Behaviorism, also known as behavioral psychology, is a theory of learning based on the idea that all behavior is acquired through conditioned reflexes, which occur through interaction with the environment. Emotions are both reactive to behavior and can be constantly influenced by subsequent behavior.
The Russell's circumplex model of affect: arousal is represented by the vertical axis and valence by the horizontal axis.
The stylistic variations conveyed through emotions are both difficult to quantify and highly subjective. To overcome these difficulties, studies from affective science are employed that aim to identify the characteristics required for emotional communication, and to differentiate emotions from one another.
In this project, the Russell's circumplex model (RCM) of affect is used as a reference, which distributes emotions in a two-dimensional circular space. Seven emotions are chosen from the RCM diagram for the dancers to perform. The chosen emotions widely span the arousal and valence axes of the Russell's Circumplex Model of affect.
The body language data of different performers expressing 7 representative emotions were collected separately as a sample database and used for the following data model training.
Based on Laban Motion Analysis (LMA), a large number of body and motion features can be extracted and analyzed. These features provide a mechanism to not only derive emotional coordinates for new input movements, but also to stylize movements to express different emotions without reference to training data. This decoupling of training data and new input actions removes the need for manual processing and action registration.
We extracted human motion with 37 basic spatiotemporal features f = {fi}; i = 1,...,37, corresponding to the four Laban motion analysis components (Body, Effort, Shape, Space). These derived features are normalized to the range [0,1].
To this end, motion  synthesis in the system is a process of continuous optimization with multiple modified motions to make sure to get closer to the target emotion. In the scene of the system, the optimized motion mode is mapped to the animals or plants in the scene, and an interactive experience can be formed with the user.
The data of emotional intensity, pleasure level and emotional retention time of different users in different scenarios will be presented in a visual form, and each user can get an exclusive emotional spectrum in this immersive VR experience.
In this immersive experience, the priority is the virtual reality immersion and visual quality, with the HTC Vive 2 serving as the chosen test equipment for the setup
