Abstract

There are infinitely many possible hypotheses that are internally consistent and are compatible with a finite set of data. One of the criteria most often cited by the scientists when trying to decide between these "empirically equivalent" theories is simplicity. The principle of simplicity recommends that, in the absence of other overriding factors, the simplest theory compatible with the evidence should, at least on a prima facie basis, be preferred. This thesis is concerned with the problem of explicating the notion of simplicity and the possibility of justifying its role in scientific theory choice. ;I begin by suggesting that in order to avoid the confusions about simplicity, it is essential that we distinguish between different, and often incompatible, kinds of simplicity. I then proceed to argue that empirical simplicity, or economy of content, is the most relevant kind of simplicity in the context of theory choice. ;Following Elliott Sober, I propose that the notion of scientific problem can be characterized as the sets of questions that interest scientists. I then proceed to propose that our intuitions about simplicity ranking can largely be captured by what I call the "relevance condition" and the "expressibility condition". In brief, the former states that the theories being considered for simplicity ranking must at least be probabilistically relevant to the set of questions which interest scientists. The latter states that if all the basic propositions of one hypothesis are expressible by another using the logical apparatus and the primitive concepts and processes already employed in it, and not vice versa, then the former hypothesis is simpler than the latter. These two formal conditions are shown to be applicable to many important philosophical and scientific problems. ;And finally, I conclude by arguing that it follows from the proposed notion of simplicity that it is rational to act as if nature is simple. The principle of simplicity is, therefore, justified