ERP and fMRI datasets of addition and subtraction approximation operations

Data collection period: 2022-2023.EEG data recording and processingPreprocessing. EEGLAB and ERPLAB toolboxes were used, including (1) loading electrode position files (2) using the mean values of the left and right mastoids (M1, M2 electrodes) as a re-reference; (3) applying 0. 1 Hz high-pass filtering and 30 Hz low-pass filtering to all electrode data; (4) setting up Eventlist in ERPLAB according to different experimental conditions and assigning it to the corresponding Bin; (5) Calculate the weights of artefacts such as corrected electrooculograms using FastICA; (6) Perform automatic bad-conductance detection of the data; (7) Artefact detection; (8) Stacked average the preprocessed EEG data using ERPLAB toolkit to obtain the ERP results. ERP data analysis. The ERP results were obtained by overlay averaging the preprocessed EEG data using the ERPLAB toolkit and low-pass filtering the overlay averaged data at 20 Hz. According to topographic maps and previous studies, it was found that attentional shifts on the mental digit line mainly activate the inferior parietal cortex [6,13], so we selected (P3, PZ, P4, PO3, POZ, PO4), and performed 2 (arithmetic operation: subtraction, addition) × 3 (direction of the result: less than, equal to, or greater than) repeated-measures ANOVA, and p-values of the ANOVA were corrected using the Green-house correction, Bonferroni-corrected. Bonferroni-corrected test was used for post hoc multiple comparisons. Magnetic resonance data analysisPreprocessing. The SPM12 toolbox was used, including: (1) conversion of DICOM data to NIFTI format; (2) time-layer correction of functional image data; (3) head-motion correction; (4) alignment of T1 structural images to the MNI standard space; (5) alignment of structural images to functional images; (6) normalisation of brain imaging data to the standard space by means of the EPI template included in the SPM and voxel resampling to 3 mm × 3 mm × 3 mm; (7) spatial smoothing of the functional image using Gaussian kernel function with FWHM set to 6 mm. Analysis of fMRI data. Individual analyses were performed using a generalised linear model (GLM) to construct multiple linear regression matrices. The matrix was designed to include six types of regressions, including additive-less than, additive-equal to, additive-greater than, subtractive-less than, subtractive-equal to, and subtractive-greater than, with the six header The six head parameters were added as covariates in the analyses. Based on the contrast images generated by GLM, a 2 (arithmetic operation: addition, subtraction) × 3 (result direction: less than, equal to, greater than) repeated measures ANOVA was performed using Flexible Factorial Design, with the calibration thresholds set at voxel level p < 0.001 uncorrected, and clump level p < 0.05 FWE corrected. Signal values of significant peak interaction coordinates in ANOVA were extracted by Marsbar and analysed by 2 (arithmetic operations: addition, subtraction) × 3 (direction of results: less than, equal to, greater than) repeated measures ANOVA in SPSS. Functional connectivity analyses . The signal values of the significant peak interaction coordinates in the ANOVA were extracted by the CONN toolkit to create spherical seed points with a radius of 6 mm, and their functional connectivity with the whole brain was examined, and the strength of connectivity between the seed points and different brain regions was examined by a 2 (arithmetic operation: addition, subtraction) × 3 (direction of the result: less than, equal to, or greater than) repeated-measurement ANOVA analysis, with the calibration threshold set at the voxel level of p < 0.001 uncorrected, cluster level p < 0.05 FWE corrected.EEG and MRI correlation analysisCharacterisation similarities . This included (1) calculation of Representational Dissimilarity Matrices (RDM) for ERP using a continuous 10ms time window based on preprocessed EEG data; (2) calculation of RDM for fMRI based on preprocessed functional image data by means of a spherical searchlight with a radius of 3mm; and (3) The time points corresponding to the N1, P2, and P3b wave peaks were selected, and the Spearman correlation coefficient was used to construct the RSA characterisation similarity analysis by calculating the correlation between the RDM of each ERP modality and the RDM of the fMRI modality in a spherical searchlight with a radius of 3 mm. For the RSA corresponding to each time point, a temporal activity profile was obtained from the ERP data, and a spatial correlation map was obtained from the fMRI data, and the comparison results were compared. Spatial correlation maps were obtained from ERP data and fMRI data. Significance tests were performed to level the comparison results, and the calibration thresholds were set at p < 0. 001 uncorrected for the voxel level and p < 0. 05 FWE corrected for the cluster level.