As transistor scaling trends in VLSI and von Neumann computing begin to slow down, many look to distributed computing techniques as an avenue for further improvements in computing capacity. Neuromorphic computing algorithms and architectures are better suited to distributed computing approaches due to the complexity and parallelism achievable through brain-inspired neural units (neurons). Further, integrated photonics has greater potential for realization of high-bandwidth data paths, though the complexity of nonlinear elements is somewhat limited. Co-integration of optical and electrical elements allow for the design of more complex optoelectronic analog neurons which can better serve the needs of novel algorithms in neuroscience, and the development of self-supervised, hierarchical, online learning agents operating in real-time environments