{"schema_version":1,"experiment_id":"eoa_inverse_innovation_exp03_full320_20260801","cell_id":"invariant_mode_decomposition_design__archaeology_paleontology","trajectory_id":"R","attempt_index":0,"archetype_slug":"invariant_mode_decomposition_design","domain_slug":"archaeology_paleontology","decision":"CANDIDATE","problem_id":"coupled_taphonomic_distortion_of_recovered_assemblages","causal_lever_id":"mode_targeted_taphonomic_bias_correction","proposal":{"problem":"Recovered archaeological and paleontological assemblages can differ systematically from deposited assemblages because fragmentation, survival, displacement, recovery, and identification jointly transform combinations of taxa, materials, body parts, and condition states. Item-by-item corrections can therefore mistake coupled preservation patterns for past behavior or ecology.","actors_substrate":["Excavators and field paleontologists","Taphonomists and assemblage analysts","Museum and repository curators","Assemblages represented as counts or proportions across taxon, material, element, size, condition, and context states"],"observable_state":"Across contexts or experimental stages, multiple assemblage categories change together in repeatable patterns; conventional category-level corrections leave structured residuals and unstable inferred abundance rankings.","consequence":"Past behavior, community composition, mortality structure, or site formation is reconstructed from preservation-dominant patterns rather than the deposited signal.","affected_objective":"Produce defensible reconstructions of deposited assemblages while keeping unmodeled taphonomic distortion visible.","structural_mapping":[{"archetype_element":"Coupled transformation","domain_realization":"An estimated transition operator maps deposited or earlier-stage category vectors to later preserved-and-recovered category vectors.","claim_kind":"HYPOTHESIS"},{"archetype_element":"Invariant or approximately invariant directions","domain_realization":"Weighted combinations of assemblage categories recur as preservation, fragmentation, displacement, or recovery-response modes.","claim_kind":"HYPOTHESIS"},{"archetype_element":"Scalar modal response","domain_realization":"Each mode receives an estimated persistence, attenuation, or amplification value under the scoped transition.","claim_kind":"INFERENCE"},{"archetype_element":"Action-relevant mode selection","domain_realization":"Modes are retained according to held-out reconstruction benefit and their effect on substantive abundance estimates, not variance alone.","claim_kind":"INFERENCE"},{"archetype_element":"Residual visibility","domain_realization":"Observed assemblages are reconstructed through retained modes, with structured leftovers reported by category and context.","claim_kind":"INFERENCE"},{"archetype_element":"Local validity","domain_realization":"Modes are trusted only for comparable depositional, burial, exposure, recovery, and identification regimes.","claim_kind":"INFERENCE"}],"component_map":[{"component":"Transformation Scope","status":"adapted","domain_realization":"Bound the operator to specified materials, contexts, taphonomic stages, recovery methods, and time intervals."},{"component":"State-Vector Definition","status":"direct","domain_realization":"Define harmonized taxon/material/element/size/condition/context categories, including loss or unobserved states where estimable."},{"component":"Invariant Mode Basis","status":"adapted","domain_realization":"Estimate approximate preservation-response directions from the scoped transition operator."},{"component":"Modal Gain Spectrum","status":"direct","domain_realization":"Estimate and uncertainty-bound each mode's attenuation, persistence, or amplification."},{"component":"Dominant Mode Selection Rule","status":"direct","domain_realization":"Retain modes that improve held-out reconstruction and materially affect target interpretations."},{"component":"Stable/Unstable Mode Partition","status":"adapted","domain_realization":"Classify modes as reliably persistent, attenuated, amplifying, or indeterminate within the observed stage window."},{"component":"Modal Intervention Map","status":"adapted","domain_realization":"Map candidate corrections, sampling changes, and robustness analyses to the modes they alter."},{"component":"Reconstruction Residual Check","status":"direct","domain_realization":"Reconstruct held-out assemblages and inspect error magnitude and structure."},{"component":"Mode Drift Monitor","status":"direct","domain_realization":"Re-estimate directions and gains when context, recovery protocol, or condition distributions change."},{"component":"Interpretation Scope Contract","status":"direct","domain_realization":"State that modes describe scoped taphonomic transformations and are not automatically behavioral or ecological causes."},{"component":"Mode-Coupling Register","status":"direct","domain_realization":"Record near-degenerate, non-orthogonal, and intervention-linked modes that cannot be interpreted independently."},{"component":"Local Linearization Window","status":"adapted","domain_realization":"Specify the ranges of assemblage composition and taphonomic conditions over which the operator approximates observed transitions."},{"component":"Spectral Gap Threshold","status":"direct","domain_realization":"Require uncertainty-adjusted separation between retained and excluded modes before using a reduced correction."}],"mechanism_dispositions":[{"slug":"eigendecomposition_workflow","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"A complete eigenbasis is not assured for an estimated, potentially non-normal or rectangular taphonomic map; treating it as exact would overstate identifiability.","counterfactual_removal":"No material change because SVD supplies the bounded decomposition."},{"slug":"modal_sensitivity_sweep","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Perturb modal gains and correction weights to rank which modes materially change substantive reconstructions and expose cross-effects.","counterfactual_removal":"Without it, retained modes would be ranked by numerical prominence rather than inferential leverage."},{"slug":"modal_stability_analysis","disposition":"selected_supporting","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Adapt stability labels to persistence, attenuation, amplification, or indeterminacy across observed taphonomic stages and attach the validity window.","counterfactual_removal":"Analysts could apply locally estimated modal behavior beyond supported stages."},{"slug":"mode_shape_testing","disposition":"incompatible","contribution_type":"NONE","adaptation_or_rejection":"Mechanical excitation and sensor-response mode recovery do not correspond to historical assemblage formation.","counterfactual_removal":"No change."},{"slug":"network_spectral_centrality_analysis","disposition":"incompatible","contribution_type":"NONE","adaptation_or_rejection":"Node prestige in a connectivity graph does not estimate preservation transformation or deposited composition.","counterfactual_removal":"No change."},{"slug":"power_iteration_probe","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"A dominant-only approximation would conceal secondary modes and is unreliable when the retained boundary has a small gap.","counterfactual_removal":"No change because a multi-mode factorization is required."},{"slug":"principal_component_analysis","disposition":"considered_rejected","contribution_type":"NONE","adaptation_or_rejection":"Variance directions in observed assemblages need not represent deposition-to-recovery transformations.","counterfactual_removal":"No change; avoiding it preserves the process-based interpretation."},{"slug":"reduced_order_model","disposition":"selected_supporting","contribution_type":"OPERATIONAL","adaptation_or_rejection":"Create a bounded surrogate that propagates candidate deposited vectors through retained taphonomic modes for rapid correction and sensitivity checks.","counterfactual_removal":"The decomposition remains explanatory, but routine scenario comparison becomes less practical."},{"slug":"residual_reconstruction_test","disposition":"selected_load_bearing","contribution_type":"TEST_DESIGN","adaptation_or_rejection":"Choose mode count using blinded reconstruction error plus residual structure and consequence-specific tolerances.","counterfactual_removal":"There would be no empirical stopping rule or safeguard against discarding meaningful low-prominence distortion."},{"slug":"singular_value_decomposition","disposition":"selected_load_bearing","contribution_type":"CORE_CAUSAL","adaptation_or_rejection":"Factor the estimated deposition-to-recovery mapping into paired input and observed-assemblage directions with ordered gains; interpret them as approximate response modes, not exact dynamical eigenmodes.","counterfactual_removal":"The proposal loses its coupled transformation basis and collapses to category-wise correction."},{"slug":"spectral_decomposition_report","disposition":"selected_supporting","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Document mode meanings, uncertainty, coupling, residuals, and prohibited behavioral or ecological interpretations.","counterfactual_removal":"Results become easier to reify as causal accounts of the past."},{"slug":"spectral_gap_monitor","disposition":"selected_supporting","contribution_type":"SAFETY_GUARDRAIL","adaptation_or_rejection":"Track retained-boundary separation and basis rotation across contexts or protocol changes.","counterfactual_removal":"A previously valid reduction could continue unnoticed after its modal separation erodes."}],"causal_chain":["Coupled taphonomic and recovery processes transform deposited category combinations rather than isolated categories.","A scoped empirical transition operator represents those joint changes within a bounded regime.","SVD exposes paired response modes and gains hidden by raw category comparisons.","Held-out residual tests and sensitivity sweeps retain modes that both reconstruct known inputs and affect target interpretations.","Corrections and robustness analyses targeted to those modes reduce preservation-driven distortion while residual and drift checks expose failure."],"baseline":"Compare recovered counts or proportions directly, with category-specific preservation adjustments and qualitative taphonomic discussion.","nearest_rival":"A hierarchical category-level recovery model with taxon, material, context, and protocol covariates but no modal decomposition of their joint transition.","authority_safety":{"affected_parties":["Researchers whose interpretations change","Field and collections staff whose sampling practices may change","Repositories responsible for specimens and records","Descendant, Indigenous, and other communities affected by archaeological interpretation"],"decision_authority":"The project research lead and responsible repository may authorize an analytical pilot; any sampling or interpretive change remains subject to existing heritage, repository, and community governance.","authorized_first_step":"Run a preregistered, read-only pilot on one actualistic or otherwise known-input dataset and one held-out context; compare baseline, nearest rival, and modal method using reconstruction error, residual structure, and recovery of prespecified abundance contrasts.","excluded_actions":["Destructive sampling","Discarding or deprioritizing specimens","Changing excavation strategy","Publishing revised cultural or ecological claims","Extrapolating beyond represented taphonomic regimes"],"halt_rollback":"Stop operational use if held-out error exceeds the nearest rival, residuals remain structured above tolerance, the retained spectral gap overlaps zero under uncertainty, modes drift beyond threshold, or stakeholder governance is absent. Revert to the baseline interpretation and retain the modal output only as exploratory analysis."}},"negative_tests":{"strongest_counterevidence":"Taphonomic histories may be path-dependent, nonlinear, poorly observed, and site-specific, so no transportable local operator or identifiable modal basis may exist.","analogy_break":"Unlike a repeatedly observed engineered transformation, the deposited state is usually unobserved and the historical process cannot be rerun; fitted modes may encode sampling design or assemblage covariance rather than taphonomic dynamics.","failure_condition":"The approach fails if modes are unstable across resamples or contexts, gains are not identifiable, no uncertainty-adjusted retained-mode gap exists, or structured residuals concentrate in consequence-relevant categories.","problem_falsifier":"The proposed problem is falsified for the scoped setting if category-independent recovery models predict held-out known inputs as well as coupled models and no repeatable cross-category residual structure remains.","intervention_falsifier":"The modal intervention is falsified if, on blinded known-input cases, it does not improve reconstruction and prespecified abundance-contrast recovery over both the ordinary baseline and hierarchical category-level rival within stated uncertainty.","risks":["Reifying descriptive modes as distinct historical processes","Correcting away genuine behavioral or ecological structure","Instability from sparse categories or compositional constraints","False confidence from in-sample reconstruction","Excluding rare but substantively important categories","Extending modes across incompatible sites, protocols, or preservation regimes"]},"null_rationale":null,"classification":{"candidate_kind":"MECHANISM_ADAPTATION","prior_art_status":"UNSEARCHED","evidence_maturity":"HYPOTHESIS"},"revision_change_log":{"revision_kind":"ORIGINAL","prior_problem_id":null,"prior_causal_lever_id":null,"problem_changed":false,"causal_lever_changed":false,"conceptual_changes":[],"operational_changes":[],"repairs_addressed":[]},"confidence":0.78,"generator_notes":"Closed-book structural inference from the supplied packet. Empirical effectiveness, operator identifiability, and transportability are unverified."}