simplify this for me: The feature decoding analysis was performed with fMRI activity patterns in visual cortex (VC) measured while subjects viewed or imagined visual images. Our experiments consisted of the training sessions in which only natural images were presented and the test sessions in which independent sets of natural images, artificial shapes, and alphabetical letters were pre- sented. In another test session, a mental imagery task was performed. The decoders were trained using the fMRI data from the training sessions, and the trained decoders were then used to predict DNN feature values from the fMRI data of the test sessions