Most subjective reasoning is influenced by mental states. It facilitates communication between people and enhances interpersonal relationships. It can be difficult to duplicate using machine learning since complicated questions with subjective and challenging themes allow for multiple correct answers rather than just one. In my thesis, I mainly explored the kinds of diverse replies in Visual Question Answering and what elements could cause mental state references to signify the need for further exploration. Therefore, I created a user study to collect the rationales for answers given to question-image pairs. The survey study’s results offer a data set with several justifications that include mental states.