Cultural transmission mechanism classification

Telling states apart9

• Ordinary transformations may explain retelling chains without an extra recursive state

  Where this hypothesis actsHuman–AI retelling chains evaluated on independent seeds and held-out contexts

  What is proposedDetermine whether recursive mechanisms are distinguishable from compositional nulls

  With whatInstrument or assay

  HowFreeze calibrated one-step and reconstruction nulls; test held-out predictions and resource-matched ancestry-access and source-regeneration interventions

  Possible resultPossible reclassification of distinct recursion as ordinary composed transformations, retaining observed attractors

  2026-10-05

• Continuing evaluation with coarse checks may reward strategic omissions in retelling

  Where this hypothesis actsHuman–AI retelling chains under continuing or one-shot evaluation and coarse or proposition-specific verification

  What is proposedDistinguish a relational contract mechanism from composed one-step editing mechanisms

  With whatInstrument or assay

  HowCross evaluation continuity with verification specificity; compare held-out effects with independently calibrated one-shot kernels composed across rounds

  Possible resultPossible evidence for an extra contractual state if continuation-by-verifiability effects exceed component predictions

  2026-10-05

• Inherited withdrawal records may block stale copies while allowing independent support

  Where this hypothesis actsMatched cultural transmission experiments with source-specific withdrawals and repeated assertions

  What is proposedDistinguish inherited revocation-state effects from established source-monitoring effects

  With whatInstrument or assay

  HowMatch final content and ordinary source information, vary which occurrence a withdrawal targets, and compare stale retransmission with independent support

  Possible resultPossible identification or rejection of inherited revocation state as a distinct cultural transmission mechanism

  2026-10-05

• Overridden revision rights may make accurate model corrections provoke deliberate errors

  Where this hypothesis actsRecurring human–AI retelling chains with a history of exercised revision rights

  What is proposedDistinguish a proposed revision-rights mechanism from established explanatory components

  With whatInstrument or assay

  HowManipulate revision authority and correction accuracy under matched conditions; compare predictions with independently calibrated established components on held-out lineages

  Possible resultPossible rejection of a distinct mechanism if established components predict the authority-by-history contrast

  2026-10-05

• Cultural transmission may require six distinct pathways that cannot be merged

  Where this hypothesis actsHeld-out cultural transmission experiments with localized removals and selective rescue

  What is proposedDistinguish independently necessary mechanism families and merge equivalent mechanisms

  With whatInstrument or assay

  HowUse timed removals, selective rescue and intermediate-state assays; test whether reduced nonlinear kernels predict held-out isolated and combined interventions

  Possible resultExpected six necessary families, with fewer retained if merged models predict the same intervention effects

  2026-10-05

• Weak disturbances of inherited material memory may preserve multiple cultural practices

  Where this hypothesis actsPractice transmission through inherited working objects across novice generations

  What is proposedDistinguish the proposed material-coupling family from established alternatives

  With whatInstrument or assay

  HowCross inherited versus reset workpieces with demonstrations, cohort turnover and perturbation; compare outcomes with a calibrated physical-memory plus reconstruction baseline

  Possible resultPossible admission of F_MATERIAL if its added predictive gain exceeds the preregistered margin

  2026-10-05

• Relations among contributions may carry meaning through complete learner replacement

  Where this hypothesis actsRole-split cultural transmission across controlled regrouping and complete learner turnover

  What is proposedDistinguish endogenous distributed-correlation inheritance from established alternatives

  With whatInstrument or assay

  HowSwap partner contributions while preserving marginals and baseline-modeled correlations; reinstate coupling and test regeneration after complete cohort replacement

  Possible resultPossible recognition of a distinct family if established alternatives miss selective loss and recovery of joint meaning

  2026-10-05

• One fixed response to meaning loss may govern cultural reconstruction and choice

  Where this hypothesis actsA bounded panel of held-out cultural transmission experiments with matched loss and source information

  What is proposedDetermine whether selection and reconstruction constitute one causal hypothesis family

  With whatInstrument or assay

  HowCalibrate beta on isolated reconstruction trials, predict held-out choices without refitting, and experimentally remove the source-loss dependency

  Possible resultPossible support for one family if the shared law predicts every held-out block and dependency removal causes failure

  2026-10-05

• Coupled content and source-validation routes may determine cultural-lineage survival

  Where this hypothesis actsControlled online relays with separately routed content and truthful source metadata

  What is proposedDistinguish coupled content-and-provenance dependence from established transmission models

  With whatInstrument or assay

  HowRandomize content and provenance routing in matched graphs; compare held-out predictions with established models, calibrating route attention and source comprehension

  Possible resultPossible retention of F_DUAL_PATH as a distinct family if established models fail matched intervention tests

  2026-10-05