In most animals, the brain sends signals to local neural circuitry in the nerve chord to produce behaviors. Despite the central importance of these signals as an informational and anatomical bottleneck, little is known about how these signals are encoded at the neuronal level or how they control aspects of behavior. In insects, signals from the brain to the ventral nerve chord are carried by an estimated 350 pairs of bilaterally symmetric descending neurons (DNs). To date, only a handful of these descending neurons have known functions. In order to understand how DNs can control insect behaviors, we developed a method to identify descending interneuron function in an unbiased and systematic fashion in the model insect D. melanogaster. Using the red-shifted channelrhodopsin CsChrimson, we activated neurons in a collection of ~200 lines, most of which target single neurons out of a collection of DNs that fall into 60 distinct neuro-anatomical classes. Using a PCA and frequency analysis based technique described in Berman et al. (2014) we created a two-dimensional behavioral space based on the underlying postural dynamics of freely moving flies with and without red light activation. In this map, stereotyped behaviors are represented by local probability density maxima, and distinct behavioral motifs are easily distinguished. We then looked for map regions that were upregulated in CsChrimson activated animals. Using this technique, we were able to assign phenotypes to 90% of the DNs in our collection. We find that (1) DN control of stereotyped behaviors appears to be modular, (2) much DN function can be correlated with neuro-anatomy, and (3) the nature of this correlation hints at centralized control of locomotory activities. These findings, which are only apparent in a dataset of this size, have wide-ranging implications for how complex signals from the brain are encoded by descending neurons.