**Data description** The uploaded data covers the preprocessed EEG data, classification accuracies for each main analysis and scene property ratings of each participant as well as the scene descriptions. The classification accuracies for each analysis are stored in MATLAB arrays (which are contained in results structs for the property decoding analyses) and can be used to replicate the respective figure in the manuscript. The classification accuracy arrays have the following dimensions and were used in the following figures: Mean pairwise scene decoding array - File name: mean_pairwise_scene_decoding.mat - Dimensions: subjects x frequencies - Figure: 3a Scene property decoding array - File name: scene_property_decoding.mat - Dimensions: subjects x properties x frequencies (properties is a dimension with a length of 4, where 1=Openness, 2=Naturalness, 3=Clutter Level, and 4=Brightness) - Figures: 3b, 4a (colored bars are peak mean decoding accuracies for each property) Mean pairwise scene cross-decoding accuracy array - File name: mean_pairwise_scene_cross_decoding.mat - Dimensions: frequency bands x subjects x time points (frequency bands is of length 3, where 1=theta, 2=alpha and 3=beta) - Figures: 3c (only alpha) as well as S2a (only theta) and S2c (only beta) in the supplement Scene property cross-decoding accuracy array - File name: scene_property_cross_decoding.mat - Dimensions: subjects x frequency bands x properties x time points (frequency bands is of length 3, where 1=theta 2=alpha and 3=beta, properties is of length 4, where 1=Openness, 2=Naturalness, 3=Clutter Level, and 4=Brightness) - Figures: 3d (only alpha) as well as S2b (only theta) and S2d (only beta) in the supplement Shuffled property decoding - File name: shuffled_property_decoding.mat - Dimensions: subjects x frequencies - Figures: 4a (shaded bars are shuffled property decoding accuracies at the peak mean property decoding frequencies) as well as S3a in the supplement Shuffled property cross-decoding - File name: shuffled_property_cross_decoding.mat - Dimensions: subjects x time points - Figures: 4b (shaded bars are shuffled property cross-decoding accuracies at the peak mean property cross-decoding time points) as well as S3b in the supplement **Code description** The uploaded code encompasses the preprocessing, frequency decomposition, decoding analyses as well as statistical testing used for the main analyses. Our code needs the toolboxes FieldTrip (https://www.fieldtriptoolbox.org/) and CoSMoMVPA (https://www.cosmomvpa.org/) to be installed. Statistical testing of the EEG data requires the custom functions EEG_clusterstats() and EEG_clusterstats_no_poststim_freq() in the Functions folder. Also note, that we only uploaded scripts of the statistical testing for the imagery property decoding and imagery-perception property cross-decoding as examples, since the other scripts we used for statistical testing only featured minor alterations of these scripts. The output of the permutation tests are z-values of the TFCE statistics which are corrected for multiple comparisons and still need to be thresholded or converted to p-values. Refer to the documentation of the function cosmo_montecarlo_cluster_stat() for thresholding options: https://cosmomvpa.org/matlab/cosmo_montecarlo_cluster_stat.html. **Pipeline for the main analyses:** 1. **preprocessing** (using preprocess_continuous_eeg.m, but note that the data in this repository is already preprocessed) 2. **(time-)frequency decomposition** (using fieldtrip_frequency_analysis_multitaper.m for imagery and fieldtrip_time_frequency_analyis_multitaper.m for perception) 3. **decoding** (using the respective "decoding_..." script, except for the mean-pairwise scene decoding, for which we first created representational dissimilarity matrices with pairwise decoding accuracies as the distance measures using the respective "rsa_..." scripts and then EEG_RDM_averages.m to average the bottom triangle of these matrices. 4. **statistical testing** (using the respective "Cosmo_EEG_sign_test_..." script) We conducted the comparison between property (cross-)decoding and shuffled property (cross-)decoding using prop_dec_vs_shuff_plot_paired_wilc.m and prop_cross_dec_vs_shuff_plot_paired_wilc.m.