Correcting the code

The pattern of activity in specific cells in brain regions responsible for spatial navigation is comparable with a novel mathematical coding scheme which allows for very accurate localization, reports a study published online this week in Nature Neuroscience.

Grid cells are a class of neurons mainly found in the entorhinal cortex, a region important to memory and navigation. These neurons respond in an unusual, repeating pattern according to where an animal is located, responding more when an animal is at certain places. For example, grid cells are likely to be important for coding where a rat is in a complicated maze, but it is unclear how this repeating pattern might help such localization.

Sameet Srinivasan and Ila Fiete’s work suggests that grid cells might be taking advantage of the same design principle that the online retailer Amazon uses to ensure accurate order fulfilment. The Amazon warehouse, unlike a store floor, does not need to keep similar items next to each other: a size 12 T-shirt, for example, might be next to books, rather than next to identical T-shirts of other sizes. This reduces the possibility of errors when packing an order following the fundamental principle that similar items should not be placed close to each other, to avoid confusion.

Similarly, the authors suggest that grid cells code locations in such a way that similar locations are placed further apart in an abstract mathematical ‘coding space’, which means the animal is less likely to mix up similar locations.