Single-Dataset Workflow

Use one representative file first

Select one continuous .set file and a separate output directory in Step 1. Inspect event types and fields before Step 2.

Configure the text IA mapping, manually save the .mat configuration, and optionally save <base>_eyesort_ia.set. Review parsed regions and EEG.eyesort_step2_diagnostics; stop if key or assignment counts are wrong.

Build broad-to-specific labels

Start with a broad region-only label to prove that candidates exist. Add pass, fixation class, surrounding region, and saccade direction one at a time while monitoring zero-match diagnostics. Save the complete queue.

Click Apply All & Finish. EyeSort writes <base>_processed.set/.fdt automatically and may create a numbered summary CSV.

Validate the output

labeled = ~cellfun(@isempty, {EEG.event.eyesort_full_code});
table({EEG.event(labeled).eyesort_full_code}', ...
      {EEG.event(labeled).bdf_full_description}')

Spot-check original_type, current_region, region_pass_number, fixation_in_pass, is_last_in_pass, previous/next region fields, and the source X coordinate.

Once the counts and event examples match the analysis plan, reuse the saved configurations in the batch workflow.