Mismatch negativity (MMN) is the brain response to violations of a rule. The MMN peaks at about 100–250 ms after change onset. If MMN occurs so early, before conscious perception, why is surprise considered a cognitive phenomenon? Isn't it rather connected with autonomic processes that take place earlier than cognitive process. Surprise is considered as belonging to cognitive processes. Among emotion theories, the appraisal theory of emotions, considers that there can be no emotion unless I consciously 'label' it. Here I wonder - how can there be any conscious 'labeling' of surprise when MMN shows that surprise takes place earlier than conscious states? If we accept that surprise does not need conscious awareness for taking place then the appraisal theories that connect surprise with unexpected events are mistaken. See Ortony, Reisenzein and many other appraisal theorists on surprise.
As often in this area, the problem seems to be entirely created by confusions about definitions and terminology and the difficulties resulting from the use of common-sense concepts like “surprise” in scientific theories. One answer would be “Why not?” Basically you seem to object to the definition of surprise used by some theorists but definitions are neither true or false, they are merely common or unusual, and useful or not. Emotions in particular have received so many definitions that there isn't much of a basis to claim any particular one as the “correct” one anymore. Those theorists are just talking about something else than mismatch negativity, it's as simple as that.
Obviously there is a such a thing as being consciously surprised and we have a word for it: “surprise”. There is no way around that without redefining the word to something unrecognizable to most speakers of the English language. It's only with a way to clearly distinguish (1) physiological processes (2) cognitive processes and (3) subjective experiences that we can begin to explore the relationship between them, describe the physiological underpinnings of surprise and account for the experience as such. This experience is a basic fact that you can always “define away” from the scope of your interest but would still need to be explained. Said differently, the “labeling” process is an interesting object of study in itself, whether or not you are prepared to include it in the definition of surprise or want to define emotion through, say, autonomic activation.
If you just assume “surprise” to mean “mismatch negativity”, what you have is neither an empirical finding nor an interesting theory of the conscious experience of surprise. You have just deprived yourself of the ability to clearly speak about the core phenomenon of surprise. In my view, interesting questions to ask would therefore be “What reason is there to conflate surprise with mismatch negativity (or, for that matter, any other brain process)?” or “To what extent does our experience of surprise coincide/follow from mismatch negativity?”
Appraisal theory argues that a cognitive mechanism assesses the meaning of a situation before initiating an emotion. It does not argue that this mechanism operates consciously, so demonstrating that it doesn't doesn't impinge on the validity of appraisal theory at all.
Mismatch negativity may be part of the process leading to surprise, but noting that a rule has been violated is peripheral to what surprise is all about. Surprise is the emotional response.