Most animals, including humans, are able to flexible navigate the complex world. They can explore new areas, returning quickly to remembered places, and taking shortcuts. The recent discovery in neuroscience, including place cells, grid cells, head direction cells, border cells, goal-vectorial cells, etc. are revealing the secrets of navigation in our brain.

In most recent paper from the DeepMind published in Nature, the authors developed an artificial agent to test the theory that grid cells support vector-based navigation, in keeping with their overarching philosophy that algorithms used for AI can meaningfully approximate elements of the brain.

The fig. is from Andrea et al., 2018

The fig. is from Andrea et al., 2018

In addition, the emergence of grid-like representations by training recurrent neural networks to perform spatial localization by Cueva and Wei, which was published contemporaneously at ICLR. While different in scope and findings, it shows interesting results. In brief, the authors found periodic firing that conformed to the shape of the enclosure, e.g rectangular grids in a square environment and triangular in a triangular environment (fig. 2 of Cueva and Wei). This differs from the study by Banino et al. , where they found grid-like units whose firing pattern closely resembles rodent grid cells which typically show hexagonal firing patterns across different shaped environments (e.g. square and circular arena).

For further info, please read the following references.

Andrea Banino, Caswell Barry, Benigno Uria, Charles Blundell, Timothy Lillicrap, Piotr Mirowski, Alexander Pritzel, Martin J. Chadwick, Thomas Degris, Joseph Modayil, Greg Wayne, Hubert Soyer, Fabio Viola, Brian Zhang, Ross Goroshin, Neil Rabinowitz, Razvan Pascanu, Charlie Beattie, Stig Petersen, Amir Sadik, Stephen Gaffney, Helen King, Koray Kavukcuoglu, Demis Hassabis, Raia Hadsell & Dharshan Kumaran. Vector-based navigation using grid-like representations in artificial agents. Nature (2018), doi:10.1038/s41586-018-0102-6. https://www.nature.com/articles/s41586-018-0102-6

Christopher J. Cueva, Xue-Xin Wei. Emergence of grid-like representations by training recurrent neural networks to perform spatial localization. ICLR 2018. https://openreview.net/forum?id=B17JTOe0- 

The Nobel Prize in Physiology or Medicine 2014 was divided, one half awarded to John O’Keefe, the other half jointly to May-Britt Moser and Edvard I. Moser “for their discoveries of cells that constitute a positioning system in the brain“. https://www.nobelprize.org/nobel_prizes/medicine/laureates/2014/

Andrea Banino, Dharshan Kumaran, Caswell Barry. Navigating with grid-like representations in artificial agents. https://deepmind.com/blog/grid-cells/

Francesco Savelli & James J. Knierim. AI mimics brain codes for navigation. https://www.nature.com/articles/d41586-018-04992-7. doi: 10.1038/d41586-018-04992-7.

Alison Abbott. AI recreates activity patterns that brain cells use in navigation. https://www.nature.com/articles/d41586-018-05133-w

Will Knight. AI program gets really good at navigation by developing a brain-like GPS system. https://www.technologyreview.com/s/611105/robots-may-someday-explore-the-world-using-features-borrowed-from-your-brain/ 

John Rennie. Artificial Neural Nets Grow Brainlike Cells to Find Their Way. https://www.wired.com/story/artificial-neural-nets-grow-brainlike-cells-to-find-their-way/