What is an error-driven learning algorithm?

If the prediction is correct, the algorithm continues to the next instance. If the prediction is incorrect, the algorithm updates the weights.