This paper gives an analysis of some conceptual issues in the neuroscientific study of empathy. The focus will almost exclusively be on a seminal paper by Decety and Jackson (2004) on the functional architecture of empathy. The authors withstand reductionistic tendencies in the exposition of what their findings might mean for the psychology of social cognition. They are aware of the thorny conceptual issues that arise when attempting to bridge intuitive folk psychological conceptions of empathy with explanations offered by social psychology, developmental science, and, most of all, neuroscience. They defend a conception which puts emphasis on the developmental, interactional and human aspects of empathy. In the second part of the paper we will see that this overt contention is at some points at odds with the conceptual framework that underlies the presentation of scientific findings. It will appear that the method of decomposition, i.e., breaking empathy down into (mutually interacting) ‘pieces’, is difficult to reconcile with the idea that empathy should primarily be defined as an interactional phenomenon. The method of decomposition puts empathy back within the brain, whereas recent philosophical work argues that empathy needs a definition which includes both processes in the empathizing subject and in the person with whom the subject empathizes. In the final part of the paper it is asked whether, how and to what extent it does matter that professionals know about the social neuroscience of empathy and, especially, its underlying conceptual framework. It is argued that conceptual innovations that currently are emerging in social neuroscience do matter for clinical and legal practices. In spite of the limitations mentioned earlier, Decety & Jackson's developmental and interactional approach helps to overcome reductionistic and mentalistic interpretations of human empathy.