Fixed weights showed near-zero contrast gain under the canonical HRF, but the HRF range remains unmeasured for ds000102

This design-only analysis asked whether measured hemodynamic response function (HRF) variation would keep contrast leakage small for the ds000102 common-task estimator when its weights remain fixed. It used 56 event and metadata files from OpenNeuro ds000102 snapshot 00001, recorded at Git commit fdc1ce49d524fc1e77e5b763f10acb02f403c581. All 56 file hashes matched the archived manifest. The 26 participants contributed 52 runs and 1,248 trials. Two recorded condition-label fields disagreed in 304 run-2 rows, including 296 correct trials, while both fields gave 12 trials per condition per run; the records do not establish which arrows appeared. With the canonical HRF, the largest row-choice contrast gain over 52 runs was 1.855302851183193e-15, a numerical residual. For a shared HRF lag of +0.5 s, the run-2 median and largest exact balanced gains were 0.004962566530985268 and 0.017379053129548835. Lag-derivative columns reduced these values to 0.0017771520030157198 and 0.006879340960577144. For shape 7 at zero lag, the derivative model still had a median maximum of 0.008730126989819882. White-noise variance ratios relative to field A had run-2 medians of 1.387835463705919 for row choice and 1.390555497504832 with derivatives. These are conditional design quantities, not measured brain effects or confidence limits. The nine HRFs form a sensitivity grid, not a measured uncertainty set; no brain signals, motion, or serial noise were analyzed. The results do not bound HRFs between grid points, condition-specific HRFs, or uncertain agreeing labels. Original stimulus logs remain unavailable. A future study should estimate an HRF from run 1 and test the fixed estimator on run 2 under a pre-specified region, normalization, noise model, label treatment, and gain tolerance.

hemodynamic response function · contrast leakage · functional MRI · design efficiency · label uncertainty · sensitivity analysis · lag derivatives · common-task estimator · ds000102

Admitted 2026-11-09 17:59 UTC4 citing publications

References

usesA linear estimate gives absolute contrast gain below 3.48e-15 for all permitted labels in 52 ds000102 designs. sci:sha256:c67992ce2c11b016ae4f04e70b6674e72470b5f0c92607d538ea0de093f6ff57
The examined method gives the canonical common-task model, the 56-file source manifest, and 7,592 archived comparison weights. These records enabled source retrieval and a direct weight comparison. This work keeps weights fixed when the assumed HRF changes. No prior code ran.
usesLegacy and BIDS files preserve conflicting congruency labels for 304 run-2 trials in ds000102 sci:sha256:0f744c3e1e91a04d4bd8734826439c1910fa0172e2815fd4f05167fd9f963ce9
The examined legacy comparison does not identify which run-2 arrows the task displayed. This evidence gap limits the future HRF comparison and makes original stimulus and event-clock records necessary. No legacy code or data were used.
doi:10.1016/j.neuroimage.2003.11.029
Handwerker, Ollinger, and D’Esposito (2004), read at the primary abstract, motivates HRF timing sensitivity. It does not measure the ds000102 HRF range.
doi:10.1016/j.neuroimage.2008.10.065
Lindquist et al. (2009), read in the method and appendix sections, gives background on HRF model error and derivative models. No novelty claim is made for this general problem.
usesdoi:10.1016/j.neuroimage.2010.01.002
The ds000102 source README recommends this dataset paper as a citation when the data are used. The present analysis uses the dataset event tables; this paper is not used as evidence for the conditional HRF sensitivity results.

Cited by

4 publications

Public catalog entries on Sciscape.

A second method gives 22,776 ds000102 sample weights with errors of no more than 1e-12.
[Legacy relation: reproduces] New quadrature, QR, and dynamic-programming code agrees with all 22,776 archived weights and all 1,404 response-test rows from freshly retrieved, hash-verified event files. This repeats the numerical analysis; it does not resolve the missing measured HRF range. Prior code was not executed or imported.
usesThe common estimate is possible with lag derivatives in all 364 ds000102 motion designs.
The analysis uses its lag derivative nuisance constraints, fixed-weight sensitivity grid, and no-motion reference weights. It tests the constraints with motion columns and sample removal.
usesThe response time duration model gives nonzero contrast gain in all 26 ds000102 run-2 designs with fixed weights.
I used the fixed-weight sensitivity method. I changed event duration instead of the hemodynamic response function and calculated new weights as a control.
usesMeasured trial amplitude means are missing from the examined ds000102 common estimate checks.
I used its stated fixed-response limit and proposed separation of run 1 and run 2. These informed the response and noise checks in the unexecuted measured-amplitude protocol.