CAPTURING INDOOR WAYFINDING STRATEGIES AND DIFFERENCES IN SPATIAL KNOWLEDGE WITH SPACE SYNTAX 043

It has been long-known in cognitive psychology that geometric and configurational features of the environment influence human orientation and navigation (e.g., Weisman, 1981, Garling et al., 1983). Yet proper analytic characterisation of environmental features has generally been elusive in these studies. More recently, space syntax methods have been adopted to capture environmental determiners of individual human navigation decisions (Peponis et al, 1990; Haq & Zimring, 2003; Conroy Dalton, 2003). Our study extends this approach by introducing route-based space syntax measures that capture the interplay of spatial features of a building with cognitive properties (level of prior knowledge) of its users. For this purpose we reanalyse an experiment on wayfinding in a complex multi-level conference centre (Holscher et al., 2005). The empirical study compared the performance of experienced and inexperienced participants in a set of wayfinding tasks, identifying specific strategies for vertical navigation. Inexperienced participants predominantly rely on a central-point strategy, while experienced ones have more accurate knowledge and plan more effectively. They prefer a memory-efficient “floor strategy” of swift movement to the destination floor. The building and its vertical connections were segmented into cognitively plausible subsections corresponding to decision points in the corridor network. For each path segment we computed VGA-based average connectivity, integration and step depth (to target location) with Depthmap (Turner, 2004). Movement trajectories of each participant were translated to a string of visited segments for each task. Route-specific space syntax measures were computed by aggregating the values of all segments along each string, based on both maximum and mean values. Step depth between start and goal location proved to be an excellent predictor of general difficulty of a wayfinding task, since we find significant correlations (r=.77 to .87) with 5 of 6 behavioural performance variables. But more importantly, the route-based measures reveal distinct, statistically reliable differences between experienced and novice users of the building. The experienced participants showed lower mean step depth to goal along their paths, thus avoiding unnecessary route segments. By contrast, the routes chosen by novice users had significantly higher values on mean and max connectivity as well as mean and max integration. Having less precise knowledge about the setting, they rely more on locally visible information and relatively familiar building parts like main corridors and the entrance area, as reflected in their central point strategy. Overall, our relatively simple route-based measures capture important behavioural differences between users and ultimately tie them to differences in spatial knowledge and cognitive strategies.