Structs and the EEG Variable

1Fields

A struct keeps related values together under one name. Each value sits in a named field, reached with a dot.

info.subject = 'sub-01';
info.srate = 500;
info.condition = 'oddball';
info
info.srate
fieldnames(info)

2Loading a dataset

EEGLAB is already on the path and started in this lesson, and the current folder is psych390. pop_loadset reads a dataset from disk and hands it back as a struct, stored here in EEG.

EEG = pop_loadset('filename', 'sub-01.set', 'filepath', 'data');
EEG

Typing EEG lists every field with its value, or its size when the value is too big to show.

3The fields you will use most

EEG.srate  # samples per second
EEG.nbchan  # number of channels
EEG.pnts  # time points per epoch (the whole recording, before epoching)
EEG.pnts / EEG.srate  # seconds of recording
size(EEG.data)

EEG.data has 32 rows, one per channel, and 30,000 columns: 60 seconds times 500 samples per second. After the data are cut into epochs it gains a third dimension, one layer per trial, and EEG.trials counts them.

4Indexing into EEG.data

This is the matrix indexing from the last lesson, on real-sized data. The voltages are in microvolts; EEG.times gives the time of each column in milliseconds.

EEG.data(14, 1:5)  # channel 14, first five samples
EEG.times(1:5)
mean(EEG.data(14, :))

5Structs inside structs

Some fields hold struct arrays, one element per channel or per event. Pick the element with parentheses, then the field with a dot.

EEG.chanlocs
EEG.chanlocs(14).labels
EEG.event(1)
EEG.event(1).type
numel(EEG.event)

Leave out the index and you ask for all 32 labels at once. MATLAB prints them one after another; to keep them as a list, wrap the request in braces.

labels = {EEG.chanlocs.labels}

6Changing a field

EEG.setname = 'sub-01 practice';
EEG.setname

This changes the copy in memory only. The file on disk stays the same until you save it with pop_saveset (lesson 8).