SILC Showcase

Showcase March 2015: Anchoring the neural compass: Coding of local spatial reference frames in human medial parietal cortex

Share this article using our url:

Anchoring the neural compass: Coding of local spatial reference frames in human medial parietal cortex

Steven A. Marchette, Lindsay K. Vass, Jack Ryan, and Russell A. Epstein

University of Pennsylvania

To successfully navigate one must know how one is oriented within the world. Quite a lot of work in neuroscience has studied the neural mechanisms that might allow an animal to represent and maintain its orientation. Invasive recording techniques in rats have discovered “head direction” cells that seem to act like a neural compass which keeps track of the animal’s heading as it navigates and may provide the internal sense of direction (Taube, 1980). However, perhaps the number one thing that we all know intuitively about our sense of direction is that it occasionally fails. We have probably all had the experience of coming out of a subway station and being suddenly lost. In this situation, how can we get our sense of direction back?

One solution might be to look for North. Sadly, humans seem to lack a keen sense of magnetoception. Another solution that might work better would be to use your orientation relative to local (currently-visible) landmarks around you to anchor your sense of direction. For example, if you can work out that City Hall is to your right and the subway station is to your left, then there is only one direction that you could be facing. This kind of local heading representation would be useful because when you’re lost you could use it to reset your neural compass. In this experiment, we wanted to discover where this kind of orientation relative to local environment features might be represented within the human brain (Marchette, Vass, Ryan & Epstein, 2014). Based on previous neuroimaging (Baumann & Mattingley, 2010; Vass & Epstein, 2013) and neuropsychological (Aguirre, G.K. & D'Esposito, 1999) data, we hypothesized that a region of medial parietal cortex referred to as the retrosplenial complex (RSC) might support this locally-referenced representation of heading, thus providing a “dial” for the neural compass.

To test this hypothesis, we taught participants a virtual environment constructed in the commercial video game, Portal (, Valve Software, Bellevue, WA) that we designed to look for local heading codes. This environment consisted of a park containing four large rectangular buildings or “museums” with distinct textures and architectural features (Figure 1).The key feature of the design was that all of the museums had the same geometry and were laid out in a clover-leaf pattern. This dissociated the local headings within a museum, e.g., “facing the back wall”, from the global heading within the park. For example, one museum’s back wall faced to the North, whereas another had a back wall that faced to the East. This allowed us to test whether RSC coded one’s heading with respect to the local geometry of the museum or the broader frame of the park.

Figure 1
Figure 1. Map and images of the virtual environment. A. Map of the virtual park and the four museums. Each museum was oriented at a unique direction with respect to the surrounding park. Objects were displayed within alcoves, which are indicated by grey squares. Each alcove could only be viewed from one direction. B. Images of the exteriors, interiors and alcoves of each museum.

To test participants’ heading representations, we asked participants to learn and then recall the position of objects within two of the museums. Each museum had 8 unique objects which were set into alcoves so that each one was clearly associated with a particular heading. For example, if you were facing straight at the cake, then there was only one heading you could be facing (i.e., towards the left wall). Then, while participants underwent functional magnetic resonance imaging (fMRI), we asked them to imagine facing the objects and report whether a target object from the same museum would be to their left or their right. So, although participants only ever saw words within the scanner (“Facing the Cake”), we knew which heading they had imagined themselves facing (e.g., toward the left wall).

For our primary analysis, we measured the activity pattern within the RSC evoked by imagined views drawn from two of the museums. We hypothesized that if RSC was involved in representing one’s heading, then the activity patterns for two views that faced the same heading (North) should be more similar to one another than the activity patterns for pairs of views that faced different headings (North and East). First, we considered views that were all within the same museum and this was exactly what we saw (Figure 2). Because we were always looking at the similarity among different views (i.e., facing different objects), RSC must have represented information about heading that was abstract and generalized across distinct views and locations.

Figure 2
Figure 2. Coding of facing direction in retrosplenial complex (RSC) activation patterns. A. Sagittal slice showing RSC. B. Coding of local direction in the local (museum-referenced) frame in RSC. Left panel shows that pattern similarity between views that face the same direction in local space was greater than pattern similarity between views that face different local directions, both within and across museums. Right panel shows that across museum views were no more similar when they faced the same or different global direction.

Next, we considered how RSC would treat views that were located in different museums. This was important because it allowed us to distinguish between three possible coding schemes that RSC might use to define one’s heading. First, it was possible that heading codes would not generalize across museums because idiosyncratic codes were used for each museum. After all, each museum was a different building—visually distinct and with its own unique location in the world—perhaps different buildings always receive different spatial codes. Second, it was possible that the same global code would be used for both museums. In this case, a view that faced towards North should be similar to all other views facing to the North, no matter which museum they were in. Finally, it was possible that the codes would generalize based on a local heading relative to the geometry of the surrounding environment. If so, then views that faced the back walls of different museums might be treated as similar even though the back walls of the museums were all facing different global directions within the park. Our data confirmed this third possibility: views that faced the same local heading were more similar than views that faced different local headings (Figure 2). In addition, we observed no evidence for global coding in RSC: that is, North views were no more similar to one another across museums than to East, West, or South facing views.

We then took another approach to looking at our data. If RSC really was involved in reorienting oneself within a mental map of local space, then RSC should contain enough information for us to reconstruct the shape of that space. To test this, we used a multidimensional scaling technique (Kruskal & Wish, 1978) to estimate a mental map of the museum that matched the way that RSC treated different views as similar or dissimilar. Similar views should be placed close together, and dissimilar views should be further apart. This analysis reconstructed an accurate map of the museum (Figure 3)—providing evidence that RSC contained a rich representation of where views occurred within the local environment. Since all our participants saw in the scanner were the names of objects, all of the spatial information must have been retrieved from memory. To our knowledge, this is the first time that the spatial structure of an environment could be reconstructed purely from the neural correlates of the human imagination.

Figure 3
Figure 3. Activity patterns in RSC contain sufficient information about the spatial relations between views to reconstruct the spatial organization of the environment. A. A confusion matrix describing the similarity between every pair of views within the same museum was used to estimate the spatial representation within RSC. The estimated locations (colored diamonds) are close to the real locations (numbers in black outline). B. View reconstruction was also possible when considering the similarity among views in different museums.

Taken together, our data suggest that RSC represents one’s heading relative to local environmental features and its heading codes generalize across environments of similar geometry. So, what does this tell us about how we orient ourselves to the world? Our results suggest that RSC acts as the compass rose for the neural compass; it anchors the internal sense of direction to the features of the external world, allowing us to use reorient to our surroundings and recover our sense of direction when we get lost.


  • ♦ Taube, J. S., Muller, R. U. & Ranck, J. B. (1990). Head-direction cells recorded from the postsubiculum in freely moving rats. I. Description and quantitative analysis. The Journal of Neuroscience, 10, 420-435.
  • ♦ Marchette, S. A., Vass, L. K., Ryan, J., & Epstein, R. A. (2014). Anchoring the neural compass: Coding of local spatial reference frames in human medial parietal lobe. Nature Neuroscience, 17, 1598-1606.
  • ♦ Baumann, O. & Mattingley, J. B. (2010). Medial Parietal Cortex Encodes Perceived Heading Direction in Humans. The Journal of Neuroscience, 30, 12897-12901.
  • ♦ Vass, L. K. & Epstein, R. A. (2013). Abstract Representations of Location and Facing Direction in the Human Brain. The Journal of Neuroscience, 33, 6133-6142.
  • ♦ Aguirre, G. K. & D'Esposito, M. (1999). Topographical disorientation: A synthesis and taxonomy. Brain, 122, 1613-1628.
  • ♦ Kruskal, J. B. & Wish, M. (1978). Multidimensional scaling. Beverly Hills: Sage.
You are here: SILC Home Page SILC Showcase Showcase March 2015: Anchoring the neural compass: Coding of local spatial reference frames in human medial parietal cortex