Hinton spent years trying to develop biologically-plausible learning algorithms as good as backpropagation and finally concluded there probably aren't any—backpropagation is actually better, which was a major relief because it resolved a long research puzzle.