Quality, Provenance, and Responsible Interpretation

Eye and physiological signals are observations. They do not identify mental states without a defensible measurement model and external validation.

validate_eye_dataset(x)
audit_sampling_rate(x)
audit_signal_quality(x)
audit_pupil_quality(x)
audit_trial_coverage(x)
audit_missingness(x)
analysis_readiness(x)
interpretive_warnings()

The package does not automatically equate:

• fixation with attention;
• long dwell time with item difficulty;
• pupil dilation with cognitive load;
• rapid response with guessing;
• EDA with a uniquely identified emotional state;
• a statistical process component with effort or engagement.

Reproducible records

provenance_manifest(x)
write_provenance(x, "provenance.json", format = "json")
report_eye_dataset(x, "analysis-report.md", include_plots = TRUE)

Exploratory and confirmatory analyses should use separately declared feature, preprocessing, AOI, exclusion, and model specifications.