Game Design Guided by Visual Attention

This research explores the integration of human visual attention models into game design to dynamically adjust difficulty levels. The central thesis posits that by monitoring eye movements—specifically fixation and pursuit—designers can predict a player’s attentional state and strategically place game elements to either facilitate or hinder performance. The study focuses on the relationship between local image complexity, eye movement patterns, and reaction times to external stimuli.

The methodology involved a controlled psychophysical experiment with six participants using an eye tracker to record gaze during free viewing and pursuit tasks. Researchers analyzed the entropy of image segments at fixation points to quantify visual complexity. Findings indicate that during fixation, higher local image entropy correlates with slower reaction times, suggesting that complex visual environments "glue" attention and delay disengagement. Conversely, during pursuit, attention is biased toward the direction of motion, significantly reducing reaction times for targets appearing ahead of the pursuit path.

These findings were applied to an interactive 2D shooter game where difficulty was modulated by placing enemies in locations of high or low attentional cost. Testing with five participants confirmed that games designed with these principles successfully produced measurable differences in player performance. Specifically, placing enemies in high-entropy fixation zones or behind the pursuit direction increased difficulty, resulting in slower reaction times and lower scores. The study concludes that incorporating task-dependent eye movement data into game engines offers a viable, data-driven approach to adaptive difficulty, moving beyond traditional, static game design paradigms.