INTRODUCTION

Paul Churchland cemented his appointment as Ambassador of Connectionism to Philosophy with the 1986 publication of his paper “Some reductive strategies in cognitive neurobiology.” However, as Churchland tells the story in the preface to his collection of papers, A Neurocomputational Perspective, his relationship with connectionism began three years earlier, when he became acquainted with the model of the cerebellum put forward by Andras Pellionisz and Rodolfo Llinas (1979). The work of Pellionisz and Llinas foreshadows many of the arguments that Churchland makes. They argue that functions of the brain are represented in multidimensional spaces, that neural networks should therefore be treated as “geometrical objects” (323), and that “the internal language of the brain is vectorial” (330). The Pellionisz and Llinas paper also includes an argument for the superiority of neural network organization over von Neumann computer organization on the grounds that the network is more reliable and resistant to damage, a theme to which Churchland often returns.

Over the years, Churchland has applied connectionism to several areas of philosophy, notably: philosophy of mind, epistemology, philosophy of science, and ethics. Churchland's arguments in these areas have a common structure. First, he shows that the predominant positions in the field are (a) based on an assumption that the fundamental objects of study are propositions and logical inferences, and (b) have significant internal difficulties largely attributable to that assumption. Second, he presents a re-construal of the field based on connectionism, giving a “neurocomputational perspective” on the fundamental issues in the field.