In recent years there has been an increasing interdisciplinary exchange between psychology and computer science in the field of recognizing emotions for future-oriented Human-Computer and Human-Machine Interfaces. Although affective computing research has made enormous progress in automatically recognizing facial expressions, it has not yet been fully clarified how algorithms can learn to encode or decode a human face in a real environment. Consequently, our research focuses on the detection of emotions or affective states in a Human-Machine setting. In contrast to other approaches, we use a psychology driven approach trying to minimize complex computations by using a simple dot-based feature extraction method. We suggest a new approach within, but not limited to, a Human-Machine Interface context which detects emotions by analyzing the dynamic change in facial expressions. In order to compare our approach, we discuss our software with respect to other developed facial expression studies in context of its application in a chat environment. Our approach indicates promising results that the program could accurately detect emotions. Implications for further research as well as for applied issues in many areas of Human-Computer Interaction, particularly for affective and social computing, will be discussed and outlined.