The cognitive performance of the human brain depends on both device and network attributes. Regarding devices, analog operation with high dynamic range and adaptation across temporal scales lead to circuits capable of processing diverse information while learning from a continually changing environment. We present superconducting circuits with rich neuromorphic functionality and multiple plasticity mechanisms achieving these functions. Regarding network structure, dense local connectivity combined with long-range, high-bandwidth links enable efficient information integration across the network, supporting the dynamical activity associated with cognition. We present integrated photonic circuits capable of achieving such interconnection networks. By leveraging superconducting detectors, signaling between artificial neurons can be accomplished with single telecom photons, providing the energy efficiency necessary for systems with billions of neurons. This talk will describe the theoretical motivations for this approach to artificial cognition, the superconducting optoelectronic hardware comprising the circuits and networks, and our experimental progress toward realization of functioning systems. Special attention will be paid to recent developments in this field. These developments include circuit and network architectures enabling series current biasing of the neural elements as well as computationally efficient models for simulating large networks without circuit simulations of the constituent Josephson junctions on the picosecond time scale.