'dpca_code_V2' folder contains python notebooks (I used Google Colab) for RNN training, data analysis and plotting. - '2back_rnn_train.ipynb' trains and tests 7-unit and 60-unit RNNs (with circular input) and plots the PCA projections shown in Figure 4. - 'dpca_timecourse.ipynb' plots timecourses of dPCA projection of stimulus means, and timecourses of scalar transform, for both RNN and EEG (Figures 5 and 6). - 'subspace_angle.ipynb' finds angles between each pair of subspaces identified by dPCA for both RNN and EEG (Figure 7) - 'pev_calc.ipynb' finds cumulative percent explained variance of each dPC identified. - 'track_wm_info.ipynb' implements the method to track the disappearance of WM information using dPCA. - The '.py' files contain support functions. 'RNN' folder contains folders for 10 trained 7-unit RNNs and 10 trained 60-unit RNNs, and '.txt' files containing stimuli and target output sequences for training and testing the RNNs. 'data' folder contains processed data for EEG, 7-unit RNNs and 60-unit RNNs.