Dretske, Shannon’s Theory and the Interpretation of Information

Abstract

In his well known book, Knowledge and the Flow of Information, Fred Dretske (1981) attempts to build a bridge between philosophy and cognitive sciences by introducing the concept of information in the theory of knowledge. He distinguishes between sensory processes (seeing) and cognitive processes (recognizing) in terms of the different ways in which the received information is coded. In the final part of his book, he analyzes the capacity of physical systems to hold beliefs and to develop concepts on an informational basis. For these purposes, Dretske begins by examining the notion of information as characterized in Shannon’s theory. But although this is his starting point, Shannon’s theory has, he argues, two main limitations: first, it is unable to handle the information contained in individual messages and, second, it is a quantitative theory dealing only with amounts of information and ignoring its content. For these reasons, Dretske proposes some formal modifications of the standard theory to make room for individual amounts of information. On the basis of the proposed changes, he elaborates a semantic theory which attempts to capture what he considers to be the nuclear sense of the term ’information’, that is, information as something capable of yielding knowledge. In this article it is shown, first, that Dretske’s modifications suffer from some formal defects. It is then indicated precisely how these defects can be remedied in order to preserve Dretske’s general proposal. In fact, it is shown that, if the changes are introduced in a formally correct way, Shannon’s theory can express much more than what Dretske himself assumes. Finally, it is argued that the semantic character of Dretske’s theory relies neither on the definition of informational content nor on the intentionality of the natural laws underlying the transmission of information. What confers a semantic dimension to Dretske’s theory is a particular interpretation of the very nature of information, which differs widely from the interpretation usually adopted in physical sciences...