EEG in game user analysis: A framework for expertise classification during gameplay.

Area of Science:

• Neuroscience
• Human-Computer Interaction
• Cognitive Science

Background:

• Video games are culturally pervasive, with research focusing on cognitive aspects to optimize gaming experiences and design.
• Understanding player cognitive states, like expertise level, is crucial for personalized gaming and improved game development.

Purpose of the Study:

• To develop and validate a framework for classifying video game player expertise levels using wearable electroencephalography (EEG).
• To identify significant EEG channels and features that differentiate between novice and expert players.
• To evaluate the effectiveness of machine learning classifiers for expertise level prediction.

Main Methods:

• A wearable electroencephalography (EEG) headset (Emotiv EPOC) was used to record brain activity from video game players.
• Frequency domain features were extracted from EEG signals, followed by a systematic channel reduction using correlation-based attribute evaluation.
• Identified significant EEG channels (AF3, P7) and employed K-nearest neighbor (KNN) classifier for expertise level classification.

Main Results:

• A channel reduction method identified AF3 and P7 as the most significant EEG channels for expertise classification.
• Features extracted from AF3 and P7 channels showed statistically significant differences between novice and expert players (t-test).
• The K-nearest neighbor classifier achieved high accuracy, reaching 98.33% with data balancing, in distinguishing player expertise levels.

Conclusions:

• Wearable EEG technology can effectively classify video game player expertise levels.
• Specific EEG channels (AF3, P7) and their frequency domain features are critical indicators of player expertise.
• This framework offers a pathway for adaptive game design and personalized gaming experiences based on cognitive states.