{"id":"594de5a4-5069-4dee-ac3b-da912591245e","arxiv_id":"2411.17415","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper characterizes the provable Pi^1_e, Sigma^1_e, and Boolean-combination classes of the strong dependent choice system Sigma^1_i-SDC0 using beta-model reflection principles.","lead":"This paper identifies exactly which formulas of a given logical complexity are provable from a family of choice axioms in second-order arithmetic. It gives reflection-style axiom schemes that characterize these fragments, extending prior work on strong dependent choice systems.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Section 4 Claim 4.2.2 relies on an unjustified submodel property: the constructed chain does not make A0 a βi-submodel of M′, and the needed upward transfer of Σ1_{e+1} truth is not proved.","rationale":"I read the paper as giving exact proof-theoretic characterizations via chains of coded β-models. The reader's weakest assumption was Lemma 2.12; I checked that lemma and found its induction step to be compressed but repairable: statements the proof calls 'arithmetical' are actually Π1_i, but the β_i-submodel hypothesis supplies the needed transfer after Skolemizing witnesses. The more serious gap is in Section 4, Claim 4.2.2, which asserts a submodel relation that the construction does not deliver. The intended conclusion may still be correct, and there is a plausible chain-based absoluteness argument that would fix it, but that argument is absent. Because the gap is in the proof of the main general characterization rather than in an auxiliary remark, the manuscript should not be accepted without this detail being supplied. The reader's CONDITIONAL verdict is therefore appropriate, though the specific load-bearing point differs from the one emphasized by the reader.","tokens_in":12394,"tokens_out":35163,"duration_ms":315279,"concrete_test":"Write out and verify the missing chain-transfer lemma: if A0 ∈ A1 ∈ ⋯ are coded models with A_{k+1} |= β_i(A_k) for all k, M′ is the union of the elements of the A_k, and e < i, prove by induction on formula complexity that every Σ1_{e+1} sentence with parameters in A0 that holds in A0 also holds in M′. Check the induction step for Π1_e formulas: it must use that truth of Π1_e formulas propagates from A_k to A_{k+1} because A_{k+1} thinks A_k is β_i. If the step instead requires A_k to be a β_i-submodel of A_{k+1}, the direction of the chain is wrong and Claim 4.2.2 cannot be repaired as stated. A successful formal proof would settle Theorem 4.2; a counterexample to the transfer lemma would refute the current proof.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"Theorem 4.2, the characterization of B(Π1_{e+1}) consequences of Σ1_i-SDC0, rests on the compactness argument in its proof. The final step, Claim 4.2.2, states that M′ satisfies ¬π because “¬π is a Σ1_{e+1} sentence true in A0 and A0 is a βi-submodel of M′.” This submodel property does not follow from the construction. The theory T′ only gives A0 ∈ A1 and A1 |= β_i(A0), and more generally A_n ∈ A_{n+1} with A_{n+1} |= β_i(A_n). From this one can show that the elements of A_n are contained in A_{n+1}, so the A_n are nested, but it does not follow that A0 is a βi-submodel of M′ = ⋃_n (elements of A_n): a βi-submodel requires equivalence for all Σ1_i formulas with parameters in A0, and witnesses in M′ may lie outside A0. What the proof actually needs is a one-way upward transfer: a Σ1_{e+1} formula true in A0 should remain true in the union. Such a transfer is plausible by propagating Π1_e truth along the chain, since each A_{k+1} is a β_i-model from the perspective of its successor, but the paper does not supply this argument. As written, the step is a gap in the central Theorem 4.2, and the same compressed reasoning is used in Theorem 4.3. If the upward transfer cannot be proved from the chain conditions, the compactness argument for the general e < i case collapses.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the sets of Π1_e-, Σ1_e-, and B(Π1_e)-consequences of the theory Σ1_i-SDC0 of strong dependent choice for Σ1_i formulas. It claims exact characterizations in terms of reflection principles over chains of coded β-models: Theorem 2.14 treats Π1_{e+2}-consequences for i>e, Theorem 3.7 treats B(Π1_{i+1})-consequences, and Theorems 4.2 and 4.3 treat B(Π1_{e+1})- and Σ1_{e+1}-consequences for e<i. The proofs use compactness/Barwise-Schlipf constructions with chains of β_i-models, together with a hierarchy of reflection schemata β^i_eRFN and β^i_eRFN^-.","tokens_in":12727,"tokens_out":49356,"duration_ms":413945,"significance":"If correct, these are sharp, elegant characterizations of the provable fragments of Σ1_i-SDC0, extending the authors' earlier work on the Π1_2-consequences of Π1_1-CA0 and connecting the subject to local reflection principles in the style of Beklemishev. The Section 2 argument is informative and the paper is explicit about the provenance of Theorem 2.14. However, the proofs of the central results in Sections 3 and 4 contain two nontrivial transfer steps that are not justified as written; hence the results should be regarded as not yet established.","major_comments":[{"comment":"The proof of Lemma 3.5 contains a load-bearing unsupported inference. After choosing a β_i-model Y with Y |= β^i_iRFN^-(φ;n), the proof states that the condition in the square brackets, which includes Y0 |= φ, \"is Π1_i\" and therefore holds in the ambient model. This is false for φ ∈ Σ1_{i+1}: the formula Y0 |= φ is Σ1_{i+1}, not Π1_i. Consequently, the conclusion that Y0,...,Y_n,Y_{n+1} witness β^i_iRFN^-(φ;n+1) in the ambient does not follow from Y |= β^i_iRFN^-(φ;n). Since Lemma 3.5 is used in Lemma 3.6 and hence in Theorem 3.7, this gap affects the central claim of Section 3. Please supply a correct proof, for example by constructing the chain inside a β_i-model and transferring only the Π1_i matrix via β_i-submodel absoluteness.","section":"Section 3, Lemma 3.5"},{"comment":"The assertion that A0 is a β_i-submodel of M′ is not established by the construction. The theory T′ only yields A_{n+1} |= β_i(A_n) for each n, and Lemma 2.12 does not apply, since its hypothesis requires M |= β_i(A_n) for the relevant A_n. Therefore the upward transfer of the Σ1_{e+1} sentence ¬π from A0 to M′ is unproved. The same compressed step is used in Theorem 4.3. A lemma is needed showing that Π1_e truth propagates from A0 along the chain and then to the union M′, or otherwise that ¬π is true in M′. As written, this is a gap in the central Theorem 4.2.","section":"Section 4, Proof of Theorem 4.2, Claim 4.2.2"}],"minor_comments":[{"comment":"In the statement of Claim 4.2.1, \"σ∨τ\" should presumably read \"σ∨π\", and the subscript \"A_{k+e+1}\" should presumably be \"A_{k+n+1}\".","section":"Section 4, Claim 4.2.1"},{"comment":"The heading \"Introdcution\" contains a typo.","section":"Section 1 heading"},{"comment":"The base case i=1 is delegated to [4] without stating the exact result being invoked; please include the precise statement or a proof, since this lemma is central to the compactness arguments.","section":"Section 2, Lemma 2.12"},{"comment":"When applying the induction hypothesis to β^i_iRFN^-(φ;n), the proof says \"also holds by the induction hypothesis\", but the displayed implication is the n=0 case, not the induction hypothesis for the current n; please rephrase for clarity.","section":"Section 3, Lemma 3.5 proof"},{"comment":"The step \"Mk+1 satisfies ACA0 ∧ X ∈ M0 ∧ β^i_i(M0,...,Mk)\" is terse; a brief remark that the parameters M0,...,Mk and X lie in Mk+1 would improve readability.","section":"Section 2, Proof of Theorem 2.14"}],"recommendation":"major_revision","confidential_remarks":"The two gaps identified in the major comments are probably repairable, but as they stand the main theorems of Sections 3 and 4 are not fully proved. I have not been able to verify the base case of Lemma 2.12 from [4] independently; the authors should make that dependence explicit and ideally self-contained. The paper fits the journal's scope, and the results would be significant if the gaps are closed."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is a real step forward in the program on Pi^1_2 consequences of Pi^1_1-CA0. It generalizes the earlier work [4] from the single case to arbitrary i, and adds Boolean combinations and Sigma consequences. The paper is mostly clean, but Section 4 has a genuine gap in the final step of Theorem 4.2 that the authors need to address.\n\nThe genuinely new material is concentrated in Sections 3 and 4. Theorem 3.7, equating the B(Pi^1_{i+1}) consequences of Sigma^1_i-SDC0 with the reflection scheme beta_iRfn(Sigma^1_{i+1})_0, is a nice result and the proof via Lemma 3.6 is clear. The conservativity result in Theorem 3.10, with the corollary about iterated hyperjumps and determinacy, is also valuable. And Lemma 2.13, which gives a computability-theoretic strengthening of the known Theorem 2.14, is a useful tool.\n\nThe problem is Claim 4.2.2 in the proof of Theorem 4.2. The proof says that since not-pi is a Sigma^1_{e+1} sentence true in A0 and A0 is a beta_i-submodel of M', M' also satisfies not-pi. That submodel property does not follow from the construction. The chain gives A0 in A1 and A1 |= beta_i(A0), and more generally A_n in A_{n+1} with A_{n+1} |= beta_i(A_n). That makes A0 a beta_i-model from the perspective of A1, not necessarily from the perspective of the union M'. The proof would need an upward transfer of Sigma^1_{e+1} truth from A0 to M'. That is plausibly provable by propagating Pi^1_e truth along the chain, but the paper does not supply the argument. As written, the step is a gap in the central theorem of Section 4, and the same compressed reasoning appears in Theorem 4.3. There are also some typos: tau appears where pi is meant, and the index A_{k+e+1} looks like it should be A_{k+n+1}. These are minor individually, but they do not help.\n\nThe stress-test note is accurate: the chain does not make A0 a beta_i-submodel of M'.\n\nWho is this for? People working in reverse mathematics and the proof theory of second-order arithmetic. The results are plausible and likely correct, but Section 4 is not ready in its current form. The paper deserves a serious referee because the core ideas are good, the gap is localized, and the rest of the paper is careful. I would send it to review, with the expectation that the referees will ask the authors to fix Section 4.","headline":"Genuine progress on consequence classes of Sigma^1_i-SDC0, but Section 4's main proof has a gap that needs fixing.","tokens_in":119,"tokens_out":11627,"would_cite":true,"duration_ms":218221,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["03F35","03B30","03E15"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper proves that the $\\Pi^1_{e+2}$, $\\Sigma^1_{e+1}$, and Boolean-combination consequences of $\\Sigma^1_i$-$\\mathsf{SDC}_0$ are exactly the sentences provable from specific $\\beta$-model reflection schemes, so the provable fragments…","keywords":["strong dependent choice","beta-models","reflection principles","reverse mathematics","second-order arithmetic","conservativity","Pi-1-1-CA0","Boolean combinations"],"falsifier":"A concrete falsifier is to construct a model $\\mathcal{M}$ of $\\mathsf{ACA}_0$ and a subcollection $\\mathcal{S}'$ closed under $\\mathcal{M}$'s $\\beta_{i+1}$-models such that the induced $\\omega$-submodel is not a $\\beta_{i+1}$-submodel of $\\mathcal{M}$ and does not satisfy $\\Sigma^1_{i+1}$-$\\mathsf{SDC}_0$; since the preservation lemma is the load-bearing step in the compactness arguments behind the characterizations, such a counterexample would refute the paper's central theorems.","tokens_in":14,"feed_emoji":"🔁","tokens_out":20791,"duration_ms":215173,"temperature":0.7,"pith_summary":"This paper aims to identify exactly which sentences of limited logical complexity are provable from the theory $\\Sigma^1_i$-$\\mathsf{SDC}_0$, the strong dependent choice scheme for $\\Sigma^1_i$ formulas in second-order arithmetic. The claimed answer is that each fragment is governed by a reflection principle built from finite chains of $\\beta_i$-models: the $\\Pi^1_{e+2}$ consequences are exactly the theorems of $\\mathsf{ACA}_0$ plus the chain schemes $\\beta^i_e\\mathrm{RFN}(n)$, the Boolean combinations of $\\Pi^1_{i+1}$ consequences are exactly the theorems of the parameter-free reflection scheme $\\beta_i\\mathrm{Rfn}(\\Sigma^1_{i+1})_0$, and for $e<i$ the $\\mathsf{B}(\\Pi^1_{e+1})$ and $\\Sigma^1_{e+1}$ fragments are captured by the corresponding schemes $\\beta^i_e\\mathrm{RFN}^-$. The significance is that the result would show that the apparent extra strength of strong dependent choice over these fragments is an illusion: every provable sentence in these classes follows from reflection alone, with no hidden axioms. It also provides a precise reverse-mathematical calibration of a theory that is not finitely axiomatizable.","feed_headline":"Dependent choice's provable fragments are exactly beta-model reflection","feed_subtitle":"The Pi-e, Sigma-e, and Boolean consequence classes coincide with reflection along finite chains of beta-models.","key_machinery":"The central objects are coded $\\beta_i$-models: countable $\\omega$-models that agree with the ambient universe on all $\\Sigma^1_i$ formulas, equivalently that reflect $\\Pi^1_i$ truths. The machinery is the finite chain $\\beta^i_e(M_0,\\dots,M_n)$, which requires $M_0\\in M_1\\in\\cdots\\in M_n$, makes each earlier model a $\\beta_i$-model according to later members of the chain, and adds a $\\beta_e$-condition on the last model; the corresponding reflection sentences $\\beta^i_e\\mathrm{RFN}(\\sigma;n;\\tau)$ assert that every set belongs to the first model of such a chain with prescribed sentences $\\sigma$ and $\\tau$ holding at the endpoints. The load-bearing step is a preservation lemma: a subuniverse closed under finding $\\beta_i$-models is itself a $\\beta_i$-submodel and a model of $\\Sigma^1_i$-$\\mathsf{SDC}_0$. This lemma is what lets the compactness argument convert a failure of provability into a model of $\\Sigma^1_i$-$\\mathsf{SDC}_0$ with the target sentence false.","core_discovery":"Working over $\\mathsf{ACA}_0$, the paper proves that the $\\mathsf{B}(\\Pi^1_{i+1})$ sentences provable from $\\Sigma^1_i$-$\\mathsf{SDC}_0$ are exactly those provable from $\\beta_i\\mathrm{Rfn}(\\Sigma^1_{i+1})_0$, whose axioms are all instances of $\\sigma \\to \\exists M(\\beta_i(M) \\land M\\models\\sigma)$ for $\\Sigma^1_{i+1}$ sentences $\\sigma$, where $\\beta_i(M)$ says that $M$ is a coded $\\beta_i$-model: an $\\omega$-model satisfying the same $\\Sigma^1_i$ formulas as the ambient universe. It also proves that for $e<i$ the $\\Pi^1_{e+2}$ consequences are exactly the theorems of $\\mathsf{ACA}_0$ plus the chain reflection schemes $\\beta^i_e\\mathrm{RFN}(n)$, and that the $\\mathsf{B}(\\Pi^1_{e+1})$ and $\\Sigma^1_{e+1}$ consequences are captured by the corresponding $\\beta^i_e\\mathrm{RFN}^-$ schemes. The direction from reflection to strong dependent choice is immediate from the known $\\beta_i$-model reflection characterization of $\\Sigma^1_i$-$\\mathsf{SDC}_0$; the converse is the main work, carried out by a compactness argument that builds a $\\beta_i$-submodel of a model of the reflection scheme and shows it satisfies $\\Sigma^1_i$-$\\mathsf{SDC}_0$.","pith_inferences":["A natural next step, not pursued here, is to apply the same chain-extraction compactness technique to other choice schemes such as $\\Sigma^1_i$-$\\mathsf{DC}$ or transfinite dependent choice, where the shape of the $\\beta$-model chain would likely determine the exact consequence classes.","The strict inclusions among finite reflection levels suggest that the number of $\\beta$-model rungs needed to prove a sentence could serve as a proof-theoretic rank measuring how far a sentence sits above the reflection base, analogous to ordinal ranks in first-order reflection hierarchies.","For $i=1$, the paper's identification of the $\\Sigma^1_2$ consequences with iterated hyperjumps hints that the higher-$i$ hierarchies may correspond to iterated $\\beta_i$-jumps, though no such correspondence is proved for $i>1$.","A testable extension is whether the parameter-free characterizations of Section 3 still hold when the ambient base theory is weakened below $\\mathsf{ACA}_0$; the present proofs work over $\\mathsf{ACA}_0$, and the exact strength needed for the compactness construction is not isolated."],"forward_implications":["Every $\\Pi^1_{e+2}$ sentence provable from $\\Sigma^1_i$-$\\mathsf{SDC}_0$ is already provable from $\\mathsf{ACA}_0$ plus the chain reflection schemes $\\beta^i_e\\mathrm{RFN}(n)$, so the $\\Pi^1_{e+2}$ fragment of strong dependent choice is exactly the union of those schemes.","The $\\mathsf{B}(\\Pi^1_{i+1})$ fragment is captured by the parameter-free reflection scheme $\\beta_i\\mathrm{Rfn}(\\Sigma^1_{i+1})_0$, which is not finitely axiomatizable.","For $e<i$, the $\\Sigma^1_{e+1}$ consequences are conservative over $\\mathsf{ACA}_0$ plus $\\beta^i_e\\mathrm{RFN}^-(n)$; in the case $i=1,e=0$, this says the $\\Sigma^1_2$ consequences of $\\Pi^1_1$-$\\mathsf{CA}_0$ coincide with iterated hyperjumps and with the determinacy levels $(\\Sigma^{0,\\emptyset}_1)^n$-$\\mathsf{Det}$.","The finite levels of the reflection hierarchy are strictly increasing: $\\beta^i_i\\mathrm{RFN}^-(n)$ is properly $\\Pi^1_{e+2}$-weaker than $\\beta^i_e\\mathrm{RFN}(n+1)$, so each additional rung of the $\\beta$-model chain proves new $\\Pi^1_{e+2}$ sentences.","Because $\\Sigma^1_i$-$\\mathsf{SDC}_0$ is $\\Pi^1_4$-conservative over $\\Pi^1_i$-$\\mathsf{CA}_0$, the same characterizations transfer to $\\Pi^1_i$-$\\mathsf{CA}_0$ for $e=0,1,2$."],"supporting_citations":[{"why":"Supplies the definition of $\\Sigma^1_i$-$\\mathsf{SDC}_0$, the $\\beta_i$-model reflection theorem used throughout, and the $\\Pi^1_4$-conservativity over $\\Pi^1_i$-$\\mathsf{CA}_0$.","marker":"[3]"},{"why":"Proves the base case $i=1$ of the preservation lemma and introduces the chain reflection formulas $\\beta^i_e\\mathrm{RFN}(n;\\tau)$ that the present paper extends.","marker":"[4]"},{"why":"Introduces the reflection sentences $\\beta^i_e\\mathrm{RFN}(n)$ as $\\psi_e(i,n)$ and proves a syntactic reflection equivalence that Theorem 2.14 generalizes to the standard $\\beta$-model setting.","marker":"[2]"}],"fun_headline_variants":["SDC consequences equal beta-model reflection","Strong dependent choice collapses to beta-model reflection","Beta-model reflection captures SDC's provable truths","Reflection schemes characterize SDC's consequences","Finite beta-model chains axiomatize SDC's consequences"],"cache_read_input_tokens":15360,"weakest_assumption_plain":"The whole characterization rests on a preservation lemma: any collection of sets closed under passing to $\\beta_i$-models is itself a $\\beta_i$-submodel and satisfies strong dependent choice; the induction proving that lemma assumes truth of a $\\Pi^1_i$ formula inside a coded model is absolute between the subcollection and the ambient universe, and the base case is imported from an earlier paper rather than proved here.","fun_headline_variants_meta":{"raw":{"variants":["SDC consequences equal beta-model reflection","Strong dependent choice collapses to beta-model reflection","Beta-model reflection captures SDC's provable truths","Reflection schemes characterize SDC's consequences","Finite beta-model chains axiomatize SDC's consequences"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001082,"raw_usage":{"total_tokens":4528,"prompt_tokens":950,"completion_tokens":3578,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":566,"completion_tokens_details":{"reasoning_tokens":3507}},"tokens_in":566,"tokens_out":3578,"duration_ms":22852,"temperature":1.0,"reasoning_tokens":3507,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T12:11:04.448403+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete falsifier is to construct a model $\\mathcal{M}$ of $\\mathsf{ACA}_0$ and a subcollection $\\mathcal{S}'$ closed under $\\mathcal{M}$'s $\\beta_{i+1}$-models such that the induced $\\omega$-submodel is not a $\\beta_{i+1}$-submodel of $\\mathcal{M}$ and does not satisfy $\\Sigma^1_{i+1}$-$\\mathsf{SDC}_0$; since the preservation lemma is the load-bearing step in the compactness arguments behind the characterizations, such a counterexample would refute the paper's central theorems.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the definition of $\\Sigma^1_i$-$\\mathsf{SDC}_0$, the $\\beta_i$-model reflection theorem used throughout, and the $\\Pi^1_4$-conservativity over $\\Pi^1_i$-$\\mathsf{CA}_0$."},{"cited_title":"On the $\\Pi^1_2$ consequences of $\\Pi^1_1$-$\\mathsf{CA}_0$","cited_arxiv_id":"2402.07136","evidence_quote":"Proves the base case $i=1$ of the preservation lemma and introduces the chain reflection formulas $\\beta^i_e\\mathrm{RFN}(n;\\tau)$ that the present paper extends."}],"review_version":1}