Feeling good and in the zone? Or maybe you are hot and bothered? Irritated and frustrated? Or maybe sad and melancholic? While all kinds of games exist for many different moods, it might be a good idea for a video game to adjust the difficulty level based on the player’s feelings. Because being angry at a game all the time might not be as fun or good for you.
Scientists in South Korea at the Gwangju Institute of Science and Technology have come up with a pretty intriguing way to do this. Researchers developed a dynamic difficulty model that adapts to player emotions and tweaks it accordingly to ensure player satisfaction is maximized. Because who doesn’t like maximum satisfaction?
Game developers have long understood the need to balance game difficulty and player progression, trying to find a sweet spot that’s neither too hard nor too easy to ensure the gaming experience feels good. While settings can usually be changed, this often requires the player to manually adjust the setting. The Korean scientists propose something much more dynamic.
Their model involves training dynamic difficulty adaptation (DDA) agents using machine learning collected data from human players, who then adjust the game’s difficulty to maximize one of four different aspects related to a player’s satisfaction : Challenge, Competence, Flow, and Valence.
The scientists used a fighting game for their model and to train their DDA agents, as the human players played the fighting game against AI opponents, generated data for the agents, and the humans also had to answer a questionnaire about their experiences. Using an algorithm called Monte Carlo Tree Search, each DDA agent uses actual game data and simulated data to fine-tune and optimize the opposing AI’s combat style to maximize a specific emotion or “affective state”.
Associate Professor Kyung-Joong Kim, who led the study, said one advantage of their approach is that the player doesn’t need to be monitored with external sensors to detect their emotions. “Once trained, our model can only estimate players’ states using in-game features,” he said.
The study was small, with just 20 volunteers, but the team said the DDA agents produced AIs that enhanced the overall player experience. However, fighting games offer the most direct feedback, so the question arises as to how it could be used for other types of games, but the professor had an answer for that.
‘Commercial game companies already have huge amounts of player data. They can use this data to model the players and solve various problems related to game balance with our approach,’ said Professor Kim.
Their paper documenting the model, “Diversifying dynamic difference adjustagent by integration player state models into Monte-Carlo tree search,” will be published in Expert Systems With Applications on November 1. However, for those interested, it is already available online and can be found here.