Development and usability evaluation of an adaptive cognitive training game based on raven’s coloured progressive matrices

Abstract

This study proposes an intelligent adaptive cognitive training system based on a soft computing approach to enhance non-verbal reasoning skills in children with mild intellectual disabilities (MID). The system integrates Raven’s Coloured Progressive Matrices (RCPM) into a mobile puzzle-based learning environment called Cognitia. Unlike conventional educational games with static difficulty levels, the proposed system employs a fuzzy inference system (FIS) to dynamically adjust task difficulty based on user performance metrics, including normalized completion time, error frequency, and level of assistance. A Mamdani-type fuzzy model with defined membership functions and rule-based reasoning is utilized to handle uncertainty and variability in user behavior, enabling personalized and human-like decision-making in difficulty adaptation. The system was developed using the Game Development Life Cycle (GDLC) framework and implemented on the Android platform. Experimental evaluation was conducted through usability testing and real-world deployment involving 12 students with MID in a special education setting. The results indicate that the proposed adaptive mechanism successfully maintains an optimal challenge level, achieving a task completion rate of 92% and a user acceptance score of 84.38%. Furthermore, qualitative feedback from teachers confirms that the system is accessible, engaging, and pedagogically relevant. This study contributes to the field of soft computing by demonstrating the practical implementation of a fuzzy-based adaptive difficulty model in an educational game context. The findings highlight the effectiveness of integrating lightweight computational intelligence into cognitive training systems to support inclusive and personalized learning environments.