Constraints and Theory Refinements — 2014–2024

These are not side notes. They are first-class nodes because they determine where the genealogy must stop, branch, or be redrawn.

C1 — No single low-arousal biomarker

The 2019 systematic review/meta-analysis of 66 studies (N=10,227) found task-related effects for some autonomic measures, but no case-control difference in heart rate or baseline measures overall. Correlational results were also measure-dependent.

Resolution: model autonomic findings as system-, task-, and design-dependent rather than as a unitary biomarker.

C2 — Eye-neglect is incomplete as an explanation

The 2020 young-adult variant study did not support a simple eye-region-specific deficit, while later child work did find reduced eye fixation under some CU/LPE operationalizations. The 2024 multi-method study found no high/low-CU difference in recognition accuracy regardless of measure.

Resolution: treat gaze allocation as one candidate component of emotional processing, not a sufficient explanation of CU-related emotional deficits.

C3 — CU specificity is inconsistent across developmental samples

Preschool work found atypical facial responding in children with conduct problems, but CP+CU and CP-only groups were not consistently separable on those reactions.

Resolution: distinguish conduct-problem-wide emotional-processing abnormalities from CU-specific effects.

C4 — Neural activity is not uniformly reduced

Some neural results fit reduced responsivity accounts, while other LPE/CU findings show increased prefrontal recruitment during socioemotional processing.

Resolution: code direction, task, region, developmental stage, and subtype for every neural result; prohibit generic 'underactivation' summaries.

C5 — Mechanism change is not clinical efficacy

cTBS can alter selected emotion-recognition outcomes; facial-mimicry instructions can alter recognition; parent-child interventions can alter sadness recognition or physiological response. None of those proximal changes alone establishes durable reduction in serious antisocial behavior.

Resolution: preserve separate inference gates for manipulation, proximal outcome, distal clinical outcome, durability, and real-world effectiveness.

C6 — Cross-age comparison is not developmental continuity

Child and young-adult samples can reveal developmental differences, but separate age samples do not prove within-person continuity.

Resolution: reserve 'developmental pathway' claims for longitudinal evidence; label cross-age comparisons as developmental contrasts.

C7 — Physiology can operate as susceptibility, not fixed deficit

Longitudinal work linking fear reactivity, parental distress, prosocial behavior, and conduct problems shows physiology can moderate environmental effects.

Resolution: retain a transactional pathway in parallel with trait/neural pathways.

HRV Biofeedback — LF/HF interpretation constraint (v0.3.5)

Petridou's dissertation treats decreases in LF/HF under stress as evidence of improved autonomic/emotional regulation. Modern HRV guidance cautions that LF/HF is not a direct measure of sympathovagal balance and LF is not a pure sympathetic index. Atlas consequence: encode the observed ratio change separately from the author's emotion-regulation interpretation.

HRV Biofeedback — endpoint/trait distinction (v0.3.5)

The dissertation's intervention results primarily concern physiological outcomes (SC and LF/HF) moderated by antisocial behavior, CU traits, and anxiety. The 2026 keynote's stronger statement that HRV biofeedback decreases CU traits is not established by the located dissertation results. This remains an open evidentiary gate.

HRV Biofeedback — design constraints (v0.3.5)

The intervention uses a waiting-list comparator and has no follow-up. Nonspecific intervention effects and durability are therefore unresolved. The dissertation also contains inconsistent recruitment/sample counts across abstract and methods; use N=110 pre/post randomized and N=106 final physiological sample for the intervention analysis unless the underlying dataset/paper clarifies otherwise.