Construct a model RDM

This example shows how to create RDMs from arbitrary data. A common use case for this is to construct a “model” RDM to RSA against the brain data. In this example, we will create a RDM based on the length of the words shown during an EEG experiment.

# Import required packages
import mne
import mne_rsa

MNE-Python contains a build-in data loader for the kiloword dataset, which is used here as an example dataset. Since we only need the words shown during the experiment, which are in the metadata, we can pass preload=False to prevent MNE-Python from loading the EEG data, which is a nice speed gain.

data_path = mne.datasets.kiloword.data_path(verbose=True)
epochs = mne.read_epochs(data_path / "kword_metadata-epo.fif", preload=False)

# Show the metadata of 10 random epochs
print(epochs.metadata.sample(10))

Using default location ~/mne_data for kiloword...
Downloading file 'MNE-kiloword-data.tar.gz' from 'https://osf.io/qkvf9/download?version=1' to '/home/vanvlm1/mne_data'.

  0%|                                              | 0.00/25.5M [00:00<?, ?B/s]
  1%|▍                                     | 327k/25.5M [00:00<00:08, 2.83MB/s]
  3%|█                                     | 740k/25.5M [00:00<00:07, 3.23MB/s]
  5%|██                                   | 1.38M/25.5M [00:00<00:05, 4.24MB/s]
  7%|██▌                                  | 1.81M/25.5M [00:00<00:06, 3.79MB/s]
 17%|██████▎                              | 4.30M/25.5M [00:00<00:02, 10.4MB/s]
 26%|█████████▋                           | 6.67M/25.5M [00:00<00:01, 14.5MB/s]
 36%|█████████████▎                       | 9.18M/25.5M [00:00<00:00, 17.7MB/s]
 44%|████████████████▎                    | 11.3M/25.5M [00:00<00:00, 18.5MB/s]
 54%|███████████████████▊                 | 13.7M/25.5M [00:00<00:00, 20.1MB/s]
 63%|███████████████████████▍             | 16.1M/25.5M [00:01<00:00, 21.5MB/s]
 72%|██████████████████████████▌          | 18.3M/25.5M [00:01<00:00, 21.0MB/s]
 81%|██████████████████████████████▏      | 20.8M/25.5M [00:01<00:00, 22.0MB/s]
 91%|█████████████████████████████████▌   | 23.1M/25.5M [00:01<00:00, 20.4MB/s]
100%|████████████████████████████████████▉| 25.5M/25.5M [00:01<00:00, 21.3MB/s]
  0%|                                              | 0.00/25.5M [00:00<?, ?B/s]
100%|██████████████████████████████████████| 25.5M/25.5M [00:00<00:00, 104GB/s]
Untarring contents of '/home/vanvlm1/mne_data/MNE-kiloword-data.tar.gz' to '/home/vanvlm1/mne_data'
Download complete in 03s (24.3 MB)
Reading /home/vanvlm1/mne_data/MNE-kiloword-data/kword_metadata-epo.fif ...
Isotrak not found
    Found the data of interest:
        t =    -100.00 ...     920.00 ms
        0 CTF compensation matrices available
Adding metadata with 8 columns
960 matching events found
No baseline correction applied
0 projection items activated
         WORD  Concreteness  WordFrequency  OrthographicDistance  NumberOfLetters  BigramFrequency  ConsonantVowelProportion  VisualComplexity
547     trust          2.45       2.828015                  1.60              5.0       510.000000                  0.800000         53.993651
324     tonic          5.30       1.845098                  1.70              5.0       702.600000                  0.600000         56.868425
335     flail          4.65       0.903090                  1.85              5.0       472.200000                  0.600000         48.824559
419   charter          4.40       2.155336                  1.60              7.0       919.000000                  0.714286         60.155132
754      chin          5.85       2.660865                  1.45              4.0       637.750000                  0.750000         59.599664
741      tree          6.70       3.111934                  1.65              4.0       741.250000                  0.500000         62.240529
806     sheen          4.00       1.662758                  1.80              5.0       592.000000                  0.600000         72.735896
78      bench          5.75       2.539076                  1.75              5.0       633.800000                  0.800000         72.166025
593    appeal          2.90       2.915927                  2.40              6.0       340.333333                  0.500000         73.575297
435  material          5.35       3.235781                  2.45              8.0       987.750000                  0.500000         62.126647

Now we are ready to create the “model” RDM, which will encode the difference in length between the words shown during the experiment.

rdm = mne_rsa.compute_rdm(epochs.metadata.NumberOfLetters, metric="euclidean")

# Plot the RDM
fig = mne_rsa.plot_rdms(rdm, title="Word length RDM")
fig.set_size_inches(3, 3)  # Make figure a little bigger to show axis properly