Learner Agency Across Scales: An Integrative Perspective on Self-Regulated Learning in Algorithmically Mediated Environments

Synthesis: Learner agency across scales — a theoretical/integrative paper proposing an account of learner agency that extends SRL beyond task-focused AI Regulation in Education by connecting it with complex systems science and Freirean critical pedagogy. It foregrounds three dimensions of learner agency — regulation, integration, and positioning — and a cross-scale scaffold supporting learners to move between them in algorithmically mediated environments.

Key Findings

• Three dimensions of learner agency. Regulation (the self as an active managing process, drawing on social-cognitive SRL); Integration (the self as an organizing center across time and contexts, drawing on complex systems science); Positioning (the self as a critical interpreter, drawing on Freirean critical Pedagogies and Teaching Strategies). Agency lies in the coordination/movement between these, not any one alone.
• Two limitations of SRL. SRL has progressively narrowed to task-focused regulation (leaving broader life conditions and structural positions under-theorized), and its operationalization in educational technology has narrowed the construct to optimizable behavioral proxies (e.g., time management, strategy use, effort allocation).
• Cross-scale scaffold. A design orientation with four functions — surfacing patterns across scales, opening conditions to critique, returning interpretation to action, and centering the learner as the agent of connection. It extends SRL from "agency within conditions" to "agency over conditions" (questioning what metrics measure and whose purposes they encode).
• Critical stance on Learning Analytics and Generative AI. Algorithmically mediated environments encode assumptions about what counts as engagement and who succeeds; supporting agency requires learners to interpret their data, question what it does not capture, and act on that interpretation — a critique that has been largely disconnected from SRL theory.

What this means for practice

• Researchers. Treat regulation, integration, and positioning as a proposed integrative account to operationalize and test, not as an established model: the paper advances theory across SRL, complex systems science, and Freirean critical pedagogy without collecting data.
• Learning analytics designers. Design for cross-scale movement rather than optimizing behavior at one scale, using the scaffold's four functions — surface patterns across scales, open conditions to critique, return interpretation to action, and center the learner as the agent of connection.
• Learning analytics designers. Give learners access to their own data together with interpretive support, so dashboards expose what the metrics measure and what they leave out instead of positioning the learner as the subject of the system's intelligence.
• Instructional designers. Do not read behavioral proxies such as session frequency, resource access, or submission timing as evidence of self-regulation; extend designs from "agency within conditions" to "agency over conditions" by having learners question the purposes those conditions encode.

Limitations

• Conceptual article with no participants, intervention, or data: the three-dimension account and the cross-scale scaffold are proposals, and the paper offers no evidence that the scaffold changes learner AI Regulation in Education or design practice.
• The critique of how learning analytics operationalizes SRL rests on secondary sources (for example, de Barba et al., 2020; Matcha et al., 2019) rather than new analysis of dashboard behavior or learner data.
• The three dimensions are named but not operationalized: no measures, indicators, or decision rules are specified by which regulation, integration, and positioning could be distinguished empirically.
• The integration of SRL, complex systems science, and Freirean critical pedagogy is a single author's synthesis in a sole-authored Learning Letters article, so readers get one reading of each tradition rather than a negotiated integration across scholars.

Citation

de Barba, P. G. (2026). Learner Agency Across Scales: An Integrative Perspective on Self-Regulated Learning in Algorithmically Mediated Environments. Learning Letters, 7, Article 75.