Multi-Level Integration for Predictive Inference: mPFC Connectivity in Action Anticipation Across Representational Profiles

This dataset contains behavioral and neuroimaging data from a table tennis serve anticipation study investigating how different training pathways (observation-based vs. execution-based) shape neural connectivity architecture supporting predictive inference. Behavioral data includes pre-post training accuracy, kinematic encoding model outputs (overlap and alignment coefficients), and raw kinematic feature data for individual participants across point-light and full-body video conditions. fMRI data comprises first-level contrast images and second-level statistical maps from whole-brain activation analyses (including conjunction and interaction effects), alongside Dynamic Causal Modeling (DCM) results with ROI masks, extracted time-series, and estimated connectivity parameters for all three groups. Analysis code provides Python scripts for behavioral modeling (kinematic encoding/readout models, statistical analyses, and visualization) and MATLAB scripts for fMRI analyses (second-level GLM, group comparisons, DCM model specification and estimation). All data are organized by analysis type (Behavioral data/code, fMRI data/code) to facilitate reproducibility of the reported findings.本数据集源自一项乒乓球发球预判研究，该研究旨在探究不同训练路径（基于观察的训练与基于执行的训练）如何塑造支撑预测性推理的神经连接架构。 行为数据包含训练前后的准确率、运动学编码模型的输出结果（重叠系数与对齐系数），以及所有受试者在点光源视频与全身视频两种实验条件下的原始运动学特征数据。 功能磁共振成像（fMRI）数据涵盖全脑激活分析（包含联合效应与交互效应）所得的一级对比图像与二级统计图谱，同时附带三组受试者的动态因果建模（Dynamic Causal Modeling, DCM）结果，具体包括感兴趣区（Region of Interest, ROI）掩码、提取的时间序列以及估计得到的连接参数。 分析代码包含用于行为建模的Python脚本（运动学编码/读出模型、统计分析与可视化工具），以及用于功能磁共振成像分析的MATLAB脚本（二级通用线性模型（General Linear Model, GLM）、组间比较、动态因果建模的模型设定与参数估计）。 所有数据均按分析类型（行为数据/代码、功能磁共振成像数据/代码）进行分类整理，以保障研究结果的可重复性。