{"id":"e8e8f525-3714-4d00-8cfa-ea05c960b7c9","arxiv_id":"2507.21171","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Risks are best modeled in BFO as externally grounded roles, not dispositions, with a sufficient condition based on agent aversion.","lead":"This philosophy paper argues that risk should be classified as a type of role rather than a disposition in the Basic Formal Ontology, a widely used framework for organizing scientific data. It offers a testable sufficient condition: a role becomes a risk when an agent or group has an aversion to its possible realizations.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Sufficient condition turns on a cross-bearer grounding relation that BFO does not define; until this relation is axiomatized, the claim that Risk is a subclass of BFO:Role is not fully supported.","rationale":"The reader and I identify the same load-bearing point: the cross-bearer grounding relation is what separates the Risk-as-Role account from the Risk-as-Disposition account. In the dispositional approach, the disposition itself is internally grounded and only its risk status is externally grounded; in the role approach, the risk role's very existence is externally grounded in an aversion inhering in another entity. That contrast is the core of the paper's argument, so the relation must be formally supported. The paper's own listed limitations in Section VI — lack of competency questions and unresolved cases such as plants and non-material risk bearers — primarily affect necessity and practical evaluation, not this sufficiency step. I do not see an internal contradiction in the informal narrative; the problem is that BFO's Role definition does not obviously license the required grounding relation, and the paper provides no axioms to bridge that gap. A machine-checked or at least explicit formalization would settle the question. The CONDITIONAL verdict already given is therefore appropriate, and I would not change it.","tokens_in":10554,"tokens_out":12673,"duration_ms":167520,"concrete_test":"Formalize the ceiling scenario in OWL using the official BFO 2020 axioms (or BFO 2.0) plus the definitions quoted in Section III. Assert that ceiling0 hasRole hazard0, Paul has disposition aversion0, and introduce an object property grounds with domain Aversion and range Role, including the paper's sufficient condition as a definition of Risk. Run a reasoner such as HermiT and check whether (a) the ontology is consistent, (b) hazard0 is inferred to be a Role, and (c) the grounds assertion is compatible with BFO's role axiom that a role exists because its own bearer is in special circumstances. If the ontology is inconsistent or hazard0 cannot be classified as a Role without adding a non-BFO axiom, the cross-bearer grounding step fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's sufficient condition (Section V) rests on the move in Section IV.C: 'Paul's mental disposition aversion0 ... grounds the existence of the risk role hazard0 in ceiling0.' This is a grounding relation between two specifically dependent continuants inhering in different material entities. BFO's definition of Role (Section III) instead makes a role's existence depend on its bearer being in some special physical, social, or institutional set of circumstances. The quoted definition does not include a relation by which an aversion in one agent can ground a role in a distinct material entity. The paper cites [15] for the notion of grounding, but supplies no axioms that link an aversion in one bearer to a role in another bearer, and no account of how such cross-bearer grounding is compatible with the role definition's requirement of a single bearer. If 'externally grounded in an aversion' is a new primitive, the sufficient condition cannot be checked against BFO. If it is shorthand for the bearer being in circumstances that involve the aversion, then the condition should be reformulated in terms of the BFO role definition rather than left as an unanalyzed grounding relation. Either way, the central classification of Risk as a subclass of Role is not yet established, because the distinguishing step between the dispositional and role-based accounts is exactly this external grounding.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper argues that the category Risk should be modeled in the Basic Formal Ontology (BFO) as a subclass of BFO:Role rather than BFO:Disposition. It develops a ceiling-collapse example involving Paul, ceiling0, the disposition instability0, and the risk role hazard0, and it uses this example to motivate a sufficient condition: if r is a role and r is externally grounded in an aversion that an agent or group of agents has toward the possible realizations of r, then r is a risk. The paper also addresses merely possible processes by introducing axioms such as (O2) and (O3), and it closes by acknowledging that the proposal is exploratory, that it may not generalize to plant or financial risks, and that future work should evaluate the model with competency questions and real data.","tokens_in":10825,"tokens_out":3266,"duration_ms":37860,"significance":"If the central claim were fully supported, the paper would make a useful contribution to applied ontology by offering a principled way to distinguish risks from underlying physical dispositions, capturing both the agent-relativity and the unrealized-possibility aspects of risk within a top-level ontology. The paper is commendably clear in contrasting the role-based and disposition-based accounts, and it provides explicit OWL-style axioms for two key problematic cases: unrealized processes and intentional directedness. The authors are also honest about the exploratory nature of the proposal and its current limitations. However, the central sufficient condition relies on an under-specified notion of external grounding that is not yet integrated with BFO's formal definition of Role, and the generalization from a single example is not yet supported by a general argument.","major_comments":[{"comment":"The sufficient condition in Section V depends crucially on the claim in Section IV.C that \"Paul's mental disposition aversion0 ... grounds the existence of the risk role hazard0 in ceiling0.\" This is a grounding relation between a specifically dependent continuant inhering in one material entity (Paul) and a role inhering in a distinct material entity (ceiling0). BFO's definition of Role, quoted in Section III, requires that the role exist because its single bearer is in some special physical, social, or institutional set of circumstances; it does not include a primitive relation by which an aversion in one bearer grounds a role in another bearer. The paper cites [15] for grounding but gives no axioms characterizing external grounding or explaining how such cross-bearer grounding is compatible with the role definition's single-bearer requirement. Without this, the sufficient condition cannot be checked against BFO, and the central classification of Risk as a subclass of BFO:Role is not yet established.","section":"Section IV.C and Section V"},{"comment":"The phrase \"externally grounded in an aversion\" is not formally defined. The informal gloss in Section III -- that grounding is a relation in which one entity exists because of another -- is too weak to determine when a role satisfies condition (ii). For instance, it is unclear whether the grounding relation is intended to be a new BFO relation, a defined relation in terms of the bearer's circumstances, or a relation imported from some other formal theory. The authors should either provide explicit axioms for external grounding and for the intentional directedness of aversion, or reformulate condition (ii) directly in terms of the BFO role definition (e.g., the bearer is in circumstances that include an agent's aversion toward the role's possible realizations). This is a load-bearing point because the distinction between the role account and the dispositional account rests on exactly this grounding relation.","section":"Section V"},{"comment":"The sufficient condition is generalized from a single illustrative example, and the authors themselves note that the model may not extend to plant risks or financial risks and that competency questions from domain experts are lacking. Since the paper's title and abstract assert that Risk is a subclass of BFO:Role, a universal claim, the argument needs at least a sketch of why the condition applies across the intended range of cases, or a restriction of the claim to a clearly delimited domain. As it stands, the example establishes at most that one particular risk role can be modeled in this way, not that all risks are roles.","section":"Section V and Section VI"}],"minor_comments":[{"comment":"In the description of the collapse process, \"loose cohesion\" should be \"lose cohesion\".","section":"Section III.C"},{"comment":"The word \"readibility\" in the caption of Fig. 2 should be \"readability\".","section":"Section V"},{"comment":"The manuscript references Fig. 1 and Fig. 2, but the figures do not appear in the text provided; please ensure they are included and legible, since the diagrams are important for following the two competing models.","section":"Figures 1 and 2"},{"comment":"The definitions of disposition and role are attributed to reference [15], but [15] is a paper on grounding realizable entities; the canonical BFO definitions are usually found in [6] or [16]. Please check the citation.","section":"Section III"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and addresses a genuine gap in the ontology of risk. The main technical obstacle is the unformalized external grounding relation; if the authors can provide axioms or a BFO-compatible reformulation of condition (ii), the contribution would be substantially strengthened. The authors' explicit acknowledgment of limitations is a positive sign, but the current version stops short of making the central claim fully defensible."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Read the Donato/Barton paper. The short version: it argues risk should be a BFO:Role subclass rather than BFO:Disposition, with a sufficient condition: a role externally grounded in an aversion by an agent or group toward possible realizations of the role. That's a real departure from the authors' earlier dispositional account (Grenier/Barton 2021), and it's worked out with a concrete ceiling example. The paper is honest and clearly argued; it does not oversell. It explicitly labels the condition sufficient, not necessary, and flags the plant case and non-material risk bearers as unresolved.\n\nWhat's genuinely good: the contrast between the two accounts is sharp, and the realization story is the strongest part. In the dispositional account, instability0 would realize a collapse; in the role account, hazard0 realizes a damaging process involving the agent. That distinction gives content to the role proposal. The handling of merely possible processes via (O2) is sensible and respects BFO's constraint against classes without instances. The paper also acknowledges the lack of competency questions and notes this is exploratory.\n\nThe soft spot is exactly where the stress-test points: 'externally grounded in an aversion' is doing all the work, but the paper never gives axioms for the grounding relation between an aversion inhering in one person and a role inhering in a distinct material entity. BFO's role definition says a role exists because its single bearer is in some special physical/social/institutional circumstances. The paper could plausibly unpack 'externally grounded' as shorthand for the bearer being in circumstances that include the aversion, but it doesn't say that, and it doesn't show how the grounding relation composes with BFO's definition. Since the case for Risk-as-Role over Risk-as-Disposition turns precisely on this step, the sufficient condition is not fully established. That's a real gap, but proportionate: the paper is exploratory and says so; the fix is likely a reformulation, not a new framework.\n\nWho gets value: applied ontologists working in BFO, especially on risk registers or safety databases. It's a niche but legitimate contribution, and it deserves a serious referee; I would send it out rather than desk reject. My own verdict: conditional. I'd want the grounding clarified and ideally a second worked example, but I don't see a fatal flaw.","headline":"A clear, honest exploratory proposal to classify risk as BFO:Role grounded in agent aversion; worth refereeing, but the load-bearing grounding relation needs formal work before the classification is established.","tokens_in":11278,"tokens_out":2964,"would_cite":false,"duration_ms":35695,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper argues that a risk is a role, not a disposition, in the Basic Formal Ontology, and that a role externally grounded in an agent's aversion toward its possible realizations is a risk.","keywords":["ontology","risk","Basic Formal Ontology","role","disposition","realizable entity","aversion","external grounding"],"falsifier":"Ask whether a situation is still a risk when no agent or group has any aversion to the possible outcome, for example an unstable ceiling in an uninhabited building with no one who cares; if such cases are counted as risks in ordinary risk registries, the sufficient condition fails, and the paper itself flags the plant-burning case as a possible counterexample to the conative approach.","tokens_in":10347,"feed_emoji":"⚠️","tokens_out":7613,"duration_ms":74541,"temperature":0.7,"pith_summary":"This paper argues that a risk is not a physical disposition of a hazardous thing but a role that exists only because some agent or group of agents has an aversion to the harm the thing could cause. Working inside the Basic Formal Ontology, the authors classify Risk as a subclass of BFO:Role and give a sufficient condition: a role externally grounded in an agent's aversion toward its possible realizations is a risk. The payoff is a clean separation between the objective hazard (a ceiling's disposition to collapse) and the risk (the role of that ceiling, grounded in Paul's aversion to being damaged), so that the same collapse is a risk for one person and an opportunity for another. The paper matters for risk registers and databases because it offers a precise, logically expressible account of what a risk entry denotes and why risks are agent-relative.","feed_headline":"Risks are roles grounded in our aversions","feed_subtitle":"In BFO terms, a risk is a role grounded in an agent's aversion, so the same hazard can be a risk or an opportunity.","key_machinery":"The load-bearing machinery is the BFO distinction between disposition and role, both realizable entities: a disposition exists in virtue of its bearer's physical make-up, while a role exists because some external context obtains. The paper adds a grounding relation, one entity exists because of another, and combines it with intentionality: the risk role hazard0 exists in ceiling0 because Paul's mental disposition aversion0 is directed toward processes of damaging him. To keep unrealized risks representable in BFO, the paper uses axioms of the form 'realized in only (Collapsing and has participant value ceiling0)', and it distinguishes the risk's realization (a damaging process, a change in Paul's health SDCs) from the disposition's realization (a collapse process, spatial and SDC changes of the ceiling).","core_discovery":"The paper's central claim is that risk is a realizable entity of type BFO:Role, not BFO:Disposition. On this view, a risk role r exists in a material bearer only when an agent (or group) has an aversion directed toward the possible realizations of r, and this aversion externally grounds the role: in the ceiling example, Paul's mental disposition aversion0 grounds the existence of hazard0 in ceiling0. The risk role and the underlying physical disposition are distinct entities with distinct realizations, so instability0 realizes in a collapse process while hazard0 realizes in a damaging process, and the risk role is a risk for as long as it exists, whereas a disposition would only accidentally count as a risk. The paper therefore proposes the sufficient condition that (i) r is a role and (ii) r is externally grounded in an aversion toward its possible realizations imply that r is a risk, and it identifies the intentional direction of the aversion as what determines how many and which risks there are.","pith_inferences":["The same cross-bearer grounding mechanism could be applied to other conative states: if risk is a role grounded in aversion, then opportunity is presumably a role grounded in desire, making risk and opportunity two sides of the same relational structure rather than independent kinds.","If this analysis is right, risk registers should be modeled with the risk role as a first-class entity that links a bearer, an aversion, and a target process type; aggregating risks across agents or even across aversions of one agent would require summing distinct roles, which may explain why risk aggregation is hard in practice.","A testable extension: represent an aversion as an information artifact in a BFO-based electronic health record and check whether the derived risk roles align with clinicians' risk assessments; divergence would indicate the aversion needs to be refined or the grounding relation is not sufficient.","The paper's own plant example suggests the conative condition may need to be generalized to teleology (a plant's goal of living); if so, the role category might still hold but the grounding would no longer be a mental aversion, loosening the link between risk and subjectivity."],"forward_implications":["A single physical hazard can support several distinct risk roles, one for each aversion an agent has: Paul's health and Paul's clothes give rise to separate roles hazard0 and hazard1 in the same ceiling.","Risk becomes agent-relative by construction: the same ceiling collapse is a risk for Paul and an opportunity for Vladimir, so no risk register can be complete without recording whose aversion grounds each entry.","A risk can exist even though it is never realized: the role hazard0 and the disposition instability0 exist whether or not collapsing0 or damaging0 ever occurs, as long as the aversion exists.","Risks disappear when the grounding aversion disappears, even if the hazardous disposition remains: if Paul stops caring, hazard0 ceases to exist while instability0 persists.","Because the realization of the risk is a damaging process rather than the collapse process, risk assessment and prevention should target the damaging process and the agent-relative context, not just the physical failure mode."],"supporting_citations":[{"why":"Supplies the BFO categories (continuant, occurrent, realizable entity) within which the whole analysis is framed.","marker":"[6]"},{"why":"Provides the formal definitions of disposition and role as well as the grounding relation on which the sufficient condition relies.","marker":"[15]"},{"why":"The dispositional view of risk that the paper argues against; it is the rival analysis the role view must outperform.","marker":"[17]"},{"why":"The strategy of formalizing unrealized possible realizations with axioms such as (O1)/(O2), keeping merely possible risks representable.","marker":"[22]"},{"why":"Grounds the treatment of desires and aversions as dispositions that can remain latent without explicit mental processes.","marker":"[25]"},{"why":"Defines intentionality as the aboutness relation used to link an aversion to its target class of damaging processes.","marker":"[30]"},{"why":"Supplies the decomposition of events into sums of SDC changes, used to distinguish the risk realization from the physical collapse.","marker":"[20]"},{"why":"Adapts that decomposition to BFO processes, supporting the claim that collapsing0 and damaging0 are distinct realizations.","marker":"[21]"}],"fun_headline_variants":["Risk is a role, not a disposition","Aversion grounds risk roles in BFO","Risks exist only when someone is averse","Same hazard, different risk: grounded by aversion"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole account rests on the idea that an agent's aversion can bring a risk-role into existence in a different material object, a cross-entity grounding that the ontology does not define by axioms.","fun_headline_variants_meta":{"raw":{"variants":["Risk is a role, not a disposition","Aversion grounds risk roles in BFO","Risks exist only when someone is averse","Same hazard, different risk: grounded by aversion"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000291,"raw_usage":{"total_tokens":1656,"prompt_tokens":858,"completion_tokens":798,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":474,"completion_tokens_details":{"reasoning_tokens":740}},"tokens_in":474,"tokens_out":798,"duration_ms":7340,"temperature":1.0,"reasoning_tokens":740,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T14:04:42.392549+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Ask whether a situation is still a risk when no agent or group has any aversion to the possible outcome, for example an unstable ceiling in an uninhabited building with no one who cares; if such cases are counted as risks in ordinary risk registries, the sufficient condition fails, and the paper itself flags the plant-burning case as a possible counterexample to the conative approach.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the BFO categories (continuant, occurrent, realizable entity) within which the whole analysis is framed."},{"cited_title":"Grounding Realizable Entities","cited_arxiv_id":"2405.00197","evidence_quote":"Provides the formal definitions of disposition and role as well as the grounding relation on which the sufficient condition relies."},{"cited_title":"Une Ontologie Dispositionnelle du Risque,","cited_arxiv_id":null,"evidence_quote":"The dispositional view of risk that the paper argues against; it is the rival analysis the role view must outperform."},{"cited_title":"Representing Dispositions,","cited_arxiv_id":null,"evidence_quote":"The strategy of formalizing unrealized possible realizations with axioms such as (O1)/(O2), keeping merely possible risks representable."},{"cited_title":"Foundations for an ontology of belief, desire and intention,","cited_arxiv_id":null,"evidence_quote":"Grounds the treatment of desires and aversions as dispositions that can remain latent without explicit mental processes."},{"cited_title":"Intentionality,","cited_arxiv_id":null,"evidence_quote":"Defines intentionality as the aboutness relation used to link an aversion to its target class of damaging processes."},{"cited_title":"Events, their names, and their synchronic structure,","cited_arxiv_id":null,"evidence_quote":"Supplies the decomposition of events into sums of SDC changes, used to distinguish the risk realization from the physical collapse."},{"cited_title":"Two Approaches to the Identity of Processes in BFO,","cited_arxiv_id":null,"evidence_quote":"Adapts that decomposition to BFO processes, supporting the claim that collapsing0 and damaging0 are distinct realizations."}],"review_version":1}