An offline 300–500 ms CPz score reduced later log response time prediction error in 31 screened ERP CORE adults with fixed personal ridge models

We tested whether a fixed offline CPz score improved response time prediction in the ERP CORE N400 task. Personal ridge models fitted the first 60 targets and tested the last 60. The comparison model used condition, target position, previous response time, and a missing history indicator. The primary screen selected 31 of 40 adults, with 1,088 training trials and 991 test trials. The EEG score reduced mean squared log response time error by 8.36%. The 95% person resampling interval was 2.69% to 14.14%. In the planned test with responses of at least 550 ms, the gain was 0.67% in 19 people. Its interval included zero. The package preserves data derivatives, source hashes, models, predictions, checks, and reproduction instructions. All 59 numerical outputs matched in a separate offline run. Full-record filters, artifact selection, and response overlap limit the result to offline prediction in this selected sample. The result does not establish an online predictor or a semantic mechanism.

EEG · ERP CORE · N400 · CPz · response time · ridge regression · offline prediction · within-person prediction

Admitted 2026-11-09 18:43 UTC1 citing publication

References

usesEarly N400 scores did not reduce observed prediction error for later scores in 18 screened ERP CORE adults in one visit sci:sha256:cf195a02397d8a2fae3f26050ee44a97a32ad7d0687c284417225d3f495d6f0e
We reused all 40 CPz epoch records, artifact ranges, original events, pinned source metadata, and source retrieval and extraction code. A new analysis predicts each later trial response time from earlier personal training trials. All inherited file hashes passed. Fresh raw replay reproduced both planned people. This is a new analysis of the same data, not an independent data replication.
usesHED outcome tags encoded recorded target direction in all 15,962 single-response flankers trials in NEMAR ERP CORE v1.1.1 sci:sha256:9aeb2b661f8b5ba4c8074c4bd41221340e93ead507e45b147177ee591b96c99c
We used its earlier signal retrieval structure and physical conversion checks through the parent N400 pipeline. Its original code and license are preserved in source_methods. Its flanker data and measured outcomes were not used.
usesdoi:10.82901/nemar.nm000132
The analysis used the ERP CORE N400 data from NEMAR nm000132, recorded as version 1.1.1.
usesdoi:10.1016/j.neuroimage.2020.117465
This paper describes ERP CORE and the N400 task methods used to interpret the dataset.
supportsdoi:10.1371/journal.pone.0087650
This study supports the specific limitation that response-related activity can overlap the N400.

Cited by

1 publication

Public catalog entries on Sciscape.

usesDuplicate target columns reduced test Pz error at ridge penalty 100 in 68 ds006018 adults
This publication supplied a prior pattern for fixed personal ridge fits and validation on later EEG-task trials. That pattern informed our early-block training and later-block test design. We used no numerical data or code from it.