Source localization of deviance detection and regularity encoding in the auditory brain

Resumen

Resumen de la Tesis: Our auditory environment is wealth of continuously flowing information. From the whole set of acoustic inputs entering our sensory system we must create trustworthy mental representations of our world. In order to do so, our auditory system encodes regular acoustic features, stores them in sensory memory as auditory objects, and continuously compares such regularities with the incoming sensory input. Since mismatching events might carry extremely relevant information for the accomplishment of our goals, novel sounds or changes must be detected fast in an automatic and unconscious fashion, thus allowing for the reallocation of attentional resources and the proper adjustment of our behavior. Sudden deviations in our acoustic environment evoke the mismatch-negativity (MMN), an auditory evoked potential (AEP) generated between 100 and 250 ms after change onset in supratemporal and prefrontal cortices. Experimentally the MMN can be elicited in an ¿oddball¿ paradigm, where novel or infrequent stimuli (deviants) are interspersed in a regular sequence of repetitive sounds (standards) characterized by a particular acoustic feature (frequency, intensity, location, pace), or by more complex auditory regularities like patterns, abstract rules or feature combinations. Operationally, the MMN is obtained by subtracting the evoked activity to standards from that of the deviant sounds. However, recent studies challenged the notion that human deviance detection is solely indexed by the MMN. Simple auditory deviations in the early time range of the middle-latency responses (MLR), evoked between 20 and 50 ms after sound onset, produce amplitude modulations that are thought to reflect a very early mechanism of regularity encoding and deviance detection. The objective of the present PhD thesis is to examine the neuronal sources underlying auditory regularity encoding, and the subsequent detection of regularity-violating events in early (MLR) and late (MMN) time ranges. Specifically, in study I aimed to show a separation between deviance-related MLR and MMN source generators. Using an oddball paradigm, frequency changes elicited enhanced responses in both the MLR and MMN time range. Our magnetoencephalographic (MEG) source modeling revealed that deviance-relate MLR sources were generated in primary auditory areas, whereas MMN generators were located in secondary regions. In study II, the goal was to probe that MLR and the later MMN deviance detection mechanism are devoted to the processing of different levels of acoustic regularity. Using a sophisticated oddball design with both local and global changes, it was observed that complex regularities were encoded in the time range of the MMN only, and generated in secondary regions. Early deviance detection mechanisms did not show enhanced responses to complex regularity violations, thus suggesting that early and late mechanisms are devoted to different levels of regularity encoding. Finally, the third study aimed to show the neuronal sources involved in the encoding of acoustic features. A roving-standard paradigm was employed were trains of repeated tones are presented. Results indicated that both repetition suppression and repetition enhancement underlie auditory memory trace formation, and source generators are located in both typically auditory and non-auditory high-order regions. In conclusion, results presented in the current thesis indicate that early mechanisms of deviance detection exist in time intervals preceding the MMN and are generated in the primary auditory cortex, thus paralleling previous animal findings showing stimulus-specific adaptation of neurons located in primary regions. Moreover, results suggest that regularity encoding is not only a pervasive phenomenon, but is organized hierarchically with lower mechanisms devoted to the encoding of simple features and high-order regions engaged in complex regularity processing. In support for a hierarchical organization of regularity encoding, results suggest that high-order non-auditory regions of the human brain participate in the formation of new echoic memory traces. Such findings are in line with the notion that auditory perception is based on hierarchically organized sensory systems whose goal is to predict future events on the basis of previously encoded regularities. To do so, error and predictive signals are passed through organized processing stages.