CONICYT – Universidad de Valparaíso
The First Workshop on Models and Idealizations in Science will be devoted to philosophical problems arising from the use of models in the natural and social sciences. Topics include modelling and simulations, abstractions, idealizations and approximations in science, among others.
(University of South Carolina/University of Helsinki)
Abstraction, Analogical Reasoning, and Big Data in Modeling Biological Systems
The prevalent view on abstraction among philosophers of science is that of omission. Whereas idealizations are thought to introduce distortions to a scientific representation, abstraction is understood in terms of abstracting away from the details of a system (Thomson-Jones 2005, Godfrey-Smith 2009). Within mechanistic philosophy of biology Levy and Bechtel (2013) have recently argued that a model is a highly selective depiction of the underlying mechanism of a phenomenon, taking into account only those features that make a difference.
I will argue that the idea of abstraction as omission does not often capture what goes on in actual model construction. From the perspective of modeling heuristic, a distinction should be made between cases in which one abstracts away from most of the details of a system of interest from those that start from an abstract mathematical template describing a general pattern of interaction. Such general model templates are often adapted from other scientific disciplines—such as physics and engineering—by way of analogical reasoning. I will illustrate this point by examining cases from systems and synthetic biology. Moreover, I will discuss how big data and associated computational methods have taken systems biology from abstract mathematical modeling towards “detection” of patterns.
(Universidad Complutense de Madrid)
Scientific Representation, and Information Transfer
[Joint work with Agnes Bolinska,
University of Pittsburgh]
We propose a way to quantify the amount of information transmitted by models and other representational vehicles that perform a similar function (in the larger class of epistemic or cognitive representations) using the concepts of entropy, noise and equivocation, first introduced in Shannon’s (1948) Mathematical Theory of Communication and applied in an epistemological context by Dretske (1981). In information theory, entropy is used to measure the average amount of information contained in a signal. Because signals often travel through noisy communication channels, some information originating at the source may not reach the receiver; the amount of information that fails to be transmitted is known as the equivocation, while the amount of the information arriving at the receiver that did not originate in the source is called noise. We show how these concepts can be applied to scientific models (and other epistemic or cognitive representations). We then apply the inferential conception of scientific representation to show that information carrying is functionally an legitimate means of representation. In other words, on a suitably deflationary inferential account, models understood as information carriers are fully fledged scientific representations.