Empowerment and Relevant Goal Information as Alternatives to Graph-Theoretic Centrality for Navigational Decision Making

Abstract

City planners and architects employ graph-theoretic measures to analyse models of the built environment and predict human navigational behaviour. A recent breakthrough in the neuroscience of spatial cognition has shown that activation in the human hippocampus tracks the change in centrality for subjects navigating a virtual Soho in a fMRI scanner. Based on a well understood information-theoretic framework for modelling intelligent behaviour under cognitive constraints, the existing measures empowerment and relevant goal information, and novel quantity relevant goal information uptake were applied to a graph of the Soho street network navigated in the experiment. Empowerment, relevant goal information and relevant goal information uptake are shown to correlate with graph centrality for the primal graph, and to a lesser extent with centrality for the dual graph as used in the Soho experiment. These results, consistent with the hypothesis, provide preliminary evidence that human navigation employs an empowerment maximisation strategy, and to the author's knowledge, linking empowerment and relevant goal information with empirical neuroscience in a collaborative study for the first time.