eeg-behavior

To link single-trial EEG to reaction time, you can use single-trial regression with ROI mean amplitude as the predictor and RT as the outcome. This requires epoched EEG data and a behavioral log with trial-wise RT aligned to trial metadata.

The best way to compute reaction time and accuracy summaries is using IQR-based outlier removal for RT distributions, alongside condition-wise summaries and accuracy metrics like d-prime. This produces guardrailed descriptive statistics and behavior-stage JSON outputs.

Yes, you can test if an EEG effect predicts behavioral accuracy using single-trial regression, median-split, or across-subject correlation. The EEG predictor is the ROI mean amplitude, and the outcome is trial-level accuracy or RT.

You need epoched data in the epoch-stage/ directory, a DATASET_BRIEF.md for marker-to-condition mapping, and an ANALYSIS_PLAN.md for declared behavioral claims. Additionally, behavioral sources must contain trial-wise RT, response, and correctness data.

To analyze the speed-accuracy tradeoff, you can use conditional accuracy and delta-style analyses. These methods describe how performance changes across RT quantiles, providing insights into performance dynamics within your EEG experiment data.

Alternatives to single-trial regression for brain-behavior correlation include median-split analysis and across-subject correlation. These methods also link trial-level EEG predictors to behavioral outcomes like RT or accuracy.