Observation, Theory and Domains

Abstract

In this chapter we consider how abandoning the full Theory-Dependence of Observation Thesis (TDOT) affects our view of classification. We define a scientific Theory (capital-T) as something distinct from the notion that phenomena are observed based on prior criteria of salience to an observer, and adopt the Bogen—Woodward notion of a phenomenon as a pattern in data. A phenomenon, including a classification, is the explicandum that Theory explains. We then consider the question whether Theory from outside a domain of investigation counts as theory-dependence within the domain, and thus ask what a domain is in science. We set up a “domain conundrum” — how can a science get started when there is no Theory of its domain?

As all sciences are based upon facts, known, or to be known from experience, so are they, in their early state of developement [sic], matters of pure observation. It is only when we have acquired the power of generalising these facts, when such generalisations agree among themselves and with every thing we see or know of nature, that the theory of a science becomes either absolutely demonstrative, or approaches so near to certainty, by the force of analogical reasoning, that it is not contradicted by anything known. The case of natural history, then, is precisely this; in its early stages it is a science of observation; in its latter, it is one of demonstration. [William Swainson, 1834 ¹]