London taxi drivers leverage regional boundaries to optimise route choices and improve their navigation skill across three decades

The world is defined by boundaries. They segment our experience of time and space, and in virtual environments have been shown to impact navigational choices. Here, we test the impact of boundaries on route choices in a real-world environment (London, UK) with a group of expert navigators: licensed London taxi drivers who are required to memorise the layout of over 26,000 streets to obtain their licence. After presenting photographs of a start location and a goal location, taxi drivers were asked to either accept or reject a third target street as forming part of the direct route or not. Performance increased across the adult life-span period in this group (age: 34 to 67). Taxi drivers were faster and more accurate when the target location formed part of a street network boundary (e.g. streets on the edge of the London neighbourhood Soho). Our results are consistent with taxi drivers exploiting the graph structure of the street network to plan routes, as well as consistent with the formation of hierarchical state representations to reduce the dimensionality of the planning problem. Taken together, we show that navigational skill can improve over decades of exposure and that experts exploit regional boundaries for optimal choices, providing a scaffolding over which to form action plans.