Are Digital Characters Suitable for Emotion Recognition Tasks? An Evaluation Study Using MetaHumans

Dementia is often accompanied by impairments in social cognition, such as reduced empathy and difficulties in emotion recognition, which negatively impact daily functioning and increase caregiver burden. Early detection of these deficits is crucial, yet traditional assessments relying on paper-and-pencil tests or actor-based stimuli face limitations in scalability and standardization. Advances in digital technologies, particularly high-fidelity digital characters such as MetaHumans, offer a more flexible and ecologically valid alternative. This pilot study compared MetaHuman digital characters with a validated set of human actor videos in an emotion recognition task involving 20 participants. Six basic emotions (Joy, Sadness, Anger, Fear, Surprise, and Disgust) were presented, and performance was evaluated through recognition accuracy, response times, perceived naturalness, and subjective ratings. MetaHumans achieved significantly higher accuracy and comparable response times, although actors were rated slightly more natural. These findings suggest that MetaHumans can provide standardized, reproducible, and cost-effective stimuli for assessing social cognition.