ACT-R (Adaptive Control of Thought-Rational) cognitive architecture provides the foundational framework for modeling how humans maintain goal orientation in the face of distractions. This architecture assumes that cognition emerges from the interaction of specialized modules including:

• Declarative memory for facts

• Procedural memory for skills

• Working memory buffers that process information in real-time

The architecture's strength lies in its ability to make quantitative predictions about human behavior while maintaining neurobiological plausibility.

Recent advances in ACT-R modeling have specifically addressed intrinsic motivation and goal persistence. Studies show that pattern discovery mechanisms within ACT-R can effectively model intellectual curiosity and sustained engagement. The architecture's utility learning mechanisms naturally represent how interest and motivation evolve through experience, providing a computational foundation for understanding why students abandon learning goals.

Effective learning requires continuous monitoring and adjustment of one's cognitive strategies, motivation, and behavior. Digital environments often disrupt this self-regulatory cycle. Working memory capacity, attention span, and anxiety levels represent core cognitive parameters that influence task completion and goal maintenance. These executive functions interact dynamically with environmental factors, creating individual cognitive profiles that determine learning effectiveness.

When learners face excessive information and choices, their working memory becomes overwhelmed, preventing meaningful learning and goal maintenance. The research demonstrates that these parameters can be estimated through behavioral analysis, enabling personalized interventions.

The intervention strategies incorporate research on goal hierarchies, goal conflict resolution, and goal abandonment patterns to maintain long-term learning objectives. Goal disengagement involves cognitive, affective, and behavioral components that must be addressed holistically.

The research reveals that semantic relevance, rather than personal interest alone, drives effective information retrievalin digital learning environments. This finding challenges conventional assumptions about motivation and suggests that cognitive models must account for goal-directed behavior patterns rather than simple preference-based interactions.

Statistical analysis shows that semantic alignment with task objectives accounts for 49% of variance in learning outcomes. This finding indicates that relevance detection is not merely helpful but fundamental to learning effectiveness.

Based on these models, goal abandonment in digital learning occurs through:

1. Metacognitive failure - Learners lose awareness of their actual progress vs. perceived progress

2. Motivational decay - Without feedback on alignment between activities and goals, intrinsic motivation fades. The cognitive modeling reveals that motivation influences multiple downstream factors including attention span, working memory efficiency, and susceptibility to distraction.

3. Interest drift - Algorithm-driven content pulls attention away from original learning objectives. Research demonstrates that interest operates as a comparative and temporal construct that emerges through continuous evaluation of present experiences against prior interactions.

4. Self-regulation breakdown - Lack of structured self-monitoring prevents course correction

1. JR Anderson: How can the human mind occur in the physical universe. Oxford University Press, New York, 2007

2. M Csikszentmihalyi: Flow: the psychology of optimal experience. Harper & Row, New York, 1990

3. N H R. Kodikara, J Morita: Developing a tutoring system based on behavior logging and personalized cognitive modeling. IJCNN 2024 Workshop: Towards Realizing Whole-Brain Computational Models Guided by Cognitive Models (WBCM-CogM), 2024

4. J Morita, K Miwa, A Maehigashi, H Terai, K Kojima, FE Ritter: Cognitive modeling of automation adaptation in a time critical task. Frontiers in Psychology 11, 2149

5. J Morita, T Pitakchokchai, GB Raj, Y Yamamoto, H Yuhashi, T Koguchi: Regulating ruminative web browsing based on the counterbalance modeling approach. Frontiers in Artificial Intelligence 5, 741610

6. J Morita, M Kano, S Shimojo, Y Ohmoto, Y Hayashi: Model-Based Support for Collaborative Concept Mapping in Open-ended Domains. International Conference on Intelligent Tutoring Systems, 404-411, 2023

7. K Nagashima, J Nishikawa, J Morita: Modeling task immersion based on goal activation mechanism. Artificial Life and Robotics 30 (1), 72-87, 2024

8. K Nagashima, J Morita, Y Takeuchi: Intrinsic motivation in cognitive architecture: intellectual curiosity originated from pattern discovery, Frontiers in Artificial Intelligence 7, 1397860, 2024

9. K  Nagashima and J Morita: Model of Anxiety Behavior Based on Time Perception and Anticipation in ACT-R. Proceedings of 2025 International Conference on Cognitive Modeling, 2025

10. S. Sakai, K. Itabashi and J. Morita: Estimating personal model parameters from utterances in model-based reminiscence. 10th International Conference on Affective Computing and Intelligent Interaction (ACII), Nara, Japan, 2022, pp. 1-8, 2022.

11. K. Shimbori, M. Shirasuna, and J. Morita: Estimating Emotion Related Parameters for Inter- and Intra-Individual Variability. Proceedings of 2025 International Conference on Cognitive Modeling, 2025