A Connectionist Approach to Spatial Cognition

Abstract

Geographers became interested in environmental cognition and the relevance of behavioral concepts in geography approximately 30 years ago. As ideas related to spatial cognition became known to a greater number of geographers (Downs and Stea 1973; Ittelson 1973), their proper place within geography became the subject of some debate (Tuan 1975; Graham 1976; Bunting and Guelke 1979; Downs 1979, 1981; Rushton 1979; Saarinen 1979). Lloyd’s (1982) review of early progress made by behavioral geographers pointed out two general weaknesses. First, much of the early research was not based on explicit cognitive theories on how spatial knowledge is encoded, structured in human memory, or used to make decision. This was not because geographers refused to acknowledge and investigate spatial cognition theory. Research topics such as imagery had only just become a popular and important topic for psychologists (Cohen 1977; Solso 1979) because the behaviorist paradigm had not considered introspection to be an essential part of psychology. Subjects such as consciousness, mental states, and images were not considered to be worthy research topics (Woodworth 1948). Geographers who were trying to investigate cognitive issues at this time were not completely aware of the theoretical progress that had only recently been made in cognitive psychology. This chapter considers a theoretical approach that could provide a fresh perspective for many geographic problems. Connectionist theories and the neural network models that provide the practical context for such theories could provide useful insights for a variety of geographic problems.