Research On Crystallized Intelligence And Fluid Intelligence Based On Multimodal Neuroimaging

Intelligence always plays an important role in personal development and social life,and research on intelligence quotient(IQ)has always been a hot spot in academia.IQ is an abstract concept that cannot be directly measured.It is a comprehensive manifes-tation of individual abilities such as individual memory,thinking,and strain.Therefore,it is necessary to explore IQ in different cognitive domains during the research process.With the development of neuroimaging,more researchers are trying to explore the neural mechanism of intelligence with the help of emerging technical means.However,most of these studies were based on a single modality of neural data,and different brain imaging technologies have their own advantages and disadvantages.If the neural data of multiple modalities could be combined,it is believed that the relationship between the brain and intelligence can be better revealed.In this paper,by combining with MRI and electroencephalogram(EEG)data,we studied the relationship between intelligence and some related brain variables,such as white matter tracts(mean Tract-based Fractional Anisotropy,mTBFA),gray matter corti-cal volume(CV),and the peak oscillatory frequency(P_?)in the thalamic cortex,reflect-ing the processing speed of the central nervous system.Our research sample included 62healthy individuals from Cuba,and we collected their EEG and MRI data,and their WAIS-III intelligence scale,including four indicators:perceptual organization(PO),processing speed(PS),language understanding(VC)and working memory(WM).We adopted a reg-ularized multiple indicator multiple causes(MIMIC)model to assess the relation between Intelligence with mTBFA and CV in different ROI,and P_?.MIMIC is a special form of structural equation model,including the measurement model and structure model.We first constructed a measurement model using four IQ indicators(PO,PS,VC and WM)to obtain two latent variables that could reflect IQ in different cognitive domains:fluid intelligence(FI),reflected the PO and PS,and crystal intelligence(CI),reaction of VC and WM.Next,we built a three-layer MIMIC model where the first layer includes three kinds of modalities of brain index as well as the age,the second layer is variables(FI and CI)able to reflect IQ,and the third layer is the measurement of four IQ indexes.We focused on finding different brain indicators by affecting latent variables and further impacting IQ indicator index.To this end,we utilized regularization constraint to fit our model and obtained ideal results.The model revealed that some part of mTBFA and CV,P_?,and age had an impact on IQ,and these findings were consistent with previous studies.In addition,we also specifi-cally pointed out that brain indicators that affect FI include:corticospinal tract,Uncinate fasciculus;temporal lobe,parietal lobe,thalamus,subcortical structures,cerebellum and age.Brain indicators that affect CI include:Superior longitudinal fasciculus,Forceps minor;temporal lobe,occipital lobe,thalamus,subcortical structures,Cerebellum and P_?.The results obtained by using three modal data and the IQ scales to construct a causal model revealed the influence of different brain mechanisms on IQ in different cognitive domains,and hope to provide new research direction for cognitive and neuroscience re-search in the future.