{"id":"4a4ed557-ceee-429c-a381-fb753081aad6","arxiv_id":"2412.15711","paper_version":2,"verdict":"UNVERDICTED","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A review arguing that the brain's network structure is an empirical hypothesis about dynamic relevance, not an established fact.","lead":"This paper asks whether the brain really works as a network or whether graph descriptions are just a convenient way to summarize brain data. It argues that showing a network structure in brain anatomy or activity does not prove that network properties drive brain dynamics.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The dynamics-only criterion is the load-bearing assumption: the paper itself concedes that some networkness may surface only in functional contexts, so the criterion is too narrow to settle whether network properties count.","rationale":"The reader's weakest assumption identifies exactly the same soft spot: bare dynamics may not be separable from function, and networkness may become visible only in task-driven or functional contexts. The manuscript itself supplies supporting admissions: Section 1 brackets function, and Section 3.3 states that key issues about annealed network structure are better treated in a functional framework. This is not an external disagreement with consensus; it is an internal gap between the paper's dynamic-relevance criterion and its own stated ambition to decide when network properties explain brain function. The concern is load-bearing because the central claim is a criterion claim, not merely a review of results. However, the reader's UNVERDICTED verdict already reflects the absence of a testable novel claim, and the paper is explicitly presented as the first part of a two-part project. Thus the concern reinforces the current verdict rather than moving it: the paper should remain UNVERDICTED until the criterion is broadened or the companion paper supplies the missing functional analysis. A concrete empirical test on HCP-style data would settle whether the dynamics-only criterion misses functionally relevant network structure.","tokens_in":41088,"tokens_out":6041,"duration_ms":62594,"concrete_test":"On a dataset with structural connectomes plus resting-state and task-based recordings from the same subjects (e.g., HCP), compute a network measure whose claimed relevance is functional rather than purely dynamical, such as structural controllability of a node or a task-evoked network motif. Test whether this measure predicts task performance after regressing out resting-state dynamical markers (e.g., metastability, avalanche scaling exponents, dynamic functional connectivity statistics). If the measure retains predictive power, a network property can count functionally without showing up in the bare dynamics used by the paper, and the dynamics-only criterion is too narrow. If no residual prediction remains, the bracketing of function is harmless for that measure.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim turns on treating 'network properties count' as equivalent to 'network properties affect bare dynamics.' Section 1 explicitly brackets function: 'we treat dynamics in a way that is divorced from considerations of neural systems' ability to perform a given task.' Section 3.3 then concedes that 'most key issues associated with annealed brain network structure... are better treated in a genuinely functional rather than in a purely dynamical framework.' These two commitments are in tension. If a network property is functionally load-bearing but leaves no detectable trace in the resting or stimulus-free dynamics examined here, the paper's criterion would classify it as non-networkness even though the system does 'work as a network' in the sense the abstract itself lists as decisive. The review therefore establishes at most a sufficient condition for networkness (dynamic relevance), not a necessary one. The companion paper can address functional cases, but the criterion as stated in this part is already dynamics-only and will not be corrected by adding more functional examples unless the criterion itself is revised.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper, the first part of a two-part review, argues that representing brain anatomy and activity as a complex network does not by itself imply that the brain actually operates as a network. It introduces the notion of 'networkness' as the property that network characteristics are consequential, reviews conditions for a system's reducibility to a network structure (connectedness, discretisability, structure preservation, intrinsicality), and examines a wide range of dynamical evidence—activity spreading, synchronisation, criticality and cascade models, frustration, and the effects of quenched versus annealed disorder—to assess when network structure demonstrably affects brain dynamics. The paper explicitly restricts Part I to bare, task-independent dynamics and defers genuinely functional analysis to a companion paper.","tokens_in":41246,"tokens_out":8281,"duration_ms":78174,"significance":"The paper's central thesis is important because it challenges the widespread, often implicit reification of network representations in neuroscience: rather than asking whether network measures exist, it asks whether network properties make a difference to brain dynamics or function. The manuscript is commendably cautious and self-aware, explicitly flagging conflicting evidence (e.g., Morrell et al. 2021 on external inputs generating apparent criticality, and reports of non-universal epidemic-model predictions), listing caveats about model ingredients, and repeatedly indicating that many structure-dynamics questions remain open. It also provides a valuable conceptual framework—time-scale separation between node dynamics, network dynamics, and processes on networks; quenched versus annealed disorder; the role of topology, geometry, and dimension—that will likely be pedagogically useful for the field. However, the central conceptual notion of 'networkness' is not given a crisp operational definition, and the paper's conclusion is explicitly conditioned on a companion paper that is not yet available.","major_comments":[{"comment":"The abstract states that the paper 'first define(s) the meaning of networkness,' but §2.1 lists four conditions (connectedness, discretisability, structure preservation, intrinsicality) without stating whether they are individually necessary, jointly sufficient, or simply desiderata, and §2.2.1 introduces two 'elements' (coupling strength and the coupling matrix) without formally connecting them to the §2.1 conditions. Section 3 then identifies 'dynamic relevance of network structure' as 'signs of networkness' without an explicit bridge between that sign and the earlier conditions. As a result, the central claim—that network properties count only when they demonstrably affect dynamics—lacks a testable criterion that a reader could apply to a concrete system. Please state the intended definition precisely (e.g., 'a network property is genuinely network-like iff removing or altering it changes an observable dynamical or functional quantity') and clarify the logical status of the §2.1 conditions relative to that criterion.","section":"§1, §3.3, §4"},{"comment":"The paper restricts Part I to 'bare dynamics' (§1) and explicitly concedes that 'most key issues associated with annealed brain network structure ... are better treated in a genuinely functional rather than in a purely dynamical framework' (§3.3). This concession means that the abstract's general negative thesis—that a complex network representation 'does not entail that the brain actually works as a network'—is here supported only for the dynamical domain, not for the functional domain that is listed in the abstract as part of what makes network properties 'count.' The paper should state in the abstract and in the concluding remarks that the dynamical analysis provides at most a sufficient condition for networkness, not a necessary one, and that a full verdict awaits the companion paper. Without this qualification, the reader may over-interpret the title and abstract as settling the entire networkness question.","section":"§1, §3.3, §4"},{"comment":"The manuscript depends on the unpublished companion paper (Papo and Buldú, in preparation) for the formal definition and examination of 'genuine functional brain activity' and for key aspects of annealed structure (§1, §3.3, and §4). This dependence makes the present contribution incomplete as a standalone argument: the 'meaning of networkness' promised in the abstract is not fully delivered until the companion appears. Please either include a concise statement of the functional criteria in the concluding remarks, or cite a published, preprint, or otherwise accessible version of the companion paper; without this, the reader cannot assess whether the functional evidence might reverse or qualify the dynamical conclusions.","section":"§1, §3.3, §4"}],"minor_comments":[{"comment":"The paper is very long and dense; it would benefit from a short 'definitions and scope' box or a glossary collecting the many technical terms (networkness, quenched versus annealed disorder, locality, frustration, self-averaging) that are introduced at various points and used throughout.","section":"General"},{"comment":"There is a typo in 'Morevoer' in the paragraph on the coupling factor σ; please correct.","section":"§2.2.1"},{"comment":"The word 'non-equilbirium' should be 'non-equilibrium' in the paragraph on the meaning of observed structure.","section":"§3.2.4"},{"comment":"The word 'parametres' is a non-standard spelling; 'parameters' is intended.","section":"§3.4.2"},{"comment":"The phrase 'exstinguished by fluctuations' should be 'extinguished by fluctuations.'","section":"§Quenched network structure and brain criticality"},{"comment":"Equation [8.2] is labeled in a way that may confuse readers; please renumber it as a separate equation (e.g., [9]) and adjust subsequent references.","section":"§3.2.2, equation numbering"}],"recommendation":"major_revision","confidential_remarks":"The paper is a well-written and unusually self-critical review, but its conceptual framework depends on a companion paper that is currently 'in preparation.' Given the journal's likely readership, the editorial decision should weigh whether the two-part structure is acceptable without the companion being publicly available. I also note that the review is single-authored in terms of its main thesis and that the field is wide; however, I do not see evidence of inappropriate citation practices beyond the authors' occasional reliance on their own prior work."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The one thing to know: this is a review and position piece, not a new empirical or formal result. Its useful core is the four conditions for networkness (connectedness, discretisability, structure preservation, intrinsicality) and the framing question: network properties count only when they affect dynamics, not just when measurable in a reconstructed graph. That is a genuinely useful corrective to the field's default of treating graph metrics as mechanisms.\n\nWhat it does well: it is careful and honest about open questions. It flags Morrell et al. on externally driven scaling, acknowledges the limits of epidemic spreading models, and discusses coarse-graining dependence of connectome topology. The internal logic is coherent, and the tone is appropriately cautious.\n\nWhere the soft spots are: because it is a review, the central claim rests on selective synthesis rather than derivation or a falsifiable prediction. That is normal for the genre, so the value is framing. More substantively, the stress-test concern is real. The paper defines networkness by dynamic relevance while explicitly bracketing function: \"we treat dynamics in a way that is divorced from considerations of neural systems' ability to perform a given task.\" Section 3.3 concedes that key annealed-structure issues are \"better treated in a genuinely functional rather than in a purely dynamical framework.\" If a network property is functionally load-bearing but leaves no trace in the bare dynamics examined here, the paper's criterion would classify it as non-networkness even though the system does work as a network in the sense the abstract itself lists as decisive. The review establishes at most a sufficient condition for networkness, not a necessary one. That is not fatal, but it should be stated plainly.\n\nAlso, the companion paper is a real dependency: several sections defer functional treatment to \"Papo and Buldú, in preparation.\" That makes this part incomplete on its own terms, but it is an incompleteness, not circularity.\n\nWho this is for: readers in network neuroscience who want a careful map of the structure-dynamics question and a vocabulary for debating whether network properties count. It deserves a serious referee, but the referee should push the authors to narrow their claims and fix the dynamics-function tension. I would cite it as a reference for the networkness criterion and would bring it to a reading group if the group is interested in conceptual foundations.\n\nRecommendation: send to peer review, with the expectation of heavy revision on the scope of the criterion. Given the genre, this is a reasonable submission.","headline":"A careful conceptual review that usefully frames networkness via four conditions, but its dynamics-only criterion is too narrow on its own admissions and the paper is a synthesis rather than a new result.","tokens_in":41754,"tokens_out":1548,"would_cite":true,"duration_ms":13441,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["05C82","37N25","92B20"],"pacs":[],"model":"deepseek-v4-flash","headline":"Mapping the brain as a network does not make it one.","keywords":["brain dynamics","networkness","complex networks","quenched disorder","synchronisation","criticality","structure-dynamics relationship","functional connectivity"],"falsifier":"A decisive observation would be a brain phenomenon whose dynamics remain unchanged when the reconstructed connectivity matrix is randomly rewired while node dynamics, inputs, and recording scales are held fixed; such invariance would show that the specific network topology does not count for that phenomenon.","tokens_in":40886,"feed_emoji":"🧠","tokens_out":7026,"duration_ms":53775,"temperature":0.7,"pith_summary":"This paper argues that applying graph theory to brain anatomy and activity is not the same as showing that the brain works as a network. The real question is whether network properties count: whether the topology, geometry, and disorder of a reconstructed graph demonstrably shape brain dynamics. The paper defines networkness as a property that must be earned through dynamic relevance, and reviews evidence across synchronisation, criticality, activity spreading, and pattern formation to ask where that condition is met. A fair reading is that descriptive network measures are only as meaningful as the dynamics they can explain, and that networkness is scale- and phenomenon-specific.","feed_headline":"Network maps don't prove the brain is a network","feed_subtitle":"A dynamics-focused review asks when network structure truly shapes brain activity, and when it is just a convenient description.","key_machinery":"The central object is the notion of networkness, defined by whether a system's behaviour can be reduced to its network structure without loss of information and by whether the structure's properties influence dynamics. The paper's analytical tools are the quenched/annealed disorder distinction, the master stability function and Kuramoto synchronisation framework, epidemic spreading models with absorbing-state transitions, and the concept of frustration; each is used to test where topology makes a dynamical difference.","core_discovery":"The central claim is that networkness is not an automatic consequence of measuring connectivity; a brain system genuinely behaves as a network only when its network structure makes a causal dynamical difference. The paper grounds this in conditions for reducibility to a network (connectedness, discretisability, structure preservation, intrinsicality) and in a distinction between quenched disorder, the static anatomical connectivity, and annealed disorder, the activity-induced connectivity. It then surveys dynamical processes—synchronisation, criticality, epidemic spreading, frustration, topological transitions—to locate where network structure changes the physics rather than merely labelling it. The conclusion is that the brain may be describable as a network at some scales and for some phenomena, but that networkness is a scale- and phenomenon-specific achievement, not a generic property of brain tissue.","pith_inferences":["If dynamic relevance is the right criterion, then task-driven or functional contexts may be where networkness becomes visible; a dynamics-only review may understate networkness that is function-dependent, a possibility the paper brackets at the outset.","Networkness may be a graded property: the same brain region could behave as a network for one dynamical process, such as synchronisation, and as a continuous field for another, such as ephaptic coupling, so the question would shift from 'is the brain a network?' to 'for which processes and at which scales is it one?'","A testable extension of the review's framework is to compare perturbation-response predictions of network models with continuum neural-field models on the same data; wherever the predictions diverge, the network representation either earns or loses its claim to dynamic relevance.","The conceptual apparatus could transfer to other biological and social systems that use graph descriptions, such as gene-regulatory or ecological networks, where the same gap between descriptive structure and operating principle may appear."],"forward_implications":["Purely descriptive graph metrics such as clustering, small-worldness, and modularity do not, by themselves, demonstrate that the brain is a network.","Network structure should be treated as a hypothesis that can fail at specific scales: a graph may be a useful data-compression device without being the brain's operating principle.","Brain criticality studies gain a sharper prediction: if hierarchical modular topology is the cause of scale-free avalanches, then perturbing modular structure should alter critical exponents, not just the descriptive statistics.","The quenched/annealed distinction gives a concrete way to decide whether anatomy or activity-induced coupling is the load-bearing structure for a given dynamical phenomenon."],"supporting_citations":[{"why":"Supplies the concern that network representations may be robust or fragile depending on how they are constructed, motivating the networkness criterion.","marker":"Atmanspacher and beim Graben, 2007"},{"why":"Standard reference for treating brain anatomy and dynamics as graphs, the position the paper critically examines.","marker":"Bullmore and Sporns, 2009"},{"why":"Establishes the distinction between bare dynamics and genuine functional brain activity that frames the review's scope.","marker":"Papo, 2019"},{"why":"Documents how parcellation and network reconstruction choices affect the reified network structure.","marker":"Korhonen et al., 2021"},{"why":"Shows in oscillator networks that disordered dynamics can encode topological identity, giving a concrete topology-dynamics relation.","marker":"Timme, 2006"},{"why":"Provides the result that hierarchical modular networks yield Griffiths-phase extended criticality without fine tuning, a key candidate for genuine networkness.","marker":"Moretti and Muñoz, 2013"},{"why":"Derives the scaling relation between node response time and weighted degree, used to test whether topology shapes perturbation propagation.","marker":"Hens et al., 2019"},{"why":"Shows frustrated synchronisation regimes on modular and human-connectome-like networks, linking spectral structure to dynamics.","marker":"Villegas et al., 2014"},{"why":"Frames activity propagation as an ecological invasion process in spatially extended neural fields, a contrasting continuum viewpoint.","marker":"Bressloff, 2012"}],"fun_headline_variants":["Brain as network? Only when dynamics make it so","Networkness is a dynamical achievement, not a given","Scale matters: brain network status is conditional","Connectivity maps don't guarantee brain network behavior","Does brain activity truly behave as a network?"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The analysis assumes that 'bare dynamics' can be meaningfully separated from function and that dynamic relevance is the right criterion for judging networkness, since Section 1 explicitly sets aside the brain's ability to perform tasks.","fun_headline_variants_meta":{"raw":{"variants":["Brain as network? Only when dynamics make it so","Networkness is a dynamical achievement, not a given","Scale matters: brain network status is conditional","Connectivity maps don't guarantee brain network behavior","Does brain activity truly behave as a network?"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000271,"raw_usage":{"total_tokens":1593,"prompt_tokens":877,"completion_tokens":716,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":493,"completion_tokens_details":{"reasoning_tokens":645}},"tokens_in":493,"tokens_out":716,"duration_ms":6342,"temperature":1.0,"reasoning_tokens":645,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T11:09:08.475211+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A decisive observation would be a brain phenomenon whose dynamics remain unchanged when the reconstructed connectivity matrix is randomly rewired while node dynamics, inputs, and recording scales are held fixed; such invariance would show that the specific network topology does not count for that phenomenon.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the distinction between bare dynamics and genuine functional brain activity that frames the review's scope."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Shows in oscillator networks that disordered dynamics can encode topological identity, giving a concrete topology-dynamics relation."}],"review_version":1}