Neuroscience students from Massachusetts Institute of Technology (MIT) have developed a computer model with the potential to predict the emotions of others. This model can predict a range of different emotions, including joy, gratitude, bewilderment, regret, and embarrassment, and is approaching human social intelligence.
To build this model, the students designed it to predict the emotions of people facing the dilemma of the prisoner, a classic scenario from the theory of games in which two people must decide whether to cooperate with a partner or betray them. The researchers took several factors into account, including the desires of the person, their expectations in a given situation, and the presence of witnesses to their actions, which can affect emotional reactions.
Rebecca Saxe, a professor of brain and cognitive sciences at MIT and senior author of the study, explained that the model was based on “very common, basic intuitions, and we said that we can take a very simple grammar and create a model that will learn to predict emotions from such signs.”
The model considers desires, expectations, and the influence of observers on individual behavior, evaluates motives, compares outcomes with expectations, and predicts emotions based on these factors. This model, which imitates human social intelligence, has surpassed other emotion prediction models and will be adapted for wider use.
When interacting with another person, it is common to spend part of the time trying to predict their emotions in relation to what is said or done. This task requires a cognitive skill called Theory of Mind, which helps draw conclusions about beliefs, desires, intentions, and emotions of others. The MIT students have used their knowledge of how people intuitively understand other people’s emotions to design a model that approaches this aspect of human social intelligence.