ABSTRACT
The mismatch negativity (MMN), a well-known index of automatic acoustic change detection, is also a sensitive indicator of long-term memory traces for native language sounds. This may suggest that it is a more general index of the existence of memory traces or neuronal assemblies for learned engrams that play a role in cognitive processing. Here, we review recent work done to test this idea. When comparing MMNs to words and meaningless pseudowords, we found larger MMN amplitudes for words than for meaningless items, thus indicating the presence of long-term memory networks for the former, a finding now replicated by different groups using different languages. The MMN also shows differences in the brain response to individual words that even reflect aspects of the referential semantics of these words. This suggests that cortical memory networks of individual lexical items can be investigated using the MMN. In other studies, we found evidence that the MMN reflects automatic syntactic processing commencing as early as ~100 ms after relevant information becomes available in the acoustic input. In summary, neurophysiological imaging of the MMN response provides a unique opportunity to see subtle spatiotemporal dynamics of language processing in the human cortex in lexical, semantic, and syntactic domains. The chapter briefly reviews the current state of the art in the neurophysiology of language, provides motivations for studying language function with the MMN, and offers a detailed review of findings which are discussed in the framework of distributed action-perception networks underlying language and other cognitive processes.
