Philosophy and statistics have studied two causal species, deterministic and probabilistic. There's a third species, however, hitherto unanalysed: underdeterministic causal phenomena, which are non-deterministic yet non-probabilistic. Here, I formulate a framework for modelling them. -/- Consider a simple case. If I go out, I may stumble into you but also may miss you. If I don’t go out, we won't meet. I go out. We meet. My going out is a cause of our encounter even if there was no determinate (...) probability of us meeting conditional on my going out. The cause is neither deterministic (it didn't necessitate the effect) nor probabilistic (the relevant conditional probabilities are undefined). Rather, it's underdeterministic: it raises the modal status of the effect from causally impossible to possible. -/- Here, I won't offer a theory of such token causes but develop the prerequisite for any such theory: the underdeterministic framework. The framework is like the deterministic structural-equations framework but with one consequential difference---an equation can return multiple values. This change allows me to define causal possibility and necessity, and corresponding notions of interventionist might- and would-counterfactuals. I also define conditional independence, which obeys the graphoid axioms, and prove that underdeterministic models satisfy the causal Markov condition. The framework can causally model situations that other frameworks cannot: decision-making under bounded uncertainty, games with multiple equilibria, infinite fair lotteries, and any other non-deterministic situation where indeterminacies are essentially non-probabilistic, or where we have a reason not to use probabilities. (shrink)

John D. Norton is responsible for a number of influential views in contemporary philosophy of science. This paper will discuss two of them. The material theory of induction claims that inductive arguments are ultimately justified by their material features, not their formal features. Thus, while a deductive argument can be valid irrespective of the content of the propositions that make up the argument, an inductive argument about, say, apples, will be justified (or not) depending on facts about apples. The argument (...) view of thought experiments claims that thought experiments are arguments, and that they function epistemically however arguments do. These two views have generated a great deal of discussion, although there hasn’t been much written about their combination. I argue that despite some interesting harmonies, there is a serious tension between them. I consider several options for easing this tension, before suggesting a set of changes to the argument view that I take to be consistent with Norton’s fundamental philosophical commitments, and which retain what seems intuitively correct about the argument view. These changes require that we move away from a unitary epistemology of thought experiments and towards a more pluralist position. (shrink)

Imagination is important for many things in science: solving problems, interpreting data, designing studies, etc. Philosophers of imagination typically account for the productive role played by imagination in science by focusing on how imagination is constrained, e.g., by using self-imposed rules to infer logically, or model events accurately. But the constraints offered by these philosophers either constrain too much, or not enough, and they can never account for uses of imagination that are needed to break today’s constraints in order to (...) make progress tomorrow. Thus, epistemology of imagination needs to make room for an element of epistemological anarchy. (shrink)

This paper considers Norton’s Material Theory of Induction. The material theory aims inter alia to neutralize Hume’s Problem of Induction. The purpose of the paper is to evaluate the material theory's capacity to achieve this end. After pulling apart two versions of the theory, I argue that neither version satisfactorily neutralizes the problem.

Historical inductions, that is, the pessimistic metainduction and the problem of unconceived alternatives, are critically analyzed via John D. Norton’s material theory of induction and subsequently rejected as noncogent arguments. It is suggested that the material theory is amenable to a local version of the pessimistic metainduction, for example, in the context of some medical studies.

There are two important traditions in the philosophy of induction. According to one tradition, which has dominated for the last couple of centuries, inductive arguments are warranted by rules. Bayesianism is the most popular view within this tradition. Rules of induction provide functional accounts of inductive support, but no rule is universal; hence, no rule is by itself an accurate model of inductive support. According to another tradition, inductive arguments are not warranted by rules but by matters of fact. Norton’s (...) material theory of induction (MTI) is an influential view within this tradition. Norton’s MTI fails to provide an account of inductive support, but it provides a good account of inductive warrant that can help us define the domain of validity of each rule of induction. Despite their limitations, both approaches to induction are illuminating important aspects of how relations of inductive support work. In this article I present a hybrid theory of induction (HTI), in which I acknowledge and articulate the role of both rules and matters of fact in our understanding of inductive support. According to the HTI, rules of induction accurately describe relations of inductive support when they are warranted, and a rule of induction is warranted if the right facts about the matter of the induction obtain. Crucially, the HTI allows us to address the main challenges that each tradition faces while retaining their strengths, thus obtaining a functional and accurate account of inductive support. The HTI also provides a useful general framework to examine and understand induction, to make sense of how different rules of induction can coexist and to tackle some problems in epistemology. Moreover, the HTI allows us to clarify and resolve some current debates on induction, like the apparent divide between Norton and rule-based theorists. (shrink)

Philosophers have debated at length the epistemological status of scientific thought experiments. I contend that the literature on this topic still lacks a common conceptual framework, a lacuna that produces radical disagreement among the participants in this debate. To remedy this problem, I suggest focusing on the distinction between the internal and the external validity of an experiment, which is also crucial for thought experiments. I then develop an account of both kinds of validity in the context of thought experiments. (...) I show that we can naturally conceptualise internal validity in terms of games of make-believe. Then, I argue that external validity is best defined as accurate representation of a target system. Finally, I turn back to the current debate on thought experiments and show that my diarchic account provides a general framework that can be shared by the competing philosophical views, as well as a fruitful guide for their reconciliation. (shrink)

According to John D. Norton's Material Theory of Induction, all inductive inferences are justified by local facts, rather than their formal features or some grand principles of nature's uniformity. Recently, Richard Dawid (Found Phys 45(9):1101–1109, 2015) has offered a challenge to this theory: in an adaptation of Norton's own celebrated "Dome" thought experiment, it seems that there are certain inductions that are intuitively reasonable, but which do not have any local facts that could serve to justify them in accordance with (...) Norton's requirements. Dawid's suggestion is that “raw induction” might have a limited but important role for such inferences. I argue that the Material Theory can accommodate such inductions, because there are local facts concerning the combinatoric features of the induction’s target populations that can licence the inferences in an analogous way to existing examples of material induction. Since my arguments are largely independent of the details of the Dome, Norton's theory emerges as surprisingly robust against criticisms of excessive narrowness. (shrink)

Sadi Carnot’s 1824 Reflections on the Motive Power of Fire created the new science of thermodynamics. It succeeded in its audacious goal of finding a very general theory of the efficiency of heat engines, by introducing and exploiting the strange and unexpected notion of a thermodynamically reversible process. The notion is internally contradictory. It requires the states of these processes to be both in unchanging equilibrium, with a perfect balance of driving forces, while also changing. The work of Sadi’s father, (...) Lazare Carnot, on the efficiency of ordinary machines provided Sadi with a template of a very general theory of the efficiency of ordinary machines; and a characterization of the most efficient processes in them as those that minimize differences of driving forces and can be run in reverse. Lazare’s work could provide these resources because of its choice of a dissipative ontology of inelastic collisions among hard bodies. This historically ill-fated choice meant that Lazare’s machines were analogous to Sadi’s heat engines in their key aspects: they are built from essentially dissipative processes. Lazare’s strategies for controlling dissipation and optimizing his machines were transferrable to the analogous problems Sadi found in heat engines. The unanswerable historical question is whether Sadi would have sought a general theory of heat engines at all or found these general theoretical devices without the template provided in analogy by the prior work of Lazare. (shrink)

In this paper, we offer a descriptive theory of analogical reasoning in mathematics, stating general conditions under which an analogy may provide genuine inductive support to a mathematical conjecture (over and above fulfilling the merely heuristic role of ‘suggesting’ a conjecture in the psychological sense). The proposed conditions generalize the criteria of Hesse in her influential work on analogical reasoning in the empirical sciences. By reference to several case studies, we argue that the account proposed in this paper does a (...) better job in vindicating the use of analogical inference in mathematics than the prominent alternative defended by Bartha. (shrink)

This paper defends the thesis of learning from non-causal models: viz. that the study of some model can prompt justified changes in one’s confidence in empirical hypotheses about a real-world target in the absence of any known or predicted similarity between model and target with regards to their causal features. Recognizing that we can learn from non-causal models matters not only to our understanding of past scientific achievements, but also to contemporary debates in the philosophy of science. At one end (...) of the philosophical spectrum, my thesis undermines the views of those who, like Cartwright (Erkenntnis 70:45–58, 2009), follow Hesse (Models and Analogies in Science, Notre Dame, University of Notre Dame Press, 1963) in restricting the possibility of learning from models to only those situations where a model identifies some causal factors present in the target. At the other end of the spectrum, my thesis also helps undermine some extremely permissive positions, e.g., Grüne-Yanoff’s (Erkenntnis 70(1):81–99, 2009, Philos Sci 80(5): 850–861, 2013) claim that learning from a model is possible even in the absence of any similarity at all between model and target. The thesis that we can learn from non-causal models offers a cautious middle ground between these two extremes. (shrink)

This paper addresses the role of analogical reasoning in archaeoastronomy - the discipline which studies the connections between the ancient monuments and the heavens. Archaeoastronomy is a highly interdisciplinary science, placed at the border between the humanities – especially archaeology – and the scientific approach to cultural heritage. As a consequence, its scientific foundations are a delicate matter. We plan to investigate here the question of what constitutes the evidence for analogical inferences in archaeoastronomy and to what extent one can (...) achieve confirmation of archaeoastronomical hypotheses by means of such analogies. Our claim will be that, when deployed in accordance with the methodology articulated in this paper, analogies can be a highly effective epistemic tool for generating and supporting hypotheses about the relation of archaeological sites with astronomical events. (shrink)

Arguments from non-causal analogy form a distinctive class of analogical arguments in science not recognized in authoritative classifications by, e.g., Hesse and Bartha. In this paper, I illustrate this novel class of scientific analogies by means of historical examples from physics, biology and economics, at the same time emphasizing their broader significance for contemporary debates in epistemology.

Suppose a theory T entails hypotheses H and $$H'$$, neither of which entails the other. A number of authors have argued that a piece of evidence E “indirectly confirms” H when E confirms either T or $$H'$$. But there has been a protracted and unsettled debate about whether indirect confirmation is a sound inference procedure. Skeptics argue that the procedure employs conditions of confirmation that jointly lead to absurdity. Proponents argue that this criticism is unfounded or that its import is (...) exaggerated. I will argue that no side has the story quite right, and some have the story quite wrong. Indirect confirmation, as characterized above, is unsound, and a good chunk of this paper will be concerned with showing why most extant defenses of the procedure err. On the other hand, when certain probabilistic (in)dependence relations hold between T, H, and $$H'$$, indirect confirmation can work, for reasons that trace back to Reichenbach’s principle of the common cause. I illustrate these matters with some contemporary and historical examples, with a particular focus on Kepler’s use of data about mars’s elliptical orbit to justify a claim about earth’s. (shrink)

According to a widely shared generic conception of inferential justification—‘the standard conception’—an agent is inferentially justified in believing that p only if she has antecedently justified beliefs in all the non-redundant premises of a good argument for p. This conception tends to serve as the starting-point in contemporary debates about the nature and scope of inferential justification: as neutral common ground between various competing, more specific, conceptions. But it’s a deeply problematic starting-point. This paper explores three questions that haven’t been (...) given the attention they deserve, that complicate the application of the standard conception to cases, and that reveal it to be underspecified at the core—in ways that aren’t resolved but inherited by more specific (extant) versions of it. The goal isn’t to answer the questions, but to articulate them, explain what turns on them, and invite a critical re-examination of the standard conception. (shrink)

We propose a new account of calibration according to which calibrating a technique shows that the technique does what it is supposed to do. To motivate our account, we examine an early twentieth-century debate about chlorophyll chemistry and Mikhail Tswett’s use of chromatographic adsorption analysis to study it. We argue that Tswett’s experiments established that his technique was reliable in the special case of chlorophyll without relying on either a theory or a standard calibration experiment. We suggest that Tswett broke (...) the experimenters’ regress by appealing to material facts in the common ground for chemists at the time. (shrink)

Otto Neurath’s empiricist methodology of economics and his contributions to politi- cal economy have gained increasing attention in recent years. We connect this research with contemporary debates regarding the epistemological status of thought experiments by reconstructing Neurath’s utopias as linchpins of thought experiments. In our three reconstructed examples of different uses of utopias/dystopias in thought experiments we employ a reformulation of Häggqvist’s model for thought experiments and we argue that: (1) Our reformulation of Häggqvist’s model more adequately complies with many (...) uses of thought experiments, especially with the open-ended discussions of utopias and dys- topias in thought experiments. (2) As a strict logical empiricist, Neurath is committed to a strictly empiricist account of thought experiments. John Norton’s empiricist argument view can indeed account for the justifications of empirical beliefs and genuine discoveries targeted by scientific utopianism in three distinct (yet connected) ways, all of which Neur- ath already contemplated: (2.I) Dealing with utopias and thought experiments on a regular basis increases creativity and inventiveness. (2.II) Particular ways of presenting knowl- edge facilitate scientific discovery and social progress. (2.III) The use of utopias in thought experiments can prompt conceptual change and allow access to new phenomena. We con- clude by highlighting that, even though thought experiments support a positive attitude for exploring new social possibilities, Neurath points out that active decisions are unavoidable. The exploration of alternatives and the awareness of a need for decisions in policy discus- sion avert a technocratic outlook in social science. (shrink)

Otto Neurath’s empiricist methodology of economics and his contributions to political economy have gained increasing attention in recent years. We connect this research with contemporary debates regarding the epistemological status of thought experiments by reconstructing Neurath’s utopias as linchpins of thought experiments. In our three reconstructed examples of different uses of utopias/dystopias in thought experiments we employ a reformulation of Häggqvist’s model for thought experiments and we argue that: (1) Our reformulation of Häggqvist’s model more adequately complies with many uses (...) of thought experiments, especially with the open-ended discussions of utopias and dystopias in thought experiments. (2) As a strict logical empiricist, Neurath is committed to a strictly empiricist account of thought experiments. John Norton’s empiricist argument view can indeed account for the justifications of empirical beliefs and genuine discoveries targeted by scientific utopianism in three distinct (yet connected) ways, all of which Neurath already contemplated: (2.I) Dealing with utopias and thought experiments on a regular basis increases creativity and inventiveness. (2.II) Particular ways of presenting knowledge facilitate scientific discovery and social progress. (2.III) The use of utopias in thought experiments can prompt conceptual change and allow access to new phenomena. We conclude by highlighting that, even though thought experiments support a positive attitude for exploring new social possibilities, Neurath points out that active decisions are unavoidable. The exploration of alternatives and the awareness of a need for decisions in policy discussion avert a technocratic outlook in social science. (shrink)

It is common knowledge that the Aristotelian idea of an unmoved mover was abandoned definitively with the advent of modern science and, in particular, Newton’s precise formulation of mechanics. Here I show that the essential attribute of an unmoved mover is not incompatible with such mechanics; quite the contrary, it makes this possible. The unmoved mover model proposed does not involve supertasks, and leads both to an outrageous form of indeterminism and a new, accountable form of interaction. The process presents (...) a more precise characterization of the crucial going-to-the-limit operation. It has long been acknowledged in the existing literature that, theoretically, in infinite Newtonian systems, masses can move from rest to motion through supertasks. Numerous minor variations on the original schemes have already been published. Against this backdrop, this paper introduces three significant additions: 1) It formulates for the first time a limit postulate for systematically addressing infinite systems; 2) It shows that an Aristotelian unmoved mover is possible in systems of infinitely many particles that interact with each other solely by contact collision; 3) It shows how interaction at a distance can emerge in systems of infinitely many particles that interact with each other solely by contact. (shrink)

Virtually all philosophers of science have construed fundamental theories as descriptions of entities, properties, and/or structures. Call this the “descriptive-ontological” view. I argue that this view is incorrect, at least insofar as physical theories are concerned. I propose a novel construal of theories that I call the “prescriptive-dynamical” view. The central tenet of this view, roughly put, is that the _essential_ content of fundamental physical theories is a _prescription for interfacing with natural systems and translating local data into compact theoretical (...) language_. The descriptive-ontological aspects of theories, if any, are taken as _inessential_ content on this view: they do not contribute to the predictive success of the theory. Rather than describing _what is there_, the essence of a physical theory is to tell us _what to do_ when interfacing with a physical system. (shrink)

As inductive decision-making procedures, the inferences made by machine learning programs are subject to underdetermination by evidence and bear inductive risk. One strategy for overcoming these challenges is guided by a presumption in philosophy of science that inductive inferences can and should be value-free. Applied to machine learning programs, the strategy assumes that the influence of values is restricted to data and decision outcomes, thereby omitting internal value-laden design choice points. In this paper, I apply arguments from feminist philosophy of (...) science to machine learning programs to make the case that the resources required to respond to these inductive challenges render critical aspects of their design constitutively value-laden. I demonstrate these points specifically in the case of recidivism algorithms, arguing that contemporary debates concerning fairness in criminal justice risk-assessment programs are best understood as iterations of traditional arguments from inductive risk and demarcation, and thereby establish the value-laden nature of automated decision-making programs. Finally, in light of these points, I address opportunities for relocating the value-free ideal in machine learning and the limitations that accompany them. (shrink)

One of the most lively debates on scientific understanding is standardly presented as a controversy between the so-called factivists, who argue that understanding implies truth, and the non-factivists whose position is that truth is neither necessary nor sufficient for understanding. A closer look at the debate, however, reveals that the borderline between factivism and non-factivism is not as clear-cut as it looks at first glance. Some of those who claim to be quasi-factivists come suspiciously close to the position of their (...) opponents, the non-factivist, from whom they pretend to differ. The non-factivist, in turn, acknowledges that some sort of ‘answering to the facts’ is indispensable for understanding. This paper discusses an example of convergence of the initially rival positions in the debate on understanding and truth: the use of the same substitute for truth by the quasi-factivist Kareem Khalifa and the non-factivists Henk de Regt and Victor Gijsbers. It is argued that the use of ‘effectiveness’ as a substitute for truth by both parties is not an occasional coincidence of terms, it rather speaks about a deeper similarity which have important implications for understanding the essential features of scientific understanding. (shrink)

This paper discusses the role that appeals to theoretical simplicity have played in the debate between nativists and empiricists in cognitive science. Both sides have been keen to make use of such appeals in defence of their respective positions about the structure and ontogeny of the human mind. Focusing on the standard simplicity argument employed by empiricist-minded philosophers and cognitive scientists—what I call “the argument for minimal innateness”—I identify various problems with such arguments—in particular, the apparent arbitrariness of the relevant (...) notions of simplicity at work. I then argue that simplicity ought not be seen as a theoretical desideratum in its own right, but rather as a stand-in for other desirable features of theories. In this deflationary vein, I argue that the best way of interpreting the argument for minimal innateness is to view it as an indirect appeal to various potential biological constraints on the amount of innate structure that can wired into the human mind. I then consider how nativists may respond to this biologized version of the argument, and discuss the role that similar biological concerns have played in recent nativist theorizing in the Minimalist Programme in generative linguistics. (shrink)

It is often argued that while biases routinely influence the generation of scientific theories, a subsequent rational evaluation of such theories will ensure that biases do not affect which theories are ultimately accepted. Against this line of thought, this paper shows that the existence of certain kinds of biases at the generation-stage implies the existence of biases at the evaluation-stage. The key argumentative move is to recognize that a scientist who comes up with a new theory about some phenomena has (...) thereby gained an unusual type of evidence, viz. information about the space of theories that could be true of the phenomena. It follows that if there is bias in the generation of scientific theories in a given domain, then the rational evaluation of theories with reference to the total evidence in that domain will also be biased. (shrink)

Experimental archaeology is often understood both as testing hypotheses about processes shaping the archaeological record and as generating tacit knowledge. Considering lithic technologies, I examine the relationship between these conceptions. Experimental archaeology is usefully understood via “maker’s knowledge”: archaeological experiments generate embodied know-how enabling archaeological hypotheses to be grasped and challenged, and further, well-positioning archaeologists to generate integrated interpretations. Finally, experimental archaeology involves “material speculation”: the constraints and affordances of archaeologists and their materials shape productive exploration of the capacities of (...) objects and human skill in ways relevant to archaeological questions. (shrink)

Reverse inference is a crucial inferential strategy used in cognitive neuroscience to derive conclusions about the engagement of cognitive processes from patterns of brain activation. While widely employed in experimental studies, it is now viewed with increasing scepticism within the neuroscience community. One problem with reverse inference is that it is logically invalid, being an instance of abduction in Peirce’s sense. In this paper, we offer the first systematic analysis of reverse inference as a form of abductive reasoning and highlight (...) some relevant implications for the current debate. We start by formalising an important distinction that has been entirely neglected in the literature, namely the distinction between weak and strong reverse inference. Then, we rely on case studies from recent neuroscientific research to systematically discuss the role and limits of both strong and weak reverse inference; in particular, we offer the first exploration of weak reverse inference as a discovery strategy within cognitive neuroscience. (shrink)

En este artículo defiendo que el problema humeano de la justificación racional de la inducción no ha contribuido a la comprensión filosófica de este tipo de inferencias. La teoría material de la inducción propuesta por Norton (2003) permite analizar el razonamiento inductivo más allá del laberinto justificativo humeano, pero con un corte demasiado alto: asumir que las inducciones dependen intrínsecamente de cómo es localmente la realidad. En su lugar, propongo una teoría de la inducción en la que las inducciones dependen (...) no de tesis ontológicas, sino de las capacidades cognitivas para comprimir información de un modo epistémicamente efectivo. (shrink)

The main purpose of this paper is to refute the metaphysicians ‘methodological continuation’ argument supporting epistemic realism in metaphysics. This argument aims to show that scientific realists have to accept that metaphysics is as rationally justified as science given that they both employ inference to the best explanation, i.e. that metaphysics and science are methodologically continuous. I argue that the reasons given by scientific realists as to why inference to the best explanation is reliable in science do not constitute a (...) reason to believe that it is reliable in metaphysics. The justification of IBE in science and the justification of IBE in metaphysics are two distinct issues with only superficial similarities, and one cannot rely on one for the other. This becomes especially clear when one analyses the debate about the legitimacy of IBE that has taken place between realists and empiricists. The metaphysician seeking to piggyback on the realist defense of IBE in science by the methodological continuation argument presupposes that the defense is straightforwardly applicable to metaphysics. I will argue that it is, in fact, not. The favoured defenses of IBE in scientific realism make extensive use of empirical considerations, predictive power and inductive evidence, all of which are paradigmatically absent in the metaphysical context. Furthermore, I argue that the metaphysician, even if the realist would concede to the methodological continuation argument, fails to offer any agreed upon conclusions resulting from its application in metaphysics. As a result, the scientific realist is not committed to believing that there is metaphysical knowledge. (shrink)

Philosophers and scientists have often adopted the orthodox version of calibration which involves standardizing an instrument using a known phenomenon. The essential link between theoretical concepts and empirical data, in the philosophy of measurement, has generated the formulation of principles of coordination, synthetic a priori, and revisables. Operationalist thinking wanted to limit the scope of concepts to operations of measurement that are actually achievable. The coherentist perspective in the philosophy of measurement has operated a recovery of coordinationist epistemology and operationalism, (...) relying on a minimal number of ontological principles. Models of an instrument involve a commitment to separation between idealized theories and material things. However, philosophers and metrologists have advocated the requirement of a rich theoretical content in the modeling of measuring instruments. According to other contributions, the epistemic privilege of measurement precedes a background theory and its robustness lies in the free contact with empirical data. Moreover, the applicability regime of a theory dictates its boundary conditions, which guide the experimenter in the design of measuring instruments and provide the basis for operationalizing the meaning of theoretical terms. I support operational pluralism, measurement operations involving different physical indicators, accompanied by dynamic coherentism. The James Webb Space Telescope calibration program is a significant case. Observations of calibration stars by various methods are used to calculate the factors that convert a measurement in instrumental units to physical units. (shrink)

Inferences from the absence of evidence to something are common in ordinary speech, but when used in scientific argumentations are usually considered deficient or outright false. Yet, as demonstrated here with the help of various examples, archaeologists frequently use inferences and reasoning from absence, often allowing it a status on par with inferences from tangible evidence. This discrepancy has not been examined so far. The article analyses it drawing on philosophical discussions concerning the validity of inference from absence, using probabilistic (...) models that were originally developed to show that such inferences are weak and inconclusive. The analysis reveals that inference from absence can indeed be justified in many important situations of archaeological research, such as excavations carried out to explore the past existence and time-span of sedentary human habitation. The justification is closely related to the fact that archaeology explores the human past via its material remains. The same analysis points to instances where inference from absence can have comparable validity in other historical sciences, and to research questions in which archaeological inference from absence will be problematic or totally unwarranted. (shrink)

As our epistemic ambitions grow, the common and scientific endeavours are becoming increasingly dependent on Machine Learning (ML). The field rests on a single experimental paradigm, which consists of splitting the available data into a training and testing set and using the latter to measure how well the trained ML model generalises to unseen samples. If the model reaches acceptable accuracy, an a posteriori contract comes into effect between humans and the model, supposedly allowing its deployment to target environments. Yet (...) the latter part of the contract depends on human inductive predictions or generalisations, which infer a uniformity between the trained ML model and the targets. The paper asks how we justify the contract between human and machine learning. It is argued that the justification becomes a pressing issue when we use ML to reach ‘elsewheres’ in space and time or deploy ML models in non-benign environments. The paper argues that the only viable version of the contract can be based on optimality (instead of on reliability which cannot be justified without circularity) and aligns this position with Schurz’s optimality justification. It is shown that when dealing with inaccessible/unstable ground-truths (‘elsewheres’ and non-benign targets), the optimality justification undergoes a slight change, which should reflect critically on our epistemic ambitions. Therefore, the study of ML robustness should involve not only heuristics that lead to acceptable accuracies on testing sets. The justification of human inductive predictions or generalisations about the uniformity between ML models and targets should be included as well. Without it, the assumptions about inductive risk minimisation in ML are not addressed in full. (shrink)

Das Dissertationsprojekt soll eine Verbindung zwischen zwei augenscheinlich separaten Debattensträngen in der kontemporären Wissenschaftstheorie erforschen. Das ist zum einen die Debatte um die humesche Doktrin, der zufolge es keine de-re notwendigen Verbindungen respektive modale Fakten in der natürlichen Welt gibt. Diese Debatte betrifft die Grundlagen der Naturwissenschaften sowie der Natur selbst. Und das ist zum anderen die Debatte um die materiale Theorie der Induktion, welche auf John D. Norton zurückgeht und die Grundlagen der Rationalität und Logik betrifft. Nach dieser Theorie (...) werden induktive Schlüsse qua relevanter, lokaler Hintergrundfakten zuverlässig. Norton bringt viele - in meinen Augen überzeugende - Beispiele für diese Theorie, lässt uns aber im Unklaren darüber, wie diese Hintergrundfakten induktive Schlüsse lizensieren können. Ich werde argumentieren, dass es sich bei diesen relevanten Fakten zumindest in einigen paradigmatischen Fällen um irreduzibel-modale oder modal grundierte Fakten handeln muss, ansonsten können diese die ihnen zugedachte Funktion nicht erfüllen. Infolgedessen steht der Humeanismus vor einer materialen Form des Induktionsproblems. -/- Meine Forschungshypothese lautet, dass es sich bei dem materialen Induktionsproblem um das eigentliche und zentrale Problem des Humeanismus handelt. Zum einen werden viele altbekannte Probleme des Humeanismus durch das materiale Induktionsproblem motiviert und bestärkt. Zum anderen begründet das materiale Induktionsproblem ein neues und tiefgreifendes Problem des Humeanismus. Es stellt sich darin, dass der Humeanismus den empirischen Erfolg unserer verfügbaren wissenschaftlichen und alltäglichen Gesetzesannahmen nicht zirkelfrei erklären kann. Darauf aufbauend formuliere ich ein Super-Wunderargument, das im Kern in einer Inferenz von den bisherigen empirischen Erfolgen unserer verfügbaren Gesetzesannahmen auf die Wahrheit des Anti-Humeanismus besteht. Dieses Argument steht im Zentrum meiner Arbeit. Alles Weitere wird im Kontext dieses Argumentes diskutiert und illustriert. Zu diesen Untersuchungen gehören die schrittweise Motivation des Argumentes durch das materiale Induktionsproblem, die Unterscheidung und Diskussion verschiedener logischer Formen des Argumentes, das Herausarbeiten der Vorteile sowie die Antizipation von möglichen Einwänden gegen das Argument. (shrink)

In light of the recent credibility crisis in psychology, this paper argues for a greater emphasis on theorizing in scientific research. Although reliable experimental evidence, preregistration, methodological rigor, and new computational frameworks for modeling are important, scientific progress also relies on properly functioning theories. However, the current understanding of the role of theorizing in psychology is lacking, which may lead to future crises. Theories should not be viewed as mere speculations or simple inductive generalizations. To address this issue, the author (...) introduces a framework called “cognitive metascience,” which studies the processes and results of evaluating scientific practice. This study should proceed both qualitatively, as in traditional science and technology studies and cognitive science, and quantitatively, by analyzing scientific discourse using language technology. By analyzing theories as cognitive artifacts that support cognitive tasks, this paper aims to shed more light on their nature. This perspective reveals that multiple distinct theories serve entirely different roles, and studying these roles, along with their epistemic vices and virtues, can provide insight into how theorizing should proceed. The author urges a change in research culture to appreciate the variety of distinct theories and to systematically advance scientific progress. . (shrink)