Publication-ready figures

Why do I need this step?

Honesty matters even more than looks. Two maps of two conditions drawn on different color scales suggest a difference that is not there, or hide one that is. Reviewers notice.

This lesson takes two figures from the previous lessons and prepares them for publication: a vector format, a fixed size, a readable font, one color scale, and a data file that backs up what the figure shows.

How do I do it in EEGLAB?

Start with the D1 epochs clean dataset loaded, as in “Starting point”. For the maps, use D1 epochs clean scalp, the 30-channel copy without EOG1 and EOG2. If it is not open, make it as in step 0 of the scalp maps lesson (Edit › then Select data). The GUI is enough for maps of one dataset. Comparing two conditions in one figure needs the script, as in the channel ERP lesson.

1. Find the scale limit

1. With D1 epochs clean scalp selected, open Plot › then ERP map series › then In 2-D.
2. Enter 100 200 300 400 in Plotting ERP scalp maps at these latencies and [1 4] in Plot geometry (rows,col.); [] -> near square. Click Ok.
3. Read the top value on the color bar and round it up to a whole number. That is your limit.

2. Fix the scale and export

1. Open the same window again. In the additional options field, enter 'electrodes', 'on', 'maplimits', [-5 5], replacing 5 with your limit. Use the same limit for every map you will compare with this one, even from another dataset.
2. In the figure window, open File › then Save As and choose PDF or SVG as the file type. This is MATLAB’s own menu, not an EEGLAB menu.
3. Open the file and zoom in: lines and letters stay sharp, and you can select the label text. That is what “vector” means.

3. Export the numbers behind the figure

1. Open File › then Export › then Data and ICA activity to text file.
2. In Output file name, enter erp_all.txt.
3. Check Export ERP average instead of trials:, and leave Transpose matrix (elec -> cols):, Export channel labels/component numbers: and Export time values: checked as they are. Click Ok.
4. You get a text file: the first row holds channel names, the first column holds time in milliseconds, and each following row is one time point. Put it in the paper’s supplementary material, or plot from it in another program.

How do I write it as code?

EEG = pop_loadset('D1_epochs_clean.set', pwd);
pos = eeg_getepochevent(EEG, 'type', {'square'}, 'timewin', [-5 5], 'fieldname', 'position');
EEG_pos1 = pop_select(EEG, 'trial', find(pos == 1));
EEG_pos2 = pop_select(EEG, 'trial', find(pos == 2));
EEG_pos1.setname = 'D1 epochs clean pos1';
EEG_pos2.setname = 'D1 epochs clean pos2';
chan = find(strcmp({EEG.chanlocs.labels}, 'Pz'));
scalp = find(~ismember({EEG.chanlocs.labels}, {'EOG1', 'EOG2'}));   % EOG is not on the average reference
erp1 = mean(EEG_pos1.data, 3);
erp2 = mean(EEG_pos2.data, 3);

% Figure 1: Pz waveforms at single-column size, 9-pt font, white background
fig1 = figure('Color', 'w', 'Units', 'centimeters', 'Position', [2 2 9 7]);
hold on;
plot(EEG.times, erp1(chan, :), 'Color', [0 0.45 0.74], 'LineWidth', 1.2);
plot(EEG.times, erp2(chan, :), 'Color', [0.85 0.33 0.10], 'LineWidth', 1.2, 'LineStyle', '--');
xline(0, 'k:'); yline(0, 'k-');
xlim([-200 800]);
set(gca, 'YDir', 'normal', 'FontName', 'Arial', 'FontSize', 9, 'TickDir', 'out', 'Box', 'off');
xlabel('Time (ms)'); ylabel('Potential (\muV)');
legend({sprintf('Position 1 (n = %d)', EEG_pos1.trials), ...
        sprintf('Position 2 (n = %d)', EEG_pos2.trials)}, 'Location', 'northeast', 'Box', 'off');
title('Pz', 'FontWeight', 'normal');
exportgraphics(fig1, 'fig_pz_erp.pdf', 'ContentType', 'vector', 'BackgroundColor', 'white');

% Figure 2: 300 to 500 ms maps for both conditions on one shared symmetric scale
win = EEG.times >= 300 & EEG.times <= 500;
map1 = mean(erp1(scalp, win), 2);
map2 = mean(erp2(scalp, win), 2);
lim = ceil(max(abs([map1; map2])));
fprintf('Shared color scale: -%d to %d uV\n', lim, lim);
fig2 = figure('Color', 'w', 'Units', 'centimeters', 'Position', [2 2 12 6]);
subplot(1, 2, 1);
topoplot(map1, EEG.chanlocs(scalp), 'maplimits', [-lim lim], 'electrodes', 'on');
title('Position 1', 'FontName', 'Arial', 'FontSize', 9, 'FontWeight', 'normal');
ax = subplot(1, 2, 2);
topoplot(map2, EEG.chanlocs(scalp), 'maplimits', [-lim lim], 'electrodes', 'on');
title('Position 2', 'FontName', 'Arial', 'FontSize', 9, 'FontWeight', 'normal');
axPos = ax.Position;
cb = colorbar('FontName', 'Arial', 'FontSize', 9);
ax.Position = axPos;   % keep both maps the same size
cb.Position = [0.92 0.25 0.02 0.5];
ylabel(cb, '\muV');
exportgraphics(fig2, 'fig_maps_300_500.pdf', 'ContentType', 'vector', 'BackgroundColor', 'white');
exportgraphics(fig2, 'fig_maps_300_500.png', 'Resolution', 600, 'BackgroundColor', 'white');

% The numbers behind the figures: one row per time point, one column per channel
pop_export(EEG_pos1, 'erp_pos1.txt', 'erp', 'on', 'transpose', 'on', 'elec', 'on', 'time', 'on', 'precision', 4);
pop_export(EEG_pos2, 'erp_pos2.txt', 'erp', 'on', 'transpose', 'on', 'elec', 'on', 'time', 'on', 'precision', 4);

Which value should I choose?

How do I check it worked?

• Zoom the PDF to 400%: waveforms and letters stay sharp, and you can select the axis text with the cursor.
• The page size of fig_pz_erp.pdf is about 9 × 7 cm in the file properties.
• The two maps in fig_maps_300_500.pdf have 30 electrode dots each, none below the eyes, are the same size and share one color bar whose limits match the number printed by the Shared color scale line.
• erp_pos1.txt has a header row, then 384 rows (one point every 7.8 ms from -1000 to 1992 ms), and 33 columns: time, then 32 channels.

What mistakes do people make?

• Screenshots. They have your screen’s resolution and degrade in print. Always export the figure from MATLAB.
• Drawing large, then shrinking. A 10-pt font on a 25-cm figure becomes 3.6 pt once shrunk to 9 cm. Set the size before you plot.
• One color scale per map. If you plot each condition in its own window, EEGLAB picks each window’s limits from its own data. Set maplimits to the same value in every window.
• EEGLAB’s light blue background. EEGLAB plot windows default to a very light blue. Make the background white on export, as the script does.
• An incomplete caption. Under every ERP figure, state the channel, reference, filters, trials per condition and polarity. Without them the figure cannot be interpreted.
• Polishing the data. Do not add smoothing or crop the axis to enlarge a difference. What you export must match the data file.

How do I write it in my paper?

A template: edit the data, settings and versions to match what you ran.

Figures were exported as vector PDF files with MATLAB exportgraphics; scalp maps of both conditions (30 average-referenced scalp channels) share one symmetric color scale of ±[X] µV (topoplot, EEGLAB 2026.1).

Going further

• EEGLAB Tutorial: https://eeglab.org/tutorials/11_Scripting/EEG_scalp_measures.html (11. Write scripts > d. Advanced plots)
• EEGLAB Tutorial: https://eeglab.org/tutorials/misc/Exporting_Data.html (Reference Topics > Exporting Data: Exporting data to an ASCII text file)