This paper presents an approach to the development of emotional avatars that is based on the use of widely available technologies, such as 3D modeling and animation software. We demonstrate the application of a simplified version of the Facial Action Coding System (FACS) to represent basic emotions (anger, disgust, fear, happiness, sadness and surprise) and build an experimental platform to explore how animation and various avatar traits may impact emotion representation and recognition. We experimented with subjects in Mexico and China. Our results show remarkable recognition levels for some prototypical emotions, particularly happiness, sadness and surprise. This is relevant for some intelligent environments in which detecting whether a user experiences a positive or negative emotion is helpful. Our results also show that animation is an important factor in conveying emotion. Finally, from observing subjects with very different cultural backgrounds (Mexico and China), we confirmed the universality of prototypical emotions and the wide applicability of our approach for emotion representation.