{
  "harness": "sj-harness 0.8.3",
  "job": "job:1a7cf8892dea5d60655f45af36dcf19b",
  "kind": "adversarial_review",
  "bundle": "sha256:3dbc1535acb9898d951b666a7f8fb35cad8288d403456bb1079926d1556a3ffe",
  "claims": [
    {
      "local_id": "C1",
      "claim_id": "claim:0138b53828b5365f56517b9a35c26ba0200f0527ef796694953cab4133b26efa",
      "verdict": "minor_issues",
      "significance": "minor",
      "reason": "Numbers reproduce independently and the claim correctly calls this a non-detection; robust to n-matched pools, 2-6 mm regions and broader time/vestibular term sets. Issues: the registered criterion needs ~29 pp (near the 37 pp ceiling) for time perception at n=44, so the test had little power; the 10 mm 'anterior insula' region is mostly peri-insular (54% of hit coordinates >3 mm outside the insula; base rate 62.8%); title and Discussion read stronger than the hedged claim."
    },
    {
      "local_id": "C2",
      "claim_id": "claim:ddcbb310ff76f8a417c351f00692932b9f0e102e38a1f5e93ab8f3f152d1d852",
      "verdict": "major_issues",
      "significance": "minor",
      "reason": "The interval said to exclude an anterior preference (-5.98 to 3.61 pp) and 'no function shows an anterior preference' depend on the overlapping 10 mm regions (85.6% of posterior-hit articles are also anterior-hit) and the additive scale: with 2 mm regions interoception A-P is +7.82 pp (1.08 to 14.57), 4 mm +6.18 (-0.23 to 12.58), while ratios favour posterior. Equal z for interoception, the vestibular posterior lean (robust across radii, but only 6th-10th percentile vs random terms) and uninformative time perception hold."
    },
    {
      "local_id": "C3",
      "claim_id": "claim:b62b20cd0689c0bdbf37a45473c8e5f114b4dad04dfef50a80235388c9deeb80",
      "verdict": "minor_issues",
      "significance": "minor",
      "reason": "Core negative holds (mean-z percentile 0.71; 0.67/0.60 at K=1/3; near-zero at 2-6 mm; direct depression terms percentile 0.25). Second sentence overstated: mood's anterior difference +2.78 pp has CI -0.41 to 5.85 (posterior only, leans posterior); sadness's association is radius-dependent (z ~0.6 at 2 mm); sadness alone passes the cut-off at K=1 (0.9546); 'sad' largely tracks face/emotion paradigms."
    }
  ]
}
