Before Method: Implicit Laboratory Epistemology and the Architecture of Scientific Inquiry

by Prof. Dr. Oleg Maltsev

Abstract

Research methodology is commonly presented as beginning with problem formulation, research design, and method selection. Yet methods do not enter inquiry neutrally. Before researchers decide how to observe, measure, compare, classify, or interpret an object, they already operate with assumptions about what counts as evidence, what warrants a conclusion, where interpretation is legitimate, what qualifies as error, and what would justify revision of an existing explanation.

Building on Susan Ramlo’s (2026) Q-methodological study of academics’ divergent views of subjectivity in the natural sciences, this article develops a broader conceptual framework for examining such assumptions within concrete research systems. It introduces implicit laboratory epistemology (ILE), conceptually described as the implicit philosophy of the laboratory: the task-specific configuration of epistemic assumptions that governs evidential selection, interpretive permission, anomaly classification, and revision criteria in a particular investigation.

The argument is positioned in relation to Polanyi’s account of tacit knowing, Knorr Cetina’s epistemic cultures, Bourdieu’s epistemic reflexivity, and contemporary work on reflexivity in AI-augmented research. The article does not claim that implicit epistemic assumptions have previously gone unnoticed. Its contribution is more specific: it relocates the unit of analysis from individual belief or disciplinary culture to the concrete research system operating under a particular task, and it proposes epistemic explication as a prospective research-design procedure for making the governing rules of knowledge production explicit before they become invisible premises of methodological judgment.

The central claim is that method is not epistemically self-sufficient. Rigorous inquiry therefore requires not only methodological transparency but also transparency about the rules by which observations become warranted claims.

Keywords: philosophy of science; research methodology; epistemology; objectivity; subjectivity; epistemic cultures; epistemic reflexivity; scientific evidence; implicit laboratory epistemology; epistemic explication.

Introduction: The Problem Before Method

Scientific research is ordinarily described through a familiar sequence. A problem is identified, a research question is formulated, an object of inquiry is specified, methods are selected, data are produced or collected, analysis is conducted, and conclusions are drawn. The sequence is useful, but it conceals a prior layer of research architecture.

A researcher does not approach method selection without already possessing some conception of what a satisfactory answer would have to look like. Long before a statistical procedure, experimental protocol, interview strategy, archival method, simulation, or observational design begins to operate, decisions have already been made—sometimes consciously, often tacitly—about what kind of material may count as evidence, how disagreement among observations should be treated, when professional judgment is permissible, and what would constitute a serious challenge to the initial explanation.

This prior layer matters because the same method can be used within substantially different epistemic arrangements. Two researchers may employ the same software, analyze the same dataset, or follow the same nominal research technique while assigning different epistemic weight to outliers, contextual information, expert interpretation, replication, convergence across sources, or anomalous cases. Methodological sameness therefore does not guarantee epistemic sameness.

The claim of this article is not that researchers possess hidden assumptions—a proposition with extensive antecedents in the philosophy, sociology, and methodology of science. Rather, the argument advances two narrower propositions.

First, epistemic assumptions can be analyzed as a task-specific configuration of a concrete research system, rather than only as personal beliefs held by individual researchers or as historically stabilized cultures shared across scientific fields.

Second, this configuration can be made explicit prospectively through a procedure termed epistemic explication, undertaken before methodological choices become treated as self-evident warrants for scientific conclusions.

The phrase before method should therefore be understood analytically rather than as a rigid chronological claim. Research design is often iterative. Assumptions, methods, data, and interpretations may be revised repeatedly. The point is that epistemic commitments are logically prior to the use of a method as a warrant: whenever a method is treated as capable of producing convincing evidence, some prior judgment has already been made about why that form of evidence deserves epistemic authority.

The practical problem is that this judgment is rarely documented with the same precision as sampling procedures, statistical models, laboratory protocols, or source-selection criteria.

This article proposes that it should be.

From Tacit Knowing to Epistemic Cultures and Reflexivity

The proposed concept enters an intellectual landscape that is already well developed. Any claim concerning unarticulated assumptions in science must therefore be distinguished carefully from several established traditions.

Michael Polanyi’s account of tacit knowing provides an obvious starting point. His work challenged the ideal of scientific knowing as completely detached, rule-governed, and exhaustively explicit. Scientific practice depends in part on embodied skill, acquired judgment, inherited practices, and forms of knowing that cannot be completely reduced to formal propositions (Polanyi, 1966/2009). The importance of Polanyi for the present argument is fundamental: not everything that governs competent scientific action is fully articulated by the actor.

Yet tacit knowledge is broader than the concept proposed here. A scientist may tacitly know how to recognize a technically successful preparation, how to operate an instrument, or when an experimental pattern looks unusual. Implicit laboratory epistemology, by contrast, refers specifically to the assumptions governing the epistemic status of what the research system produces: why one observation counts as stronger evidence than another, where interpretation is permitted, what qualifies as an anomaly, and what would justify revision of a conclusion.

Karin Knorr Cetina’s concept of epistemic cultures comes even closer to this territory. Her comparative studies of scientific practice demonstrated that scientific fields develop different arrangements, mechanisms, instruments, empirical relations, and procedures through which knowledge is produced and warranted (Knorr Cetina, 1999). Her later formulation explicitly describes epistemic cultures as concerned with the “machineries of knowledge construction” embedded in laboratory micropractices and other sites of knowledge production (Knorr Cetina, 2007).

The present framework does not replace this concept. It changes the unit and scale of analysis.

An epistemic culture concerns relatively durable collective arrangements through which a field or community produces knowledge. Implicit laboratory epistemology concerns the operative epistemic configuration of a particular research system under a particular task. A molecular biologist and a physicist may participate in distinct epistemic cultures, as Knorr Cetina demonstrated. But two studies within the same disciplinary culture may nevertheless differ substantially in the evidential thresholds they impose, the interpretive authority they grant researchers, or the way they classify anomalies.

The concept is also related to epistemic reflexivity. Bourdieu and Wacquant (1992) insisted that the intellectual conditions of knowledge production themselves must become objects of systematic scrutiny. More recent methodological work has developed epistemic reflexivity as a means of examining how researchers’ theoretical, disciplinary, cultural, and professional positions influence research practice (Guttormsen & Moore, 2023).

The distinction proposed here is therefore not between reflexive and non-reflexive science. It concerns the timing, unit of analysis, and operational target of reflexivity.

Epistemic reflexivity asks researchers to interrogate the conditions that shape how they know. Epistemic explication asks a particular research project to specify, prospectively and revisably, the rules by which it intends to recognize, weight, interpret, challenge, and revise knowledge claims.

This difference becomes especially important in contemporary AI-augmented research. Kendeou, Veletsianos, and Whetung (2026) argue that generative AI makes epistemic reflexivity increasingly necessary because AI systems can influence research framing, methodological choice, interpretation, and disciplinary convergence. Their intervention reinforces the urgency of examining epistemic assumptions, but it also clarifies the need for an additional design-level instrument capable of documenting those assumptions within specific investigations.

The present contribution is intended to occupy that narrower space.

Ramlo’s Study as a Diagnostic Case

Susan Ramlo’s (2026) study provides a useful empirical entry point because it examines academics’ assumptions about subjectivity in the natural sciences rather than merely discussing philosophical positions in the abstract.

Using Q methodology, participants sorted 40 statements addressing objectivity, empirical evidence, interpretation, bias, scientific communication, human judgment, and the social dimensions of knowledge production. Analysis of 28 Q sorts produced a two-factor solution representing two distinguishable viewpoints. As is characteristic of Q methodology, these factors should not be interpreted as population proportions or as evidence that all academics belong to one of two psychological types. The purpose is instead to identify coherent structures of viewpoint.

Ramlo labels the first orientation The Ideal is Objective Science. This position emphasizes empirical rigor, the reduction of subjective influence, statistical control, predictive capacity, and formal mechanisms of scientific scrutiny. It nevertheless accepts the revisability of scientific knowledge and therefore should not be reduced to a simplistic belief in immutable scientific truth.

The second orientation, Scientists as Human Beings, gives greater weight to the unavoidable presence of human judgment, interpretation, context, communication, and collaboration in scientific practice. It also exhibits greater caution toward assuming that expert consensus or peer review automatically guarantees epistemic reliability.

The significance of these results for the present argument lies not in choosing between the two positions. Both are recognizable attempts to solve the same problem: how can scientific knowledge remain reliable when it is produced by human beings?

More important is Ramlo’s observation that participants did not generally present themselves as explicit adherents of Popper, Kuhn, or another philosophical tradition. The philosophical resemblance becomes visible through subsequent interpretation. Epistemic orientation can therefore operate without being fully articulated as philosophy.

This finding does not establish the universal existence of a single hidden philosophy possessed by every scientist, nor does it justify extending Ramlo’s two viewpoints across all disciplines. It does, however, provide empirical support for a narrower proposition: academics can hold coherent assumptions about the production and validation of scientific knowledge without explicitly formulating those assumptions as a philosophy of science.

That proposition opens the question addressed here: what happens when such assumptions enter the architecture of a concrete investigation?

Implicit Laboratory Epistemology: Definition and Mechanism

This article proposes the term implicit laboratory epistemology (ILE) for the task-specific configuration of epistemic assumptions operating within a particular research system.

Its conceptual description is the implicit philosophy of the laboratory.

The term laboratory is deliberately broader than a physical room containing scientific instruments. An archive can function as a laboratory. So can an archaeological site, a crime scene, a clinical environment, an anthropological field, a corpus of texts, a computational simulation, a historical record, or a digital platform. What matters is not the architecture of the room but the organization of a setting in which traces, observations, measurements, testimonies, or representations are transformed into warranted claims.

ILE consists of at least four interdependent dimensions.

The first concerns evidential selection. A research system must determine which observations, traces, measurements, cases, testimonies, or outputs can legitimately function as evidence and how much weight each class should carry.

The second concerns interpretive permission. Every research system establishes, explicitly or implicitly, where professional judgment may enter the transition from observation to classification, from classification to explanation, and from explanation to conclusion. Even an investigation designed to minimize interpretation necessarily contains decisions about operationalization, measurement, inclusion, exclusion, and relevance.

The third concerns anomaly classification. Research systems must distinguish meaningful deviations from error, noise, artifacts, or irrelevant variation. That distinction is not supplied automatically by the deviation itself. It is produced through theoretical, methodological, statistical, and practical judgments.

The fourth concerns revision criteria. A research system must possess some threshold at which contrary evidence becomes sufficiently serious to require modification, suspension, or abandonment of an existing explanation.

These four dimensions operate together. Evidential selection affects what can become anomalous; interpretive permission affects how anomalies are classified; anomaly classification influences whether revision is triggered; and revision criteria determine how resistant or responsive a research system is to conflicting observations.

The implicit philosophy of the laboratory is therefore not an ornamental worldview surrounding scientific practice. It is an operational architecture of epistemic decisions.

Nor should it be treated as a stable personality characteristic. The same researcher may employ different epistemic configurations in different projects. A scholar may demand strict quantitative convergence in one investigation while assigning considerable epistemic value to expert interpretation or a singular critical case in another. ILE belongs first to the research system under a task, not to a psychological taxonomy of researchers.

Method, Evidence, Interpretation, and Error

Once this distinction is made, method itself appears differently.

A method does not merely extract information from a pre-given reality. It establishes conditions of visibility. Every methodological design makes some features of an object more accessible while rendering others less accessible, less measurable, less comparable, or irrelevant to the chosen form of inquiry.

Statistical analysis may reveal stable distributions and associations while reducing the epistemic visibility of singular cases. Long-term observation can disclose temporal context and situational dynamics while weakening forms of experimental control. Archival reconstruction can reveal trajectories across time while remaining dependent on the survival, selection, and interpretation of documentary traces. Interviews provide access to participants’ representations and experiences but do not make those representations identical to the events they describe.

This does not imply methodological relativism. It implies methodological specificity.

The appropriate question is therefore not only, “Which method should be used?” It is also, “Why is the class of evidence made available by this method epistemically adequate for the claim we intend to make?”

That second question cannot be answered by the method alone.

Consider two investigators using the same dataset and statistical procedure. One treats a small number of deviating cases as noise because the central objective is to identify a stable population-level pattern. The other treats the same cases as theoretically significant because the claimed model is universal and even one well-established counterexample would alter the explanatory structure. No statistical program can determine which epistemic objective the investigation ought to possess.

The same issue appears in interpretation. The traditional opposition between “objective data” and “subjective interpretation” is frequently too coarse to describe actual research. Interpretive judgment may enter when a variable is operationalized, when a historical source is authenticated, when a clinical response is classified, when a qualitative code is created, when a statistical model is selected, or when an anomalous observation is judged theoretically meaningful.

The relevant methodological problem is therefore not simply whether interpretation exists.

It is whether the research system can identify where interpretive judgment enters, what authority it receives, and what mechanisms constrain it.

This leads to an operational reformulation of objectivity. For methodological purposes, objectivity need not be defined as the disappearance of the researcher. It can instead be understood as a property of a research system in which consequential decisions are sufficiently visible to be criticized, tested, constrained, replicated where appropriate, or challenged by alternative evidence.

Under this interpretation, experimental controls, statistical testing, replication, independent review, source triangulation, open data, transparent coding decisions, and documented analytic choices are different mechanisms serving a common function: they reduce the amount of epistemically consequential but invisible discretion.

Error should be understood in the same way.

A deviation does not arrive already classified as an error. It may result from instrument failure, data corruption, an inappropriate model, sampling conditions, procedural inconsistency, or a previously unknown property of the object. The same observation may be discarded as noise under one epistemic configuration and investigated as a potentially transformative anomaly under another.

The methodological question is therefore not merely, “Where did the error occur?”

A prior question is required:
“By what rule did this research system decide that the observation was an error rather than evidence?”

This question reveals the implicit philosophy of the laboratory at work.

Epistemic Explication as a Prospective Research Procedure

If epistemic assumptions influence method selection, evidential weighting, interpretation, anomaly classification, and revision, methodological transparency should include a procedure for making those assumptions visible.

This article proposes epistemic explication for that purpose.

Epistemic explication is a prospective and revisable research-design procedure through which investigators specify the operative rules by which a particular study will recognize, weight, interpret, challenge, and revise knowledge claims.

It is not a declaration of allegiance to a philosophical school. A researcher does not satisfy the requirement by announcing that the project is realist, constructivist, Popperian, Kuhnian, positivist, or interpretivist. Such labels may be useful, but they remain too general to reveal the decisions that will govern a concrete investigation.

Nor is epistemic explication equivalent to preregistration. Preregistration can specify hypotheses, sampling plans, variables, exclusion criteria, and analytic strategies. Epistemic explication addresses a different level: why those forms of evidence and those decision rules are epistemically appropriate for the claims being pursued. The two procedures can therefore complement one another.

At minimum, an epistemic-explication statement should make six domains explicit: what will count as knowledge about the object; what classes of material will count as evidence and how conflicts among them will be weighted; where interpretive judgment enters the investigation; where professional discretion is permitted and how it will be constrained; how error and meaningful anomaly will be distinguished; and what kind of finding would require revision of the initial explanatory model.

These domains are not intended as a universal checklist mechanically applicable to every discipline. Their function is diagnostic. They force a research project to expose the epistemic rules that might otherwise remain embedded in routine methodological choices.

Crucially, explication must remain revisable.

Research can produce genuinely unexpected conditions. A new instrument may reveal a limitation in the initial evidential hierarchy. A field investigation may show that a classification system is inadequate. A contradictory source may force revision of assumptions about reliability. If epistemic explication were frozen at the beginning of a project, it could become another ritual of formal compliance.

Its purpose is the opposite.

Initial assumptions should be stated before they disappear into routine practice, but significant revisions should then be recorded as part of an epistemic audit trail explaining what changed, why it changed, and how that alteration affected subsequent interpretation.

This converts reflexivity from an exclusively retrospective activity into a component of research architecture.

Research Design, AI-Augmented Inquiry, Scientific Disagreement, and Methodological Audit

The practical value of epistemic explication becomes clearer when applied to contemporary research workflows.

Ramlo’s own study provides a particularly revealing example. In constructing the initial concourse for the Q study, ChatGPT based on GPT-3.5 was used to generate candidate statements from researcher-designed prompts. The resulting material was then reviewed, edited, deduplicated, conceptually organized, and reduced by the researcher to a final Q sample. Participants subsequently imposed their own configurations on those statements through Q sorting, after which factor analysis and researcher interpretation produced the reported viewpoints (Ramlo, 2025, 2026).

This workflow is important because it makes epistemic mediation unusually visible.

The final research material was not produced by the researcher alone, by the AI system alone, by the participants alone, or by statistical analysis alone. It emerged through a sequence of differently authorized decisions: literature selection, prompt construction, machine generation, researcher selection, participant sorting, factor extraction, factor retention, and interpretive characterization.

The methodological issue is therefore not whether AI introduced “subjectivity” into an otherwise neutral process. The research process already contained multiple sites of human and instrumental judgment.

The more useful question is:

Which decisions were granted epistemic authority at each stage, and by what justification?

This question will become increasingly important as generative AI enters literature review, hypothesis generation, coding, modeling, interpretation, writing, and peer review. Kendeou et al. (2026) have recently argued that AI-augmented research requires stronger epistemic reflexivity because generative systems may narrow intellectual diversity and normalize particular framings, methods, and interpretive pathways.

Epistemic explication complements that concern by demanding project-level specificity. A researcher using AI should be able to state whether the system is acting as a linguistic aid, search instrument, idea generator, analytic partner, adversarial critic, classifier, or interpretive contributor, and how outputs from that role will be granted or denied evidential authority.

The same framework can clarify scientific disagreement.

Researchers sometimes continue to exchange additional data when their actual disagreement concerns the rules by which data become convincing. One side may treat expert consensus as strong evidence; another may treat consensus as an institutional state requiring independent scrutiny. One may regard a singular case as irrelevant to a population-level claim; another may consider the same case fatal to a universal explanation. One may remove context to improve comparability; another may regard contextual removal as destruction of the phenomenon being studied.

Such disputes cannot always be resolved simply by adding observations.

The disagreement may lie one level higher, in the epistemic architecture that assigns weight and meaning to observations.

Epistemic explication therefore offers a diagnostic procedure for scientific controversies: before asking which side has “more evidence,” researchers can ask whether the parties share the same criteria for what evidence is capable of doing.

Finally, ILE can become an object of methodological audit.

A conventional audit asks whether a study followed its declared procedure. An epistemic audit asks whether the actual architecture of scientific judgment corresponds to the architecture the study claims to use.

A project may claim to minimize interpretation while relying on expert classification at a decisive stage. It may declare openness to falsification while employing exclusion criteria that systematically remove contradictory cases. It may claim empirical primacy while allowing a prior theoretical framework to determine which observations can become visible as data.

None of these conditions automatically invalidates a study.

The problem arises when epistemically consequential decisions remain undisclosed.

A mature audit should therefore evaluate not only methodological compliance but also epistemic congruence between declared principles and operative practice.

Limitations, Implications, and Conclusion

The proposed framework has important limits.

Implicit laboratory epistemology should not become a catch-all explanation for scientific behavior. Many research decisions are driven by cost, access, instrumentation, ethics, institutional requirements, disciplinary conventions, available datasets, or technical constraints rather than by philosophical commitments in any strong sense.

The framework also does not assume that every research project possesses a perfectly coherent epistemic structure. Real investigations may contain unresolved tensions. A study may simultaneously value standardized measurement and contextual interpretation; it may use probabilistic reasoning in one stage and expert judgment in another. Epistemic explication is intended partly to reveal such tensions rather than artificially eliminate them.

Nor does the argument imply that every hidden decision should become explicit. Polanyi’s work remains an important warning against assuming that competent practice can be exhaustively formalized. Scientific expertise contains tacit dimensions that cannot simply be converted into procedural language.

The relevant distinction is between tacit knowledge that enables skilled practice and epistemically consequential assumptions whose invisibility prevents their scrutiny.

Similarly, the concept should not be used to redescribe epistemic cultures at a smaller scale. Knorr Cetina’s framework concerns historically and socially organized systems of knowledge production. ILE is intentionally more local and dynamic: it concerns the configuration operating within a specific research system as it confronts a particular problem.

This scale difference generates the principal contribution of the article.

The literature has already shown that scientific knowing contains tacit dimensions, that disciplines possess different epistemic cultures, and that researchers must become reflexive about the conditions under which knowledge is produced. The present argument adds a design-level proposition:

the epistemic rules of a particular investigation can themselves be treated as an explicit, prospective, revisable component of research architecture.

This proposition changes the practical relationship between philosophy of science and scientific methodology.

Philosophy of science is usually encountered as an interpretive discipline capable of explaining what science is, how scientific knowledge changes, what objectivity means, or how evidence warrants belief. Under the framework proposed here, philosophy of science also acquires a position inside research design.

Before a project relies on a method, it should be capable of explaining why the outputs of that method are entitled to function as evidence for the intended claim.

Before an anomaly is removed, the project should be capable of explaining why its epistemic architecture treats the anomaly as noise rather than as a threat to the model.

Before expert judgment is invoked, the project should identify the authority granted to that judgment and the procedures by which it can be challenged.

Before a conclusion becomes stabilized, the project should identify what evidence would have been capable of destabilizing it.

Method remains indispensable.

But method is not epistemically self-justifying.

A method can tell researchers how an operation was performed. It cannot, by itself, determine why the result of that operation deserves the precise epistemic status subsequently assigned to it.

This is the problem before method.

The concept of implicit laboratory epistemology names the operative configuration through which a research system makes those assignments. Epistemic explication is proposed as the procedure for making that configuration visible before its assumptions harden into invisible methodological premises.

The practical aim is therefore neither to eliminate the researcher nor to celebrate subjectivity.

It is to construct research systems in which consequential epistemic decisions can be seen.

Such visibility does not guarantee truth.

It does something more modest and methodologically indispensable: it makes the architecture by which claims about reality become warranted available for examination, criticism, correction, and improvement.

The resulting open question is no longer whether scientists possess implicit philosophies of science. Existing work gives strong reasons to expect that scientific knowing is shaped by tacit assumptions, disciplinary arrangements, professional judgment, and reflexive positioning.

The more consequential question is now institutional and methodological:

Should serious research protocols require an epistemic-explication statement, alongside descriptions of methods, data sources, analytic procedures, and validation strategies, so that the rules by which evidence becomes knowledge are themselves open to scientific scrutiny?

References

Bourdieu, P., & Wacquant, L. J. D. (1992). An Invitation to Reflexive Sociology. University of Chicago Press.

Daston, L., & Galison, P. (2007). Objectivity. Zone Books.

Guttormsen, D. S. A., & Moore, F. (2023). “Thinking about how we think”: Using Bourdieu’s epistemic reflexivity to reduce bias in international business research. Management International Review, 63, 531–559. https://doi.org/10.1007/s11575-023-00507-3

Keller, E. F. (2022). Making Sense of My Life in Science. Karger.

Kendeou, P., Veletsianos, G., & Whetung, C. (2026). Epistemic reflexivity for AI-augmented research. Communications Psychology, 4, 125. https://doi.org/10.1038/s44271-026-00527-3

Knorr Cetina, K. (1999). Epistemic Cultures: How the Sciences Make Knowledge. Harvard University Press.

Knorr Cetina, K. (2007). Culture in global knowledge societies: Knowledge cultures and epistemic cultures. Interdisciplinary Science Reviews, 32(4), 361–375. https://doi.org/10.1179/030801807X163571

Kuhn, T. S. (1970). The Structure of Scientific Revolutions (2nd ed.). University of Chicago Press.

Kuhn, T. S. (1977). The Essential Tension: Selected Studies in Scientific Tradition and Change. University of Chicago Press.

Longino, H. E. (1990). Science as Social Knowledge: Values and Objectivity in Scientific Inquiry. Princeton University Press.

Polanyi, M. (2009). The Tacit Dimension. University of Chicago Press. (Original work published 1966)

Popper, K. R. (1959). The Logic of Scientific Discovery. Hutchinson.

Ramlo, S. E. (2025). Integrated data collection in Q methodology: Using ChatGPT from concourse to Q-sample to Q-sort. Journal of Mixed Methods Research, 19(2), 191–205. https://doi.org/10.1177/15586898241262824

Ramlo, S. (2026). Implicit philosophies of science: Academics’ divergent views of subjectivity in scientific practice. Science & Education. https://doi.org/10.1007/s11191-026-00780-2