Study On The Interactive Patterns Of Dynamic Functional Network In Auditory Perception During Listening Task Of Utterance

Language is composed of small elements into larger structures according to the grammatical system,thereby forming a hierarchical structure of language units,such as syllables,words,phrases,and sentences.In natural languages,syllables are the basic units of speech structure that can be distinguished by hearing.Words are thought of as the smallest meaningful unit of speech that can stand by themselves.And phrases are a group of words,as part of sentence grammar,which is unit in grammar hierarchy.functions as a constituent in the syntax of a sentence,a single unit within a grammatical hierarchy.However,syntactic and semantic information are acquired in the process of language acquisition and cannot be read out from auditory stimuli,and then decoded in the brain.Under the boundary that hierarchical linguistic structures are not clearly defined by the acoustic cues.It is still unclear how the brain performs a series of sensory inputs that are allocated in time based on the hierarchical structure of external sentences to syllables,words,and phrases with such a block structure built internally.Many techniques have been developed over the decades to reveal brain mechanisms.However,the main problem lies in the limitations of advanced technology for investigating the brain mechanisms due to the trade-off between high temporal dynamics and spatial complexity of brain activity.Previous dominant functional magnetic resonance imaging(f MRI)and positron emission tomography(PET)studies focused on brain activation patterns and functions attributed to them,but lacked time descriptions of how these regions are involved with language development.In contrast,electroencephalogram and magnetoencephalogram are very suitable for describing millisecond dynamics.However,EEG/MEG signals are easily hidden in electromagnetic artifacts from joint muscle movements,which are usually several orders of magnitude larger,thus distorting the analysis.Although there is no scalp noise in cortical electroencephalogram(ECo G),it may be just because it is complex and expensive.In addition,the effect of oscillations on the modulation of brain networks has been discovered by more and more studies.To characterize such a highly complex system,the spatiotemporal features and oscillatory activities of the neural signals need to be considered.Fortunately,the recent development of blind source separation technology,especially the independent components analysis paves a new way for component decomposition,which can exclude muscle activity and blinking from cognitive components.In addition,the recent Granger causality analysis,multiple autoregression(MVAR)simulation and source information toolbox(SIFT)provide a new evaluation framework to visualize information flow in time-frequency dependent dynamic distributed functional networks.At the same time,a large number of mapping relationships between brain regions and functions were discovered based on f MRI technology with high spatial accuracy.So,this paper mainly takes advantage of the Electroencephalography(EEG)with the high temporal resolution to measure multi-channel EEG signals in the listening task of sentence and use functional Magnetic Resonance Imaging(f MRI)with high spatial resolution database to make up the spatial defects.We use the ICA method to decompose EEG data with timefrequency characteristics to extract brain activity characteristics,then locate the brain components in space,and then calculate the information flow between brain regions through the Granger causality principle.Thus,we obtain the cognitive activity process with three dimensions of time-frequency space.At the same time,we extract working memory,semantic network,syntactic network and prosody network static brain region information related to cognitive process through f MRI database as spatial filters of cognitive activities of these four networks to observe the time-frequency characteristics of the speech perception process.It is mainly to reveal the brain mechanism in auditory perception during listening task of sentence by observing the relationship between dynamic functional networks and linguistic hierarchical structures.This paper demonstrates speech perception is a process based on a linguistic hierarchical structure.Hierarchical processing is based on the size of processing units of the brain at different frequencies.Working memory is mainly carried by delta-theta synchronization as auditory perception input to store syllable information.The semantic network is mainly carried by alpha-beta synchrony for the processing of individual word meanings.The syntactic network is mainly carried by delta-theta synchronously to construct syntactic information in the unit of phrases.Prosodic networks are mainly delta-theta synchronized to carry out the construction of prosodic information in phrase units to help segment the boundaries of the sequence structures.