{"content":"Superior temporal sulcus (STS) — a core mentalizing-network region — represents the partner's affective experience specifically when the participant is responsible for the partner's outcome. The mentalizing computation is conditioned on agency: STS does not track partner reward prediction errors uniformly, but rather engages selectively when the participant's own choice produced the partner's outcome, consistent with self-attributed responsibility recruiting other-modeling.","format":"text/plain"}
{"content":"Participants did not engage with the lottery task in a value-sensitive way — choices were inattentive or random — so the behavioural measures carry no information about preference or affect.","format":"text/plain"}
{"content":"The happiness and choice differences between Social and Partner conditions reflect a social-context-driven shift in risk attitude — participants become more or less risk averse when choosing on another person's behalf — rather than responsibility-contingent interpersonal guilt.","format":"text/plain"}
{"content":"If superior temporal sulcus represents the partner's affective experience selectively under responsibility, then in a model-based GLM with trial-by-trial partner RPE regressors split by who-chose, the parametric modulator for partner RPE in Social trials (subject-decided) should yield a significant cluster in left STS, while the parametric modulator for partner RPE in Partner trials (partner-decided) should not. This conditional engagement, rather than a main effect of partner RPE, is the diagnostic prediction.","format":"text/plain"}
—
urn:uuid:68f77387-69ad-4b26-b766-7ed6a6c1658c
prediction-behavioral-guilt-effect
{"content":"If interpersonal guilt is a real affective phenomenon contingent on responsibility, then in a within-subject lottery paradigm where the same partner outcomes occur under Social and Partner conditions, momentary happiness should drop more after negative partner outcomes in Social trials (participant chose) than in Partner trials (partner chose). This is operationalised as a positive interaction term between partnerWon (low partner outcome) and subjDecided (Social condition) in a mixed-effects regression of happiness on outcomes and responsibility.","format":"text/plain"}
—
urn:uuid:6eb8393f-a95e-4360-91d9-442d08200190
prediction-responsibility-model-wins-comparison
{"content":"If happiness incorporates partner reward prediction errors with a responsibility-weighted rule, then a computational model of happiness that includes a social_pRPE regressor (partner RPE × subject-decided indicator) should fit participant happiness ratings better than nested alternatives that omit the responsibility interaction. Formally: the Responsibility model (and Responsibility Redux model) should achieve lower cross-validated error than baseline (own-reward-only) and additive social-utility models in both Study 1 and Study 2; and the fitted social_pRPE weight should be significantly greater than zero.","format":"text/plain"}
—
urn:uuid:7808e12b-12e9-4340-997a-f4eeb2da40d7
insula-guilt-replicates-yu-koban-signature
{"content":"The anterior insula guilt effect is consistent with a previously published neural guilt signature map (Yu & Koban), providing convergent validity for the insula as a guilt-tracking region.","format":"text/plain"}
—
urn:uuid:7a27dc0a-904f-485f-99ae-2b1ac54ca7c2
hypothesis-insula-tracks-interpersonal-guilt
{"content":"Anterior insula is the neural substrate of interpersonal guilt: when a participant bears responsibility for a choice that yields a negative outcome for another person, anterior insula activity tracks the resulting responsibility-contingent affect. The region is recruited specifically by the conjunction of self-agency over the choice and harm to the partner, distinguishing guilt from generic empathy for partner loss (which would be present in the Partner condition too) and from regret about one's own loss (which would be present in Solo trials too).","format":"text/plain"}
{"content":"The happiness cost observed in the Social condition is general agency aversion — the unpleasantness of being the decision-maker as such — and is not contingent on responsibility for a negative outcome befalling the partner.","format":"text/plain"}
{"content":"Momentary happiness in social decision contexts is governed by a computational rule in which partner reward prediction errors that arise from the participant's own choices (social_pRPEs) carry an independent, non-zero weight, distinct from the weight on partner RPEs that arise without participant agency. Responsibility modulates how partner outcomes enter the happiness computation: participants internalize partner losses as affective costs specifically when those losses can be attributed to their own decision.","format":"text/plain"}
{"content":"If anterior insula encodes interpersonal guilt, then in the trial-level GLM, anterior insula BOLD should be elevated in the Social vs Partner contrast specifically after negative partner outcomes — i.e., the (condition × partner-outcome) interaction should yield a significant positive cluster in the anterior insula, detectable with small-volume FWE correction (SVC) over an a priori anterior-insula ROI derived from prior literature. The signal should not be present for the same contrast after positive partner outcomes (specificity).","format":"text/plain"}
—
urn:uuid:aa8d444f-2f3c-4558-8074-f6347d962523
scope-two-study-design
{"content":"All findings are scoped to a two-study design: Study 1 is an fMRI study of N=40 young adults run on a 3T scanner using a within-subject lottery paradigm with three conditions (Solo, Social, Partner) over two scanning sessions; Study 2 is a behavioural-only replication with N=44 independent participants using the same task outside the scanner. Both studies use the same EV-maximising algorithmic partner (the participants are not told the partner is an algorithm). All neural claims derive from Study 1; all behavioural claims are tested in both studies. Conclusions about interpersonal guilt are conditional on this paradigm — repeated lottery decisions in a constrained risk-taking design with monetary stakes — and on the specific population sampled.","format":"text/plain"}
—
urn:uuid:cb673152-8776-46c8-8f13-c95262c2138b
partner-algorithm-deception-assumption
{"content":"The experiment partner's decisions were in fact made by an expected-value-maximizing algorithm (lottery chosen if EV_diff > 0), not by a real person, but participants were not informed of this; this deception is a structural assumption underlying all social comparison contrasts and the claim that participants felt interpersonal guilt toward a real partner.","format":"text/plain"}
—
urn:uuid:d5ba8e97-1aab-40c4-9351-c441d83c5697
alt-imaging-contrast-invalid
{"content":"The imaging pipeline and condition contrasts do not recover established effects, so differences reported between Social and Partner conditions cannot be attributed to the experimental manipulation.","format":"text/plain"}
{"content":"The happiness decrease following negative partner outcomes under participant choice is driven by the participant's own lottery outcome rather than by the partner's, and so reflects self-directed disappointment rather than interpersonal guilt.","format":"text/plain"}
{"content":"The model-based fMRI approach does not recover known neural signals, so parametric modulators derived from the computational model — including the partner reward prediction error regressors — cannot be trusted.","format":"text/plain"}