Machine learning operates at the intersection of statistics and computer science. This raises the question as to its underlying methodology. While much emphasis has been put on the close link between the process of learning from data and induction, the falsificationist component of machine learning has received minor attention. In this paper, we argue that the idea of falsification is central to the methodology of machine learning. It is commonly thought that machine learning algorithms infer general prediction rules from past (...) observations. This is akin to a statistical procedure by which estimates are obtained from a sample of data. But machine learning algorithms can also be described as choosing one prediction rule from an entire class of functions. In particular, the algorithm that determines the weights of an artificial neural network operates by empirical risk minimization and rejects prediction rules that lack empirical adequacy. It also exhibits a behavior of implicit regularization that pushes hypothesis choice toward simpler prediction rules. We argue that taking both aspects together gives rise to a falsificationist account of artificial neural networks. (shrink)

The replication (replicability, reproducibility) crisis in social psychology and clinical medicine arises from the fact that many apparently well-confirmed experimental results are subsequently overturned by studies that aim to replicate the original study. The culprit is widely held to be poor science: questionable research practices, failure to publish negative results, bad incentives, and even fraud. In this article I argue that the high rate of failed replications is consistent with high-quality science. We would expect this outcome if the field of (...) science in question produces a high proportion of false hypotheses prior to testing. If most of the hypotheses under test are false, then there will be many false hypotheses that are apparently supported by the outcomes of well conducted experiments and null hypothesis significance tests with a type-I error rate (α) of 5%. Failure to recognize this is to commit the fallacy of ignoring the base rate. I argue that this is a plausible diagnosis of the replication crisis and examine what lessons we thereby learn for the future conduct of science. (shrink)

Truthlikeness is a property of a theory or a proposition that represents its closeness to the truth. According to Niiniluoto, truthlikeness for quantitative deterministic laws can be defined by the Minkowski metric. I present some counterexamples to the definition and argue that it fails because it considers truthlikeness for quantitative deterministic laws to be just a function of accuracy, but an accurate law can be wrong about the actual ‘structure’ or ‘behaviour’ of the system it intends to describe. I develop (...) a modification of Niiniluoto’s proposal that defines truthlikeness for quantitative deterministic laws according to two parameters: accuracy and nomicity. The presented proposal solves the counterexamples and defines a new way of understanding scientific progress. (shrink)

The term “deep history” refers to historical accounts framed temporally not by the advent of a written record but by evolutionary events (Smail 2008; Shryock and Smail 2011). The presumption of deep history is that the events of today have a history that traces back beyond written history to events in the evolutionary past. For human kinship, though, even forming a history of kinship, let alone a deep history, remains problematic, given limited, relevant data (Trautman et al. 2011). With regard (...) to a deep history, one conjecture is that human kinship evolved from primate social systems in a gradual, more-or-less continuous manner (see Chapais 2008); another conjecture is that kinship, in accordance with the incest account of Claude Lévi-Strauss (1969) or the fanciful, tetradic account of Nicholas J. Allen (1986), “comes into existence with a leap” (Trautman et al. 2011: 176); and yet another, the account to be developed in this paper, is that kinship, as it is understood and lived by culture bearers today, is the consequence of a profound and qualitative evolutionary transformation going from an ancestral primate-like social system predicated on extensive face-to-face interaction to the relation-based social systems that characterize human societies (Read 2012). (shrink)

To identify and conceptualize research problems in science, philosophers and often scientists rely on classical accounts of problems that focus on intellectual problems defined in relation to theories. Recently, philosophers have begun to study the structures and functions of research problems not defined in relation to theories. Furthermore, scientists have long pursued research problems often labeled as practical or applied. As yet, no account of problems specifies the description of both so-called intellectual problems and so-called applied problems. This article proposes (...) a new conceptual framework of problems that accounts both for intellectual aspects of problems and for their practical or applied aspects. I illustrate the account with an example of a recent research project from evolutionary biology, and I indicate further routes by that to develop the account, especially in connection to empirical studies of science. (shrink)

Bayesianism and likelihoodism are two of the most important frameworks philosophers of science use to analyse scientific methodology. However, both frameworks face a serious objection: much scientific inquiry takes place in highly idealized frameworks where all the hypotheses are known to be false. Yet, both Bayesianism and likelihoodism seem to be based on the assumption that the goal of scientific inquiry is always truth rather than closeness to the truth. Here, I argue in favour of a verisimilitude framework for inductive (...) inference. In the verisimilitude framework, scientific inquiry is conceived of, in part, as a process where inference methods ought to be calibrated to appropriate measures of closeness to the truth. To illustrate the verisimilitude framework, I offer a reconstruction of parsimony evaluations of scientific theories, and I give a reconstruction and extended analysis of the use of parsimony inference in phylogenetics. By recasting phylogenetic inference in the verisimilitude framework, it becomes possible to both raise and address objections to phylogenetic methods that rely on parsimony. 1Introduction2Problems with the Law of Likelihood3Introducing Verisimilitude-Based Inference4Examples of Verisimilitude-Based Inference Procedures4.1Parsimony inference over theories4.2Parsimony inference in phylogenetics5Conclusion. (shrink)

I present a recent historical case from cosmology—the story of inflationary cosmology—and on its basis argue that solving explanatory problems is a reliable method for making progress in science. In particular, I claim that the success of inflationary theory at solving its predecessor’s explanatory problems justified the theory epistemically, even in advance of the development of novel predictions from the theory and the later confirmation of those predictions.

Karl Popper argued in 1974 that evolutionary theory contains no testable laws and is therefore a metaphysical research program. Four years later, he said that he had changed his mind. Here we seek to understand Popper’s initial position and his subsequent retraction. We argue, contrary to Popper’s own assessment, that he did not change his mind at all about the substance of his original claim. We also explore how Popper’s views have ramifications for contemporary discussion of the nature of laws (...) and the structure of evolutionary theory. (shrink)

Reflexivity is a well-established theoretical and methodological concept in the human sciences, and yet it is used in a confusing variety of ways. The meaning of `reflexivity' and the virtues ascribed to the concept are relative to particular theoretical and methodological commitments. This article examines several versions of the concept, and critically focuses on treatments of reflexivity as a mark of distinction or source of methodological advantage. Although reflexivity often is associated with radical epistemologies, social scientists with more conventional leanings (...) often speak of reflexivity as a methodological tool, substantive property of social systems, or source of individual enlightenment. Radical and conventional social scientists alike tend to stress the importance of being reflexive, as opposed to being unreflexive, but they do not share a coherent conception of what `being reflexive' means or entails. As an alternative to reflexive self-privileging, I recommend an ethnomethodological conception of reflexivity as an ordinary, unremarkable and unavoidable feature of action. The ethnomethodological conception does not support a particular theoretical or methodological standpoint by contrasting it to an `unreflexive' counterpart. It has little value as a critical weapon or source of epistemological advantage, which, in the present context, can have advantages of its own for promoting peace and epistemic democracy. (shrink)

The modern history of verisimilitude can be divided into three periods. The first began in 1960, when Karl Popper proposed his qualitative definition of what it is for one theory to be more truthlike than another theory, and lasted until 1974, when David Miller and Pavel Trichý published their refutation of Popper's definition. The second period started immediately with the attempt to explicate truthlikeness by means of relations of similarity or resemblance between states of affairs (or their linguistic representations); the (...) work within this similarity approach was summarized in the books of Graham Oddie [1986] and Ilkka Niiniluoto [1987]. During the subsequent third period, studies in verisimilitude have been actively continued, and interesting results and applications have been achieved, but not many dramatic novelties. While it is now obsolete to claim that truthlikeness with reasonable properties cannot be defined at all, there is still a lot of controversy about the best and least arbitrary approach to doing this. (shrink)

The paper investigates the implications of a nonaprioristic philosophy of science. It starts by developing a scheme of justification which draws its norms from the prevailing paradigm of rationality, which need not be universal or external. If the requirement for normativity is then abandoned we do not end up with a descriptive philosophy of science. The alternative to a prescriptive philosophy of science is a theoretical explanation of scientific decisions and acts. Explanation, rather than mere description, replaces justification; and the (...) paradigm of rationality becomes a scientific paradigm. The implications of these results for the discovery-justification distinction are investigated. An explanatory philosophy of science deals with the generation, as well as with the selection of scientific conjectures: both contexts have an epistemic dimension. * This paper was written under the auspices of the Wolfson Chair Extrordinary of Theoretical Physics, Tel Aviv University. (shrink)

This article revisits Karl Popper’s falsificationist methodology with respect to three tasks. The first is to illuminate and systematize Popper’s methodological views in light of his core epistemological commitments. A second and related objective is to gauge which aspects of falsificationism should be identified as “conventionalist”—a label that Popper himself uses (albeit with qualifications) but that is compromised by and, thus, stands in need of elucidation because of Popper’s idiosyncratic understanding of conventionalism. Third, by elaborating Popper’s virtue-epistemological, dialogical model of (...) rationality, I show how Popper’s conventionalism, fallibilism, and critical rationalism form a coherent system. This system allows Popper to accord science the status of a privileged source of knowledge—without naïve appeal to authority. My systematization of Popper’s views yields an intrinsically and exegetically more satisfactory reading of Popper’s falsificationism than usual presentations. Thanks to its marked flexibility and methodological liberalism, it bypasses many standard objections but still offers sound and relevant methodological advice. With its “virtue-methodological” thrust, Popper sketches an original and promising approach to methodology, the fertility of which deserves further exploration for contemporary debates. (shrink)

According to influential accounts of scientific method, such as critical rationalism, scientific knowledge grows by repeatedly testing our best hypotheses. But despite the popularity of hypothesis tests in statistical inference and science in general, their philosophical foundations remain shaky. In particular, the interpretation of non-significant results—those that do not reject the tested hypothesis—poses a major philosophical challenge. To what extent do they corroborate the tested hypothesis, or provide a reason to accept it? Popper sought for measures of corroboration that could (...) adequately answer this question. According to Popper, corroboration is different from probability-raising, and grounded in the predictive success and testability of a hypothesis. As such, corroboration becomes an indicator of the scientific value of a hypothesis and guides our practical preferences over hypotheses that have been subjected to severe tests. This article proves two impossibility results for corroboration measures based on statistical relevance. The generality of these results shows that Popper’s qualitative characterization of corroboration must be misguided. I explore what a more promising and scientifically useful concept of corroboration could look like. (shrink)

The modern history of verisimilitude can be divided into three periods. The first began in 1960, when Karl Popper proposed his qualitative definition of what it is for one theory to be more truthlike than another theory, and lasted until 1974, when David Miller and Pavel Trich published their refutation of Popper's definition. The second period started immediately with the attempt to explicate truthlikeness by means of relations of similarity or resemblance between states of affairs (or their linguistic representations); the (...) work within this similarity approach was summarized in the books of Graham Oddie [1986] and Ilkka Niiniluoto [1987]. During the subsequent third period, studies in verisimilitude have been actively continued, and interesting results and applications have been achieved, but not many dramatic novelties. While it is now obsolete to claim that truthlikeness with reasonable properties cannot be defined at all, there is still a lot of controversy about the best and least arbitrary approach to doing this. (shrink)

The paper offers a new understanding of induction in the empirical sciences, one which assimilates it to induction in geometry rather than to statistical inference. To make the point a system of notions, essential to logically sound induction, is defined. Notable among them are arbitrary object and particular property. A second aim of the paper is to bring to light a largely neglected set of assumptions shared by both induction and deduction in the empirical sciences. This is made possible by (...) appealing to the logic of common nouns and applying it to the logic of natural-kind terms.This strategy yields a new insight into the concept of natural kinds. While the strategy reveals deep affinity between empirical induction and deduction it also reveals two problems peculiar to induction. This helps to explain the intuition that induction is the more problematic of the two. The paper does not set out ‘to solve the problem of induction’. 1The paper has benefited from the support and critical comments of several friends: Steven Davis, Kevin Dunbar, Anil Gupta, Michael Hallett, Ray Jackendoff and Storrs McCall. Two friends, however, deserve special thanks, Michael Makkai and Gonzalo Reyes. Many of the points made in the paper, otherwise unacknowledged, are due to their generosity and patience. In fact, the paper can be seen as an application of the logic of kinds on which Gonzalo Reyes has been working for several years. (shrink)

In a recent paper, Dylan Evans proposed that emotions could help solve what has been known as ‘the frame problem’. In the process, he first questioned the utility of using the frame problem as a framework. After tackling this issue, he provided an alternative terminology to the frame problem—termed ‘the search hypothesis of emotion’—in order to re-examine how emotions aid rational agents. His new terminology, however, opens itself to other critiques. While accepting the basic tenets of his analysis, I question (...) (i) whether a single search theory of emotion is adequate, and (ii) whether his theory would have been better termed ‘the search hypothesis of feeling’. Finally, I extend some of the ideas developed in Evans' paper. 1. Introduction 2. Emotion, reason and ends 3. The search hypothesis of emotion revisited 4. Conclusion. (shrink)

Is anthropology a science? To put the question today amounts to a reply in the negative. The representatives of the ‘true’ sciences are not alone in suggesting a conjunctural or crippling lacuna which would preclude membership by right of the prestigious world, which, however, the name ‘humanistic sciences’ seems to demand. We should remember that some years ago Claude Lévi-Strauss caused a shudder to run through his discipline by describing it as a ‘flattering imposture’. Since then denigration has spread constantly, (...) and it no longer consists of deploring a slowness, an equivocation or a handicap which we would have the hope of remedying: it is definitely a fatality which we are from now on invited to confirm. (shrink)