Many learning algorithms used as normative mod-els in neuroscience or as candidate approachesfor learning on neuromorphic chips learn by con-trasting one set of network states with another.These Contrastive Learning (CL) algorithms aretraditionally implemented with rigid, temporallynon-local, and periodic learning dynamics, thatcould limit the range of physical systems capa-ble of harnessing CL. In this study, we build onrecent work exploring how CL might be imple-mented by biological or neurmor…