Recent decades have witnessed a surge of confidence in the benefits of applying methods of neuroscience to psychiatric research as indicated, for instance, by the Research Domain Criteria framework proposed by the NIMH. However, the initial excitement of this prospective interdisciplinary partnership has been tempered by a number of setbacks, such as increasing doubts about the reliability of neuroimaging research. In this chapter we propose that many of these challenges can be traced to problems of scientific practice—techne—that have been especially rampant in the neuroscience domain and its application to psychiatry. Additionally, following the work of G.E. Berrios on the specificity of psychiatric objects, we propose that translational neuroscience will be successful only when it embraces a more complete epistemological model of mental symptoms. Finally, we suggest that neuroscience needs to adopt a more critical stance with respect to the image of the brain as a computer and in general should be more critically aware of the influence of such models when attempting translational research. In the last two parts of the chapter, we discuss the relationships between these problems and propose some general guidelines for creating a more productive partnership between neuroscience and psychiatry.