Scientific understanding as a subject of inquiry has become widely discussed in philosophy of science and is often addressed through case studies from history of science. Even though these historical reconstructions engage with details of scientific practice, they usually provide only limited information about the gradual formation of understanding in ongoing processes of model and theory construction. Based on a qualitative ethnographic study of an ecological research project, this article shifts attention from understanding in the context of historical case studies (...) to evidence of current case studies. By taking de Regt’s contextual theory of scientific understanding into the field, it confirms core tenets of the contextual theory suggesting a normative character with respect to scientific activities. However, the case study also shows the limitations of de Regt’s latest version of this theory as an attempt to explain the development of understanding in current practice. This article provides a model representing the emergence of scientific understanding that exposes main features of scientific understanding such as its gradual formation, its relation to skills and imagination, and its capacity for knowledge selectivity. The ethnographic evidence presented here supports the claim that something unique can be learned by looking into ongoing research practices that can’t be gained by studying historical case studies. (shrink)

In the last decades, supervised machine learning has seen the widespread growth of highly complex, non-interpretable models, of which deep neural networks are the most typical representative. Due to their complexity, these models have showed an outstanding performance in a series of tasks, as in image recognition and machine translation. Recently, though, there has been an important discussion over whether those non-interpretable models are able to provide any sort of understanding whatsoever. For some scholars, only interpretable models can provide understanding. (...) More popular, however, is the idea that understanding can come from a careful analysis of the dataset or from the model’s theoretical basis. In this paper, I wish to examine the possible forms of obtaining understanding of such non-interpretable models. Two main strategies for providing understanding are analyzed. The first involves understanding without interpretability, either through external evidence for the model’s inner functioning or through analyzing the data. The second is based on the artificial production of interpretable structures, through three main forms: post hoc models, hybrid models, and quasi-interpretable structures. Finally, I consider some of the conceptual difficulties in the attempt to create explanations for these models, and their implications for understanding. (shrink)

The literature on values in science struggles with questions about how to describe and manage the role of values in scientific research. We argue that progress can be made by shifting this literature’s current emphasis. Rather than arguing about how non-epistemic values can or should figure into scientific assessment, we suggest analyzing how scientific assessment can accommodate non-epistemic values. For scientific assessment to do so, it arguably needs to incorporate goals that have been traditionally characterized as non-epistemic. Building on this (...) insight, we show how the adequacy-for-purpose framework recently developed for assessing scientific models can provide a general framework for describing scientific assessment so that it goes beyond purely epistemic considerations. Adopting this framework has significant advantages and opens the possibility of effecting a partial rapprochement between critics and proponents of the value-free ideal. (shrink)

The aim of this article is to develop an understanding-based argument for an explicitly political specification of the concept of race. It is argued that a specification of race in terms of hierarchical social positions is best equipped to guide causal reasoning about racial inequality in the public sphere. Furthermore, the article provides evidence that biological and cultural specifications of race mislead public reasoning by encouraging confusions between correlates and causes of racial inequality. The article concludes with a more general (...) case for incorporating empirical evidence about public reasoning into philosophical debates about competing specifications of the concept of race. (shrink)

In the last few years, biologists and computer scientists have claimed that the introduction of data science techniques in molecular biology has changed the characteristics and the aims of typical outputs (i.e. models) of such a discipline. In this paper we will critically examine this claim. First, we identify the received view on models and their aims in molecular biology. Models in molecular biology are mechanistic and explanatory. Next, we identify the scope and aims of data science (machine learning in (...) particular). These lie mainly in the creation of predictive models which performances increase as data set increases. Next, we will identify a tradeoff between predictive and explanatory performances by comparing the features of mechanistic and predictive models. Finally, we show how this a priori analysis of machine learning and mechanistic research applies to actual biological practice. This will be done by analyzing the publications of a consortium—The Cancer Genome Atlas—which stands at the forefront in integrating data science and molecular biology. The result will be that biologists have to deal with the tradeoff between explaining and predicting that we have identified, and hence the explanatory force of the ‘new’ biology is substantially diminished if compared to the ‘old’ biology. However, this aspect also emphasizes the existence of other research goals which make predictive force independent from explanation. (shrink)

Predictive processing (PP) has been repeatedly presented as a unificatory account of perception, action, and cognition. In this paper, we argue that this is premature: As a unifying theory, PP fails to deliver general, simple, homogeneous, and systematic explanations. By examining its current trajectory of development, we conclude that PP remains only loosely connected both to its computational framework and to its hypothetical biological underpinnings, which makes its fundamentals unclear. Instead of offering explanations that refer to the same set of (...) principles, we observe systematic equivocations in PP‐based models, or outright contradictions with its avowed principles. To make matters worse, PP‐based models are seldom empirically validated, and they are frequently offered as mere just‐so stories. The large number of PP‐based models is thus not evidence of theoretical progress in unifying perception, action, and cognition. On the contrary, we maintain that the gap between theory and its biological and computational bases contributes to the arrested development of PP as a unificatory theory. Thus, we urge the defenders of PP to focus on its critical problems instead of offering mere re‐descriptions of known phenomena, and to validate their models against possible alternative explanations that stem from different theoretical assumptions. Otherwise, PP will ultimately fail as a unified theory of cognition. (shrink)

The connection between understanding and explanation has recently been of interest to philosophers. Inglis and Mejía-Ramos (Synthese, 2019) propose that within mathematics, we should accept a functional account of explanation that characterizes explanations as those things that produce understanding. In this paper, I start with the assumption that this view of mathematical explanation is correct and consider what we can consequently learn about mathematical explanation. I argue that this view of explanation suggests that we should shift the question of explanation (...) away from why-questions and towards a “what’s going on here” question. Additionally, I argue that when we recognize the connection between understanding and explanation we naturally see how more than just proof can be explanatory. I expand this point by detailing how definitions and diagrams can be explanatory. In all, we see that when we take seriously the connection between understanding and explanation, we get a better sense of how explanation arises within mathematics. (shrink)

Undergraduate geoscience students are rarely exposed to history and philosophy of science. I will describe the experiences with a short course unfavourably placed in the first year of a bachelor of earth science. Arguments how HPS could enrich their education in many ways are sketched. One useful didactic approach is to develop a broader interest by connecting HPS themes to practical cases throughout the curriculum, and develop learning activities that allow students to reflect on their skills, methods and their field (...) in relation to other disciplines and interactions with society with abilities gained through exposure to HPS. Given support of the teaching staff, the tenets of philosophy of science in practice, of conceptual history of knowledge, and of ethics of science for society can fruitfully and directly be connected to the existing curriculum. This is ideally followed by a capstone HPS course late in the bachelor programme. (shrink)

Many have argued that the human sciences feature a unique form of understanding that is absent from the natural sciences. However, in the last decade or so, epistemologists and philosop...

Machine learning methods have recently created high expectations in the climate modelling context in view of addressing climate change, but they are often considered as non-physics-based ‘black boxes’ that may not provide any understanding. However, in many ways, understanding seems indispensable to appropriately evaluate climate models and to build confidence in climate projections. Relying on two case studies, we compare how machine learning and standard statistical techniques affect our ability to understand the climate system. For that purpose, we put five (...) evaluative criteria of understanding to work: intelligibility, representational accuracy, empirical accuracy, coherence with background knowledge, and assessment of the domain of validity. We argue that the two families of methods are part of the same continuum where these various criteria of understanding come in degrees, and that therefore machine learning methods do not necessarily constitute a radical departure from standard statistical tools, as far as understanding is concerned. (shrink)

There has been a burst of work in the last couple of decades on mechanistic explanation, as an alternative to the traditional covering-law model of scientific explanation. That work makes some interesting claims about mechanistic explanations rendering phenomena ‘intelligible’, but does not develop this idea in great depth. There has also been a growth of interest in giving an account of scientific understanding, as a complement to an account of explanation, specifically addressing a three-place relationship between explanation, world, and the (...) scientific community. The aim of this paper is to use the contextual theory of scientific understanding to build an account of understanding phenomena using mechanistic explanations. This account will be developed and illustrated by examining the mechanisms of supernovae, which will allow synthesis of treatment of the life sciences and social sciences on the one hand, where many accounts of mechanisms were originally developed, and treatment of physics on the other hand, where the contextual theory drew its original inspiration. (shrink)

Science frequently gives us multiple, compatible ways of solving the same problem or formulating the same theory. These compatible formulations change our understanding of the world, despite providing the same explanations. According to what I call "conceptualism," reformulations change our understanding by clarifying the epistemic structure of theories. I illustrate conceptualism by analyzing a typical example of symmetry-based reformulation in chemical physics. This case study poses a problem for "explanationism," the rival thesis that differences in understanding require ontic explanatory differences. (...) To defend conceptualism, I consider how prominent accounts of explanation might accommodate this case study. I argue that either they do not succeed, or they generate a skeptical challenge. (shrink)

Hempel seems to hold the following three views: (H1) Understanding is pragmatic/relativistic: Whether one understands why X happened in terms of Explanation E depends on one's beliefs and cognitive abilities; (H2) Whether a scientific explanation is good, just like whether a mathematical proof is good, is a nonpragmatic and objective issue independent of the beliefs or cognitive abilities of individuals; (H3) The goal of scientific explanation is understanding: A good scientific explanation is the one that provides understanding. Apparently, H1, H2, (...) and H3 cannot be all true. Some philosophers think that Hempel is inconsistent, while some others claim that Hempel does not actually hold H3. I argue that Hempel does hold H3 and that he can consistently hold all of H1, H2, and H3 if he endorses what I call the “understanding argument.” I also show how attributing the understanding argument to Hempel can make more sense of his D-N model and his philosophical analysis of the pragmatic aspects of scientific explanation. (shrink)

I want to combine two hitherto largely independent research projects, scientific understanding and mechanistic explanations. Understanding is not only achieved by answering why-questions, that is, by providing scientific explanations, but also by answering what-questions, that is, by providing what I call scientific descriptions. Based on this distinction, I develop three forms of understanding: understanding-what, understanding-why, and understanding-how. I argue that understanding-how is a particularly deep form of understanding, because it is based on mechanistic explanations, which answer why something happens in (...) virtue of what it is made of. I apply the three forms of understanding to two case studies: first, to the historical development of thermodynamics and, second, to the differences between the Clausius and the Boltzmann entropy in explaining thermodynamic processes. (shrink)

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)

What is the primary intellectual aim of philosophy? The standard view is that philosophy aims to provide true answers to philosophical questions. But if our aim is to settle controversy by answering such questions, our discipline is an embarrassing failure. Moreover, taking philosophy to aim at providing true answers to these questions leads to a variety of puzzles: How do we account for philosophical expertise? How is philosophical progress possible? Why do job search committees not care about the truth or (...) falsity of a candidate’s philosophical views? We argue that philosophy does not aim at discovering true answers to philosophical questions. Instead, its primary intellectual aim is understanding. We argue that many familiar aspects of philosophy become intelligible once we accept this hypothesis. (shrink)

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)

COVID-19 has substantially affected our lives during 2020. Since its beginning, several epidemiological models have been developed to investigate the specific dynamics of the disease. Early COVID-19 epidemiological models were purely statistical, based on a curve-fitting approach, and did not include causal knowledge about the disease. Yet, these models had predictive capacity; thus they were used to ground important political decisions, in virtue of the understanding of the dynamics of the pandemic that they offered. This raises a philosophical question about (...) how purely statistical models can yield understanding, and if so, what the relationship between prediction and understanding in these models is. Drawing on the model that was developed by the Institute of Health Metrics and Evaluation, we argue that early epidemiological models yielded a modality of understanding that we call descriptive understanding, which contrasts with the so-called explanatory understanding which is assumed to be the main form of scientific understanding. We spell out the exact details of how descriptive understanding works, and efficiently yields understanding of the phenomena. Finally, we vindicate the necessity of studying other modalities of understanding that go beyond the conventionally assumed explanatory understanding. (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)

Contemporary debate surrounding the nature of scientific progress has focused upon the precise role played by justification, with two realist accounts having dominated proceedings. Recently, however, a third realist account has been put forward, one which offers no role for justification at all. According to Finnur Dellsén’s (Stud Hist Philos Sci Part A 56:72–83, 2016) noetic account, science progresses when understanding increases, that is, when scientists grasp how to correctly explain or predict more aspects of the world that they could (...) before. In this paper, we argue that the noetic account is severely undermotivated. Dellsén provides three examples intended to show that understanding can increase absent the justification required for true belief to constitute knowledge. However, we demonstrate that a lack of clarity in each case allows for two contrasting interpretations, neither of which serves its intended purpose. On the first, the agent involved lacks both knowledge and understanding; and, on the second, the agent involved successfully gains both knowledge and understanding. While neither interpretation supports Dellsén’s claim that understanding can be prised apart from knowledge, we argue that, in general, agents in such cases ought to be attributed neither knowledge nor understanding. Given that the separability of knowledge and understanding is a necessary component of the noetic account, we conclude that there is little support for the idea that science progresses through increasing understanding rather than the accumulation of knowledge. (shrink)

The Modern Synthesis has received criticism for its purported gene-centrism. That criticism relies on a concept of the gene as a unit of instructional information. In this paper I discuss information concepts and endorse one, developed from Floridi, that sees information as a functional relationship between data and context. I use this concept to inspect developmental criticisms of the Modern Synthesis and argue that the instructional gene arose as an idealization practice when evolutionary biologists made comment on development. However, a (...) closer inspection of key claims shows that at least some associated with the Modern Synthesis were in fact adopting the data led definition I favour and made clear arguments for the role of developmental processes beyond genetic input. There was no instructional gene. (shrink)

Here is a puzzling phenomenon. Moral theories are typically thought to be necessary. If act utilitarianism is true, for example, then it is necessarily true. However, scientific theories are typically thought to be contingent. If quantum field theory is true, it’s not necessarily true — the world could have been Newtonian. My aim is to explore this discrepancy between domains. -/- In particular, I explore the role of what I call `internality’ intuitions in motivating necessitism about both moral and scientific (...) domains. In effect, these internality intuitions tell against the explanatory role of certain `external’ entities – like moral or scientific principles. I’ll suggest that perhaps these internality intuitions are more compelling in the moral case than the scientific case. If so, this would rationalize the combination of moral necessitism and scientific contingentism. (shrink)

In earlier work, I criticized de Regt’s contextual theory of understanding, and advertised the advantages of my own, knowledge-based account. Using the early history of the standard model in particle physics as an illustration, I instead consider the benefits of unifying these two accounts of understanding. I argue that de Regt’s account substantially improves my own account of explanatory consideration, and that my account of explanatory comparison substantially improves upon his account of explanatory evaluation. De Regt and my apparent disagreement (...) about the so-called factivity of understanding is defused through a voluntaristic approach to the scientific realism debates. (shrink)

The thesis proposes an account of the means of scientific representation focused on similarity, or more specifically, on the notion of “creative similarity”. I first distinguish between two different questions regarding the problem of representation: the question about the constituents and the question about the means of representation (following Suárez 2003; van Fraassen 2008). I argue that, although similarity is not a good candidate for constituent of representation, it can satisfactorily answer the question about the means of representation if adequately (...) characterized. To motivate this position, I dispute the main arguments offered against similarity as means of representation, namely the arguments from variety, vagueness, and misrepresentation, and contend that similarity plays a central epistemic role in practices of representing in science. The study of the role of similarity in scientific practice, I argue, requires the analysis of the uses of judgments of similarity in the construction of scientific models. I examine the cases of the Mississippi Basin Model and the San Francisco Bay Model to illustrate how judgments of similarity are directly involved in the production of epistemically fruitful models. Informed by the practical investigation developed throughout the thesis, I finally outline what I call the creative similarity account of the means of representation. The notion of “creative similarity” helps to capture the way in which similarity, as means of representation, intervenes in actual practices of representing, namely, in the form of a productive interplay of judgments of similarity and distortions (i.e. idealizations, abstractions, simplifications), which is employed by resourceful agents with the aim of understanding aspects of the natural world. (shrink)

We provide two programmatic frameworks for integrating philosophical research on understanding with complementary work in computer science, psychology, and neuroscience. First, philosophical theories of understanding have consequences about how agents should reason if they are to understand that can then be evaluated empirically by their concordance with findings in scientific studies of reasoning. Second, these studies use a multitude of explanations, and a philosophical theory of understanding is well suited to integrating these explanations in illuminating ways.

The main objective of this dissertation is to philosophically assess how the use of informational concepts in the field of classical thermostatistical physics has historically evolved from the late 1940s to the present day. I will first analyze in depth the main notions that form the conceptual basis on which 'informational physics' historically unfolded, encompassing (i) different entropy, probability and information notions, (ii) their multiple interpretative variations, and (iii) the formal, numerical and semantic-interpretative relationships among them. In the following, I (...) will assess the history of informational thermophysics during the second half of the twentieth century. Firstly, I analyse the intellectual factors that gave rise to this current in the late forties (i.e., popularization of Shannon's theory, interest in a naturalized epistemology of science, etc.), then study its consolidation in the Brillouinian and Jaynesian programs, and finally claim how Carnap (1977) and his disciples tried to criticize this tendency within the scientific community. Then, I evaluate how informational physics became a predominant intellectual current in the scientific community in the nineties, made possible by the convergence of Jaynesianism and Brillouinism in proposals such as that of Tribus and McIrvine (1971) or Bekenstein (1973) and the application of algorithmic information theory into the thermophysical domain. As a sign of its radicality at this historical stage, I explore the main proposals to include information as part of our physical reality, such as Wheeler’s (1990), Stonier’s (1990) or Landauer’s (1991), detailing the main philosophical arguments (e.g., Timpson, 2013; Lombardi et al. 2016a) against those inflationary attitudes towards information. Following this historical assessment, I systematically analyze whether the descriptive exploitation of informational concepts has historically contributed to providing us with knowledge of thermophysical reality via (i) explaining thermal processes such as equilibrium approximation, (ii) advantageously predicting thermal phenomena, or (iii) enabling understanding of thermal property such as thermodynamic entropy. I argue that these epistemic shortcomings would make it impossible to draw ontological conclusions in a justified way about the physical nature of information. In conclusion, I will argue that the historical exploitation of informational concepts has not contributed significantly to the epistemic progress of thermophysics. This would lead to characterize informational proposals as 'degenerate science' (à la Lakatos 1978a) regarding classical thermostatistical physics or as theoretically underdeveloped regarding the study of the cognitive dynamics of scientists in this physical domain. (shrink)

In this talk I present the main results from Anta (2021), namely, that the theoretical division between Boltzmannian and Gibbsian statistical mechanics should be understood as a separation in the epistemic capabilities of this physical discipline. In particular, while from the Boltzmannian framework one can generate powerful explanations of thermal processes by appealing to their microdynamics, from the Gibbsian framework one can predict observable values in a computationally effective way. Finally, I argue that this statistical mechanical schism contradicts the Hempelian (...) (1958) thesis that the predictive power of a scientific theory is directly proportional to its explanatory potential, and vice versa. (shrink)

This dissertation starts with a concise overview of what philosophers of science have written about unification and its role in scientific explanation during the last 50 years to provide the reader with some background knowledge. In order to bring unification back into the picture, I have followed two strategies, resulting respectively in Parts I and II of this dissertation. In Part I the idea of unification is used to refine and enrich the dominant causalmechanist and causal-interventionist accounts of scientific explanation. (...) In this part of the dissertation I bracket the classical ideas about unification: deduction and derivation. I do grant, for the sake of argument, that explanations are causal and argue that unification is important from within this causalist perspective. In Part II I continue my strategy of digging into scientific practice to find cases of ontological unification. But here I distance myself from the dominant literature that all explanations must be causal. I will investigate whether explanatory unification is possible in non-causal explanations. Part III contains some further reflections and conclusions. I will formulate my primary results, and I will elaborate on their implications for thinking about unification and explanation. The different forms of ontological unification were quite diverse. This relates to the method I have used. Throughout this dissertation the types of unification that were discussed emerged from digging into scientific practice. This philosophy-of-science-practice approach steered me towards a pluralistic view on unification and on explanation. In this dissertation I do not try to develop a new model of explanation and compare it to existing models. The aim is to show that there are important types of explanatory practice which cannot be properly analyzed if we neglect unification as a desideratum for explanations. (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)