When the Framework Becomes the Explanandum

​

Scientific Progress, Framework-Level Understanding, and the Limits of Within-Framework Inquiry

​

Abstract

Scientific inquiry can achieve substantial progress in explaining phenomena while leaving a different epistemic object comparatively unchanged: the framework through which those phenomena are organised as objects of inquiry. This article develops that distinction without treating normal science as deficient and without claiming that scientists lack understanding of their own disciplines. Existing work already treats scientific improvement as target-relative, distinguishes questions within and about frameworks, and investigates understanding of theoretical material itself. The problem isolated here is narrower: whether an increase in understanding achieved through a scientific framework with respect to an explanandum X warrants a corresponding claim of increased understanding of the organising framework F.

I argue that it does not do so without additional framework-directed warrant. I introduce Framework-Level Understanding (FLU) as a minimum relational architecture requiring attention to explanatory grammar, explanatory jurisdiction, bridge structure, admissibility, criteria of success, and conditions of retention and reopening. I then define Framework-as-Explanandum Backgrounding (FEB), in which a framework remains epistemically active as an organiser of productive inquiry while not functioning as a routine explanandum.

The Progress Attribution Error (PAE) is introduced as a normative evidential-error schema rather than as a claim about the historical prevalence of such reasoning. PAE occurs where an epistemic gain established for one explanandum is attributed to an enlarged explanandum without the additional warrant required by that enlargement.

Evolutionary biology serves as a deep case rather than as the foundation of the argument. Debates surrounding evolutionary developmental biology and the Extended Evolutionary Synthesis demonstrate how genuine progress on mechanisms can coexist with a distinct dispute over explanatory architecture. A non-evolutionary control from perturbative quantum field theory shows that the underlying distinction is not peculiar to evolutionary science.

The general conclusion is that within-framework understanding and framework-level understanding are distinct epistemic achievements. Once the framework itself becomes the explanandum, the scientific evidence may remain indispensable, but the question—and therefore the evidential burden—has changed.

Keywords: scientific understanding; scientific progress; scientific frameworks; explanandum; epistemic warrant; philosophy of science; evolutionary theory; Extended Evolutionary Synthesis

1. Progress in Understanding and the Framework Target

Debates about scientific progress ask what kind of epistemic improvement constitutes progress. Knowledge, truthlikeness, problem-solving capacity, predictive success, explanation, and understanding have all played major roles in this literature. Bird defends an epistemic account on which scientific progress consists fundamentally in accumulating scientific knowledge, whereas Dellsén has developed understanding-based alternatives. In his later noetic account, Dellsén explicitly formulates scientific improvement as improvement with respect to a topic or phenomenon X (Bird 2022; Dellsén 2016, 2021).

The present argument therefore does not introduce the idea that epistemic improvement must have a target.

It accepts that target-relative structure and asks what follows when the target changes.

Let F be a scientific framework and X an explanandum investigated through it. Let

U_F(X)

denote understanding of X achieved in inquiry organised through F.

The question examined here is whether

ΔU_F(X) > 0

entails

ΔFLU(F) > 0,

where FLU(F) denotes Framework-Level Understanding.

The central non-entailment is:

ΔU_F(X) > 0 ⇏ ΔFLU(F) > 0

This does not imply that understanding X is irrelevant to understanding F. Discoveries about X may expose hidden assumptions, alter causal priorities, reveal defective connections between explanatory levels, or force reconsideration of framework boundaries.

The claim is only that the two epistemic achievements are not identical and that the first does not warrant the second merely by occurring within the same scientific field.

1.1 Nested Objects of Understanding

Scientific inquiry operates on nested and overlapping objects.

A simplified sequence is:

phenomenon → mechanism → theoretical architecture → framework

This is not proposed as an ontological hierarchy. Its function is diagnostic.

A researcher may improve understanding of a mechanism while leaving unchanged the larger architecture assigning that mechanism its explanatory role. A field may explain more phenomena and integrate more evidence without thereby having reconstructed why those phenomena, mechanisms, and explanatory relations are organised in the way they are.

Disciplinary vocabulary can conceal this distinction because one name often ranges over several levels.

“Evolutionary theory,” for example, can refer to processes of population change, particular mechanisms, classes of models, historical syntheses, or the architecture coordinating evolutionary explanations. Consequently, “understanding evolution” need not denote a single epistemic achievement.

Scientific communities must aggregate local achievements under disciplinary names. Nothing is objectionable about saying that evolutionary biology, cosmology, genetics, or neuroscience has progressed.

The problem begins only when an epistemic achievement established for one target is used as warrant for a stronger claim concerning another.

The relevant question is therefore:

To which explanandum has the epistemic gain actually been established?

1.2 The Contribution

This article does not claim novelty for target-relative scientific progress, the philosophical analysis of scientific frameworks, the possibility of understanding theories themselves, or the use of frameworks as units of scientific analysis.

Its contribution is narrower.

It isolates the warrant problem created by a change of explanandum: whether an increase in understanding established for an object X through an organising framework F warrants a corresponding claim about understanding F itself.

The proposed answer is negative without additional framework-directed warrant.

Framework-Level Understanding supplies a minimum diagnostic architecture for specifying what the enlarged target contains; Framework-as-Explanandum Backgrounding identifies the condition in which the framework remains epistemically active without being the routine target of inquiry; and the Progress Attribution Error identifies the invalid evidential move that occurs when an enlargement of explanandum is not matched by an enlargement of warrant.

2. Within-Framework Inquiry and Framework-as-Explanandum Inquiry

The distinction between questions formulated within a framework and questions concerning the framework itself has clear philosophical predecessors.

Carnap distinguished internal from external questions in his discussion of linguistic frameworks. Maxwell subsequently used the language of questions and assertions within a framework and about a framework while analysing conceptual frameworks, ontology, and rules of confirmation and disconfirmation (Carnap 1950; Maxwell 1962).

Other major traditions likewise developed higher-order units of scientific analysis. Kuhn analysed paradigms and disciplinary matrices; Lakatos developed the methodology of scientific research programmes; Laudan treated research traditions as broader structures within which theories and problem-solving practices develop (Kuhn 1962/2012; Lakatos 1970; Laudan 1977).

The contribution proposed here is therefore not the discovery of higher-order scientific structures.

It concerns what follows for claims of understanding when such a structure becomes the explanandum.

2.1 What Counts as a Scientific Framework?

For the purposes of this article:

A scientific framework is a sufficiently stable organising architecture of inquiry that coordinates explanatory objects and relations, distributes methodological or explanatory roles, constrains admissible problems and standards of success, and possesses identifiable conditions of continuity and revision across a research practice.

A framework is therefore broader than a single local hypothesis but need not encompass an entire discipline, culture, or worldview.

This definition imposes a constraint on individuation.

A proposed framework must be historically and epistemically traceable. At minimum, an analysis should identify:

1. organising commitments that persist across more than one local hypothesis;

2. effects on actual research practice;

3. criteria by which some changes count as changes within the framework;

4. and at least provisional conditions under which changes would count as changes of it.

Framework boundaries may remain contested. The requirement is not perfect boundary sharpness. It is sufficient constraint to prevent the analyst from constructing F arbitrarily.

Framework individuation should therefore be understood as plural but constrained.

The argument does not assume that every scientific field possesses one uniquely correct framework boundary waiting to be discovered. Different legitimate analytical purposes may individuate partially overlapping frameworks at different scales. What prevents this pluralism from becoming arbitrary is the traceability requirement: each proposed F must correspond to organising relations identifiable in scientific practice, historical continuity, methodological commitments, explanatory allocation, or revision behaviour.

PluralityOfFrameworkIndividuation ≠ ArbitraryFrameworkConstruction

Two analyses may legitimately draw different framework boundaries. What neither may do is define F solely so that a preferred philosophical conclusion follows.

Framework individuation is itself subject to warrant.

2.2 Within-Framework Inquiry

Within-Framework Inquiry treats F primarily as an operative condition of research.

It asks:

• What caused this transition?

• Which mechanism accounts for the result?

• Which model predicts more accurately?

• Which parameter values fit the evidence?

• How do two processes interact?

• Which experiment discriminates between competing hypotheses?

These questions can be exceptionally difficult.

Calling them “within-framework” does not make them routine, superficial, or epistemically subordinate.

Nor does within-framework inquiry require consensus. A mature science may contain intense disputes about causal mechanisms, measurements, model structure, explanatory weight, and interpretation while sharing enough architecture to make those disputes intelligible as parts of one scientific enterprise.

The relevant structure is:

F → Q(X)

The framework helps organise a question whose primary target is something other than the framework.

2.3 Framework-as-Explanandum Inquiry

The relation changes when the framework itself becomes the object to be understood.

The questions now include:

• What explanatory objects does the framework organise?

• Which causal or methodological roles does it allocate?

• Which processes receive primary, enabling, constraining, or derivative status?

• Which transitions between explanatory levels are licensed?

• Which questions are treated as admissible?

• What counts as a successful result?

• Which revisions preserve framework identity?

• What conditions justify reopening the architecture itself?

The transition is:

F: organising condition → F: explanandum

This is the Change-of-Explanandum Principle:

When a scientific framework itself becomes the object to be understood, successful inquiry conducted through that framework remains relevant but is not sufficient merely by virtue of being successful. The explanatory target has changed.

No paradigm replacement follows from this principle.

Framework-level inquiry can conclude:

Retain(F)

Clarify(F)

Extend(F)

Redistribute(F)

or, in some cases,

Replace(F)

The required shift is from:

Q(X)

to:

Q(F)

not necessarily from one framework to another.

2.4 The Evidential Burden Changes

Within-framework evidence remains indispensable.

Experiments, mechanisms, predictions, anomalies, model comparisons, and failed explanations can all constrain claims about F.

The distinction is therefore not science versus philosophy.

Rather:

the same evidence may acquire a different evidential role when the explanandum changes.

Suppose a field repeatedly improves models using variables a, b, c.

This may substantially increase understanding of X.

But if the question becomes:

Why are a, b, c privileged as explanatory variables?

additional measurements of a, b, c do not automatically settle the variable-selection architecture.

The second question may require comparative analysis, historical reconstruction, cross-level evidence, or explicit examination of alternatives.

The epistemic burden changed because the explanandum changed.

3. Framework-Level Understanding

Existing philosophy of science already addresses understanding of phenomena, explanations, theories, theoretical representations, and subject matters.

Dellsén's objectual account treats understanding in terms of grasping a sufficiently accurate and comprehensive model of a phenomenon's dependence relations. De Regt develops a contextual account centred on intelligibility and conceptual tools, including the historical and contextual variability of criteria for intelligibility (Dellsén 2020; de Regt 2017).

Nor is understanding theoretical material itself unprecedented. Vorms explicitly frames scientific understanding as including understanding of the theoretical material itself and distinguishes representational from computational aspects of theory understanding (Vorms 2007).

The target isolated here is different.

It is not merely a theory's content, logical structure, representational format, or usability. It is the organising architecture of inquiry within which explanatory objects, roles, admissibility conditions, success criteria, and revision relations are coordinated.

The proposal borrows the general idea of structured or dependency-based grasp while deliberately changing the type of object to which it is applied. Dellsén's objectual account is formulated around phenomena or topics X, including their dependence relations; FLU should therefore be understood as an analogous but distinct construction rather than as a direct instance of Dellsén's account.

For the class of questions examined here, adequate Framework-Level Understanding requires at least a sufficiently accurate grasp of six interdependent dimensions:

FLU(F) ⊇ Grasp[G, J, B, A, C, R]

where:

• G = explanatory grammar

• J = explanatory jurisdiction

• B = bridge structure

• A = admissibility regime

• C = criteria of success

• R = retention and reopening conditions

The list is deliberately non-exhaustive.

Instrumentation, representational practices, exemplars, values, institutional arrangements, or other elements may need to be added in particular cases.

3.1 Explanatory Grammar — G

Explanatory grammar concerns the recurrent organisation of explanatory relevance.

It asks:

• Which kinds of causes or dependencies count as explanatory?

• Are explanations primarily mechanistic, statistical, historical, structural, dynamical, functional, or mixed?

• Which distinctions organise the problem-space?

• What counts as an explanatory unit?

• Which relations are treated as primary?

“Grammar” does not imply literal syntax.

It denotes patterned constraints on what can perform explanatory work.

Understanding G therefore means grasping more than terminology. It means grasping how the framework structures explanatory significance.

3.2 Explanatory Jurisdiction — J

Frameworks do not merely contain mechanisms.

They allocate roles among them.

Jurisdiction asks:

• Which process explains which phenomena?

• Which mechanisms are primary, enabling, constraining, or derivative?

• Where do explanatory levels begin and end?

• When are mechanisms complementary rather than competing?

• Which disputes alter causal weights without altering framework architecture?

One may understand M₁ and M₂ increasingly well while continuing to disagree about:

J(M₁, M₂)

Understanding the mechanisms and understanding the distribution of explanatory jurisdiction are related but distinct achievements.

3.3 Bridge Structure — B

Many scientific explanations cross levels or scales:

molecular → cellular

individual → population

micro → macro

Such transitions require warrant.

Bridge structure includes:

• mappings between variables;

• scale relations;

• reduction or emergence assumptions;

• correspondence rules;

• idealisations;

• inferential premises;

• and criteria determining when evidence at one level supports claims at another.

Thus:

Evidence(Lᵢ) ⇏ Warrant(Lᵢ → Lⱼ)

The bridge may be strong, weak, provisional, contested, or absent.

3.4 Admissibility Regime — A

No science treats every formulable question as an equally legitimate research problem.

The framework therefore helps structure admissibility.

Relevant questions include:

• Which problems are scientifically tractable?

• Which variables may enter explanation?

• Which evidence is recognised?

• Which questions are premature, external, or methodologically inaccessible?

• Under what conditions could an excluded question become admissible?

• Is a current non-answer represented as a methodological limitation or as a limitation on meaningful inquiry?

Framework-Level Understanding does not require abolishing boundaries.

It requires understanding their epistemic status.

3.5 Criteria of Success — C

Scientific research requires standards by which one result counts as an improvement over another.

These may include:

• predictive accuracy;

• empirical fit;

• causal resolution;

• robustness;

• explanatory scope;

• unification;

• intervention;

• formal tractability.

Framework success is significant evidence.

But successful performance under the framework's operative standards does not automatically answer every question about the architecture producing and evaluating that success.

A framework may perform exceptionally under C_F while still containing a disputed bridge, an unclear jurisdictional boundary, or poorly articulated reopening conditions.

3.6 Retention and Reopening — R

Scientific frameworks survive substantial change.

They can absorb new mechanisms, measurements, subfields, parameters, and causal emphases.

Two questions therefore arise:

What can change while F remains F?

and:

What would justify reopening F itself?

Kuhn, Lakatos, and Laudan provide different resources for understanding scientific continuity under revision. Lakatos analyses scientific research programmes as persisting methodological structures; Laudan's research traditions permit theory change inside a broader unit of scientific development.

FLU therefore requires some account of:

• permissible revision;

• anomaly or failure conditions;

• boundary disputes;

• comparative tests;

• and the distinction between local adjustment and architectural change.

A framework that can absorb every possible outcome without relevant revision differs epistemically from one with identifiable failure conditions.

Yet a framework that reopened at every anomaly could not sustain cumulative research.

Understanding R therefore requires grasping both stability and corrigibility.

3.7 Competent Use and Objectual Framework Understanding

A strong objection now arises.

Surely competent scientific practice already requires understanding the framework.

This should be partly accepted.

A first-rate scientist cannot operate successfully while knowing nothing about admissible questions, explanatory forms, relevant evidence, accepted methods, or standards of success.

Thus:

CompetentUse(F) ⇒ PartialGrasp(F)

is plausible.

But it does not follow that:

CompetentUse(F) ⇏ ComprehensiveObjectualFLU(F)

The distinction is between framework uptake sufficient for successful practice and explicit reconstruction of the framework as the object whose interdependent structure is being understood.

Recent work reinforces the need not to treat understanding in science as one homogeneous epistemic state. Boge, Schuster, and Stoll distinguish individual scientists' understanding—including forms that may remain objectual or difficult to articulate—from scientific understanding that must be communicable within a scientific community (Boge, Schuster, and Stoll 2026).

Their distinction concerns communicability rather than the X/F explanandum distinction developed here, but it supplies adjacent prior art for separating different epistemic achievements within scientific cognition.

A scientist may therefore understand how to use a bridge without having made the warrant for that bridge an explicit object of inquiry.

A community may reproduce disciplinary standards reliably without representing them as one unified dependency structure.

FLU asks whether the organising architecture itself has become the explanandum.

3.8 Why These Dimensions? An Omission Test

The six dimensions are not proposed merely because each is philosophically interesting. Their inclusion can be tested by asking what type of framework-level understanding becomes unavailable if one dimension is omitted.

If G, explanatory grammar, is omitted, one may know the framework's claims and mechanisms without knowing what kinds of relations are permitted to perform explanatory work.

If J, explanatory jurisdiction, is omitted, one may understand each mechanism individually without understanding how explanatory authority is distributed among them.

If B, bridge structure, is omitted, one may understand the objects at several levels without knowing what licenses inferential movement between those levels.

If A, admissibility, is omitted, one may know how accepted problems are solved without understanding how the framework determines which questions, variables, or forms of evidence can enter inquiry in the first place.

If C, criteria of success, is omitted, one may catalogue scientific results without understanding the standards by which those results count as improvement within the framework.

If R, retention and reopening, is omitted, one may describe the framework at a moment in time without understanding what preserves its identity under revision or what could make its organising architecture itself revisable.

The six dimensions therefore correspond to six distinguishable information deficits:

−G ⇒ explanatory-form deficit −J ⇒ jurisdiction deficit −B ⇒ cross-level warrant deficit −A ⇒ problem-space deficit −C ⇒ progress-criterion deficit −R ⇒ continuity/corrigibility deficit

This does not establish that the list is exhaustive.

It establishes the weaker but sufficient point that none of the six dimensions is redundant for the particular form of Framework-Level Understanding developed here.

Accordingly,

FLU(F) ⊇ Grasp[G, J, B, A, C, R]

should be read as a minimum diagnostic profile whose components perform non-interchangeable epistemic functions, not as a complete ontology of scientific frameworks.

4. Normalisation Without Epistemic Failure

Framework-Level Understanding would become implausible if every absence of framework analysis counted as a defect.

Mature science requires selective stability.

A scientific field can pursue difficult specialised questions partly because it does not reopen every organising commitment in every investigation. Kuhn's account of normal science remains a central precedent for this relation between comparatively stable disciplinary commitments and productive puzzle-solving.

Framework stability can therefore be an epistemic resource.

A population geneticist need not reconstruct the identity of evolutionary theory before estimating a parameter. A developmental biologist need not reassess the Modern Synthesis before examining a regulatory interaction.

The research is not deficient because F is not its explanandum.

It is appropriately targeted.

4.1 Framework Normalisation

A framework is normalised when its central commitments function mainly as stable conditions for selecting problems, interpreting evidence, allocating explanatory roles, and evaluating solutions rather than as routine first-order explananda.

Normalisation does not mean unanimity.

A field can contain severe internal disagreement.

Nor does normalisation mean immutability.

A normalised framework can remain corrigible.

The critical distinction is:

not presently investigated as an explanandum ≠ immune from investigation as an explanandum.

The first is frequently ordinary scientific specialisation.

The second would raise a separate problem of epistemic closure.

4.2 Background Is Not Inactivity

A normalised framework continues to organise inquiry even when it is not explicitly discussed.

It affects:

• which objects are grouped together;

• which variables receive explanatory importance;

• what counts as an anomaly;

• what evidence is relevant;

• which disputes remain internal;

• and what constitutes a satisfactory solution.

Therefore:

BackgroundStatus(F) ≠ EpistemicInactivity(F)

No psychological inference follows.

A scientist may be fully capable of articulating framework assumptions when required. Normalisation concerns the structure and distribution of inquiry, not the mental limitations of individual researchers.

5. Framework-as-Explanandum Backgrounding

A framework may remain active as an organiser of inquiry while not functioning as a routine object of inquiry.

I call this Framework-as-Explanandum Backgrounding (FEB):

FEB(F) ⇔ F remains an organising structure while F is not a routine explanandum of inquiry conducted through it.

FEB is not blindness.

It is not suppression.

It is not dogmatism.

It describes a distribution of epistemic labour.

Most research may target:

Q(X)

while a smaller class targets:

Q(F)

A scientifically productive framework can support this arrangement because successful normal inquiry generates specialised methods, models, datasets, instruments, and increasingly tractable local problems.

The claim is not that success necessarily causes backgrounding.

It is simply that substantial within-framework productivity is compatible with comparatively little routine framework-as-explanandum activity.

5.1 Reactivation

Backgrounding is reversible.

A new mechanism may not fit inherited explanatory jurisdiction.

A cross-level inference may become controversial.

An anomaly may expose an architectural assumption.

Two subfields may discover that their local disagreement actually concerns incompatible bridge structures.

At such points:

F: background organiser → F: foreground explanandum

Reactivation need not result in replacement.

It may yield clarification, extension, redistribution, partial reconstruction, or confirmation.

The analytically important event is the change in explanatory target.

6. Evolutionary Science as a Deep Case

Evolutionary biology is used here as an illustrative deep case, not as the foundation of the general argument and not as evidence of scientific uniqueness.

The argument of Sections 1–5 remains intact if this section is removed.

The case is useful because evolutionary biology combines substantial within-field scientific progress with explicit disputes concerning the architecture of evolutionary explanation.

It also permits a worked demonstration of how the space of relevant explanatory questions changes when evolutionary theory moves from organising condition to explanandum.

6.1 The Modern Synthesis as an Organising Architecture

The Modern Synthesis integrated major strands of genetics, systematics, comparative morphology, palaeontology, and evolutionary theory during the twentieth century. Müller characterises it as the prevailing theoretical framework produced through major disciplinary integrations in the 1930s and 1940s (Müller 2007).

In the vocabulary proposed here—not as terminology attributed to its historical architects—the synthesis can be reconstructed as organising relations among:

G, J, B, A, C, R.

It enabled specialised evolutionary problems to be pursued without requiring the identity of evolutionary theory to be reconstructed in every study.

That is a case of framework normalisation.

6.2 Evo-Devo and Explanatory Jurisdiction

Evolutionary developmental biology generates two distinguishable questions:

How do developmental mechanisms operate?

and:

What explanatory role should development occupy within evolutionary theory?

Müller's analysis characterises evo-devo as investigating mechanistic relations between development and evolutionary change and argues that its theoretical implications take evolutionary theory beyond important boundaries of the Modern Synthesis (Müller 2007).

For present purposes, the substantive correctness of that conclusion is not required.

The important point is that:

ΔU(developmental mechanisms) > 0

does not by itself settle:

J_evolution(development)

The empirical discoveries constrain the architecture-level question.

They do not make the two questions identical.

6.3 Framework Reactivation and the EES Debate

The 2014 Nature exchange on whether evolutionary theory required a “rethink” makes the distinction especially visible. The debate explicitly concerned disagreement over which evolutionary processes should be regarded as fundamental and whether phenomena emphasised by proponents of an Extended Evolutionary Synthesis required substantial reorganisation of evolutionary theory or were already accommodated within existing theory (Laland et al. 2014).

This is framework-level disagreement.

In the vocabulary developed here it concerns especially:

J, B, R.

What belongs at the explanatory core?

How should levels and processes be connected?

How much can the inherited architecture absorb while retaining identity?

The framework itself becomes an explicit explanandum.

Yet replacement does not follow.

That is precisely why the case is useful: it demonstrates framework reactivation without making scientific revolution a prerequisite.

6.4 What Changes When Evolutionary Theory Becomes the Explanandum? A Worked Demonstration

The distinction can now be made more concrete. Sections 6.2 and 6.3 show that progress on evolutionary mechanisms and disagreement over evolutionary architecture are not identical epistemic events. A further question is what actually changes in inquiry once evolutionary theory itself, rather than an evolutionary phenomenon, becomes the explanandum.

Within evolutionary biology, investigators may ask what generates heritable variation, what changes population composition, how selection, drift, mutation, recombination, developmental processes, plasticity, or other mechanisms contribute to evolutionary transformation, and how these processes interact. Progress can occur by discovering a new mechanism, refining an existing one, redistributing explanatory importance among several mechanisms, or establishing previously unknown relations among them.

Such inquiry can therefore involve what may be called, in a limited diagnostic sense, ontological substitution or redistribution: one proposed biological process is replaced, supplemented, constrained, or repositioned by another biological process while the explanandum remains biological transformation.

Nothing is epistemically defective about this mode of inquiry. If the question is how populations, inherited structures, lineages, or species change, then the relevant explanatory burden is principally biological. Better evidence concerning biological mechanisms is precisely what the question requires.

But now change the explanandum.

Instead of asking:

What processes explain evolutionary transformation?

ask:

What is evolutionary theory as an organising explanatory framework?

The second question does not merely request another mechanism.

Additional measurements of mutation rates, selection coefficients, developmental interactions, or population change may remain relevant, but they no longer exhaust the evidential burden. A different question-space becomes available:

• Which explanatory form organises the relevant mechanisms?

• Which processes receive primary, enabling, constraining, or derivative roles?

• Which transitions between explanatory levels are licensed?

• Which claims possess warrant, and for which explananda?

• How did particular explanatory commitments acquire or lose scientific authority?

• Which bridge propositions permit conclusions to move between explanatory levels or domains?

• Which historical or institutional carriers preserved, translated, or amplified particular interpretations?

• Which changes preserve the identity of the framework, and which would constitute substantial reconstruction?

• Which rival explanatory grammars could organise some of the same phenomena differently?

The first-order FLU architecture remains the six-dimensional structure already specified in this article. Questions concerning scientific authority, historical formation, institutional carriers, or rival grammars become relevant only where the enlarged framework-level claim is itself historical, institutional, comparative, or genealogical. They are not introduced here as additional mandatory dimensions of FLU.

These are not alternative biological causes of evolutionary change.

Scientific authority is not a mechanism of speciation. Institutional transmission is not a substitute for mutation. Intellectual genealogy does not compete with natural selection. A rival explanatory grammar does not become a biological process merely because it is used to analyse biological phenomena.

The relevant transition is therefore not:

biological mechanism → epistemic mechanism

as though epistemology had replaced ontology inside the same explanation.

It is:

biological explanandum → framework explanandum

and consequently:

mechanism-directed explanatory vocabulary → framework-directed explanatory vocabulary.

This distinction clarifies the role of Framework-Level Understanding. The dimensions introduced earlier—explanatory grammar, jurisdiction, bridge structure, admissibility, criteria of success, and retention or reopening conditions—do not constitute additional mechanisms within evolutionary biology. They specify features of evolutionary theory that become direct objects of inquiry only when the target has been enlarged from evolutionary phenomena to the architecture through which those phenomena are organised.

The contrast can therefore be stated in two question-spaces.

Within-framework inquiry

How does evolutionary transformation occur?

The appropriate answer-space contains biological processes, mechanisms, relations, constraints, models, and evidence.

Framework-as-explanandum inquiry

What makes these processes, mechanisms, relations, constraints, and standards parts of this organised explanatory framework, and what warrants the roles assigned to them?

The appropriate answer-space now also contains explanatory architecture, jurisdiction, bridge warrant, historical formation, scientific authority, framework continuity, and alternatives.

The second inquiry does not supersede the first. It becomes relevant because the explanandum has changed.

This is the practical meaning of the Change-of-Explanandum Principle. Scientific progress can continue within evolutionary biology while the framework remains principally an organising condition of that progress. Once evolutionary theory itself becomes the explanandum, the same scientific achievements remain evidence, but they are repositioned within a larger evidential problem.

6.5 Evidential Limit

The evolutionary case does not establish that:

• evolutionary biology is dogmatic;

• the Modern Synthesis is false;

• the Extended Evolutionary Synthesis is correct;

• development was suppressed;

• evolutionary scientists lack framework understanding;

• or evolutionary theory is unique.

It establishes only that a successful science can contain both:

genuine progress on mechanisms

and:

genuine disagreement about explanatory architecture.

The first does not settle the second.

6.6 A Non-Evolutionary Control: Perturbative Quantum Field Theory

The distinction is not peculiar to evolutionary biology.

Antoniou and Thébault analyse perturbative quantum field theory as a framework for theories, with individual theories such as quantum electrodynamics and quantum chromodynamics specified within that framework. Their case is especially useful because it begins from empirically successful theories while identifying a higher-order problem concerning the relation between theories, mathematical idealisations, theoretical models, and empirical adequacy. They consequently suggest that philosophical analysis may sometimes need to look beyond the individual theory as the unit of analysis when evaluating empirical adequacy (Antoniou and Thébault 2025).

The case does not instantiate the six-dimensional FLU architecture in full, and it should not be presented as confirmation of that architecture.

Its role is narrower.

It supplies an independent control for the central non-entailment:

exceptional success of inquiry within an established scientific architecture ⇏ settlement of every higher-order question concerning that architecture.

The evolutionary case therefore illustrates one form of framework reactivation; the QFT case shows that the underlying epistemic distinction is not generated by evolutionary controversy itself.

7. The Progress Attribution Error

The central error can now be stated precisely.

The Progress Attribution Error (PAE) is introduced here as a normative evidential-error schema, not as an empirical claim about how frequently scientists, disciplines, or institutions actually reason in this way.

PAE occurs whenever an epistemic gain established for one explanandum is used, without additional warrant appropriate to the enlarged target, to support a claim of epistemic gain concerning another explanandum.

In the present case:

PAE: ΔU_F(X) > 0 ∴ ΔFLU(F) > 0

The antecedent may be entirely true.

The inferential step may nevertheless be unwarranted.

Whether this pattern is historically frequent, rare, discipline-specific, or institutionally concentrated is a separate empirical question.

Nothing in the argument requires a prevalence claim.

The thesis is conditional:

Where the target of an epistemic claim is enlarged, the warrant must be adequate to the enlarged explanandum.

7.1 Why This Is Not the Triviality That X ≠ F

The substantive point is not merely that different objects are different.

Scientific disciplines aggregate epistemic achievements across nested objects under common names. Such aggregation is indispensable.

PAE has a more demanding structure:

1. epistemic improvement is validly established for E₁;

2. E₁ is nested within, organised by, or conventionally grouped with E₂;

3. the improvement claim is enlarged from E₁ to E₂;

4. warrant appropriate to the enlarged target is not supplied.

The issue is therefore not linguistic ambiguity.

It is:

scope enlargement without warrant enlargement.

7.2 Internal Progress Remains Genuine

Suppose a framework supports increasingly successful investigation.

Researchers identify new mechanisms, improve causal discrimination, resolve anomalies, increase predictive reliability, and integrate evidence.

Those achievements remain scientific achievements.

PAE begins only if they are used automatically to establish that:

G, J, B, A, C, R

have themselves become correspondingly better understood.

That may be true.

But it must be demonstrated.

Internal results can improve FLU. A mechanism may expose a jurisdiction problem. An anomaly may reveal a hidden bridge assumption. A failed prediction may alter reopening conditions.

The non-entailment prohibits only automatic transfer.

7.3 Success Is Evidence, Not Exhaustive Understanding

A highly productive framework has important evidence in its favour.

A framework that systematically failed to generate reliable inquiry would face an obvious epistemic problem.

But productivity alone does not exhaust every question about:

• explanatory allocation;

• cross-level warrant;

• boundaries;

• admissibility;

• success standards;

• or revision conditions.

Thus:

Success through a framework is evidence about the framework, but it is not an exhaustive substitute for understanding its architecture.

7.4 PAE Is Not Psychological

No scientist needs consciously to perform the inference.

PAE can appear in:

• textbook narratives;

• retrospective histories;

• institutional descriptions;

• interdisciplinary arguments;

• public claims about scientific maturity;

• philosophical accounts of a discipline.

The test is structural:

Was an epistemic gain established for one target used as warrant for a stronger claim about another target without satisfying the enlarged burden?

No accusation of bad faith is required.

7.5 Correcting PAE

The correction is not permanent foundationalism.

It is:

Match evidential burden to explanandum.

If the claim concerns a mechanism, mechanism-level evidence may suffice.

If it concerns jurisdiction, relations among mechanisms must be assessed.

If it concerns a bridge, the warrant connecting levels must be evaluated.

If it concerns framework continuity, retention conditions become relevant.

If it concerns reopening, failure conditions and alternatives matter.

Not every scientific investigation requires FLU.

Every framework-level claim does require warrant appropriate to a framework-level target.

8. What the Argument Does Not Claim

The limits below are constitutive of the argument.

8.1 Normal Science Is Not Epistemically Defective

Normalisation permits cumulative specialisation. The article does not advocate continuous foundational reopening.

8.2 Scientists Need Not Lack Framework Understanding

Expert practice ordinarily contains partial, tacit, and often sophisticated framework knowledge. The distinction concerns operational uptake versus framework architecture as an explicit objectual target.

8.3 Framework Analysts Possess No Automatic Superiority

A philosopher, historian, or scientist can reconstruct a framework incorrectly. Changing the explanandum does not confer truth.

8.4 Framework Inquiry Does Not Replace Empirical Science

Experiments, models, mechanisms, anomalies, and predictions constrain framework-level conclusions.

8.5 Every Investigation Need Not Reopen F

FLU is target-relative. It becomes required when the claim itself concerns F.

8.6 Framework Persistence Does Not Imply Dogmatism

A stable framework may persist because it remains empirically and explanatorily successful.

8.7 Reopening Does Not Imply Replacement

Framework-level inquiry can rationally vindicate an existing framework.

8.8 The Article Does Not Resolve the EES Controversy

The evolutionary example demonstrates the distinction between mechanism questions and architecture questions. It does not adjudicate the substantive evolutionary debate.

8.9 Evolution Is Not Claimed to Be Unique

The argument is general and does not depend on uniqueness.

8.10 G, J, B, A, C, R Are Not Exhaustive

They constitute a minimum diagnostic architecture for this problem, not a complete ontology of scientific frameworks.

8.11 Framework Boundaries Cannot Be Freely Stipulated

Framework individuation must be constrained by persistent organising commitments and research practice.

8.12 FLU Does Not Require a Framework-Free Standpoint

If F₁ is analysed using resources organised by F₂, no infinite regress follows.

The current explanandum is F₁.

If F₂ later becomes the explanandum, a new inquiry begins.

Framework-Level Understanding is target-relative, not a demand for epistemic self-foundation.

8.13 Higher-Order Analysis Is Not Intrinsically Deeper

A correct mechanism-level explanation may be a greater epistemic achievement than a weak metatheoretical interpretation.

The relevant criterion is:

AdequacyRelativeToExplanandum.

9. When the Framework Becomes the Explanandum

The argument reduces to three propositions.

First:

ΔU_F(X) > 0 ⇏ ΔFLU(F) > 0

An increase in understanding established through F with respect to X does not automatically constitute an increase in understanding of F.

Second:

F: organising condition → F: explanandum

is an epistemically meaningful shift even when F is not replaced.

Third:

PAE = scope enlargement of epistemic achievement without corresponding enlargement of warrant.

Together these claims distinguish two achievements that scientific language can easily compress into one.

A field may become extremely good at using a framework. Its models may improve, its measurements become more precise, its causal accounts deepen, and its predictions become more reliable.

All of this can constitute genuine scientific understanding.

Yet if the question becomes:

What architecture makes these achievements members of one organised explanatory programme?

then the explanandum is different.

The inquiry must now address how explanatory grammar, jurisdiction, bridges, admissibility, standards of success, and revision conditions interact.

This is Framework-Level Understanding.

The required shift is therefore not primarily:

F₁ → F₂

It is:

Q(X | F) → Q(F)

A framework may survive that inquiry unchanged.

It may emerge better justified precisely because it has been made an explanandum and survived the corresponding burden.

The article therefore does not oppose normal science, scientific specialisation, or established theoretical frameworks.

Its demand is narrower:

Do not enlarge the scope of an epistemic achievement without enlarging the warrant to match the explanandum.

Once the framework becomes the explanandum, the science may remain intact.

The question has changed.

References

Antoniou, Antonis, and Karim P. Y. Thébault. 2025. “Theories without Models: Uncontrolled Idealizations in Particle Physics.” Synthese 205, article 3.

Bird, Alexander. 2022. Knowing Science. Oxford: Oxford University Press.

Boge, Florian J., Annika Schuster, and Frauke Stoll. 2026. “Disentangling Scientific and Scientists’ Understanding.” Synthese 207, article 243.

Carnap, Rudolf. 1950. “Empiricism, Semantics, and Ontology.” Revue Internationale de Philosophie 4: 20–40.

Dellsén, Finnur. 2016. “Scientific Progress: Knowledge versus Understanding.” Studies in History and Philosophy of Science Part A 56: 72–83.

Dellsén, Finnur. 2020. “Beyond Explanation: Understanding as Dependency Modelling.” British Journal for the Philosophy of Science 71 (4): 1261–1286.

Dellsén, Finnur. 2021. “Understanding Scientific Progress: The Noetic Account.” Synthese 199: 11249–11278.

de Regt, Henk W. 2017. Understanding Scientific Understanding. Oxford: Oxford University Press.

Kuhn, Thomas S. 2012 [1962]. The Structure of Scientific Revolutions. 4th ed. Chicago: University of Chicago Press.

Lakatos, Imre. 1970. “Falsification and the Methodology of Scientific Research Programmes.” In Criticism and the Growth of Knowledge, edited by Imre Lakatos and Alan Musgrave, 91–196. Cambridge: Cambridge University Press.

Laudan, Larry. 1977. Progress and Its Problems: Towards a Theory of Scientific Growth. Berkeley: University of California Press.

Laland, Kevin N., Tobias Uller, Marc Feldman, et al. 2014. “Does Evolutionary Theory Need a Rethink?” Nature 514: 161–164.

Maxwell, Grover. 1962. “Theories, Frameworks, and Ontology.” Philosophy of Science 29 (2): 132–138.

Müller, Gerd B. 2007. “Evo–devo: Extending the Evolutionary Synthesis.” Nature Reviews Genetics 8: 943–949.

Vorms, Marion. 2007. “Understanding Theories: Formats Matter.” Paper presented at the European Philosophy of Science Association Congress, Madrid, 13–17 November 2007.

Related Source and Reference Pages

This article asks when inquiry must shift from explanation through a framework to explanation of the framework itself. The sources below connect that general Change-of-Explanandum architecture to the historical reconstruction of developmental form and to the positioning page that keeps framework-level analysis distinct from historical evidence.

The Epistemic Career of Developmental Form

This article reconstructs developmental form as a thin explanatory structure of historically organized becoming and examines its historical and epistemic career. It separates pre-Darwinian developmental form from Darwinian scientific reauthorization, historical portability from epistemic transferability, source warrant from bridge warrant, intellectual and institutional carriers from adoption, and function from genealogy. The article does not claim that Darwin originated developmental form, that evolutionary theory became a master ontology of modernity, or that institutionalization establishes truth. Its purpose is to show why explanatory form, scientific warrant, authority, transmission, institutional use, and rival explanatory grammars require separate evidential treatment.

Knowledge Through a Framework, Knowledge of the Framework

This positioning page coordinates two independently warranted studies on target-sensitive scientific understanding. It distinguishes progress achieved through an organizing framework from progress in understanding that framework itself and relates the general Change-of-Explanandum / Framework-Level Understanding architecture to an independent historical-epistemic reconstruction of developmental form. The papers are analytically ordered but evidentially independent: the historical study does not validate Framework-Level Understanding, and the general framework argument does not depend on its historical conclusions. The project’s governing discipline is that a change in explanandum requires a corresponding change in evidential burden.