Perception of Difficulty in 2D Platformers Using Graph Grammars

Abstract

This paper summarizes the findings of a study on difficulty perception conducted with players of 2D side-scrolling platform games which level design is automatically created. The primary goal of our research, is to offer players a suitable experience that matches their abilities and the difficulty of the games they play. The study presented here was conducted to test the calculations and formulas designed for a method that adapts the difficulty of a level according to the player skills through level design. We created a new method that combines Dynamic Difficulty Adjustment and the principle of Graph Grammars to create levels with a specific degree of difficulty, the novelty of our method lies on the use of Graph Grammars as a tool for creating multipath levels in 2D platformers; an implementation of Graph Grammars applied to 2D platformers; the creation of formulas that numerically estimate difficulty and insights about the perception of difficulty. Results show that the difficulty estimation of our method and the difficulty perceived by participants of our experiment has a correlation coefficient of 0.75, with a linear relationship and a strong correlation between both variables, demonstrating that the approach is heading to the right direction. In addition, difficulty and performance have a correlation of -0.69, which shows an inversely proportional relationship, more difficult levels have lower performance than easier levels; for this particular research this was an expected result, confirming the method is doing the calculations properly