{
 "paper": "gadeke-2026-guilt-insula",
 "reviewed_on": "2026-09-12",
 "model": "supplied:runs/gadeke-2026-guilt-insula/edge-review.answer.json",
 "note": "Relations for claims that had none, asked of the claim set as it stands. Validated by edges.edges_from_raw, so the direction rules and the corpus vocabulary hold exactly as they do for edge-inference.",
 "edges": [
  {
   "source": "social-vs-partner-roi-risky-safe-coefficients",
   "target": "only-precuneus-tpj-showed-positive",
   "relation": "part-of",
   "why": "Claim 68 is the Social-vs-Partner half of the decision-phase Risky-Safe ROI regression (Appendix 1-table 6, span tables-019) whose two comparisons claim 45 states as one whole when it says only precuneus and TPJ were positive in both."
  },
  {
   "source": "social-vs-solo-roi-risky-safe-coefficients",
   "target": "only-precuneus-tpj-showed-positive",
   "relation": "part-of",
   "why": "Claim 69 is the Social-vs-Solo half of the same decision-phase Risky-Safe ROI regression (Appendix 1-table 5, span tables-016), the second of the two comparisons claim 45 states together."
  },
  {
   "source": "risk-aversion-parameter-not-differ-between",
   "target": "participants-slightly-more-risk-averse",
   "relation": "validates",
   "why": "The null difference in rho between gain and loss trials is the control that justifies pooling across them, and claim 54's Social-vs-Solo comparison of rho is computed on those pooled estimates."
  },
  {
   "source": "left-ifg-insula-connectivity-subthreshold",
   "target": "connectivity-between-guilt-responsibility-related-outcome-ph",
   "relation": "tests",
   "why": "The subthreshold left IFG cluster from the insula-seeded gPPI (span results-143) is a further result of the seed-to-voxel analysis claim 14 predicts would show a Condition-by-Choice interaction in prefrontal cortex."
  },
  {
   "source": "among-computational-models-fitted-momentary",
   "target": "responsibility-redux-model-incorporating-expected",
   "relation": "part-of",
   "why": "The AIC row of the happiness-model comparison is one measure of the Responsibility Redux model's fit that claim 63 states as a whole when it says the model predicted momentary happiness well."
  },
  {
   "source": "participants-own-reward-prediction-errors",
   "target": "partner-reward-prediction-errors-resulting",
   "relation": "qualifies",
   "why": "Claim 55 asserts that the partner's prediction errors arising from the participant's own choices contribute to happiness; claim 51 bounds that contribution by showing the participant's own prediction errors weigh more heavily in the same model."
  }
 ],
 "dropped_already_connected": []
}