Group conditions and select accepted trials

On ERPERF > Averaging, select an event-group/trial-selection profile or open its editor. Profiles are stored in ERPERF_event-groups.json. The reserved no-grouping profile preserves the original condition/output convention.

Group or pass through conditions

A profile can merge source conditions into named output conditions. When groups are explicitly defined, each source condition must be assigned once or explicitly excluded; incomplete mappings are not silently accepted.

You can also leave the output-group table empty to keep source conditions separate and apply only trial constraints. Explicit exclusions remain meaningful. Use a distinct output suffix so the result can be identified later.

Grouping does not rewrite the original epoch condition file or its artefact decisions. It operates on trials accepted by artefact assessment.

Trial constraints

• Minimum event duration uses the timing intervals associated with the extracted trials. This is not the epoch duration.
• Source-condition constraints run before merging and control source contributions.
• Grouped-output constraints run after merging and control final counts.
• Minimum balances the listed conditions to their smallest available accepted count. A fixed count requires that many trials in every listed cell.
• Selection can use deterministic random order with a saved seed, or first/last accepted trials.

For example, retaining conditions 11, 21 and 31 separately while equalizing their counts requires a pass-through profile with a constraint on those conditions—not three new merged groups. Whether balancing is scientifically appropriate depends on the measure and study question.

A condition with no usable trials cannot supply a positive balanced count. Inspect the decisions and timing selection instead of interpreting failed balancing as an averaging result.

Outputs and exact reuse

Profile-specific averages and optional condition APP files carry the suffix:

subject.SES.<suffix>.at11.ar
subject.SES.<suffix>.APP11
subject.SES.<suffix>.trial-selection.json

The trial-selection manifest records original trial indices, source/output conditions, selection method/seed and relevant input identities. A completed matching manifest can be reused. Changes to artefact decisions, grouping, constraints or timing can invalidate it.

The global .APP.SES remains the canonical original-trial container; it is not reduced to just one grouping profile’s retained trials.

When WAVELET uses the same compatible completed ERPERF profile, it can reuse the exact selected trials, rather than independently drawing an equal-sized sample. Check its log for the explicit reuse statement and manifest path.

Use Retained Trials with the same branch and profile to inspect the final counts.