From Guidance to Autonomy: Temporal Dynamics of Self-Regulated Inquiry and the Role of Learner-Generated Products

Developing learners’ autonomy in conducting scientific inquiry requires instructional guidance that should fade as learners extend self‑regulation. This study investigates how learners internalise inquiry tactics while engaging with a simulation‑based learning environment that delivers just‑in‑time prompts based on an autonomy transition model. Thirty‑nine participants completed three successive tasks. Key tactics, ranging from identifying variables to generating rules, were operationally defined within log files and coded videos. For every tactic instance, we marked macro‑ and micro‑regulation phases, coded guidance sources (system prompt, prompt‐review, note‑review, information access), and assigned a four‑level autonomy score. Chi‑square tests showed demand for guidance varied by tactic: prompts were most frequent during conceptualisation, whereas experimentation tactics were often self‑initiated. Temporal visualisation and sequence clustering revealed three autonomy trajectories. Cluster 1 moved from prompts to note‑review yet reached autonomy; Some learners in Cluster 3 achieved autonomy by Task C; Cluster 2 never reviewed self-generated notes, leading low-knowledge learners to rely on prompts. Review of self-generated notes was a pivotal intermediate scaffold and often preceded upward shifts in autonomy. The findings refine design principles for adaptive guidance systems: early conceptual prompts, dynamic detection of product reuse, and strategic withdrawal when notes stabilise. Implications for AI‑driven tutors are discussed.