Cognitive Maps

Introduction

This entry discusses cognitive maps and the methodologies used to define them. The concept is traced from its emergence in psychology through attempts to operationalize it and use it in disciplines such as planning, behavioural geography, artificial intelligence, and computer science. Assessment tasks include sketch mapping, written and verbal descriptions, orientation and direction estimation, interpoint distance estimation, establishing frames of reference, establishing configurational or layout knowledge using trilateration or non-metric multidimensional scaling, and completion of navigation or wayfinding tasks. Future research directions involve more work on spatial cognition and spatial abilities, research at macrospatial scales, evaluation of potential contribution of virtual environments, and investigation of the neurobiology of place cells.

Definition of Terms and Background

Although cognitive maps have been used for environmental knowing and wayfinding throughout the entirety of human history, they have only become a matter of scientific experimentation and analysis since the advent of Tolman's place learning theory (Tolman, 1948). This theory suggests that a cognitive map develops in the long-term memory of humans and other animals. Continuing multidisciplinary efforts have been made to examine the content, validity, and reliability of these internal representations.

Defined as one's internal representation of the experienced world, the concept of a cognitive map has spread among many disciplines. Beyond the original work in psychology, the first application of a ‘cognitive map’ was made by planner Kevin Lynch (1960). He examined what people knew about environments, suggesting that knowledge [Page 224]depended on environmental legibility. Legibility was defined as the ease with which an environment could be perceived, comprehended, and used. To determine what people knew about environments, Lynch asked them to externalize their knowledge (of selected cities) by producing ‘sketch maps’. These were examined to find which features (landmarks and other reference points, paths and boundaries, and neighbourhoods or districts) were included. Using the sketches made by many individuals, he produced (for specific cities) a composite sketch (or ‘city image’) of those locational, path, and district features that were represented by the majority of the participants.

Following Lynch's efforts, geographers became interested in the cognitive map concept. Initially termed ‘mental maps’ (Gould, 1966), these were cartographic representations of the rank orders of stated preferences for living in places. The rank orders were aggregated, and a cartographic isoline map of the places – or a map represented as a continuous surface using lines of uniform preference value – was constructed. Remarkable regional differences in preferences were found, and the results were often interpreted as a regionalized ‘view of the world’ – such as ‘a Californian view of the US’.

Piaget and Inhelder (1967) provided both a theoretical structure and empirical evidence that cognitive maps develop over time as age and intellectual maturity advanced. Their developmental theory of knowledge acquisition was made explicitly spatial by Hart and Moore (1973) and Siegel and White (1975), who argued that there was a continuous transformation of spatial knowledge from an egocentric structure that dominated the first two years of infancy and was epitomized by a projective form of representation, to a topological knowledge structure as children advanced to pre-operational learning stages, to a semi-metric and metric understanding as children passed through concrete operational and abstract stages of thinking and reasoning. In recent years, Montello (1998) vigorously challenged these ideas, not for their relevance to the spatial knowledge acquisition of children as they age, but in terms of adult learning about new environments. His argument suggests that adult humans have the ability to reason abstractly about space and to represent it metrically, and, thus, would not need to go through the earlier stages outlined in developmental theories.

...