Emegs Help Logo Version

Flow control, review passes and reuse

Open Configure Study > ERPERF > Flow control. Three independent choices determine a run: its processing range, its interaction mode and its reuse policy.

Select work, not just a starting point

In Batch processing steps, select one step or a contiguous range (for example, with Shift). Earlier checkpoints are prerequisites: they are located and validated, not automatically recalculated. Later steps are deferred.

With ICA enabled, the overview lists eleven steps. Without ICA, the list omits ICA-specific work and renumbers the visible rows. Use names when following instructions.

For example, to review thresholds across subjects:

  1. Run Calculate statistical artefact parameters.
  2. Run only Decide statistical artefact thresholds, in Review decisions.
  3. Run Finish averaging and results after decisions are saved.

To review components after a separate unattended training pass, end that pass at Calculate ICA decomposition, then select Decide ICA component rejection. A subsequent Apply ICA correction pass uses the saved selections.

Interaction mode

Mode User interaction
Unattended Stored settings and automatic/saved decisions; no settings or decision dialogs
Review decisions Stored settings, with selected decision checkpoints presented for review
Fully interactive Decision review plus settings prompts for selected steps

Selecting a decision step alone does not request a dialog. Conversely, reuse does not suppress review explicitly requested for a selected decision step. Inspect the log to distinguish a restored decision from a newly calculated one.

Reuse compatible completed work

This option applies to single-file and batch processing. It requests reuse of compatible completed checkpoints inside the selected range. Required checkpoints before that range remain prerequisites even when reuse is disabled.

A matching filename is not sufficient proof of compatibility. Input identity, channel order, sample rate, settings and completion metadata can prevent reuse. A diagnostic tells you which prerequisite or setting needs attention.

For example, changing experimental epoch length can reuse a compatible accepted ICA correction. Changing ICA training filters cannot be made compatible by starting later at epoch extraction. Include the required training/calculation steps or restore the original settings.

Older checkpoints may lack full provenance. When a compatibility fallback is used, the log warns about what could not be verified. It is not an assurance that all settings matched.

Stop, review rejection and restarting

Use Stop to cancel the run and wait for the worker to finish. A rejected subject during ICA review can advance the batch without treating the whole batch as cancelled. Read the retention/retraining choices carefully.

After interruption, start the appropriate range again. Only valid completed checkpoints can be reused; a partial calculation is not automatically complete. A recoverable ICA optimizer checkpoint is separate from an accepted correction.

For alternatives, use a processing branch instead of overwriting an established result. If deliberately removing outcomes, use the cleanup preview.