This article centers on Hume’s position on the intelligibility of natural philosophy. To that end, the controversy surrounding universal gravitation shall be scrutinized. It is very well-known that Hume sides with the Newtonian experimentalist approach rather than with the Leibnizian demand for intelligibility. However, what is not clear is Hume’s overall position on the intelligibility of natural philosophy. It shall be argued that Hume declines Leibniz’s principle of intelligibility. However, Hume does not eschew intelligibility altogether; his concept of causation itself (...) stipulates mechanical intelligibility. (shrink)

Simple idealized models seem to provide more understanding than opaque, complex, and hyper-realistic models. However, an increasing number of scientists are going in the opposite direction by utilizing opaque machine learning models to make predictions and draw inferences, suggesting that scientists are opting for models that have less potential for understanding. Are scientists trading understanding for some other epistemic or pragmatic good when they choose a machine learning model? Or are the assumptions behind why minimal models provide understanding misguided? In (...) this paper, using the case of deep neural networks, I argue that it is not the complexity or black box nature of a model that limits how much understanding the model provides. Instead, it is a lack of scientific and empirical evidence supporting the link that connects a model to the target phenomenon that primarily prohibits understanding. (shrink)

Understanding is an ability manifested by grasping relations of a phenomenon and articulating new explanations. Hence, scientific understanding is inextricably intertwined with and not possible without explanation, and understanding is not a type of propositional knowledge. Anna Elisabeth Höhl provides a novel philosophical account of scientific understanding by developing and defending necessary and sufficient conditions for the understanding that scientists achieve of the phenomena they are researching. This account of scientific understanding is based on and supported by a detailed investigation (...) of an episode from scientific practice in biology. (shrink)

These preprints were automatically compiled into a PDF from the collection of papers deposited in PhilSci-Archive in conjunction with the PSA 2018.

I examine situations in which we say that different subjects have ‘different’, ‘competing’, or ‘conflicting understandings’ of a phenomenon. In order to make sense of such situations, we should turn our attention to an often neglected ambiguity in the word ‘understanding’. Whereas the notion of understanding that is typically discussed in philosophy is, to use Elgin’s terms, tethered to the facts, there is another notion of understanding that is not tethered in the same way. This latter notion is relevant because, (...) typically, talk of two subjects having ‘different’, ‘competing’, or ‘conflicting understandings’ of a phenomenon does not entail any commitment to the proposition that these subjects understand the phenomenon in the tethered sense of the word. This paper aims, first, to analyze the non-tethered notion of understanding, second, to clarify its relationship to the tethered notion, third, to explore what exactly goes on when ‘different’, ‘competing’, or ‘conflicting understandings’ clash and, fourth, to discuss the significance of such situations in our epistemic practices. In particular, I argue for a version of scientific pluralism according to which such situations are important because they help scientific communities achieve their fundamental epistemic goals—most importantly, the goal of understanding the world in the tethered sense. (shrink)

This chapter defends a (minimal) realist conception of progress in scientific understanding in the face of the ubiquitous plurality of perspectives in science. The argument turns on the counterfactual-dependence framework of explanation and understanding, which is illustrated and evidenced with reference to different explanations of the rainbow.

The problem of epistemic opacity in Artificial Intelligence is often characterised as a problem of intransparent algorithms that give rise to intransparent models. However, the degrees of transparency of an AI model should not be taken as an absolute measure of the properties of its algorithms but of the model’s degree of intelligibility to human users. Its epistemically relevant elements are to be specified on various levels above and beyond the computational one. In order to elucidate this claim, I first (...) contrast computer models and their claims to algorithm-based universality with cybernetics-style analogue models and their claims to structural isomorphism between elements of model and target system. While analogue models aim at perceptually or conceptually accessible model-target relations, computer models give rise to a specific kind of underdetermination in these relations that needs to be addressed in specific ways. I then undertake a comparison between two contemporary AI approaches that, although related, distinctly align with the above modelling paradigms and represent distinct strategies towards model intelligibility: Deep Neural Networks and Predictive Processing. I conclude that their respective degrees of epistemic transparency primarily depend on the underlying purposes of modelling, not on their computational properties. (shrink)

I argue that philosophical issues concerning reductive explanations help constrain the construction of quantum theories with appropriate state spaces. I illustrate this general proposal with two examples of restricting attention to physical states in quantum theories: regular states and symmetry-invariant states. 1Introduction2Background2.1 Physical states2.2 Reductive explanations3The Proposed ‘Correspondence Principle’4Example: Regularity5Example: Symmetry-Invariance6Conclusion: Heuristics and Discovery.

Interest in the demarcation problem is undergoing a boom after being shelved and even given up for dead. Nevertheless, despite current philosophical discussions, there are no substantial advances i...

We argue that artificial networks are explainable and offer a novel theory of interpretability. Two sets of conceptual questions are prominent in theoretical engagements with artificial neural networks, especially in the context of medical artificial intelligence: Are networks explainable, and if so, what does it mean to explain the output of a network? And what does it mean for a network to be interpretable? We argue that accounts of “explanation” tailored specifically to neural networks have ineffectively reinvented the wheel. In (...) response to, we show how four familiar accounts of explanation apply to neural networks as they would to any scientific phenomenon. We diagnose the confusion about explaining neural networks within the machine learning literature as an equivocation on “explainability,” “understandability” and “interpretability.” To remedy this, we distinguish between these notions, and answer by offering a theory and typology of interpretation in machine learning. Interpretation is something one does to an explanation with the aim of producing another, more understandable, explanation. As with explanation, there are various concepts and methods involved in interpretation: Total or Partial, Global or Local, and Approximative or Isomorphic. Our account of “interpretability” is consistent with uses in the machine learning literature, in keeping with the philosophy of explanation and understanding, and pays special attention to medical artificial intelligence systems. (shrink)

In recent years, scientific understanding has become a focus of attention in philosophy of science. Since understanding is typically associated with the pragmatic and psychological dimensions of explanation, shifting the focus from explanation to understanding may induce a shift from accounts that embody normative ideals to accounts that provide accurate descriptions of scientific practice. Not surprisingly, many ‘friends of understanding’ sympathize with a naturalistic approach to the philosophy of science. However, this raises the question of whether the proposed theories of (...) understanding can still have normative power. In this paper I address this question by examining two theories of scientific understanding: Jan Faye’s pragmatic-rhetorical theory and my own contextual theory of scientific understanding. I argue that both theories leave room for normativity, despite their naturalistic tendencies. The normative power of my contextual theory is illustrated with a case study of the chemical revolution. (shrink)

While the relation between visualization and scientific understanding has been a topic of long-standing discussion, recent developments in physics have pushed the boundaries of this debate to new and still unexplored realms. For it is claimed that, in certain theories of quantum gravity, spacetime ‘disappears’: and this suggests that one may have sensible physical theories in which spacetime is completely absent. This makes the philosophical question whether such theories are intelligible, even more pressing. And if such theories are intelligible, the (...) question then is how they manage to do so. In this paper, we adapt the contextual theory of scientific understanding, developed by one of us, to fit the novel challenges posed by physical theories without spacetime. We construe understanding as a matter of skill rather than just knowledge. The appeal is thus to understanding, rather than explanation, because we will be concerned with the tools that scientists have at their disposal for understanding these theories. Our central thesis is that such physical theories can provide scientific understanding, and that such understanding does not require spacetimes of any sort. Our argument consists of four consecutive steps: (a) We argue, from the general theory of scientific understanding, that although visualization is an oft-used tool for understanding, it is not a necessary condition for it; (b) we criticise certain metaphysical preconceptions which can stand in the way of recognising how intelligibility without spacetime can be had; (c) we catalogue tools for rendering theories without a spacetime intelligible; and (d) we give examples of cases in which understanding is attained without a spacetime, and explain what kind of understanding these examples provide. (shrink)

In a recent paper, Erasmus et al. (2021) defend the idea that the ambiguity of the term “explanation” in explainable AI (XAI) can be solved by adopting any of four different extant accounts of explanation in the philosophy of science: the Deductive Nomological, Inductive Statistical, Causal Mechanical, and New Mechanist models. In this chapter, I show that the authors’ claim that these accounts can be applied to deep neural networks as they would to any natural phenomenon is mistaken. I also (...) provide a more general argument as to why the notion of explainability as it is currently used in the XAI literature bears little resemblance to the traditional concept of scientific explanation. It would be more fruitful to use the label “understandable AI” to avoid the confusion that surrounds the goal and purposes of XAI. In the second half of the chapter, I argue for a pragmatic conception of understanding that is better suited to play the central role attributed to explanation in XAI. Following De Regt (2017) and Kuorikoski and Ylikoski (2015), the conditions of satisfaction for understanding an ML system are fleshed out in terms of an agent’s success in using the system. (shrink)

I conceptualise the role of dualities in quantum gravity, in terms of their functions for theory construction. I distinguish between two functions of duality in physical practice: namely, discovering and describing ‘equivalent physics’, versus suggesting ‘new physics’. I dub these the ‘theoretical’ versus the ‘heuristic’ functions of dualities. The distinction seems to have gone largely unnoticed in the philosophical literature: and it exists both for dualities, and for the more general relation of theoretical equivalence. The paper develops the heuristic function (...) of dualities: illustrating how they can be used, if one has any luck, to find and formulate new theories. I also point to the different physical commitments about the theories in question that underlie these two functions. I show how a recently developed schema for dualities articulates the differences between the two functions. (shrink)

In this paper we have two aims: first, to draw attention to the close connexion between interpretation and scientific understanding; second, to give a detailed account of how theories without a spacetime can be interpreted, and so of how they can be understood. In order to do so, we of course need an account of what is meant by a theory ‘without a spacetime’: which we also provide in this paper. We describe three tools, used by physicists, aimed at constructing (...) interpretations which are adequate for the goal of understanding. We analyse examples from high-energy physics illustrating how physicists use these tools to construct interpretations and thereby attain understanding. The examples are: the ’t Hooft approximation of gauge theories, random matrix models, causal sets, loop quantum gravity, and group field theory. (shrink)

While the relation between visualization and scientific understanding has been a topic of long-standing discussion, recent developments in physics have pushed the boundaries of this debate to new and still unexplored realms. For it is claimed that, in certain theories of quantum gravity, spacetime ‘disappears’: and this suggests that one may have sensible physical theories in which spacetime is completely absent. This makes the philosophical question whether such theories are intelligible, even more pressing. And if such theories are intelligible, the (...) question then is how they manage to do so. In this paper, we adapt the contextual theory of scientific understanding, developed by one of us, to fit the novel challenges posed by physical theories without spacetime. We construe understanding as a matter of skill rather than just knowledge. The appeal is thus to understanding, rather than explanation, because we will be concerned with the tools that scientists have at their disposal for understanding these theories. Our central thesis is that such physical theories can provide scientific understanding, and that such understanding does not require spacetimes of any sort. Our argument consists of four consecutive steps: We argue, from the general theory of scientific understanding, that although visualization is an oft-used tool for understanding, it is not a necessary condition for it; we criticise certain metaphysical preconceptions which can stand in the way of recognising how intelligibility without spacetime can be had; we catalogue tools for rendering theories without a spacetime intelligible; and we give examples of cases in which understanding is attained without a spacetime, and explain what kind of understanding these examples provide. (shrink)

Understanding natural phenomena is an important aim of science. Since the turn of the millennium the notion of scientific understanding has been a hot topic of debate in the philosophy of science. A bone of contention in this debate is the role of truth and representational accuracy in scientific understanding. So-called factivists and non-factivists disagree about the extent to which the theories and models that are used to achieve understanding must be true or accurate. In this paper we address this (...) issue by examining a case from the practice of synthetic chemistry. We investigate how understanding is obtained in this field by means of an in-depth analysis of the famous synthesis of periplanone B by W. Clark Still. It turns out that highly idealized models—that are representationally inaccurate and sometimes even inconsistent—and qualitative concepts are essential for understanding the synthetic pathway and accordingly for achieving the synthesis. We compare the results of our case study to various factivist and non-factivist accounts of how idealizations may contribute to scientific understanding and conclude that non-factivism offers a more plausible interpretation of the practice of synthetic chemistry. Moreover, our case study supports a central thesis of the non-factivist theory of scientific understanding developed by De Regt, namely that scientific understanding requires intelligibility rather than representational accuracy, and that idealization is one way to enhance intelligibility. (shrink)

Deep neural networks have become increasingly successful in applications from biology to cosmology to social science. Trained DNNs, moreover, correspond to models that ideally allow the prediction of new phenomena. Building in part on the literature on ‘eXplainable AI’, I here argue that these models are instrumental in a sense that makes them non-explanatory, and that their automated generation is opaque in a unique way. This combination implies the possibility of an unprecedented gap between discovery and explanation: When unsupervised models (...) are successfully used in exploratory contexts, scientists face a whole new challenge in forming the concepts required for understanding underlying mechanisms. (shrink)

The use of machine learning instead of traditional models in neuroscience raises significant questions about the epistemic benefits of the newer methods. I draw on the literature on model intelligibility in the philosophy of science to offer some benchmarks for the interpretability of artificial neural networks used as a predictive tool in neuroscience. Following two case studies on the use of ANN’s to model motor cortex and the visual system, I argue that the benefit of providing the scientist with understanding (...) of the brain trades off against the predictive accuracy of the models. This trade-off between prediction and understanding is better explained by a non-factivist account of scientific understanding. (shrink)

Within eighteenth-century debates on animal cognition we can distinguish at least three main theoretical positions: (i) Buffon’s mechanism, (ii) Reimarus’ theory of instincts, and (iii) the sensationalism of Condillac and Leroy. In this paper, I adopt a philosophical perspective on this debate and argue that in order to fully understand the justification Buffon, Reimarus, Condillac, and Leroy gave for their respective theories, we must pay special attention to the theoretical virtues these naturalists alluded to while justifying their position. These theoretical (...) virtues have received little to no attention in the literature on eighteenth-century animal cognition, but figure prominently in the justification of the mechanist, instinctive, and sensationalist theories of animal behavior. Through my philosophical study of the role of theoretical virtues in eighteenth-century debates on animal cognition, we obtain a deeper understanding of how theoretical virtues were conceptualized in eighteenth-century science and how they influenced the justification of theories of animal cognition. (shrink)

The Minimalist Program in generative linguistics is predicated on the idea that simplicity is a defining property of the human language faculty, on the one hand; on the other, a central aim of linguistic theorising. Worryingly, however, justifications for either claim are hard to come by in the literature. We sketch a proposal that would allow for both shortcomings to be addressed, and that furthermore honours the program’s declared commitment to naturalism. We begin by teasing apart and clarifying the different (...) conceptions of simplicity underlying generative inquiry, in both ontological and theoretical capacities. We then trace a path towards a more robust justification for each type of simplicity principle, drawing on recent work in cognitive science and in philosophy of science, respectively. The resulting proposal hinges on the idea that simplicity is an evolved, virtuous cognitive bias—one that is a condition of our scientific understanding and, ultimately, of successful scientific practice. Finally, we make a case for why minimalists should take this proposal seriously, on the one hand; and for why generative linguistics would make for an interesting case study for philosophy of science, on the other. (shrink)

“Explanation” refers to a wide range of activities, with a family resemblance between them. Most satisfactory explanations in a discipline for a domain fail to satisfy some general desiderata, while fulfilling others. This can happen in various ways. Why? An idealizing response would be to say that in real science explanations fall short along some dimensions, so that for any explanatory failure there is a conceivable improvement that addresses its shortcomings. The improvement may be more accurate causally or possess more (...) unifying power, or deliver deeper understanding. We formulate a drastically less idealizing response. We argue that there are typically trade-offs in explanation, so that in strengthening one explanatory virtue one will usually weaken another. Scientific explanations, if this is correct, are constrained by such trade-offs. Particular trade-offs are appropriate for particular explanatory vehicles. There are the overarching equations of theoretical physics, which produce unification at the expense of causal detail; there are theoretical models of phenomena that occupy a middle ground between generality and the detailed workings of particular cases and get closer to explaining the workings of specific systems at the expense of unification. Sometimes experiments aim at general causal patterns at the expense of particular detail; and sometimes they are designed to give us information of particular detail at the expense of generality. There are further trade-offs associated with other vehicles of explanation. We provide examples from physics and biology. (shrink)

In this paper significant challenges are raised with respect to the view that explanation essentially involves unification. These objections are raised specifically with respect to the well-known versions of unificationism developed and defended by Michael Friedman and Philip Kitcher. The objections involve the explanatory regress argument and the concepts of reduction and scientific understanding. Essentially, the contention made here is that these versions of unificationism wrongly assume that reduction secures understanding.

Metaphors are found all throughout science: in published papers, working hypotheses, policy documents, lecture slides, grant proposals, and press releases. They serve different functions, but perhaps most striking is the way they enable understanding, of a theory, phenomenon, or idea. In this paper, we leverage recent advances on the nature of metaphor and the nature of understanding to explore how they accomplish this feat. We attempt to shift the focus away from the epistemic value of the content of metaphors, to (...) the epistemic value of the metaphor’s consequences. Many famous scientific metaphors are epistemically good, not primarily because of what they say about the world, but because of how they cause us to think. Specifically, metaphors increase understanding either by improving our sets of representations, or by making it easier for us to encode and process data about complex subjects by changing how we are disposed to conceptualize those subjects. This view hints towards new positions concerning testimonial understanding, factivity, abilities, discovery via metaphor, and the relation between metaphors and models. (shrink)

In this paper, we will explore one way in which understanding can—and, we will argue, should—be valuable. We will do this by drawing on what has been said about the different ways knowledge can be valuable. Our main contribution will be to identify one heretofore undiscussed way knowledge could be valuable, but isn’t—specifically, having value to someone other than the understander. We suggest that it is a desideratum on an account of understanding that understanding have the specified type of value; (...) our basis for this claim will come from recent work in cognitive psychology. This desideratum can then be used to measure the success of various accounts of understanding. We argue that accounts of understanding that have a particular structure will predict why understanding has that sort of value. For good measure, we then engage in a bit of a literature review, investigating which extant accounts of understanding satisfy this desideratum. (shrink)

One can find in the literature two sets of views concerning the relationship between understanding and explanation: that one understands only if 1) one has knowledge of causes and 2) that knowledge is provided by an explanation. Taken together, these tenets characterize what I call the narrow knowledge account of understanding. While the first tenet has recently come under severe attack, the second has been more resistant to change. I argue that we have good reasons to reject it on the (...) basis of theoretical models that provide how-possibly explanations. These models, while they do not explain in the strict sense, afford understanding. In response, I propose an alternative epistemology of understanding, broad KAU, that takes cases of theoretical modelling into account. (shrink)

This article compares treatments of the Stern–Gerlach experiment across different physical theories, building up to a novel analysis of electron spin measurement in the context of classical Dirac field theory. Modeling the electron as a classical rigid body or point particle, we can explain why the entire electron is always found at just one location on the detector but we cannot explain why there are only two locations where the electron is ever found. Using non-relativistic or relativistic quantum mechanics, we (...) can explain both uniqueness and discreteness. Moving to more fundamental physics, both features can be explained within a quantum theory of the Dirac field. In a classical theory of the Dirac field, the rotating charge of the electron can split into two pieces that each hit the detector at a different location. In this classical context, we can explain a feature of electron spin that is often described as distinctively quantum but we cannot explain another feature that could be explained within any of the other theories. (shrink)

What exactly do we need in order to enjoy the cognitive benefit that is supposed to be provided by an explanation? Some philosophers :15–37, 2012, Episteme 10:1–17, 2013, Eur J Philos Sci 5:377–385, 2015, Understanding, explanation, and scientific knowledge, Cambridge University Press, Cambridge, 2017) would say that all that we need is to know the explanation. Others :1–26, 2012; Strevens in Stud Hist Philos Sci Part A 44:510–515, 2013) would say that achieving understanding with the help of an explanation requires (...) more than that, that it requires a grasping or an understanding of the explanation. My aim in this paper is to come up with a new answer to this problem by exploring the shortcomings of the received view of understanding. In my view, besides having testimonial knowledge of an explanation, obtaining explanatory understanding requires full cognitive access to the explanation. (shrink)

We argue that mechanistic models elaborated by machine learning cannot be explanatory by discussing the relation between mechanistic models, explanation and the notion of intelligibility of models. We show that the ability of biologists to understand the model that they work with severely constrains their capacity of turning the model into an explanatory model. The more a mechanistic model is complex, the less explanatory it will be. Since machine learning increases its performances when more components are added, then it generates (...) models which are not intelligible, and hence not explanatory. (shrink)

A lot of philosophical energy has been devoted recently in trying to determine if mathematics can contribute to our understanding of physical phenomena. Not many philosophers are interested, though, if the converse makes sense, i.e., if our cognitive interaction (scientific or otherwise) with the physical world can be helpful (in an explanatory or non-explanatory way) in our efforts to make sense of mathematical facts. My aim in this paper is to try to fill this important lacuna in the recent literature. (...) My answer to the question of this paper is negative. As I will argue, there are serious problems with the main reasons for believing in the first place that it is possible to have physical understanding of mathematical facts. (shrink)

Highly idealized models may serve various epistemic functions, notably explanation, in virtue of representing the world. Inferentialism provides a prima facie compelling characterization of what constitutes the representation relation. In this paper, I argue that what I call factive inferentialism does not provide a satisfactory solution to the puzzle of model-based—factive—explanation. In particular, I show that making explanatory counterfactual inferences is not a sufficient guide for accurate representation, factivity, or realism. I conclude by calling for a more explicit specification of (...) model-world mismatches and properties imputation. (shrink)

In this paper we have two aims: first, to draw attention to the close connexion between interpretation and scientific understanding; second, to give a detailed account of how theories without a spacetime can be interpreted, and so of how they can be understood. In order to do so, we of course need an account of what is meant by a theory ‘without a spacetime’: which we also provide in this paper. We describe three tools, used by physicists, aimed at constructing (...) interpretations which are adequate for the goal of understanding. We analyse examples from high-energy physics illustrating how physicists use these tools to construct interpretations and thereby attain understanding. The examples are: the ’t Hooft approximation of gauge theories, random matrix models, causal sets, loop quantum gravity, and group field theory. (shrink)