{"id":"dde0c504-4bda-49d1-ade9-2a2333ca054a","arxiv_id":"2607.09782","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.5,"correctness_risk":"low","formal_verification":"none","parameter_count":2,"one_line_summary":"Discretized continuous-strategy human experiments confirm that observed angular-momentum manifolds are statistically significant linear combinations of theoretical eigenmanifolds from the Nash Jacobian.","lead":"Human continuous-strategy price and location games support the claim that steady-state dynamics are linear superpositions of eigenmanifolds built from Nash-equilibrium eigenvectors. The result extends a discrete-strategy framework to continuous strategy spaces via discretization and offers a unified spectral description of observed cycles.","discovery_kind":"extension","skeptic_critique":{"model":"grok-4.5","headline":"No significant objection identified beyond the Reader's already-flagged linearization-far-from-NE assumption.","rationale":"The central claim is an empirical statement about linear dependence between two concrete vectors (theoretical σ_k and experimental L-bar) under a fixed measurement protocol. Tables 5–8 document that dependence at extreme significance for every treatment and every bin count examined. The only substantive vulnerability is the one the Reader already named: whether the basis itself remains meaningful far from equilibrium. Because that vulnerability is already priced into the CONDITIONAL verdict, and because no stronger internal inconsistency or measurement artifact appears, the verdict and confidence level should stay exactly as the Reader set them. Public code and a continuum argument would still be desirable, but they are not required to accept the statistical pattern that is actually claimed.","tokens_in":20066,"tokens_out":443,"duration_ms":5303,"concrete_test":"Recompute the full-model F-tests of Table 5 after replacing the theoretical eigenmanifolds with those obtained from best-response or logit dynamics (instead of pure replicator) on the same N-bin grids; if the p-values remain <0.01 across the 60 cells, the linearization-plus-replicator assumption is not decisive for the headline claim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The Reader correctly isolates the softest point: that the eigenmanifold basis obtained from the Jacobian of replicator dynamics after uniform binning of a continuous strategy interval continues to span the observed angular-momentum vector even when the empirical distribution remains far from the Nash equilibrium for long stretches (§2.2.4). The paper's only defense is analogy to earlier discrete-strategy experiments; no continuum-limit argument or alternative-dynamics robustness check is supplied. That concern is real but already fully articulated by the Reader and does not, by itself, overturn the statistical pattern reported in Tables 5–8. No additional, independent load-bearing flaw (e.g., circularity in the definition of L-bar, multicollinearity artifact, or data-selection bias) is visible in the manuscript.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.5","summary":"The paper tests the eigenmanifold hypothesis for continuous-strategy evolutionary games: after uniform discretization of the strategy interval into N bins, the time-averaged experimental angular-momentum vector L-bar is a linear combination of theoretical eigenmanifold vectors σ_k constructed from the eigenvectors of the Jacobian of replicator dynamics at the Nash equilibrium (Eq. 1). Using six human-subject treatments from CFH2021 and CFGS2026 (plus a CF2003 illustration), the authors report that full-model F-tests reject the null for all 60 treatment-by-bin combinations (Table 5, p < 0.01, often far smaller), that the principal eigenmanifold (largest imaginary eigenvalue) is individually significant and has a large marginal contribution (Tables 6–7), and that the secondary eigenmanifold likewise contributes (Table 8). Heatmaps, velocity-field projections, and comparisons with discrete-strategy experiments are offered as supporting visualizations.","tokens_in":20256,"tokens_out":1116,"duration_ms":8911,"significance":"If the result holds, it supplies a concrete, falsifiable bridge between continuous-strategy laboratory data and the spectral structure of evolutionary dynamics, extending a program previously confined to discrete games. Strengths include transparent reuse of independent public data sets, an explicit and reproducible measurement protocol for experimental angular momentum, systematic robustness checks across ten bin sizes, and leave-one-manifold-out tests that quantify the contribution of the principal and secondary modes. The velocity-field and heatmap presentations also give a more microscopic description of price cycles than earlier qualitative reports. These features make the work a useful empirical contribution to evolutionary game theory and experimental economics, provided the linearization-after-discretization step is accepted or further justified.","major_comments":[{"comment":"§2.2.4 and the construction in §1.3.2–1.3.3: the central claim rests on the assertion that the eigenmanifold basis obtained from the Jacobian of replicator dynamics after uniform binning remains an adequate spanning set even when the empirical distribution stays far from the Nash equilibrium for long periods. The only defense offered is analogy to earlier discrete-strategy experiments. A continuum-limit argument, a quantitative measure of distance-to-equilibrium, or at least a robustness check under an alternative dynamics (best-response or logit) would substantially strengthen the load-bearing assumption; without it the statistical pattern in Tables 5–8 is suggestive but not fully conclusive for continuous strategy spaces.","section":null},{"comment":"Tables 5–8: sixty separate F-tests (and nested comparisons) are reported without multiple-testing correction or a hierarchical/mixed-effects model that treats bin_num as a repeated factor within treatment. While every individual p-value is tiny, a formal accounting for the multiplicity and for dependence across bin sizes is needed before the claim of “without exception” robustness can be taken at face value.","section":null},{"comment":"Eq. (1) and footnote 4: the regression fixes the coefficient of one member of each conjugate pair to zero to avoid multicollinearity. The manuscript should report the condition numbers of the design matrices (or variance-inflation factors) and confirm that the remaining columns remain linearly independent for large N; otherwise the reported F-statistics could be inflated by near-collinearity among the higher-order eigenmanifolds.","section":null}],"minor_comments":[{"comment":"Abstract and §1.1: several grammatical slips (“this hypotheses has been supported,” “supported in significant”) should be corrected.","section":null},{"comment":"Table 1 and Figures 5–6: the discrete-strategy comparison panels are helpful, but the figure captions should state explicitly which discrete experiments share the same N so that visual comparability is unambiguous.","section":null},{"comment":"§3.1.2, Eq. (10): the factor π in the eigencycle definition is conventional; a one-sentence remark on its origin (or a pointer to the earlier discrete papers) would aid readers new to the framework.","section":null},{"comment":"Figure 4 and Eq. (4)–(5): the theoretical velocity uses only the principal manifold while the experimental velocity uses the full L-bar; the text should note this asymmetry when claiming qualitative agreement.","section":null},{"comment":"References: several arXiv preprints and Chinese theses are cited; where journal versions exist they should be preferred, and DOIs should be supplied for the CFGS2026 and CFH2021 source papers.","section":null}],"recommendation":"major_revision","confidential_remarks":"The manuscript is part of a long-running program by the same group; the continuous-strategy data are independent, so novelty is real, but the self-citation density is high. The work is a natural fit for a journal that publishes experimental evolutionary game theory; the linearization-far-from-NE issue is the only point that, if left unaddressed, would keep me from recommending acceptance."},"author_rebuttal":null,"desk_editor":{"model":"grok-4.5","letter":"The new result is straightforward: after uniform binning of continuous price/location strategies, the observed angular-momentum vector L-bar is a statistically significant linear combination of the eigenmanifolds built from the Jacobian of replicator dynamics at the Nash equilibrium. Six independent treatments (CFH2021 discrete-time, CFGS2026 continuous-time, plus the older CF2003 posted-offer data) and ten bin sizes (N=10…100) all give full-model F-test p-values below 0.01, many far smaller; leave-one-out tests confirm both the principal and secondary manifolds contribute. That is the first systematic continuous-strategy confirmation of a claim the group had already established for discrete games.\n\nWhat works: the measurement protocol is transparent and reusable (Eqs. 10–16), the heatmaps and velocity-field projections make the cycle direction and active regions visible without extra fitting, and the data provenance is clean—public lab experiments, not re-analyzed private runs. Self-citation is heavy because the method is theirs, but the continuous data sets are independent, so the circularity burden is mild rather than load-bearing.\n\nSoft spots, in proportion: (1) the only defense of linearization far from NE is analogy to earlier discrete experiments (§2.2.4); no continuum-limit argument or alternative-dynamics check appears. (2) No multiple-testing correction across the 60 regressions and no public code. Neither overturns the pattern in Tables 5–8, but both should be fixed before anyone treats the continuum claim as settled.\n\nThis is for people who already care about spectral structure in evolutionary games or who run continuous-strategy experiments and want a uniform language that matches the discrete case. It is not a field-reorganizing paper, but it is honest computational work that closes an open question the authors themselves posed. I would send it to referees; the central claim is supported at a level that deserves serious scrutiny rather than desk rejection.","headline":"Solid empirical extension of the authors' eigenmanifold framework to continuous-strategy lab data; the stats hold up, the continuum/linearization defense is thin but already flagged.","tokens_in":20855,"tokens_out":518,"would_cite":true,"duration_ms":5730,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.5","headline":"Human continuous-strategy game dynamics are linear superpositions of eigenmanifolds fixed by the Nash Jacobian.","keywords":["eigenmanifold","evolutionary game dynamics","continuous strategy","Nash equilibrium","human-subject experiment","angular momentum","replicator dynamics","price cycles"],"falsifier":"Re-run any of the six continuous-strategy experiments, recompute the experimental manifold vector, and obtain a full-model F-test p-value that remains above 0.05 for fine discretizations (N≥50) or that fails to rise after the principal eigenmanifold is removed.","tokens_in":20948,"feed_emoji":"♻️","tokens_out":761,"duration_ms":6726,"temperature":0.7,"pith_summary":"After Nash equilibrium is located, the open question is how the game actually moves. This paper tests a structural hypothesis already checked on many discrete-strategy laboratory games: that the steady-state motion of a population is a linear combination of eigenmanifolds built from the eigenvectors of the Jacobian of the evolutionary dynamics at the Nash point. The authors take six continuous-strategy human experiments (price and location games, discrete and continuous time), discretize the strategy interval into N bins, compute the theoretical eigenmanifolds from the replicator Jacobian, measure the corresponding experimental manifold vectors from angular-momentum time series, and run ordinary-least-squares F-tests. Across every treatment and every discretization from N=10 to N=100 the full model is significant, the dominant (highest-frequency) eigenmanifold is both present and indispensable, and the secondary eigenmanifold also contributes. The same objects yield heat-map net-flow patterns and one- and two-dimensional velocity fields that recover the observed price cycles without extra free parameters. Continuous-strategy games therefore obey the same spectral organization that discrete games do, so the Nash equilibrium remains the organizing center of the dynamics rather than merely a static rest point.","feed_headline":"Continuous game cycles are linear eigenmanifold sums","feed_subtitle":"Human price and location experiments match spectral predictions from the Nash Jacobian","key_machinery":"The eigenmanifold vector σ_k: for each complex eigenvector of the Jacobian at Nash, the set of signed two-dimensional eigencycles σ^{mn}=π∥η_m∥∥η_n∥sin(arg(η_m)-arg(η_n)) ordered by subspace index. These theoretical vectors form a basis onto which the experimentally measured angular-momentum vector L-bar is regressed.","core_discovery":"The dynamic structure of continuous-strategy human-subject games is a statistically significant linear superposition of eigenmanifolds whose shapes are fixed solely by the eigenvector structure of the Jacobian of the game dynamics at the Nash equilibrium. Across six laboratory treatments and every discretization N=10…100 the full-model F-tests reject the null of no linear relation (all p<0.01, many p≪10^{-20}), the principal eigenmanifold is both significant and indispensable, and the secondary eigenmanifold retains a measurable marginal contribution.","pith_inferences":[],"forward_implications":[],"fun_headline_variants":["Human continuous games form linear eigenmanifold sums","Lab data confirm continuous games as Nash eigenmanifolds","Continuous strategy experiments validate eigenmanifold structure","Nash Jacobian eigenvectors shape human continuous game cycles","Eigenmanifolds explain continuous human-subject game dynamics"],"cache_read_input_tokens":16512,"weakest_assumption_plain":"Linearizing the dynamics about the Nash point after binning the continuous strategy interval still describes the observed motion even when the population stays far from equilibrium for long stretches.","fun_headline_variants_meta":{"raw":{"variants":["Human continuous games form linear eigenmanifold sums","Lab data confirm continuous games as Nash eigenmanifolds","Continuous strategy experiments validate eigenmanifold structure","Nash Jacobian eigenvectors shape human continuous game cycles","Eigenmanifolds explain continuous human-subject game dynamics"]},"model":"grok-4.5","effort":"low","cost_usd":0.00318,"raw_usage":{"total_tokens":986,"prompt_tokens":636,"num_sources_used":0,"completion_tokens":72,"cost_in_usd_ticks":31800000,"prompt_tokens_details":{"text_tokens":636,"audio_tokens":0,"image_tokens":0,"cached_tokens":128},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":278,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":636,"tokens_out":72,"duration_ms":3779,"temperature":1.0,"reasoning_tokens":278,"cache_read_input_tokens":128,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-07-14T15:48:57.205753+00:00","model_set":{"reader":"grok-4.5"},"falsifier":"Re-run any of the six continuous-strategy experiments, recompute the experimental manifold vector, and obtain a full-model F-test p-value that remains above 0.05 for fine discretizations (N≥50) or that fails to rise after the principal eigenmanifold is removed.","supporting_citations":[],"review_version":1}