Marion Clara Vorms

Theories are among the main achievements of scientific inquiry and appear as the repositories of scientific knowledge. This chapter is devoted to an examination of the notion of theory as a unit of analysis for the study of scientific knowledge. Most of the analysis consists in presenting and criticizing two major proposals made by philosophers of science, the “received view,” commonly attributed to logical empiricism, and the “semantic view of theories,” which became the new orthodoxy in the 1960s. Both proposals aim at formal reconstructions of theories. The shared assumptions underlying this common project will be questioned. Alternative ways of construing scientific theorizing will be sketched, notably those which are more “agent-centered” and put forward the way scientists use and understand their theories in practice.

Linkage (or genetic) maps are graphs, which are intended to represent the linear ordering of genes on the chromosomes. They are constructed on the basis of statistical data concerning the transmission of genes. The invention of this technique in 1913 was driven by Morgan’s group’s adoption of a set of hypotheses concerning the physical mechanism of heredity. These hypotheses were themselves grounded in Morgan’s defense of the chromosome theory of heredity, according to which chromosomes are the physical basis of genes. In this paper, I analyze the 1919 debate between William Castle and Morgan’s group, about the construction of genetic maps. The official issue of the debate concerns the arrangement of genes on chromosomes. However, the disputants tend to carry out the discussions about how one should model the data in order to draw predictions concerning the transmission of genes; the debate does not bear on the data themselves, nor does it focus on the hypotheses explaining these data. The main criteria that are appealed to by the protagonists are simplicity and predictive efficacy. However, I show that both parties’ assessments of the simplicity and predictive efficacy of different ways of modeling the data themselves depend on background theoretical positions. I aim at clarifying how preference for a given model and theoretical commitments articulate.

During the last few decades, models have become the centre of attention in both cognitive science and philosophy of science. In cognitive science, the claim that humans reason with mental models, rather than mentally manipulate linguistic symbols, is the majority view. Similarly, philosophers of science almost unanimously acknowledge that models have to be taken as a central unit of analysis. Moreover, some philosophers of science and cognitive scientists have suggested that the cognitive hypothesis of mental models is a promising way of accounting for the use of models in science. However, once the importance of models in cognition as well as in science has been acknowledged, much more needs to be said about how models enable agents to make predictions, and to understand the world. In this paper, our goal (as a cognitive scientist, working on causal reasoning, and a philosopher of science, working on models and representations) is twofold. We would like to further develop the notion of mental models, and to explore the parallels between mental models as a concept in cognitive science, and models in science. While acknowledging that the parallel move towards models in cognitive science and in philosophy of science is in the right direction, we think that: i. the notion of mental models needs to be clarified in order to serve as a useful tool, ii. the relation between the hypothesis of mental models and the use of models in science is still to be clarified. First, we will briefly recall a few points about the mental model hypothesis, on the one hand, and the model-centred view of science, on the other hand. Then, we will present our parallel criticisms, and put forward our own proposals

According to the “story-model” of jurors’ decision-making, as advocated by Pennington and Hastie (1986, 1988, 1992, 1993), jurors in criminal trials make sense of the evidence through the construction of a mental representation of the events, rather than through the estimation and combination of probabilities. This ‘story’ consists in a causal explanatory scenario of the crime, and is supposed to drive the jurors’ choice of a verdict. As suggested by Heller (2006), the story-model can be described as a legal application of the simulation heuristic (Kahneman and Tversky 1982), according to which people determine the likelihood of an event based on how easy it is to picture the event mentally: the easier to mentally simulate the prosecution scenario, the higher the conviction rate. The primary goal of this paper is to present the main tenets of Pennington and Hastie’s (1986, 1988, 1992, 1993) “story-model” of jurors’ decision-making, and to draw a few criticisms thereof, in the light of an analysis of evidential reasoning. While acknowledging that some fundamental reasons for adopting this model are well-grounded, and make it a plausible account of jurors’ reasoning, we raise some issues concerning its core theses. In particular, we show that the claim that the evaluation of the credibility of the evidence is mediated by story construction, and determined by the coherence of the story, is not tenable as such, and needs to be complemented by a more probabilistically centered approach.

Computer simulations are widely used in current scientific practice, as a tool to obtain information about various phenomena. Scientists accordingly rely on the outputs of computer simulations to make statements about the empirical world. In that sense, simulations seem to enable scientists to acquire empirical knowledge. The aim of this paper is to assess whether computer simulations actually allow for the production of empirical knowledge, and how. It provides an epistemological analysis of present-day empirical science, to which the traditional epistemological categories cannot apply in any simple way. Our strategy consists in acknowledging the complexity of scientific practice, and trying to assess its rationality. Hence, while we are careful not to abstract away from the details of scientific practice, our approach is not strictly descriptive: our goal is to state in what conditions empirical science can rely on computer simulations. In order to do so, we need to adopt a renewed epistemological framework, whose categories would enable us to give a finer-grained, and better-fitted analysis of the rationality of scientific practice.

In this paper, I construe scientific understanding not only as understanding the phenomena by means of some theoretical material (theory, law or model), but more fundamentally as understanding the theoretical material itself that is supposed to explain the phenomena. De Regt and Dieks (2005) emphasise the contextual aspects of the intelligibility of theories, showing that it depends on their "virtues", on the historical standards of intelligibility, and on the particular "skills" of their users. My paper aims at continuing this proposal, first by giving a more precise definition of one's understanding of a theory and then by emphasising the importance, for this issue, of the particular formats in which a theory is expressed and hence grasped by its users. To defend this, I take the example of the versions of classical mechanics (variational versus vectorial) and the various formats of representation of its main principles and models. What does "understanding a theory" mean? At first sight, we could say that it amounts to having a clear view of the logical relations between its core principles and theorems. This kind of understanding, though global, is quite abstract: one can understand the logical structure of a theory without being able to connect it to the phenomena. Moreover, this definition depends on how one 115 construes the structure of theories: it will vary according to whether one defines theories as logical sets of statements with interpretative rules (following the "syntactic conception" of theories) or as families of models ("semantic conception"). I thus suggest that there is another sense of "understanding a theory" that itself has two aspects. To understand a theory, one has to understand both what the theory says or means and how it works; in other words, one has to grasp the phenomena by means of its conceptual apparatus (representational aspect) and to be able to manipulate it and make it fit the phenomena (computational aspect). I claim that these are essentially contextual and practical matters, and that the particular format in which the theoretical content is displayed is crucial to them. Following Humphreys' proposal (2004), I claim that one never accesses to a theory as a whole. Be it a set of statements or a class of models, in practice, it is always displayed in some particular equations, statements, images, graphs, diagrams. Humphreys' proposal of the notion of "template" to complement the classical "units of analysis" of science, like theories and models, may be a good candidate to study the relationship between the representational and computational aspects of understanding: a template is a "concrete piece of syntax" (most of the time an equation, but I suggest that Humphreys' claim could be extended to other formats) that has both a representational and computational function. With the example of classical mechanics, I show how these two functions are interrelated and, as Humphreys suggests, sometimes in tension with each other. Adressing these issues by focusing on the particular formats that are dealt with in practice may enlight this problematic relationship.

In this paper, I construe scientific understanding not only as understanding the phenomena by means of some theoretical material (theory, law or model), but more fundamentally as understanding the theoretical material itself that is supposed to explain the phenomena. De Regt and Dieks (2005) emphasise the contextual aspects of the intelligibility of theories, showing that it depends on their ―virtues‖, on the historical standards of intelligibility, and on the particular ―skills‖of their users. My paper aims at continuing this proposal, first by giving a more precise definition of one's understanding of a theory and then by emphasising the importance, for this issue, of the particular formats in which a theory is expressed and hence grasped by its users. To defend this, I take the example of the versions of classical mechanics and the various formats of representation of its main principles and models. What does ―understanding a theory‖ mean? At first sight, we could say that it amounts to having a clear view of the logical relations between its core principles and theorems. This kind of understanding, though global, is quite abstract: one can understand the logical structure of a theory without being able to connect it to the phenomena. Moreover, this definition depends on how one construes the structure of theories: it will vary according to whether one defines theories as logical sets of statements with interpretative rules (following the ―syntactic conception‖ of theories) or as families of models (―semantic conception‖). I thus suggest that there is another sense of ―understanding a theory‖ that itself has two aspects. To understand a theory, one has to understand both what the theory says or means and how it works; in other words, one has to understand it as representing the phenomena (representational aspect) and to be able to manipulate it and make it fit the phenomena (computational aspect). I claim that these are essentially contextual and practical matters, and that the particular format in which the theoretical content is displayed is crucial to them. Following Humphreys' proposal (2004), I claim that one never accesses to a theory as a whole. Be it a set of statements or a class of models, in practice, it is always displayed in some particular equations, statements, images, graphs, diagrams. Humphreys' proposal of the notion of ―template‖ to complement the classical ―units of analysis‖ of science, like theories and models, may be a good candidate to study the relationship between the representational and computational aspects of understanding: a template is a ―concrete piece of syntax‖ (most of the time an equation, but I suggest that Humphreys' claim could be extended to other formats) that has both a representational and computational function. With the example of classical mechanics, I show how these two functions are interrelated and, as Humphreys suggests, are sometimes in tension with each other. Adressing these issues by focusing on the particular formats that are dealt with in practice may enlight this problematic relationship.

I analyse the double function of models (representing the phenomena, and being a tool for calculating and predicting them) from a cognitive point of view. Taking the same approach as Ronald Giere, I nevertheless argue that he is to much committed to an abstract conception of theories and that one should give more attention to the particular formats in which models are expressed and grasped. By taking the example of Classical Mechanics, I show that a model, as an abstract entity, doesn't represent anything if not presented in some particular format, be it linguistic, diagrammatic, or pictorial. This format is crucial to the relation between the two main functions of models.

On appelle « biais de confirmation » la tendance supposée des humains à sélectionner les informations qui vont dans le sens de ce qu’ils croient (ou veulent croire) et à interpréter celles dont ils disposent en faveur de leurs hypothèses favorites. Cet article vise à porter un regard critique sur certains usages de cette notion, et plus généralement sur le recours aux « biais cognitifs » pour expliquer tout un ensemble de phénomènes sociaux interprétés comme les marques d’une forme d’irrationalité. Remettant en question le caractère probant des résultats expérimentaux destinés à mettre en évidence une déviation qui nous conduirait systématiquement à des représentations inexactes, je suggère que l’approche bayésienne en psychologie du raisonnement permet de rendre compte de certains de ces résultats en faisant l’hypothèse que nos inférences se conforment à une telle norme.

This paper explores ‘reasonable doubt’ as an enlightening notion to think of reasoning and decision-making generally, beyond the judicial domain. The paper starts from a decision-theoretic understanding of the notion, whereby it can be defined in terms of degrees of belief and a probabilistic confirmation threshold for action. It then highlights some of the limits of this notion, and proposes a richer analysis of epistemic states and reasoning through the lens of ‘reasonable doubt’, which in turn is likely to supplement the DT framework. The strategy consists in fighting on two fronts: with DT, the paper claims that there is no absolute notion of ‘reasonable doubt’ but, pace DT, it shows that reasonable doubt cannot be accounted for only in terms of degrees of belief and probabilistic threshold. We argue that the lens of reasonable doubt sheds light on aspects of belief dynamics, as well as of the nature of epistemic attitudes, which are often obscured by belief-centred approaches. In particular, when it comes to acknowledging the necessary ignorance and irreducible uncertainty that we face in our everyday-life decisions, studying the various facets of doubt rather than focusing on what can be believed, enables one to do justice to the richness and diversity of the mental states in play.

La notion de témoignage est au cœur d’une branche de la théorie de la connaissance, appelée « épistémologie du témoignage », qui s’interroge sur le bien-fondé des croyances acquises par le biais d’autrui – par opposition aux seules capacités sensorielles et intellectuelles de l’individu. Le témoignage est aussi un des modes de preuve les plus importants dans le cadre judiciaire, aussi bien au cours de l’enquête et de l’instruction que du procès, et a fait l’objet de nombreuses réflexions de la part de juristes, en particulier dans les pays de common law. Par-delà les différences disciplinaires, les distinctions des juristes sont à même d’apporter un éclairage nouveau à la réflexion épistémologique, laquelle invite à reconsidérer certaines définitions juridiques.

The book edited by Roman Frigg and Matthew C. Hunter is a great example of interdisciplinary collaborative work, bringing together contributions by scholars of science and of art, around the topic of representation. The collection consists of eleven essays, seven of which were presented in early form at a conference organized by the two editors at the London School of Economics and the Courtauld Institute of Art in June 2006; the other four have been added subsequently. The result is a high-standard, remarkably edited book

Models such as the simple pendulum, isolated populations, and perfectly rational agents, play a central role in theorising. It is now widely acknowledged that a study of scientific representation should focus on the role of such imaginary entities in scientists’ reasoning. However, the question is most of the time cast as follows: How can fictional or abstract entities represent the phenomena? In this paper, I show that this question is not well posed. First, I clarify the notion of representation, and I emphasise the importance of what I call the “format” of a representation for the inferences agents can draw from it. Then, I show that the very same model can be presented under different formats, which do not enable scientists to perform the same inferences. Assuming that the main function of a representation is to allow one to draw predictions and explanations of the phenomena by reasoning with it, I conclude that imaginary models in abstracto are not used as representations: scientists always reason with formatted representations. Therefore, the problem of scientific representation does not lie in the relationship of imaginary entities with real systems. One should rather focus on the variety of the formats that are used in scientific practice.

In this paper, I propose a study of the invention and development of the technique of genetic mapping in the 1920's. I show that what is usually taken as one and the same theory (Classical Genetics) is in fact the result of the articulation of various levels of explanations corresponding to two different disciplines, with different methods and representational practices -- namely Mendelian theory and cytology. The merging of these two disciplinary frameworks is embodied in the very rules underlying the construction and interpretation of genetic maps. Moreover, the debates between the geneticists around how to display data within these maps reveal that they have different understanding of the articulation of these disciplines (and different theoretical commitments)

Linkage (or genetic) maps are graphs, which are intended to represent the linear ordering of genes on the chromosomes. They are constructed on the basis of statistical data concerning the transmission of genes. The invention of this technique in 1913 was driven by Morgan's group's adoption of a set of hypotheses concerning the physical mechanism of heredity. These hypotheses were themselves grounded in Morgan's defense of the chromosome theory of heredity, according to which chromosomes are the physical basis of genes. In this paper, I analyze the 1919 debate between William Castle and Morgan's group, about the construction of genetic maps. The official issue of the debate concerns the arrangement of genes on chromosomes. However, the disputants tend to carry out the discussions about how one should model the data in order to draw predictions concerning the transmission of genes; the debate does not bear on the data themselves, nor does it focus on the hypotheses explaining these data. The main criteria that are appealed to by the protagonists are simplicity and predictive efficacy. However, I show that both parties' assessments of the simplicity and predictive efficacy of different ways of modeling the data themselves depend on background theoretical positions. I aim at clarifying how preference for a given model and theoretical commitments articulate