[2F4-OS-33] OS-33

1:45 PM - 2:00 PM JST(4:45 AM - 5:00 AM UTC)

[2F4-OS-33-02] Goal-Management Companion AI Using Affective States as Internal Control Variables: A Meta-Cognitive Modeling Approach

〇JUNYA MORITA¹, Umito Sorita¹, Nilupul Randika Kodikara¹, Joy KARMOKER¹, Nethmee Rasanga Gunasekara¹ (1. Shizuoka University)

Keywords:

Cognitive Architecture,Interest,Metacognitive Support

Conventional affective computing has primarily developed around the estimation of emotional states and externally induced interventions via emotion elicitation. However, in human intellectual activity, emotion functions as an internal state of cognitive control that regulates attentional allocation, memory retrieval strategies, persistence in exploration, and information selection. This study treats emotion as a control variable and proposes a design framework for an intelligent system that accompanies the goal-achievement process while providing metacognitive intervention. The framework assumes mechanisms that infer intrinsic states corresponding to user interest and cognitive load from Web browsing behavior, and that these states modulate goal activation and memory retrieval thresholds. As a foundational experiment, Web-based inquiry learning behavior of university students was recorded. The results indicate that interest dynamically fluctuates during the learning process, and that different fluctuation patterns produce differences in behavioral structure, including exploration depth and dwell time. These findings highlight the importance of capturing learners' interest during learning and supporting goal selection and regulation based on interest.