Baseline correction left early leakage from 13 late EEG templates under a two-pass 1 Hz high-pass filter
We made 13 test signals from public auditory oddball EEG in OpenNeuro ds003061, version 1.1.2. Each signal was zero before 200 ms. We used a second-order Butterworth high-pass filter in both directions. The single-pass cutoff was 1 Hz. We then removed the mean from -200 to 0 ms. The mean output from 0 to 100 ms was -0.698 microvolts. The participant bootstrap 95% interval was -0.835 to -0.561 microvolts. A causal filter with the same frequency magnitude response gave exactly zero output before 200 ms. The package includes code, input statistics, output, source hashes, figures, and raw retrieval instructions. This benchmark measures a known filter artifact in specified late inputs. It does not establish early brain activity.
EEG · signal processing · high-pass filter · baseline correction · temporal leakage · filter artifact · negative control
References
3- usesdoi:10.18112/openneuro.ds003061.v1.1.2
Source EEG for the 13 participant templates, version 1.1.2.
- doi:10.1111/psyp.12437
- doi:10.1016/j.jneumeth.2016.01.002
Cited by
5 publicationsPublic catalog entries on Sciscape.
- usesMean response slopes were positive for previous target gaps and smaller for next gaps in 49 EEG and 16 fMRI adults during visual tasks
The filter leakage audit motivates the causal high-pass control.
- usesThe Pz target interval slope had a 95% interval that included zero in 10 eligible adults in ds003061 version 1.1.2
We used the result that baseline removal can leave earlier filter output from constructed late EEG inputs. This result limits the interpretation of baseline voltage in our report. It does not establish the cause of the measured interval slopes.
- usesResponse terms gave an uncertain mean late Pz contrast change in 13 adults from ds003061 v1.1.2 under fixed additive models
This work supplied the preserved BIDS metadata, fixed Git commit, annex keys, MD5 values, SHA-256 values, and raw EEG URLs. We checked the pinned metadata and pointers, downloaded all raw EEG files again, and checked all file identifiers. Its filter test informed the choice to add no high-pass filter.
- usesThe mean target-gap slope for 300–600 ms Pz EEG voltage was uncertain in 13 participants from ds003061 version 1.1.2
This publication supplied the fixed ds003061 version 1.1.2 raw manifest, BIDS metadata, and retrieval helper. We checked all 39 raw files and adapted the helper to extract single-target epochs and event history.
- usesA centered FIR filter with 1691 taps changed 0 to 100 ms Pz means in 13 adults when later input was held constant
The isolated late-wave filter test informed our question. We tested recorded continuous input and derived exact deadline errors and conditional bounds.