According to a widespread view, Thomas Kuhn’s model of scientific development would relegate rationality to a second plane, openly flirting with irrationalist positions. The intent of this article is to clarify this aspect of his thinking and refute this common interpretation. I begin by analysing the nature of values in Kuhn’s model and how they are connected to rationality. For Kuhn, a theory is chosen rationally when: i) the evaluation is based on values characteristic of science; ii) a theory is (...) considered better the more it manifests these values; and iii) the scientist chooses the best-evaluated theory. The second part of this article deals with the thesis of the variability of values. According to Kuhn, the examples through which epistemic values are presented vary for each person, and for this reason individuals interpret these criteria differently. Consequently, two scientists, using the same values, can come to a rational disagreement over which theory to choose. Finally, I point out the limitations of this notion of rationality for the explanation of consensus formation, and the corresponding demand for a sociological theory that reconnects individual rationality with convergence of opinions. (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)

Thomas Kuhn is mostly known for his contributions to the philosophy of science. However, it was chiefly to investigations in philosophy of language that he dedicated the last part of his career. The aim of this paper is to present a systematic view of Kuhn’s main ideas on this subject. I start by describing his theory of concept, in particular what he says about kind terms. Such terms, acquired in blocks that form contrast sets or “taxonomies,” are learned through ostensible (...) definitions. Next, I analyze Kuhn’s concepts of “meaning” and “truth.” Finally, I discuss a second type of kinds terms, not fully investigated by him, which is learned together with scientific laws. Resumo Thomas Kuhn é conhecido principalmente por suas contribuições para a filosofia da ciência. No entanto, foi em grande medida à filosofia da linguagem que ele dedicou a última parte de sua carreira. O objetivo deste artigo é apresentar uma visão sistemática das principais ideias de Kuhn sobre esse assunto. Começo descrevendo sua teoria do conceito, em particular o que ele diz sobre termos para espécie. Tais termos, aprendidos em blocos que formam conjuntos de contraste ou “taxonomias”, são aprendidos através de definições ostensivas. Em seguida, analiso as concepções de Kuhn de “significado” e “verdade”. Finalmente, discuto um segundo tipo de termo para espécies, não totalmente investigado por ele, que é aprendido junto com as leis científicas. (shrink)

The Structure of Scientific Revolutions represented a milestone in the attempt to understand scientific development based on empirical observations. However, in the next decades after the publication of his book, history, psychology, and sociology became increasingly marginal in Kuhn’s discussions. In his last articles, Kuhn even suggested that philosophers should pay less attention to empirical data and focus more on “first principles.” The purpose of this article is, first, to describe this radical transformation in Kuhn’s methodological approach, from his initial (...) naturalism to his later and more strict philosophical stance. Next, I present some of the alleged justifications for explaining this transition, such as his interest in problems more firmly attached to the philosophical tradition and a desire for greater acceptance within this community. Although these factors certainly played a role in explaining Kuhn’s change, I also believe that an important theoretical component exerts a fundamental function in this transition: the idea that scientific development is a kind of evolutionary process, which is better understood through the use of abstract theoretical models, instead of sparse observations of scientific activity. (shrink)

In The Structure of Scientific Revolutions, Thomas Kuhn resorts to concepts from several disciplines in order to describe the general patterns of scientific development. This blend of disciplines can be explained in part by Kuhn's intellectual path, from physics to history and then to philosophy of science; but it also points to a deeper methodological problem, which is the question of what is the real unity of analysis in his model of science. The primary intention of this article is, thus, (...) to give a solution to this difficulty. The answer, I believe, rests on identifying three fundamental units present in Kuhn's theory of scientific development. They are, respectively, the individual, responsible for producing evidence, spreading information, and choosing theories; the community, a set of scientists investigating a series of phenomena; and the groups, individuals with similar behavior but with looser institutional or social ties — a usually neglected category in Kuhnian literature, but equally fundamental for the final outcome of scientific debates. After investigating these categories in detail, I propose a way of integrating them into a general model for explaining the resolution of scientific controversies. Finally, I try to resolve the apparent conflict among disciplinary vocabularies by offering an account of the function of sociological, psychological, and epistemological concepts for describing controversies, and some of the methodologies appropriate for each of these tasks. (shrink)

Changes of theories are major events in science. Two main types of questions may be asked about them: i) how do scientists choose new theories?, and ii) how is consensus formed? Generally, philosophers do not distinguish these two questions. Kuhn, on the contrary, offers very different answers to each of these questions. Theory-choice, on the one hand, is explained through the application of epistemic criteria, such as accuracy and consistency; nonetheless, because these values do not prescribe a single choice, consensus (...) formation, on the other hand, is explained through a series of socio-epistemic mechanisms, namely: scientific pedagogy, diffusion and production of knowledge within the community, and restructuring of the scientific field. These mechanisms are the basis of Kuhn’s social epistemology, in that they are not restricted to sociology nor epistemology, encompassing both social interactions and epistemic evaluations of theories. Keywords: Thomas Kuhn, consensus formation, social epistemology. (shrink)