Adaptive Cognitive Intervention Architecture: An Exploratory Computational Framework for Precision Reading...

Area of Science:

• Cognitive Science
• Educational Technology
• Machine Learning

Background:

• Reading comprehension is vital in higher education, but student responses to metacognitive interventions vary significantly.
• Understanding individual cognitive-response processes is key to improving learning outcomes.

Purpose of the Study:

• To develop and evaluate an adaptive metacognitive reading system using precision-learning architecture.
• To model learner heterogeneity, predict comprehension gains, and simulate instructional optimization.

Main Methods:

• Quasi-experimental longitudinal design with an 8-week metacognitive reading intervention.
• Latent-profile analysis to identify learner phenotypes (High Responders, Strategic Improvers, etc.).
• Explainable machine learning (XGBoost) for predicting comprehension gains (R²=0.61), Markov modeling, Bayesian inference, and reinforcement learning.

Main Results:

• The adapted Metacognitive Awareness Inventory (MAI) showed good reliability (Cronbach's α: 0.83-0.89).
• Four distinct learner-response phenotypes were identified, highlighting significant heterogeneity.
• Machine learning models accurately predicted individualized comprehension gains, with XGBoost performing best.

Conclusions:

• The proposed Adaptive Cognitive Intervention Architecture (ACIA) provides a computational framework for understanding learner differences.
• The system can model learner heterogeneity and predict comprehension improvements.
• It offers a method for simulating post hoc computational optimization for adaptive learning environments.