Reseach On Cognitive Control In Language Information Processing And Its Neural Mechanism

Cognitive control is a construct in contemporary cognitive neuroscience that refers to the top-down mental processes for the plan, manipulation and control of information processing depending on the current goals. Cognitive control is associated with such basic cognitive processes as attention allocation, goal maintenance, conflict control and error processing, and thus is fundamental for higher order cognitive abilities such as perception, thought, and language. As language is the basic capacity that distinguishes human being from other animals, study into the cognitive control mechanism of language is of great significance. Language ability compromises multiple dimensions such as language production, learning and comprehension. Cognitive control mechanism may penetrate into each dimension and play a key role in inhibition control in bilingual language production, feedback-controlled language learning, and contextual control of language comprehension. Thus, using event-related potentials(ERP), functional magnetic resonance imaging(fMRI) together with neuropeptide oxytocin control, the present study explored the functional features and neuroinformation mechanism of cognitive control in the above three dimensions of language processing. The major contributions of this dissertation are as follows:1. Based on comparisons of language switching activations against different baseline conditions in language production, the study explored the comprehensive neural network of inhibition control in language switching. In a picture-naming paradigm, the study compared brain activations in mixed naming against control baseline, L1 and L2 naming baseline respectively. Result shows that inhibition control in language switching activates a frontal-parietal-subcortical network that is employed in general cognitive control. The activation spans from the dorsolateral prefrontal cortex(DLPFC) and pre-supplementary motion area(SMA), supramarginal gyrus down to anterior cingulate cortex(ACC) and the caudate. After receiving from the ACC the error monitoring information of the automatically activated non-target language, DLPFC executes the decision to inhibit it. The caudate is involved in this process by participating in the inhibition or activation of the target language on task demand through the frontal-basal ganglia route. The pre-SMA serves as a control trade-off of the ACC error monitoring function. The DLPFC-parietal route is involved in maintaining the working memory and target representation and also inhibiting the phonetic activation of the non-target language. Meanwhile, switching between different script language also employed additional structures beyond the general cognitive control network, such as left lingual gyrus and bilateral fusiform gyrus, to main the correspondence between meaning representation and phonological encoding.2. Based on a series of feedback-controlled learning tasks, the study proposed the neuromechanims of feedback control in language learning. Neuropeptide oxytocin versus placebo controlled subjects’ performance comparisons showed that there are cross-cultural differences in social feedback facilitation of language learning. fMRI results showed that the feedback information activates limbic structures related to memory and emotion control, and increases the functional connection of amygdala and hippocampus, caudate and insular based rewards regions. Thus by enhancing the activations in the reward and salience network in cognitive control, the feedback can modulate language learning and memory. In this process, oxytocin may be increasing the salience and reward value of anticipated social feedback to enhance to activation of salience network in cognitive control.3. Based on language context controlled lexical processing, the study examined the time course of cognitive control and allocation of cognitive resources in language comprehension. Through two ERP tests comparing the online auditory language information processing with strong constraining language context, the study revealed that vowels and tones evoked different patterns in a series of ERP components, N1, N400 and LPC. These modulations shows that cognitive control from the language context may affect pre-lexical level phonetic and phonological processing and then controls the whole process of lexical activation from lexical selection to meaning integration and reanalysis. The different ERP patterns also showed that in the early stage of lexical selection and meaning integration, more cognitive resources were allocated to the vowels but in the later stage of lexical reanalysis, more attentional resources were given to tone processing.Overall, the current results from the three dimensions of cognitive control in language processing showed that language control employs mainly a general cognitive control network that involves frontal, parietal and subcortical structures. This network may interact with brain networks for other cognitive abilities like the emotional control network through the caudate related salience network for cognitive control. The temporal course and the allocation of cognitive resource for language control mechanism show great flexibility and accuracy that tailor to the goal of language processing. Such need-based flexible cognitive control mechanism might be biologically evolved to adapt to the demand of human language processing. Future researches should examine the dynamic developments of the cognitive control mechanism in language processing.