Scientific Ontology

The Question of Scientific Ontology

Angela Bogdanova
Published by Aisentica Development
Philosophy by Bogdanova · Philosophy of Science · Research Dossier
Language: American English · Research source edition: November 7, 2026
Canonical reservation: https://philosophybybogdanova.com/dossiers/scientific-ontology/
Editorial state: BATCH READY / NOT RELEASED

Scientific ontology investigates the kinds of entities, properties, structures, processes, and relations that scientific inquiry gives us reasons to recognize. Its central question—Which entities should science lead us to accept?—demands an account of how scientific commitments arise, how evidence discriminates among them, and how they can change without making scientific achievement unintelligible. Ontology in this sense is neither an inventory of familiar objects nor a transcription of every noun in a research article. The relevant issue concerns what must obtain in the world for successful and well-supported scientific claims to be true, and what additional evidence permits belief in particular constituents of that world.

The modern debate is anchored by Quine's (1948) analysis of what there is and Carnap's (1950) account of linguistic frameworks. Both identify conceptual conditions of ontology; they disagree about how ontological questions earn their significance. Later realism and empiricism debates add the difference between accepting a theory for scientific use and believing its full description of unobservables. As Chakravartty (2017) shows in the scientific realism survey, ontological, semantic, and epistemic claims belong together but require separate arguments.

This monograph defends a case-sensitive interpretation of scientific ontological warrant. An existential commitment can be expressed by the logic of a theory, supported by intervention or measurement, constrained by structural continuity, or attached to a discipline's practical classification. These different grounds justify different strengths and kinds of belief. Their separation is the key to reading scientific ontology responsibly.

Method, Evidence Grades, and Scope

The inquiry combines historical reconstruction, conceptual analysis, the close reading of accessible scholarly reference essays, publisher-verified bibliographic descriptions of major books and articles, and comparison of documented empirical research. The primary empirical cases are the Higgs-compatible boson observations reported separately by ATLAS and CMS in 2012, the 2015 gravitational-wave event reported in 2016, and IUPAC's contemporary chemical-element definitions. Biological species serve as a disciplinary comparison grounded in Ereshefsky's review of competing species concepts.

Three evidence levels are kept distinct. First, bibliographic identity confirms authorship, title, venue, date, and the existence of a specific work. Second, accessible abstract or chapter summary supports a delimited account of an author's argument. Third, a complete accessible research report or reference discussion supports a more detailed reconstruction. A publisher synopsis is sufficient to establish the advertised thesis of a book, while it cannot establish all of the book's steps, qualifications, or examples. The source register makes these restrictions explicit.

The method is philosophical. Statistical significances in the empirical studies are claims made by their research collaborations, not freshly computed results. Illustrative hypothetical examples below are explicitly described as constructed. No interviews, laboratory work, replication studies, new empirical discoveries, or independent readings of inaccessible complete books are represented as having occurred. The aim is a complete reasoned analysis within this documentary scope, and its arguments are original formulations without claims to a new philosophical theory.

Ontology, Metaontology, and Scientific Practice

Ontology asks what exists; metaontology asks how questions of existence should be formulated and answered. Scientific ontology joins these inquiries to laboratory, observational, computational, and classificatory practices. Bricker (2014), in the ontological commitment analysis, identifies the problem of what entities a theory requires as a problem of metaontology. That distinction explains why a scientist can infer a particle signal while a philosopher must still ask how the signal licenses a commitment to a particle rather than to some more cautious representational claim.

Scientific practice is often ontologically productive. Experimental teams design detectors in anticipation of possible entities; taxonomists adopt kinds that permit stable inferences; modelers introduce ideal objects to calculate relationships inaccessible to direct observation. None of these practices settles every metaphysical interpretation of its products. A good ontology of science must explain how the same research can successfully manipulate physical phenomena while remaining open about philosophical interpretations of the entities involved.

Accordingly, analysis moves among three questions: What does a scientific description assert? What must be real for that assertion to be true under its intended interpretation? What grounds a commitment to the asserted thing? Treating the three as one question conceals the difference between meaning, logical consequence, and warrant. A responsible ontology keeps all three visible.

The Quinean Criterion of Commitment

Quine's (1948) classic ontological argument connects the commitments of a regimented theory to the values its bound variables must take for the theory's sentences to be true. The claim matters because scientific language is filled with expressions that can be treated literally, paraphrased, or translated into alternative formal representations. A statement about electrons normally presents electrons as existing, whereas a statement about a mathematical approximation need not present the approximation as a material object.

The criterion is powerful as a discipline of bookkeeping. If a theory interpreted literally entails that there are entities of type E, the interpreter can identify the commitment rather than disguising it in metaphor. Bricker's (2014) discussion of quantifier accounts also emphasizes the complications: formal regimentation, paraphrase, intentional contexts, and rival notions of commitment can change what the test returns. A formal semantic test therefore needs an independently defended reading of the theory.

Consider a constructed example: a simulation includes 10,000 numerical 'particles' to represent a fluid. Quantification over simulated particles is internal to the algorithm. It would be a mistake to infer that the model asserts 10,000 microscopic physical particles in the tank. To make the ontological move, an argument must connect the simulation variables to intended empirical referents. The grammar of existence is a beginning of the philosophical inquiry, not its experimental conclusion.

Carnap and the Choice of Conceptual Framework

Carnap (1950), in Empiricism, Semantics, and Ontology, differentiates internal questions asked within an adopted linguistic framework from external practical questions about the usefulness of adopting that framework. Once numbers, physical things, or properties are admitted within a framework, specific existential questions receive the forms of answer authorized by its rules and evidence. Choosing a framework is a matter of methodological appraisal involving utility, simplicity, and purpose.

The contrast matters for scientific ontology because scientists routinely revise the terms in which their problems are expressed. A chemist may use atoms and molecules; a statistical physicist may work with ensembles; a population biologist may identify lineages and ecological relations. A scientist can acknowledge the reality of experimentally established phenomena while evaluating alternative descriptive frameworks without declaring every framework object a new metaphysical constituent.

Carnap's perspective also encounters a serious problem. If rival frameworks make incompatible assertions about unobservable reality, the contest cannot always be resolved by evaluating linguistic convenience alone. Empirical findings may favor particular causal descriptions and reject others. The more an internal statement is tied to intervention and prediction beyond the framework in which it originated, the stronger the pressure to treat it as a claim about the world. Scientific ontology therefore inherits both Carnap's alertness to conceptual choice and a demand for evidence-sensitive adjudication.

From Theory Acceptance to Ontological Belief

Acceptance is a practical scientific act: researchers may use a theory, trust a model within a domain, or employ its equations when designing an instrument. Belief that the theory's entities exist is a further epistemic stance. Van Fraassen's (1980) constructive empiricist account makes this difference central by proposing empirical adequacy as the aim of science and as the content of belief involved in theory acceptance. His formulation permits the literal interpretation of discourse about unobservables while withholding belief in its complete truth.

Scientific realism responds that sustained explanation, predictive achievement, and instrumental success may justify taking at least significant parts of a theory as descriptions of the world. Chakravartty's (2017) realist argument map distinguishes the ontological dimension of realism from its semantic and epistemic dimensions. This shows why realism and antirealism cannot be decided by presenting a list of scientific terms.

My assessment is that no single move from practical adoption to ontological belief is justified merely by adoption itself. Belief requires an evidential account of what makes the posited entity necessary, well-confirmed, manipulable, or retained across independent ways of finding out. Where such support is strong, abstention may become increasingly costly; where support is weak, full ontological assent is premature even when a model is extraordinarily useful.

Observability and Its Changing Boundaries

The distinction between observable and unobservable is philosophically important because instruments extend access beyond human senses and make unobservable entities explanatory targets. Van Fraassen (1980) treats observation and empirical adequacy as central to constructive empiricism, whereas scientific realists commonly endorse warranted belief in parts of the unobservable structure of the world. Chakravartty's (2017) analysis of observability records how even the observer-relative meaning of 'observable' is disputed.

Science itself relies on elaborate chains of mediation. The display of a detector consists of signals, numbers, calibrations, filters, backgrounds, and statistical inferences. It does not present its causal sources without interpretation. Tal's (2020) account of measurement models explains why instrument indications and measurement outcomes must be distinguished. These outcomes may remain well-supported while an interpretation of their microscopic source is revised.

Treating direct visual access as the unique condition of existence would make the ontology of modern physics implausibly dependent on human anatomy. But treating every instrument output as a direct entity observation would erase the very inferential work that makes instruments reliable. The more productive question is how different measurement channels, error models, and interventions converge on a stable causal target.

Intervention and Entity Realism

Hacking (1983) gives manipulation a central evidential role in Experimentation and Scientific Realism. Experimental scientists use electrons, for example, to generate effects, to probe other phenomena, and to create controlled arrangements. This provides a distinct source of confidence from the fact that a theory predicts an outcome. An entity with reliably exploitable causal powers acquires a practical profile resistant to some changes of theoretical description.

Intervention nevertheless requires interpretation. An instrument may operate through a causal process whose detailed mechanism remains disputed; two theories can sometimes account for an intervention equally well. Manipulation supports realism most strongly when the alleged entity is involved across different, independently calibrated interventions and when alternatives have been seriously tested. The achievement is then more than mere predictive accommodation, although it is still less than infallible knowledge of every property assigned to the entity.

The original inference advanced here is a separation between control and complete characterization. A scientist may be justified in believing that a manipulable cause exists while remaining uncertain about the theoretical properties ascribed to it. This preserves Hacking's experimental insight without treating ontological confidence as an all-or-nothing endorsement of a surrounding theory.

Explanation, Success, and the Realist Inference

The no-miracles tradition asks why scientifically successful theories would predict, explain, and support interventions unless they represented important features of reality. The question is substantial because accidental success becomes less plausible as novel tests and independent research programs accumulate. Scientific ontology gains a positive argument from this coherence: the success of a highly constrained explanatory picture is evidence relevant to the entities and relations on which it depends.

Yet success is not self-interpreting. A model may succeed because its mathematical relations fit a domain while some elements of its imagined mechanism are false. Cartwright's (1983) critique of idealized laws describes the centrality of idealization in scientific explanation. Chakravartty (2017) notes in selective realist strategies that realists themselves disagree about which components of successful theories deserve commitment. Whole-theory truth and domain-specific reliable structure are different epistemic conclusions.

Scientific ontology should therefore examine the explanatory work of each posited entity. If eliminating the entity destroys the explanation, and alternative causal models fail to preserve its independent empirical achievements, the entity's ontological standing improves. If replacing it with another representation preserves every tested consequence, the case for the specific posit weakens. This assessment is comparative and historically revisable.

Historical Error and Pessimistic Induction

Laudan (1981), in A Confutation of Convergent Realism, challenges the assumption that historical scientific success generally indicates reference and approximate truth. Past theories achieved notable empirical success while describing the world in ways subsequently rejected. The historical record makes ontological continuity a question requiring demonstration rather than a property automatically possessed by each succeeding theory.

Stanford (2006), through the unconceived alternatives problem, deepens the challenge: investigators often failed to imagine competitors that later became intelligible and scientifically superior. A present theory may be the strongest available option partly because more adequate future alternatives have not yet been formulated. If so, present success alone cannot establish that a particular metaphysical inventory will survive.

These arguments are strongest against unrestricted inferences from the success of a theory to the truth of its whole ontology. They have less force against modest claims confirmed through repeated independent intervention or against mathematical relations that remain accurate across substantial theory change. A cautious scientific ontology is thus selective, historically instructed, and able to explain why discovery can survive the replacement of explanatory stories.

Structural Realism and the Persistence of Relations

Worrall (1989), in Structural Realism: The Best of Both Worlds?, argues that scientific continuity may be found in relational structure even when the entities and mechanisms described by theories change. The proposal takes the history of scientific change seriously while retaining a realist interpretation of what successful science learns. It is especially useful when mathematical relations remain empirically powerful across different physical interpretations.

Ladyman's (2023) structural realism overview distinguishes epistemic structural realism, which emphasizes what we know of relations, from ontic structural realism, which makes a deeper claim about the ontological priority of structure. These theses require different defenses. The first is primarily an account of the reach of knowledge; the second reconsiders what the basic constituents of reality are.

A structure-focused ontology can prevent unnecessary allegiance to historical pictures of particles as miniature everyday objects. Its own challenge is specificity: a description of structure must explain the relations' empirical interpretation, how it distinguishes possible worlds, and how it relates to experimentally accessible phenomena. Invoking 'structure' without those details relocates rather than resolves the ontological question.

Naturalized Metaphysics and Disciplinary Authority

Ladyman and Ross, with Spurrett and Collier (2007), propose in Every Thing Must Go a philosophy that gives contemporary scientific knowledge a governing role in metaphysics. Their approach challenges the assumption that scientific objects must resemble the independent individuals of everyday intuition. Physics and the special sciences often reveal processes, interactions, symmetries, networks, and patterns that fit such intuitive categories imperfectly.

A naturalized approach places welcome constraints on free philosophical speculation. It directs metaphysicians to take successful scientific research seriously, to consider its mathematical commitments, and to respect what investigators actually measure. Yet science includes multiple levels of description and unresolved interpretive disputes. A philosophy that simply declares the latest scientific vocabulary to be the final furniture of the universe would violate its own historical sensitivity.

The conclusion is that scientific authority changes the standards of metaphysical argument. The philosopher bears a burden to show how a proposed ontology relates to evidence, but empirical science itself does not relieve philosophy of interpreting reference, identity, kindhood, or modality. Naturalization is most compelling as an ongoing discipline of accountability.

Models, Idealizations, and Fictional Objects

Scientific representation frequently employs systems designed for tractability rather than literal reproduction. Frigg and Nguyen (2026), in Scientific Representation, survey the problem of how a model stands for a target. Cartwright (1983) shows through idealized physics examples why the properties assigned in theoretical models may belong to an intentionally simplified construction. A frictionless surface, an exactly homogeneous population, or an isolated perfect gas can be explanatory while failing to correspond literally to a real object.

Ontology must distinguish a model object, its intended target, and the properties the model licenses us to infer about that target. For example, a computational grid cell can represent atmospheric behavior without constituting a naturally occurring physical cell with precisely the same boundaries. The object can exist as a mathematical or computational construction while its physical counterpart is an aggregate or relational approximation.

This distinction also prevents an easy antirealist overreach. The fact that models idealize does not imply that their targets or all modeled relations are fictional. Scientific success depends upon establishing which inferences survive the simplification and which require correction. Ontological judgment is located in that validated inferential relationship.

Measurement and the Ontology of Quantities

Tal (2020) distinguishes instrument indications from the measured properties inferred from them in Measurement in Science. A photodetector reading, a coordinate estimate, or an inferred mass depends on calibration, measurement modeling, and environmental assumptions. The ontology of the measured quantity emerges through reliable relations among these operations and a purported target property.

The numerical value itself and the property measured are different sorts of thing. A mass estimate is a quantity claim, not an independent particle. A probability distribution is a mathematical object used to represent uncertainty or variation, and its status cannot be inferred simply from its indispensability in a calculation. Scientists can agree on a calibrated value while differing about the metaphysics of quantities, probability, or spacetime.

Measurement provides strong ontological evidence when its target properties are stably estimated across independent methods and coherently explain instrument behavior. It also imposes humility: every measurement value presupposes a domain of validity, resolution, model assumptions, and uncertainty. The move from stable measurement to belief in an entity must record what was measured and what was inferred.

Case Study I: The 2012 Higgs-Compatible Boson

The ATLAS Collaboration (2012) reported the observation of a new particle with statistical significance of 5.9 standard deviations in combined analyses. The CMS Collaboration (2012) independently reported a new boson near 125 GeV with a local significance of 5.0 standard deviations. Both papers concerned a signal consistent with expectations for the Standard Model Higgs boson. Their experimental records exemplify how detectors, backgrounds, theoretical channels, statistical analysis, and independent apparatus contribute to a claim about a physical entity.

The immediate inference in 2012 was to the existence of a new boson compatible with the Higgs hypothesis, rather than a logically complete demonstration of every property expected of a Standard Model Higgs boson. The experimental observations were real signals in distinct detectors, and the scientific interpretation connected their decay-channel patterns to a candidate particle. Subsequent characterization is an additional scientific enterprise. Confusing the two stages would inflate the content of the original discovery reports.

This case supports a graded ontology. Independent experimental convergence can justify belief that a new physical source produced observed event patterns; identifying its spin, couplings, and precise theoretical role requires further work. Scientific ontology is strengthened by preserving the temporal sequence from the detection claim to the refined characterization of the discovered entity.

Statistical Significance and Existence Claims

The ATLAS (2012) discovery report and CMS (2012) independent experiment report present statistical tests of signal-plus-background hypotheses, not direct metaphysical proofs. A reported significance quantifies the incompatibility of observed data with a specified background-only model under particular assumptions. It is an evidential achievement whose interpretation depends on instrument modeling, event selection, systematic uncertainty, and independent validation.

A scientific inference to an entity requires both a statistical difference and an explanation connecting that difference to a causal source. A threshold convention helps a collaboration regulate false-discovery risks; it does not define existence. Conversely, an entity does not become unreal because one experiment has inadequate power to identify it. What matters for ontology is the full pattern of constrained explanatory and empirical relations.

Constructed contrast: two detectors can each register a pattern with a small estimated background probability. If one relies on a shared faulty calibration, apparent agreement may exaggerate evidence. If independent calibrations and distinct channel-specific backgrounds agree, convergence has greater ontological force. This comparison is an illustrative argument, not a claim that such a calibration defect occurred in either 2012 experiment.

Case Study II: Gravitational Waves and Black-Hole Mergers

The LIGO Scientific Collaboration and Virgo Collaboration (2016) described the observation of gravitational waves from GW150914. Their report states that two observatories registered a transient on September 14, 2015, and that the observed waveform matched the general relativistic expectation for a binary black-hole coalescence. The reported matched-filter signal-to-noise ratio was 24, with a significance above 5.1 standard deviations. These are the collaboration's findings, supported by the article's original abstract and corresponding publication record.

The ontological story has several levels. First there are instrument strain records. Second there is the inference that a propagating gravitational perturbation caused the correlated signals. Third there is the source model: a merging binary black-hole system. Fourth there are broader interpretations of gravitational waves and spacetime in metaphysics. Confidence at each level depends on particular analysis and background theory.

The study exemplifies the success of model-mediated inference. Detection was accomplished through theory-informed templates and calibration, yet the use of models does not nullify the reality of the detected event. It shows how a well-developed theory can make inaccessible processes experimentally investigable while leaving distinct philosophical questions about the ontology of spacetime open.

Ontology of Signals, Events, and Sources

The Higgs and gravitational-wave cases illustrate why an 'entity' need not mean a permanent, independently identifiable material object. Scientific descriptions also concern events, fields, processes, interactions, relations, and transient patterns. The GW150914 observational report is concerned with an event and its inferred astronomical source; the ATLAS boson analysis infers a newly produced particle from event distributions. The causal form of the evidence differs even though both reports establish strong empirical claims.

This diversity calls for ontological precision. The existence of a signal is a claim about registered data and its causal provenance; the existence of a source is a further explanatory claim; the existence of a theoretical field as a fundamental constituent may involve an additional interpretation. Failure to distinguish them can produce both exaggerated realist conclusions and exaggerated skepticism about experimental discovery.

My reading of the two cases favors ontological articulation at the level at which evidence actually operates. A particle production process, a gravitational-wave transient, and a mathematical field description may all be scientifically indispensable in different senses. Their ontological statuses must be argued for at the corresponding level of empirical and theoretical dependence.

Case Study III: Chemical Elements and Natural Kinds

IUPAC's (2025) chemical element definition distinguishes an element as a species of atoms characterized by the number of protons in the nucleus from an elementary substance made of such atoms. The distinction is compact, operationally consequential, and philosophically instructive: one familiar scientific term legitimately connects a classification of atoms to a kind of material substance. Their relation is scientifically grounded, but the semantic and ontological categories differ.

Bird and Tobin (2022), in Natural Kinds, analyze the claim that successful scientific classifications track objective divisions in the world. Chemistry supplies especially compelling examples because atomic structure constrains chemical behavior. Yet the grouping of instances into kinds can involve disciplinary purposes and choices of descriptive level. A single term's practical stability does not settle the metaphysical thesis that every kind has one timeless essence.

Chang (2012) uses the historical case of water in Is Water H2O? to examine evidence, realism, and pluralism. The documentary synopsis supports treating that work as a major historical-philosophical challenge to overly simple chemical ontology; it does not license attributing unexamined detailed arguments from its restricted chapters. Taken together, the sources show that scientific classifications may be strongly reality-constrained while their exact conceptual articulation remains a subject for philosophical assessment.

Biological Species and Ontological Pluralism

Biological classification resists a uniform account of natural kinds. Ereshefsky (2022), in the Species entry, describes major disputes over species as natural kinds, individual evolving lineages, or alternative classificatory units. Biological, phylogenetic, and ecological species concepts can track different aspects of organisms and evolutionary history. These differences are not merely verbal: they affect what researchers count, preserve, compare, and explain.

Where chemical elements can be defined through proton number, evolutionary species have lineage and reproductive histories. Their identities are shaped by descent, population processes, and practical research aims. A universal criterion imported without qualification from chemistry would erase biological facts. Conversely, a purely conventionalist interpretation would ignore the constraints that reproduction, ancestry, ecology, and evolutionary divergence place upon acceptable classifications.

Scientific ontology therefore requires domain sensitivity. Pluralism in classification can coexist with realism about organisms, ancestry, and causal processes. The challenge is to determine whether plural terms identify competing objects, different properties of the same biological systems, or multiple legitimate abstractions. This is an empirical and conceptual investigation, not a permission to treat classifications as arbitrary.

Levels of Organization and Explanatory Dependence

A single phenomenon may receive descriptions in physics, chemistry, biology, and higher-level sciences. Molecular interactions explain one aspect of a cell; ecological networks explain others. Moving between levels requires mappings and causal understanding. It does not automatically establish that every higher-level entity is either fundamental or illusory.

The naturalized metaphysics of Ladyman and Ross (2007) discusses special-science patterns, while Ereshefsky (2022) examines evolving lineages and Bird and Tobin (2022) analyze classification realism. These sources make clear that ontology across sciences must address individuation, persistence, and explanatory utility separately. The existence of an ecological population does not imply that it is elementary in the sense of a particle; its persistence may depend on organized relations among organisms and environments.

The analytical result is a distinction between dependence and dispensability. An entity can be dependent on lower-level constituents yet indispensable to reliable description and intervention at its own scale. Reductive explanations must show how relevant higher-level properties are recovered, rather than merely asserting that lower-level objects exist. Ontology becomes richer when it identifies such dependencies explicitly.

Mathematics, Structures, and Abstract Objects

Scientific theories employ sets, functions, Hilbert spaces, probability distributions, coordinate systems, and numerical solutions. Their indispensability in formulation raises a different question from the existence of electrons or cells. Quine's (1948) quantificational criterion provides a way of identifying apparent commitments to abstract objects in regimented theories, while Bricker (2014) reviews rival commitment tests. This debate concerns the semantics and metaphysics of scientific mathematics rather than the direct detection of a material mathematical object.

A physicist can use an abstract space to represent physical possibilities without committing to the claim that each coordinate point is a separate concrete substance. Nevertheless, some philosophical interpretations treat mathematical entities as indispensable and therefore ontologically serious. Alternatives seek nominalist or structural paraphrases. The disagreement cannot be resolved solely by counting symbols or equations.

Scientific ontology is best served by stating what role an abstract object plays. A mathematical entity may be a domain element, representational device, invariant relation, or indispensable explanatory resource. Those roles are not interchangeable. The logical form of a mature theory identifies a candidate commitment, and a further metaphysical account determines what kind of existence, if any, it supports.

Underdetermination and the Space of Alternatives

Empirically successful descriptions can sometimes be compatible with different accounts of underlying reality. Van Fraassen (1980), through the empirical adequacy argument, emphasizes that predictions and observations do not compel belief in the complete truth of an accepted theory. Stanford (2006) adds in the historical alternatives critique that relevant competitors may not yet have been conceived. Together these considerations constrain overconfident ontological inference.

Underdetermination is not uniform. In one situation, two models share observations but disagree about an unobservable cause. In another, the models are representational variants related by an exact transformation. In a third, different levels of description can be simultaneously true within their domains. These situations require different resolutions. The bare fact that there are two descriptions cannot establish either that they describe separate worlds or that neither refers to reality.

An ontology informed by these distinctions asks which new measurements, interventions, or theoretical constraints could discriminate among alternatives. It also records where current evidence underdetermines the metaphysical interpretation. The result is a transparent research problem rather than a denial of scientific knowledge.

Realist Selectivity and Graded Ontological Warrant

The case for realism grows stronger when a posit contributes to independently tested causal interventions, survives novel experimental contexts, supports coherent measurement across instruments, and remains identifiable through theoretical change. Chakravartty (2017) distinguishes selective realist views; Hacking (1983) emphasizes experimental manipulation; Worrall (1989) emphasizes preserved relations. Their positions vary, but each challenges the idea that a successful theory must be endorsed as one indivisible unit.

I propose here an analytical practice, not a new philosophical system: specify the kind of commitment, its evidential basis, and its current scope. In an individual case, 'electron' may be supported as an experimentally controlled causal entity, 'a model parameter' as an estimated quantity, and 'a simulated particle' as a representational object. One may reasonably have different confidence in these claims while using the same mathematical tools.

This practice makes intellectual responsibility visible. Ontological belief is strongest when warranted claims remain stable under changes in instruments, models, and descriptions. It is weaker when a claim rests on an untested explanatory layer. The distinction is a reasoned interpretation of existing literature and cases; it asserts no original discovery of a general law of scientific ontology.

What Negative Results Can and Cannot Establish

An experiment that fails to detect a predicted signal can constrain a parameter space or exclude a theory under the assumptions of the test. It normally cannot establish the absence of an entity in every possible circumstance. The interpretive force depends on sensitivity, background models, selection criteria, and the relation between a theory's prediction and its ontological commitments. Tal's (2020) measurement uncertainty analysis provides an account of these inference conditions.

A model that fails to predict a phenomenon may be wrong in its parameter values, experimental domain, causal architecture, or fundamental ontology. Those possibilities are distinguishable in follow-up work. Similarly, past rejection of a named theoretical entity does not entail that all phenomena previously explained using that entity were illusory. A historical ontology of science must preserve the recorded achievements even when its explanatory inventory changes.

The rational form of a negative ontological conclusion is conditional: given specified evidence, instruments, assumptions, and alternatives, the case for a particular posit has weakened or collapsed. A stronger universal absence claim requires its own independent justification. This approach protects both scientific discovery and scientific correction from metaphysical overstatement.

Scientific Ontology in Research Institutions and Data Systems

Ontological decisions enter research protocols and data systems through identifiers, taxonomies, measurement definitions, and choices about what counts as one object or event. A database may assign an identifier to a specimen, an experimental event, or a simulated state. Those are useful institutional commitments that make research comparable and auditable. They should be recorded with their intended reference and domain.

IUPAC's (2025) controlled chemical terminology illustrates the importance of stable definitions in interoperable scientific records. Tal (2020) on calibrated measurands shows why even numerical entries require metadata about operationalization and uncertainty. Frigg and Nguyen (2026) on representation and targets explain the conceptual problem of connecting a data structure or model to the phenomenon it represents.

Digitally codifying a classification amplifies its research consequences, but technical standardization alone does not prove metaphysical necessity. Machine-readable consistency helps investigators preserve reference through revisions; it also makes conceptual mistakes travel efficiently. Scientific ontology gains from maintaining explicit source provenance, entity definitions, version history, and qualifications beside machine-readable identifiers.

Historical Change and Ontological Continuity

Scientific concepts acquire new empirical meanings as instruments, problems, and theories change. Laudan (1981) documents realist historical challenges, while Worrall (1989) investigates continuity across change. Ontological inquiry should attend to this history because retaining a word does not guarantee retaining its referent, and replacing a word does not guarantee that its referent has disappeared.

The historical comparison must ask what survived: a mathematical relation, an experimental intervention, a systematic pattern of observations, an entity's measured properties, or a comprehensive picture of its nature. These are different forms of continuity. A later theory may preserve successful equations while changing its causal story; it may preserve the same experimental object while revising what the object is believed to be.

A philosophy that recognizes continuity and rupture together can explain progressive scientific knowledge without treating current ontology as permanent. The result is neither a simple succession of arbitrary worldviews nor a guarantee of eventual convergence to a final inventory. It is a historical account in which existence claims gain and lose warrant through identifiable scientific achievements.

Comparison with Neighboring Philosophical Inquiries

Scientific ontology occupies a particular place among related research subjects. Scientific realism asks whether our best theories are approximately true and what their success warrants. Entity realism emphasizes experimental interaction with particular unobservable entities. Ontological commitment asks what a theory presupposes within its logical and semantic form. Natural-kinds research investigates classification and kindhood. Scientific representation analyzes how a model represents a target. The boundaries are intellectually productive because each neighboring inquiry supplies a part of the larger question.

Bricker (2014) explains the metaontological task, Chakravartty (2017) reviews scientific realism, and Frigg and Nguyen (2026) analyze model representation. Their arguments show why a claim's existential presuppositions, its representational role, and the strength of belief warranted by evidence cannot be collapsed.

The present dossier synthesizes these connections around the explicit question, Which entities should science lead us to accept? It accordingly studies the passage from scientific work to justified ontological judgment rather than replacing the independent investigations reserved for adjacent dossiers.

The Place of Artificial-Era Ontology

The Aisentica Research Group's public statement presents Artificial as a philosophical category concerning a nonbiological historical order of meaning, reason, and provenance. The project declares this position in its canonical system statement. That statement is an attributable philosophical source, with a distinct subject, argument, and authorial provenance. It is important to preserve this identity when different projects use the word 'ontology.'

Scientific ontology, as treated in this dossier, concerns the entity commitments licensed by empirical and theoretically mediated scientific research. Aisentica's historical-philosophical claims can be discussed as philosophical claims about the Artificial Era and traced to their own foundational publications. They must be evaluated by their stated conceptual arguments, rather than presented as conclusions already demonstrated by particle physics, gravitational-wave experiments, or machine-learning benchmarks.

This connection illustrates a general editorial discipline: a philosophical ontology, a formal data ontology, and an empirically warranted scientific ontology are different kinds of intellectual enterprise. Their relations may be studied openly without borrowing experimental authority where the relevant evidence has not been furnished. Preserving these boundaries allows original philosophical contributions and scientific research to be understood on their own terms.

Objections and Replies

One objection says that all sufficiently successful science should be taken as literally true. The answer is the historical record: Laudan (1981) challenges convergent realism and Stanford (2006) documents unconceived alternatives give reasons to distinguish success from comprehensive truth. The selective position can accept scientific achievements while asking which components deserve lasting belief.

A second objection says that the entire ontology of science is a convenient fiction. Hacking's (1983) intervention-based argument challenges this conclusion: repeated control of an alleged entity's causal powers can make existence belief reasonable. Constructive empiricism remains an intelligible alternative in its treatment of theoretical acceptance, but it does not erase the empirical weight of stable experimental interventions.

A third objection says structural realism provides the universal answer. Ladyman's (2023) structural realism distinctions reveals why this is contested: one must explain what structure is, how its relations obtain, and whether the thesis concerns epistemic access or the furniture of reality. A fourth objection says all scientific kinds have fixed essences. Bird and Tobin (2022) review natural-kind essentialism and Ereshefsky (2022) examine species pluralism show the need for case-specific arguments. The replies support disciplined selection rather than an unrestricted metaphysical recipe.

Independent Synthesis: When Science Warrants Ontology

The investigations yield a practical hierarchy of philosophical questions. The first concerns semantic content: what the theory says under a declared interpretation. The second concerns inferential necessity: which posits are indispensable to explaining the tested results within the available alternatives. The third concerns experimental grounding: which posits participate in stable measurements, manipulations, and independent observations. The fourth concerns historical resilience: which aspects remain warranted across changing theories and research techniques. These questions are analytical checkpoints distilled from the preceding discussion, not a proposed formal model or an independently validated doctrine.

The hierarchy explains why the existence of a new boson can become well supported while its complete theoretical characterization remains open; why a computational cell can be useful without being a concrete physical cell; why a species can be real as an evolving lineage while its boundaries depend on a scientific concept; and why a gravitational-wave event can be experimentally established despite continuing metaphysical debate about spacetime.

The central conclusion is that scientific ontology is an exercise in reasoned commitment under evidential constraint. Its success is measured by the clarity with which it distinguishes what has been established, what has been inferred, what remains interpretation-dependent, and what would change a judgment. This interpretation preserves scientific realism's ambitions, empiricism's discipline, historical criticism's corrective force, and experimental research's distinctive authority.

Scholarly Audit, Limits, and Research Directions

The source audit identifies an asymmetry worth stating. The Stanford Encyclopedia essays and official IUPAC entry were readable as full accessible webpages; the original research article landing pages supplied detailed abstracts and bibliographic records, with public collaboration documentation available for cross-checking. Major philosophical books and some original articles were accessible through publisher abstracts, descriptive previews, journal catalogue entries, and scholarly cross-checks, not through unrestricted complete texts. This source-stage monograph limits attributions accordingly. It does not invent quotations, page-specific arguments, or close readings of unavailable chapters.

Future research can test the argument against detailed cases in quantum field theory, molecular biology, neuroscience, climate modeling, and machine-readable scientific knowledge graphs. A focused historical project could compare successive ontologies of one entity across laboratory notebooks, formal models, and publication archives. Such investigations would sharpen assessments of reference continuity and the evidential force of intervention.

The present dossier resolves its own philosophical and editorial questions through the research it actually documents. Remaining larger disciplinary questions are prospective research problems, not unresolved source-integrity defects in this manuscript. An external scholarly referee may be appointed in a separate process; no such review is implied by the internal decision.

Conclusion

Scientific ontology asks what kinds of things science gives us reason to believe exist. The answer emerges from the relation among the language of scientific theories, the causal structure of experiments, the representational role of models, the reliability of measurement, and the history of scientific revision. No single one of these dimensions automatically fixes the complete inventory of reality.

Quine made existential commitment logically visible; Carnap exposed the practical significance of conceptual frameworks; scientific realism and constructive empiricism clarified the difference between theory success and full belief; Hacking gave experimental intervention philosophical weight; Laudan and Stanford exposed the historical fragility of unrestricted realism; Worrall and structural realism investigated the persistence of relations across changing descriptions. The ATLAS, CMS, LIGO, IUPAC, and biological taxonomy cases show these issues operating in actual scientific practices.

The justified conclusion is specific. Science can establish compelling reasons to accept entities, events, properties, and relations that are not available to unaided observation. Such commitments gain strength through independent empirical constraints and remain proportionate to the evidence that supports them. Ontological inquiry retains its rigor by identifying exactly what was shown, what was inferred, and how future discoveries could refine the claim. This is the philosophical achievement of scientific ontology: a public, revisable account of what scientific knowledge authorizes us to say exists.