topology recognizes how data practices weave together various actors that engage with interfaces (i.a. users, teachers, hardware, local environments), work behind interfaces (i.a. web design teams, data analysis protocols), and operate beyond interfaces (i.a. data policies), to result into movements on user interfaces (Decuypere, 2021). It especially considers how these data practices constitute continuous interactions that allow no ‘rupture’ or intervention, as this would annihilate their inherent structure (see Thompson & Sellar, 2018). As social topology also seeks conditions on which these continuous interactions rely, it is a theory par excellence to map operations of feedback loops and to query if, where, when, and how interventions are possible. This line of reasoning, together with the aims of this study, informs the following research questions: 1) (how) do data practices with, on, behind, and beyond interfaces of European online learning resources establish feedback loops? 2) what forms of learning do these feedback loops establish? and 3) what are the possibilities for interventions in these feedback loops while engaging with interfaces? 

In its empirical address, the study centralizes the issue of developing new methodological approaches to study emerging learning environments (see NW 28 Special Call), while similarly drawing on the social topological plea for methods that attune to the specificity of research settings (Decuypere, 2021). That is, it deploys a ‘methodology-in-progress’ among a purposeful selection of online learning resources funded by the European Commission, including assessments and courses (i.a. Online Linguistic Support). It encompasses four ‘methodological entry points’ to query these resources, i.e. data practices beyond, behind, on, and with interfaces (ibid). Particularly, to examine data practices beyond user interfaces, it has adopted document analyses of data retainment and processing policies (i.a. Privacy Statement, Cookies, Terms and Conditions) of particular online learning resources. Second, data practices behind interfaces are queried through interviews with data analysts and project managers. Interview questions included i.a. what data are gathered, through what tools, based on which rationales, and whether and where these data were visualized ‘on’ user interfaces (e.g. dashboards). Third, data practices on user interfaces are investigated through active navigations, i.e. focused interactions with ‘the screen’. In this case, active navigations were guided by a protocol that was informed by responses of data analysts and project managers (cf. supra) and focused on instances of data collection and/or visualization. Fourth, data practices with interfaces are addressed through collaborative interviews with (human) learners, who are invited to show and discuss how, when/where they engaged with the online learning resources in their local setting. These interviews are intended to illuminate local conditions that are not captured by data but that are important for engaging with the user interface (e.g. taking analog notes, work-study schedules). Insights from these four ‘entry points’ are gathered by engaging with different online learning resources and are integrated into an analysis that aims to reconstruct feedback loops. That is, by drawing on connective strategies, the analysis addresses how data could move from 1) policies, practices, and protocols beyond and behind interfaces to 2) visualizations on interfaces, to 3) interactions with interfaces that, recurrently, ‘feed back’ into the original point of departure, i.e. (1) practices beyond interfaces. The analysis results in a tentative (methodological) claim about how feedback loops operate in online learning resources, how they can be followed, dealt with, and possibly interfered in.

The study has worked towards a threefold presentation of the findings. First, findings are rendered into visualizations of radial networks which, like ‘regular’ networks, present actors (e.g. data points, learners, data analysts) as nodes, and their interactions as lines between the nodes. Radial networks specifically give account how data practices constitute enclosed sets of interactions, i.e. as feedback loops, as well as how they might introduce gaps, i.e. ruptures (see Radial Convergences | Data Viz Project). These visualizations thus give a comprehensive overview of what actors are included, how they are related to each other, and where or when possibilities for rupture or intervention arise. Second, the results include characterizations of feedback loops that, while building on the visualizations, invoke concepts to tap into the forms of learning practices that are brought into existence. Furthermore, these characterizations detail practices that could intervene or introduce ruptures in the cyclical operations of feedback loops. Based on this, proposals for interventions are presented that explicitly map out where, when and how learners can introduce actors from their local learning setting to inspire new forms of learning that involve, new forms of thinking (see Thompson & Sellar, 2018). Recognizing that the study draws on an initial reconstruction of exemplary cases, these analyses serve as springboards to develop further methodologies that could showcase possible problematic consequences of feedback loops in European online learning resources. In positive terms, it points out possibilities for rupture, that is, for interventions on the side of (human) learners that allow more radically innovative forms of learning (to think) throughout European education spaces-times.

Decuypere, M. (2021). The Topologies of Data Practices: A Methodological Introduction. Journal of New Approaches in Educational Research, 10(1), 67-84. https://doi.org/10.7821/naer.2021.1.650 European Commission. (2020). Digital Education Action Plan [2021-2027]. Resetting education and training for the digital age. Retrieved from https://ec.europa.eu/education/education-in-the-eu/digital-education-action-plan_en European Commission. (2021). Coronavirus: Online Learning Resources. Retrieved from https://ec.europa.eu/education/resources-and-tools/coronavirus-online-learning-resources_en Knox, J., Williamson, B., & Bayne, S. (2020). Machine behaviourism: future visions of ‘learnification’ and ‘datafication’ across humans and digital technologies. Learning, Media and Technology, 45(1), 31–45. https://doi.org/10.1080/17439884.2019.1623251 Marres, N. (2012). On Some Uses and Abuses of Topology in the Social Analysis of Technology (Or the Problem with Smart Meters). Theory, Culture & Society, 29(5), 288–310. https://doi.org/10.1177/0263276412454460 Thompson, G., & Sellar, S. (2018). Datafication, testing events and the outside of thought. Learning, Media and Technology, 43(2), 139–151. https://doi.org/10.1080/17439884.2018.1444637