The algorithm by which neural networks adjust weights during training by propagating error signals backward through the network; the MIT finding suggests biological learning uses a functionally similar mechanism
