See the MediaEval 2022 webpage for information on how to register and participate.
The EmotionalMario challenge focuses on the iconic Super Mario Bros. video game and provides a multimodal data set based on a Super Mario Bros. implementation for OpenAI Gym. The data set contains for multiple players their game input, demographics, biomedical sensory input from a medical-grade device, and videos of their faces while playing the game.
Participants develop approaches to two subtasks:
With the rise of deep learning, many large leaps in research have been achieved in recent years such as human-level image recognition, text classification, and even content creation. Games and deep learning also have a rather long history together, specifically in the context of reinforcement learning. However, video games still pose a lot of challenges. Games are understood as engines of experience , and as such, they need to invoke human emotions. While emotion recognition has come a far way over the last decade , the connection between emotions and video games is still an open and interesting research question.
As games are designed to evoke emotions , we hypothesize that emotions in the player are reflected in the visuals of the video game. Simple examples are when players are happy after having mastered a particularly complicated challenge, when they are shocked by a jump scare scene in a horror game, or when they are excited after unlocking a new resource. These things can be measured by questionnaires after playing , but in the Emotional Mario task, we want to interconnect emotions and gameplay based on data instead of asking the players.
The target group for this task is diverse and broad. It includes researchers and practitioners from game design and development, game studies, machine learning, data science, artificial intelligence, and interactive multimedia. We also encourage interdisciplinary research involving people from psychology, game studies, and the humanities discussing the interrelation of biometric data, facial expressions, and gameplay. In any case, regardless of the research background, the submission will help to have a basic understanding of how we can better understand the connection between gameplay and the reaction of the player.
For the task, we provide Toadstool , a data set gathered from ten participants playing Super Mario Bros. Based on the protocols established in  we extend the data set by ten more participants. We gathered gameplay, video, and sensor data while people played Super Mario Bros. Data includes for instance heart rate, skin conductivity, videos of the players’ faces synchronized to the gameplay, but also the gameplay itself, demographics on the players and their scores and times spent in the game. For the Emotional Mario task (i) we release a training set including the original Toadstool data and new data on some additional participants (ii) an additional four participants will serve as ground truth and are to be published after the evaluation of the submitted runs.
** Informative value (i.e. is it a good summary of the gameplay), ** Accuracy (i.e. does it reflect the emotional up and downs and the skill of the play), and ** Innovation (ie. surprisingly new approach, non-linearity of the story, creative use of cuts, etc.)
Here are several research questions related to this challenge that participants can strive to answer in order to go beyond just looking at the evaluation metrics:
Please contact your task organizers with any questions on these points.
 Sylvester, T. (2013). Designing games: A guide to engineering experiences. “ O’Reilly Media, Inc.”.
 Saxena, Anvita, Ashish Khanna, and Deepak Gupta. “Emotion recognition and detection methods: A comprehensive survey.” Journal of Artificial Intelligence and Systems 2.1 (2020): 53-79.
 Abeele, V. V., Spiel, K., Nacke, L., Johnson, D., & Gerling, K. (2020). Development and validation of the player experience inventory: A scale to measure player experiences at the level of functional and psychosocial consequences. International Journal of Human-Computer Studies, 135, 102370.
 Svoren, H., Thambawita, V., Halvorsen, P., Jakobsen, P., Ceja, E. G., Noori, F. M., … Hicks, S. (2020, February 28). Toadstool: A Dataset for Training Emotional Intelligent Machines Playing Super Mario Bros. https://doi.org/10.31219/osf.io/4v9mp