Creating a shadow matching game for studies regarding spatial reasoning

Sammanfattning

Spatial reasoning is a critical cognitive ability linked to success in science, technology,engineering, and mathematics (STEM). Research indicates that spatial skills are malleableand can be improved through targeted training, including video games. This thesis presents the development and evaluation of Shadow Match, a game designed to train the ability to map three‐dimensional objects onto their two‐dimensional shadow projections. The game was implemented in Godot 4.6, featuring a shadow‐matching mechanism based on pixel comparison and Intersection over Union (IoU) to determine level completion. A study with ten participants investigated learning schedules over several weeks, collecting data oncompletion time, button presses, IoU values, and more. Results show that participants improved across sessions, with convergence in completion times regardless of schedule, suggesting that a compressed training period of four to five days with 20–30 minute sessions may be sufficient to learn the game's mechanics. Cross‐analysis with two other spatial games (Box Blox and Tumble) indicated transfer of rotation‐based spatial skills to Box Blox but not to Tumble, which relies on sequential planning. However, due to the absence of pre‐and post‐test spatial ability assessments, no definitive conclusion can be drawn regarding the game’s effectiveness in improving general spatial cognition. The findings offer preliminary evidence of transferable skill acquisition and highlight the need for validated assessment instruments in future work.