Fixed report readers failed a shared pain reversal ranking test in 114 EEG adults and 20 fMRI adults

We tested whether neural weights fitted to pain reports at fixed heat improve trial ranking when lower heat has the higher report. Separate public studies supplied scalp EEG and BOLD fMRI. Development used 39 EEG adults and eight MRI adults. A local code lock preceded new reversal scores in 114 other EEG adults and 20 other MRI adults. Mean report-minus-heat fitting gains were 0.27 points, with 95% person Student limits -0.81 to +1.35, and 5.70 points, with limits +0.68 to +10.72. Report-reader accuracies were 45.18% and 56.78%. The shared rule required above-chance report ranking and positive gain in both modalities. It failed. A second shared rule that bounded gain below five points also failed. Both two-claim Holm values were 1.0. An added sparse MRI study retained 87 adults. Its gain was 4.74 points but accuracy was 49.50%, with limits 46.52% to 52.47%. Its full transport rule failed. Refitting readers in whole-person resamples put both MRI gain ranges across zero. All negative people and fixed controls remain. A Gaussian selection example gave 40.41% reversal accuracy despite within-input correlation 0.5. Independent loops rebuilt all 221 scored person results exactly. Fresh source checks matched all EEG features and hashes, MRI metadata and selected MRI values. The package supplies a checked challenge, permitted compact inputs, code, models, all results, source pins, figures and repeat steps. No common gain or gain bound was confirmed. Source selection, small MRI training sets, cohort differences, sparse support and measurement filters limit scope. No pain-specific or causal process was established.

human neuroscience · pain reports · EEG · fMRI · trial ranking · cross-modal tests · negative results · reader uncertainty