Online research in philosophy

This paper explores the question of whether connectionist models of cognition should be considered to be scientific theories of the cognitive domain. It is argued that in traditional scientific theories, there is a fairly close connection between the theoretical (unobservable) entities postulated and the empirical observations accounted for. In connectionist models, however, hundreds of theoretical terms are postulated -- viz., nodes and connections -- that are far removed from the observable phenomena. As a result, many of the features of any given connectionist model are relatively optional. This leads to the question of what, exactly, is learned about a cognitive domain modelled by a connectionist network.

Larry Laudan has challenged the realist to come up with a program that submits realism to "those stringent empirical demands which the realist himself minimally insists on when appraising scientific theories." This paper shows how the realist can go about taking up Laudan on this challenge; and, in such a way that the realist hypothesis actually ends up being confirmed, by any empirical standards. In other words, it is shown that we can test for convergent realism, just as readily as Laudan can test for a connection between theories that are controlled by the cannons of science and their subsequent reliability.

This paper explores a new reason for preferring a model-theoretic approach to understanding the nature of scientific theories. Identifying the models in philosophers' model-theoretic accounts of theories with the concepts in cognitive scientists' accounts of categorization suggests a structure to families of models far richer than has commonly been assumed. Using classical mechanics as an example, it is argued that families of models may be "mapped" as an array with "horizontal" graded structures, multiply hierarchical "vertical" structures, and local "radial" structures. These structures promise important implications for how scientific theories are learned and used in actual scientific practice.

In this paper I distinguish various ways in which empirical claims about evolutionary and ecological models can be supported by data. I describe three basic factors bearing on confirmation of empirical claims: fit of the model to data; independent testing of various aspects of the model, and variety of evident. A brief description of the kinds of confirmation is followed by examples of each kind, drawn from a range of evolutionary and ecological theories. I conclude that the greater complexity and precision of my approach, as compared to, for instance, a Popperian approach, can facilitate detailed analysis and comparison of empirical claims.

Theoretical models are an important tool for many aspects of scientific activity. They are used, i.a., to structure data, to apply theories or even to construct new theories. But what exactly is a model? It turns out that there is no proper definition of the term "model" that covers all these aspects. Thus, I restrict myself here to evaluate the function of models in the research process while using "model" in the loose way physicists do. To this end, I distinguish four kinds of models. These are (1) models as special theories, (2) models as a substitute for a theory, (3) toy models and (4) developmental models. I argue that models of the types (3) and (4) are considerably useful in the process of theory construction. This will be demonstrated in an extended case-study from High-Energy Physics.

The goal of this article is to show that the structuralist approachprovides a powerful framework for the analysis of certain holistic phenomena in empirical theories.We focus on two aspects of holism. The first refers to the involvement of comprehensive complexes of hypothesesin the theoretical treatment of systems regarded in isolation. By contrast, the second refers to thecorrelation between the theoretical descriptions of different systems. It is demonstrated how these two aspectscan be analysed by making use of the structuralist notion of theory-nets, and how they are reflected by a refinedversion of the Ramsey sentence. Furthermore, it is argued that there exists a tight correlation between theoccurrence of these two holistic phenomena, a specific form of underdetermination of terms which occur in thefundamental principles of an empirical theory, and the shaping of the theory's protective belt. After having dealtwith these questions in abstracto, the relevance of these considerations for a better understanding of the dynamicsof empirical theories is demonstrated in a concrete case study. It refers to the role holistic phenomenaplayed in the investigation of the anomalous advance of Mercury's perihelion and in the various attempts to eliminate this anomaly.

In a scientific context, ontological commitments should be considered as supervenient over accepted scientific theories. This implies that the primarily ontological notions of reduction and emergence of entities of different kinds should be reformulated in terms of relations between existing empirical theories. For this, in turn, it is most convenient to employ a model-theoretic view of scientific theories: the identity criterion of a scientific theory is essentially given by a class of models. Accordingly, reduction and emergence are to be seen as particular kinds of relations between (some) models of different theories that subsume the same (or a similar) “experiential field”. The set-theoretical notion of an echelon-set proves to be crucial for this purpose: The domains in the models of the reduced theory are echelon-sets over the domains of the reducing theory. Finally, it is argued that emergence may plausibly be interpreted as akin to but weaker than reduction.

In spite of the ‘experimental turn’ now fashionable in the philosophy of science, the question of the structure and identity criteria of scientific theories continues to be a central issue for the philosophical analysis of empirical science. We need a precise metatheory of empirical theories to deal with this issue. Metatheoretical structuralism appears to offer the most adequate approach in this sense so far. First, some basic intuitions about what empirical theories are, and how they are structured, are laid out. Then, the main notions used by metatheoretical structuralism to analyze theories are explained, and they are illustrated by applying them to an example of a simple physical theory. Finally, it is argued that the picture of the structure and identity of empirical theories coming out of structuralistic analysis adequately corresponds to the basic intuitions stated at the beginning.

The purpose of this paper is to provide an analysis of the concept of model as it is applied in the physical sciences, and to show that this analysis is fruitful insofar as it can be used as an acceptable account of the role of models in physical explanation.A realist interpretation of theories is adopted as a point of departure. A distinction between theories and models is drawn on the basis of this interpretation. The relation between model and prototype is expressed in terms of the concepts of access and accessibility, and four conditions are proposed as an analysis of the concept of model. It is concluded that models are introduced when approximate methods are used.

The prevalence of optimality models in the literature of evolutionary biology is testimony to their popularity and importance. Evolutionary biologist R. C. Lewontin, whose criticisms of optimality models are considered here, reflects that "optimality arguments have become extremely popular in the last fifteen years, and at present represent the dominant mode of thought." Although optimality models have received little attention in the philosophical literature, these models are very interesting from a philosophical point of view. As will be argued, optimality models are central to evolutionary thought, yet they are not readily accomodated by the traditional view of scientific theories. According to the traditional view, we would expect optimality models to employ general, empirical laws of nature, but they do not. Fortunately, the semantic view of scientific theories, a recent alternative to the traditional view, more readily accomodates optimality models. As we would expect on the semantic view, optimality models can be construed as specifications of ideal systems. These specifications may be used to describe empirical systems--that is, the specifications may have empirical instances. But the specifications are not empirical claims, much less general, empirical laws. Although philosophers have yet to discuss the general features and uses of optimality models, these topics have stimulated much recent discussion among evolutionary biologists. Their discussions raise a number of precautions concerning the proper use of optimality models. Moreover, many of their caveats can be interpreted as general reminders that 1) optimality models specify ideal systems whose empirical instantiations may be quite restricted, and that 2) optimality models should not be construed as general, empirical laws. As G. F. Oster and E. O. Wilson caution, "the prudent course is to regard optimality models as provisional guides to further empirical research and not necessarily as the key to deeper laws of nature." It seems, then, that the semantic view of theories is more sensitive to the nature and limitations of optimality models than is the more traditional view of theories.