My research on navigation and spatial cognition asks three major questions: 1) What is the underlying structure of spatial knowledge?  2) How do we acquire the metric information needed for this spatial knowledge?  3) How can we help people learn new environments? 

My research draws upon innovative techniques in functional magnetic resonance imaging (fMRI) and fully-immersive virtual reality to explore these aspects of human spatial navigation.

MazeHallway     PathIntegrationEnvironment     MazeObject

What is the Structure of Spatial Knowledge?

Before determining how we acquire spatial knowledge, it is important to understand what exactly is being acquired.  We found evidence that the underlying structure of spatial knowledge is consistent with a labeled graph, which connects locations (nodes) along paths (edges) in the environment.  In contrast to pure route knowledge, we found the most frequent routes and detours had not been traveled during learning.  Contrary to purely topological knowledge, navigators traveled the shortest metric distance to a goal, rather than topologically equivalent but longer paths.  Thus, this topological structure of spatial knowledge connects locations without relying on a globally-consistent map-like structure, although it does include coarse local metric information.

From cognitive maps to cognitive graphs. PLOS ONE, 2014. [pdf]

Neural evidence supports a novel framework for spatial navigation. Psychonomic Bulletin & Review, 2013. [pdf]

How Do We Acquire Metric Information?

If the underlying labeled graph structure of spatial knowledge includes some metric information, what is the mechanism used to acquire that local metric information?

A likely candidate is path integration, the continual updating of position and orientation during movementin an environment.  We have studied path integration from a number of different angles.  For example, we found evidence that navigators preferably use body-based information about self-motion to gauge distances and angles during path integration. We also found that errors in executing a desired trajectory tend to make a larger contribution in path integration errors than do errors in encoding the outbound path.  

In the brain, we found activation that could support a homing vector for path integration, a process that keeps track of a navigator's Euclidean (straight-line) distance relative to their start location during movement in a landmark-free environment.  The hippocampus, retrosplenial cortex, and parahippocampal cortex had fMRI activation consistent with a homing vector.  These same regions also supported memory for simple straight-line translation and rotation in place by tracking the homing vectormagnitude of translation or rotation during virtual self-motion.  We also have found signals important for heading direction and have uncovered functional connections between optic flow areas—which process visual information about changes in heading—and regions important for path integration, including hippocampus and retrosplenial cortex. 

There and back again: Hippocampus and retrosplenial cortex track homing distance in human path integration. Journal of Neuroscience, 2015. [pdf]

Functional connections between optic flow areas and navigationally responsive brain regions during goal-directed navigation. NeuroImage, 2015. [pdf]

Is path integration based on an intrinsic metric or absolute distance? Attention, Perception, and Psychophysics, 2014. [pdf]

Which way and how far?  Tracking of translation and rotation information for human path integration. Under review.

How Do People Learn New Environments?

When exploring a new city, a person might learn very different thingssignpost
when walking around on their own compared to being driven in a taxi.
Several components contribute to active navigation, including    

  • motor efference
  • proprioception
  • vestibular input
  • allocation of attention
  • cognitive decision-making

To examine how these factors contribute to spatial learning, we developed the exploration-specific learning hypothesis, which posits that these factors of active learning have different effects depending on the type of spatial knowledge that is to be acquired.

We tested this hypothesis in an immersive maze environment, finding that making decisions about exploration contributed to learning the topological graph structure of the maze.  In contrast, decision-making had no effect on learning the metric distances and angles between locations.  However, body-based information did contribute to this metric survey knowledge.  These results indicate a sharp contrast between factors of active learning and their contribution to resulting spatial knowledge. With collaborators, we found that active exploration can assist blind navigators learn novel environments by allowing them to use spatial information more flexibly. 

routesActive and passive contributions to spatial learning. Psychonomic Bulletin & Review, 2012. [pdf]

Active and passive spatial learning in human navigation: Acquisition of graph knowledge. Journal of Experimental Psychology: Learning, Memory, & Cognition, 2015. [pdf]

Active and passive spatial learning in human navigation: Acquisition of survey knowledge. Journal of Experimental Psychology: Learning, Memory, and Cognition, 2013. [pdf]

Virtual environments for the transfer of navigation skills in the blind: A comparison of directed instruction vs. video game based learning approaches.  Frontiers in Human Neuroscience, 2014. [pdf]