Predictions are constantly generated from diverse sources to optimize cognitive functions in the ever-changing environment. However, the neural origin and generation process of top-down induced prediction remain elusive. We hypothesized that motor-based and memory-based predictions are mediated by distinct feedback networks from motor and memory systems to the sensory cortices. Using fMRI and a dual imagery paradigm, we showed that motor and memory upstream systems excited the auditory cortex in a content-specific manner. Moreover, the inferior and posterior parts of the parietal lobe differentially relayed predictive signals in motor-to-sensory and memory-to-sensory networks. Our results reveal the functionally distinct neural networks that mediate top-down sensory prediction and ground the neurocomputational basis of predictive processing.

1. Read review articles to understand machine learning techniques in cognitive neuroscience: multi-voxel pattern analysis (MVPA) and multivariate cross-classification (MVCC)
2. Investigated the general structure, analysis stream, and a typical searchlight decoding of The Decoding Toolbox (TDT) for MVPA
3. Wrote MATLAB codes to perform a leave-one-run-out cross-validated searchlight MVPA, including extraction of parameter estimates (“betas”) from a sphere of voxels with a certain center; partitioning of the data for training and test; classification using support vector machine (SVM); accuracy calculation of classifying test data in sphere
4. Wrote MVPA methods part of scientific manuscript

1. Cognitive neuroscience methods: brain decoding (MVPA & MVCC)
2. Cognitive neuroscience concepts: sensorimotor integration & episodic memory & mental imagery
3. Machine learning, MATLAB programming, academic writing