by Alex Masurovsky
To learn more about how grid cells work, researchers at the Max Planck Institute in Leipzig are thinking abstractly.
Grid cells and place cells work together in the brain to allow an organism to keep track of where it is in a given space. They are found in the hippocampus, a key brain area for memory functioning. Research on these cells has primarily been done in rats and other animals because researchers can insert electrodes directly into the animals’ brains to achieve higher resolution recordings. The Nobel Prize in Physiology/Medicine in 2014 went half to John O’Keefe and half split between May-Britt and Edvard I. Moser for the discovery of these cells.
Because the insertion of electrodes is considered too invasive for any in humans (for any non-medical purposes), clever researchers have found ways to study grid cells using Functional Magnetic Resonance Imaging (fMRI). However, fMRI is limited in resolution, so this method measures the activity of voxels (which are like 3D pixels of a brain image) rather than individual cells. This is considred valid because it is larger groups of cells with similar properties, rather than one cell at a time, that respond to distinctive features of the environment.
“Grid cells align to properties in the environment, like walls or landmarks,” explains one researcher at the Max Planck Institute (MPI) in Leipzig.
In the past, research has shown that voxels in humans, like individual grid cells in rats, respond to various points in an environment (such as a room) with recurring patterns at intervals of 60º of a particular angle. Theoretically, the same voxel may respond at angles of 10º, 70º and 130º from a given point.
While grid cells have typically been studied in the context of navigation, there are hints that they are responsible for storing recent events as well as associations between objects.
“What we are trying to learn is how grid cells break down the properties of objects into the most salient, interesting features.”
Researchers at the MPI Leipzig aim to study these “extra-navigational” properties of grid cells. They hypothesize: when presented with an object rather than a physical space, grid cells may map to specific features of the object.
To test this, they designed a study with stimuli designed along two axes, the way a room has a longitude and latitude. The axes in this case, however, are line thickness and shape.
The objects will transform before the eyes of research subjects, which gives the researchers the ability to measure changes in activity to fine-grained manipulations of line thickness and shape. The configuration of the object can be visualized in a 2-dimensional “object feature space.” The transformation across this 2D object feature space can be measured as an angle with a trajectory (see visualization in picture, above), which makes the study comparable to more typical, navigational studies of grid cells.
As subjects watch the objects transform, researchers will watch the activity of the subjects’ grid cells — by analyzing the data from fMRI and inferring from the activation patterns of voxels.
The researcher I spoke to summarized it this way: “what we are trying to learn is how grid cells break down the properties of objects into the most salient, interesting features.”