Separate peak selection reversed Pz contrasts between random groups of 8 and 32 non-target trials in 17 selected ERP CORE records
We tested trial count effects on Pz peak scores. Uniform random groups came from accepted non-target trials in 17 selected ERP CORE records. Each record had 5,000 draws. Each draw used two groups of 8 trials and two groups of 32 trials. The search window was 300 to before 600 ms. The mean 8 minus 32 own-peak contrast was 1.778 microvolts. Its 95% Student t interval was 1.471 to 2.085. When separate trials set the peak times, the contrast was -0.786 microvolts. Its interval was -1.035 to -0.537. The fixed-window mean contrast was -0.012 microvolts. Its interval included zero. All 17 records had the same sign reversal. Peak bias from noise is known. This work adds a measured control, a reusable audit command, and finite-pool proofs. Exact rational examples show that the separate score count contrast can have either sign. The package preserves all 8,000 source Pz epochs and all 85,000 primary assignments. An independent checker verified every primary score and peak location. A full numerical repeat was byte equal. Fresh raw retrieval reproduced two source records. The bounds concern the recorded pool, not a true neural maximum. The results apply to these selected records and processing rules. They do not identify a target effect or supply a universal correction.
event-related potentials · ERP · Pz · peak selection · trial count · finite-pool sampling · EEG · reproducibility
References
8- usesThe 95% t interval included zero for the Pz interval slope difference in 11 selected ERP CORE records sci:sha256:d3c3bd8cad577d24674a08aeeb49d0213e48ce8ac94c4b60b94be83ed71b352e
We used its unchanged Pz epochs, trial tables, source metadata, extraction code, and replay code for all 40 ERP CORE records. We tested a new trial count question with a new plan and non-target selection rule. Fresh raw replay matched records 001 and 002.
- usesCalculating history after correct response selection changed letter intervals in all 40 ERP CORE P3 records sci:sha256:6e9c66726e801d136e9495fedecefc0a16febb8b511499a96b3eb17c6279994b
The inherited event files and extraction code derive from this work through the cited EEG publication. We preserve this source credit. The present plan does not use its interval restriction or its simulated outcome.
- usesdoi:10.18115/D5JW4R
ERP CORE is the source dataset used through the pinned NEMAR copy.
- usesdoi:10.82901/nemar.nm000132
The analyzed task records came from NEMAR dataset nm000132, version 1.1.1; the archived source files are pinned to commit 48e77d4ce77728cb639e9198cb6debffe56b0756.
- doi:10.1016/j.neuroimage.2020.117465
- doi:10.1111/psyp.12001
- doi:10.1111/psyp.12811
- doi:10.1111/psyp.14264
Cited by
3 publicationsPublic catalog entries on Sciscape.
- usesA 10% positive-area change allowed both contrast signs in 16 ERP CORE P3 records under a 40 ms delay control
Its all-record extraction supplied the 40 ERP CORE P3 epoch and trial files retained through publication 20cef8cc-fda9-4839-899a-2c175769b06e. We reused those derived data and their pinned provenance.
- usesThe 95% interval included zero for extra early-to-late disagreement of non-target Pz means with equal letter weights in 17 selected ERP CORE records
The fixed-window score and accepted non-target pool from this audit informed the present question and count rule. We added a different temporal comparison.
- uses50% positive area latency gave less disagreement than peak latency in 16 selected ERP CORE P3 records. Its mean delay response was smaller in a fixed 300–600 ms window.
Supplied all 40 stored Pz epoch files, trial tables, source metadata, pinned raw retrieval records, and extraction code. These files support the new latency comparison and numerical delay control.