Emotions are undeniably a central component of human existence. In recent years, the importance of developing systems which incorporate emotions into human-computer interaction (HCI) has been widely acknowledged. However, research on emotion recognition has been dominated by studies of facial expression of emotion. In comparison, the study of EBL has received relatively little attention. Here we study the phenomena of EBL, specifically of static body postures expressing emotions, from two different perspectives. First, we have built a computational model for the recognition of four basic emotions which achieves a relatively high recognition rate (70 %). Secondly, to study perception of EBL, we examined what body parts attract the observer's attention during the perception of EBL. This is done by tracking eye movements of human subjects during the observation of static postures expressing emotions. Although invaluable information can be inferred from motion, this study will show that information about static body posture is rich enough for both automatic recognition and human perception. The present study contributes in an applicative way both to the development of automatic recognition systems of EBL and provides insight into the nature of human recognition of EBL.