Research in mathematics related affect uses a variety of theoretical frameworks. Three different dimensions have been suggested as significant to characterize concepts in this area: (1) emotional, motivational, and cognitive aspects of affect, (2) state and trait aspects of affect, and (3) physiological, psychological, and sociological level of theorizing affect. In this study, we used the information in reference lists and graph theory to identify Graph Communities (coherent clusters) of research papers published in the affect groups of CERME conferences. The four main Graph Communities identified in the analysis were Foundation (beliefs, attitudes, emotions), Self-Efficacy, Motivation, and Teacher Development. There were six smaller Graph Communities that may suggest emerging new frameworks: Academic Emotions, Metacognition, Teacher Beliefs, Resilience, Meaning, and Identity. These results suggest that of the three possible dimensions to structure the area, the distinction between cognition (beliefs), motivation, and emotions is the most important one.