The contrast weight cosine is less than zero in 9 of 26 primary ds000102 run-2 models.
This deterministic analysis tests how two conflicting event-label fields change the incongruent-minus-congruent contrast design in the ds000102 flanker task. It uses recorded event times from all 26 participants, 52 runs, and 1,248 trials; it does not analyze brain images. In run 2, `Stimulus` and `trial_type` disagree on 304 of 624 trials, including 296 correct trials. In the primary ordinary least-squares model, the cosine between the two contrast weight vectors is negative in 9 of 26 runs; its minimum, median, and maximum are -0.5000196666, 0.0530471081, and 0.4893563648. Six model changes yield 9 or 10 negative cosines. Seven participants have negative cosines in all seven models, while four have signs that change. When both runs are combined with one shared contrast and separate nuisance coefficients, all 26 primary cosines are positive (minimum 0.2349143757, median 0.5206243804, maximum 0.7558124360). These results describe specified models of recorded event times. They do not identify the displayed stimuli or measure effects on brain-response estimates. The analysis does not select a true label field, model motion or signal autocorrelation, or make population or causal claims.
functional magnetic resonance imaging · fMRI · event design · flanker task · contrast weights · label conflict · ordinary least squares · ds000102 · OpenNeuro
References
3- usesThe ds000102 run-2 congruency conflict is also present in delivered legacy OpenfMRI files sci:sha256:110f981d665e939cd3e9e1d0d31711436d8952049867c5669164d8ee7f3981ed
Its inspected delivered-legacy-file result constrained the question and interpretation. It prevented treating the conflict as an established BIDS-only error or selecting a true stimulus field from file format alone.
- usesConflicting ds000102 congruency fields yield different run-2 interference estimates, but neither encoding is verified sci:sha256:10e47bb887e96725293d19a7f6e6c66a641da7fd083bc3d83617c11c61133991
Its inspected source manifest supplied the version-pinned event URLs and expected checksums. Its label conflict supplied the two encodings tested in the new fMRI design calculation.
- doi:10.1016/j.neuroimage.2010.10.046
The source study reports the acquisition sample count used to set the assumed run length; the analysis does not reproduce that study's pipeline.
Cited by
5 publicationsPublic catalog entries on Sciscape.
- uses10 of 26 ds000102 run-2 contrast weight cosines are less than zero with six motion columns.
Its inspected primary model supplied the HRF, drift, event, and contrast settings. Its source manifest supplied expected event checksums. Its primary single-run outputs enabled the executed no-motion comparison. We added measured archive motion and fixed sample masks.
- usesThe label field changes fALFF after task regression in two raw ds000102 run-2 scans.
We used the fixed event source manifest, continuous HRF formulas, and four selected design-cosine values. Fresh public downloads agree with the reused hashes. This enabled paired image analysis without choosing a true label field.
- usesThe row model gives higher median absolute fALFF changes than the union model in two raw ds000102 run-2 scans.
The inspected event manifest and continuous HRF construction fix the event data and model definition. Signed contrast geometry motivates the executed residual-projector checks.
- usesThe common estimate has absolute label gain less than 3.26e-15 in 364 ds000102 motion designs.
Its read distinction between the common curve and coefficient weights set the joint question. Its primary HRF, drift, and sample settings defined the baseline. We compared all 52 primary no-motion cosines with its inspected outputs.
- usesA linear estimate gives absolute contrast gain below 3.48e-15 for all permitted labels in 52 ds000102 designs.
The examined result distinguishes the common regressor from its coefficient weights. Its model settings and archived sample curves enabled the new estimates and the numerical comparison. No prior code ran.