Why is there an ERN/Ne on correct trials? Response representations, stimulus-related components, and the theory of error-processing