Sabina Leonelli

I propose a framework that explicates and distinguishes the epistemic roles of data and models within empirical inquiry through consideration of their use in scientific practice. After arguing that Suppes’ characterization of data models falls short in this respect, I discuss a case of data processing within exploratory research in plant phenotyping and use it to highlight the difference between practices aimed to make data usable as evidence and practices aimed to use data to represent a specific phenomenon. I then argue that whether a set of objects functions as data or models does not depend on intrinsic differences in their physical properties, level of abstraction or the degree of human intervention involved in generating them, but rather on their distinctive roles towards identifying and characterizing the targets of investigation. The paper thus proposes a characterization of data models that builds on Suppes’ attention to data practices, without however needing to posit a fixed hierarchy of data and models or a highly exclusionary definition of data models as statistical constructs.

I propose a framework that explicates and distinguishes the epistemic roles of data and models within empirical inquiry through consideration of their use in scientific practice. After arguing that Suppes’ characterization of data models falls short in this respect, I discuss a case of data processing within exploratory research in plant phenotyping and use it to highlight the difference between practices aimed to make data usable as evidence and practices aimed to use data to represent a specific phenomenon. I then argue that whether a set of objects functions as data or models does not depend on intrinsic differences in their physical properties, level of abstraction or the degree of human intervention involved in generating them, but rather on their distinctive roles towards identifying and characterizing the targets of investigation. The paper thus proposes a characterization of data models that builds on Suppes’ attention to data practices, without however needing to posit a fixed hierarchy of data and models or a highly exclusionary definition of data models as statistical constructs.

Many biologists appeal to the so-called Krogh principle when justifying their choice of experimental organisms. The principle states that “for a large number of problems there will be some animal of choice, or a few such animals, on which it can be most conveniently studied”. Despite its popularity, the principle is often critiqued for implying unwarranted generalizations from optimal models. We argue that the Krogh principle should be interpreted in relation to the historical and scientific contexts in which it has been developed and used. We interpret the Krogh Principle as a heuristic, i.e., as a recommendation to approach biological problems through organisms where a specific trait or physiological mechanism is expected to be most distinctively displayed or most experimentally accessible. We designate these organisms “Krogh organisms.” We clarify the differences between uses of model organisms and non-standard Krogh organisms. Among these is the use of Krogh organisms as “negative models” in biomedical research, where organisms are chosen for their dissimilarity to human physiology. Importantly, the representational scope of Krogh organisms and the generalizability of their characteristics are not fixed or assumed but explored through experimental studies. Research on Krogh organisms is steeped in the comparative method characteristic of zoology and comparative physiology, in which studies of biological variation produce insights into general physiological constraints. Accordingly, we conclude that the Krogh principle exemplifies the advantages of studying biological variation as a strategy to produce generalizable insights.

This paper analyses the role of information security in shaping the dissemination and re-use of biomedical data, as well as the embedding of such data in the material, social and regulatory landscapes of research. We consider the data management practices adopted by two UK-based data linkage infrastructures: the Secure Anonymised Information Linkage, a Welsh databank that facilitates appropriate re-use of health data derived from research and routine medical practice in the region; and the Medical and Environmental Data Mash-up Infrastructure, a project bringing together researchers from the University of Exeter, the London School of Hygiene and Tropical Medicine, the Met Office and Public Health England to link and analyse complex meteorological, environmental and epidemiological data. Through an in-depth analysis of how data are sourced, processed and analysed in these two cases, we show that IS takes two distinct forms: epistemic IS, focused on protecting the reliability and reusability of data as they move across platforms and research contexts; and infrastructural IS, concerned with protecting data from external attacks, mishandling and use disruption. These two dimensions are intertwined and mutually constitutive, and yet are often perceived by researchers as being in tension with each other. We discuss how such tensions emerge when the two dimensions of IS are operationalised in ways that put them at cross purpose with each other, thus exemplifying the vulnerability of data management strategies to broader governance and technological regimes. We also show that whenever biomedical researchers manage to overcome the conflict, the interplay between epistemic and infrastructural IS prompts critical questions concerning data sources, formats, metadata and potential uses, resulting in an improved understanding of the wider context of research and the development of relevant resources. This informs and significantly improves the re-usability of biomedical data, while encouraging exploratory analyses of secondary data sources.

A heated debate surrounds the significance of reproducibility as an indicator for research quality and reliability, with many commentators linking a "crisis of reproducibility" to the rise of fraudulent, careless and unreliable practices of knowledge production. Through the analysis of discourse and practices across research fields, I point out that reproducibility is not only interpreted in different ways, but also serves a variety of epistemic functions depending on the research at hand. Given such variation, I argue that the uncritical pursuit of reproducibility as an overarching epistemic value is misleading and potentially damaging to scientific advancement. Requirements for reproducibility, however they are interpreted, are one of many available means to secure reliable research outcomes. Furthermore, there are cases where the focus on enhancing reproducibility turns out not to foster high-quality research. Scientific communities and Open Science advocates should learn from inferential reasoning from irreproducible data, and promote incentives for all researchers to explicitly and publicly discuss their methodological commitments, the ways in which they learn from mistakes and problems in everyday practice, and the strategies they use to choose which research component of any project needs to be preserved in the long term, and how.

This article considers the temporal dimension of data processing and use and the ways in which it affects the production and interpretation of knowledge claims. I start by distinguishing the time at which data collection, dissemination, and analysis occur from the time in which the phenomena for which data serve as evidence operate. Building on the analysis of two examples of data reuse from modeling and experimental practices in biology, I then argue that Dt affects how researchers select and interpret data as evidence and identify and understand phenomena.

The use of big data to investigate the spread of infectious diseases or the impact of the built environment on human wellbeing goes beyond the realm of traditional approaches to epidemiology, and includes a large variety of data objects produced by research communities with different methods and goals. This paper addresses the conditions under which researchers link, search and interpret such diverse data by focusing on “data mash-ups”—that is the linking of data from epidemiology, biomedicine, climate and environmental science, which is typically achieved by holding one or more basic parameters, such as geolocation, as invariant. We argue that this strategy works best when epidemiologists interpret localisation procedures through an idiographic perspective that recognises their context-dependence and supports a critical evaluation of the epistemic value of geolocation data whenever they are used for new research purposes. Approaching invariants as strategic constructs can foster data linkage and re-use, and support carefully-targeted predictions in ways that can meaningfully inform public health. At the same time, it explicitly signals the limitations in the scope and applicability of the original datasets incorporated into big data collections, and thus the situated nature of data linkage exercises and their predictive power.

This paper proposes an account of scientific data that makes sense of recent debates on data-driven and ‘big data’ research, while also building on the history of data production and use particularly within biology. In this view, ‘data’ is a relational category applied to research outputs that are taken, at specific moments of inquiry, to provide evidence for knowledge claims of interest to the researchers involved. They do not have truth-value in and of themselves, nor can they be seen as straightforward representations of given phenomena. Rather, they are fungible objects defined by their portability and prospective usefulness as evidence

Bogen and Woodward characterized data as embedded in the context in which they are produced (‘local’) and claims about phenomena as retaining their significance beyond that context (‘nonlocal’). This view does not fit sciences such as biology, which successfully disseminate data via packaging processes that include appropriate labels, vehicles, and human interventions. These processes enhance the evidential scope of data and ensure that claims about phenomena are understood in the same way across research communities. I conclude that the degree of locality of both data and claims about phenomena varies depending on the packaging used to make them travel and on the research setting in which they are used. †To contact the author, please write to: ESRC Centre for Genomics in Society, University of Exeter, Byrne House, St. Germans Road, EX4 4PJ Exeter, United Kingdom; e‐mail: [email protected].

Scientific knowledge production is currently affected by the dissemination of data on an unprecedented scale. Technologies for the automated production and sharing of vast amounts of data have changed the way in which data are handled and interpreted in several scientific domains, most notably molecular biology and biomedicine. In these fields, the activity of data gathering has become increasingly technology-driven, with machines such as next generation genome sequencers and mass spectrometers generating billions of data points within hours, and with little need for human supervision. Given the relative ease and low costs with which datasets can be produced (that is, once a laboratory has been able to afford ..

ArgumentWe examine the criteria used to validate the use of nonhuman organisms in North-American alcohol addiction research from the 1950s to the present day. We argue that this field, where the similarities between behaviors in humans and non-humans are particularly difficult to assess, has addressed questions of model validity by transforming the situatedness of non-human organisms into an experimental tool. We demonstrate that model validity does not hinge on the standardization of one type of organism in isolation, as often the case with genetic model organisms. Rather, organisms are viewed as necessarily situated: they cannot be understood as a model for human behavior in isolation from their environmental conditions. Hence the environment itself is standardized as part of the modeling process; and model validity is assessed with reference to the environmental conditions under which organisms are studied.

What is the best way to analyse abstraction in scientific modelling? I propose to focus on abstracting as an epistemic activity, which is achieved in different ways and for different purposes depending on the actual circumstances of modelling and the features of the models in question. This is in contrast to a more conventional use of the term ‘abstract’ as an attribute of models, which I characterise as black-boxing the ways in which abstraction is performed and to which epistemological advantage. I exemplify my claims through a detailed reconstruction of the practices involved in creating two types of models of the flowering plant Arabidopsisthaliana, currently the best-known model organism in plant biology. This leads me to distinguish between two types of abstraction processes: the ‘material abstracting’ required in the production of Arabidopsis specimens and the ‘intellectual abstracting’ characterising the elaboration of visual models of Arabidopsis genomics. Reflecting on the differences between these types of abstracting helps to pin down the epistemic skills and research commitments used by researchers to produce each model, thus clarifying how models are handled by researchers and with which epistemological implications.

We continue our discussion of the competing arguments in favour of the unified theory and the pluralistic theory of radiation advanced by three nineteenth-century pioneers: Herschel, Melloni, and Draper. Our narrative is structured by a consideration of the epistemic criteria relevant to theory-choice; the epistemic focus highlights many little-known aspects of this relatively well-known episode. We argue that the acceptance of light-heat unity in this period cannot be credibly justified on the basis of common evaluative criteria such as simplicity and standard notions of explanatory power. Whether the consensus was justified by some other criteria remains an open question.Keywords: Theory-choice; Epistemic values; Radiation; Simplicity; Explanatory power; Causal explanation.

Knowledge-making practices in biology are being strongly affected by the availability of data on an unprecedented scale, the insistence on systemic approaches and growing reliance on bioinformatics and digital infrastructures. What role does theory play within data-intensive science, and what does that tell us about scientific theories in general? To answer these questions, I focus on Open Biomedical Ontologies, digital classification tools that have become crucial to sharing results across research contexts in the biological and biomedical sciences, and argue that they constitute an example of classificatory theory. This form of theorizing emerges from classification practices in conjunction with experimental know-how and expresses the knowledge underpinning the analysis and interpretation of data disseminated online

Arabidopsis is currently the most popular and well-researched model organism in plant biology. This paper documents this plant's rise to scientific fame by focusing on two interrelated aspects of Arabidopsis research. One is the extent to which the material features of the plant have constrained research directions and enabled scientific achievements. The other is the crucial role played by the international community of Arabidopsis researchers in making it possible to grow, distribute and use plant specimen that embody these material features. I argue that at least part of the explosive development of this research community is due to its successful standardisation and to the subsequent use of Arabidopsis specimen as material models of plants. I conclude that model organisms have a double identity as both samples of nature and artifacts representing nature. It is the resulting ambivalence in their representational value that makes them attractive research tools for biologists

Hardly any ontological result of modern science is more firmly established than the fact that infrared radiation differs from light only in wavelength; this is part of the modern conception of the continuous spectrum of electromagnetic radiation reaching from radio waves to gamma radiation. Yet, like many such evident truths, the light-infrared unity was an extremely difficult thing to establish. We examine the competing arguments in favour of the unified and pluralistic theories of radiation, as put forward in the first half of the nineteenth century by three of the most important early pioneers of the study of radiation: Herschel, Melloni and Draper. In this part of the paper, we conclude that there were no compelling reasons of observational adequacy to prefer the unified theory to the pluralistic theory. Keywords: Macedonio Melloni; John William Draper; William Herschel; Infrared; Observation; Theory-choice