Asking a naive question about the McGurk effect: Why does audio [b] give more [d] percepts with visual [g] than with visual [d]?