Goal-driven deep learning is increasingly used to supplement classical modeling approaches in computational neuroscience. The strength of deep neural networks lies in their ability to autonomously learn the connectivity required for solving complex, ecologically valid tasks, obviating the need for hand-engineered or hypothesis-driven connectivity patterns. Consequently, goal-driven models may generate hypotheses about the neurocomputations underlying cortical processing. Whereas goal-driven modeling is becoming increasingly common in perception neuroscience, its application to sensorimotor control is currently hampered by the complexity of the methods required to train models comprising the closed sensation-action loop. To clear this hurdle, we introduce a modeling library that provides researchers with the tools to train complex recurrent convolutional neural networks that model sensorimotor systems.

If you use this software, please cite it using the metadata from this file.