grid cells

Bat Brains and the Nobel Prize

jhalper — Wed, 10 Dec 2014 06:53:48 +0000

Bat Brains and the Nobel Prize

Not everyone gets their research written about by this week’s Nobel Prize winners:

All mammals face the challenge of navigating in complex, three-dimensional (3D) environments, whether they are swinging from branch-to-branch in forests or burrowing underground tunnels. How does the brain maintain a sense of place and direction in 3D? In a beautiful study published on Nature's website today, Finkelstein et al. report that bats have an internal neural compass that tracks direction in 3D during both surface locomotion and flight.

That’s from Prof. May-Britt Moser in Nature’s “News and Views.” It turns out that the whole place-cell-grid-cell community is quite excited by today’s prize, not just because the Mosers are apparently well-loved, but because it boosts all the research in this field. And the fact that cells in our brain form little hexagonal grids that keep us oriented, map-like, in our surroundings is not just an important insight into the workings of our brain; it is a pyrotechnic flash of insight that changes how we understand the brain to work.

Arseny Finkelstein is in the group of Prof. Nachum Ulanovsky, and to tell the truth, their work does not need a lot of extra boosting. Watch the video of bats flying around their lab to see why: http://www.nature.com/news/bat-nav-system-enables-three-dimensional-manoeuvres-1.16475. Among other things, they show that you can really apply mathematical models to understand how our mammalian brains get their bearings.

Understanding bats’ internal compasses, by the way, could have real implications for research into human brain malfunctions, among them the sudden vertigo occasionally experienced by pilots, when they lose their sense of up and down.

jhalper Wed, 12/10/2014 - 01:53

The Physics of: Ants and Bats

jhalper — Tue, 29 Nov 2011 03:00:00 +0000

The Physics of: Ants and Bats

This week's new Weizmann science stories are on ants and bats. Two different models for investigating human behavior? Yes, but not exactly in the ways you might imagine, and so much more than that.

Dr. Ofer Feinerman, the "ant scientist," is a new member of the Physics Faculty. In his graduate research under Prof. Elisha Moses in the Physics of Complex Systems Department, Feinerman created artificial circuits out of neurons. Now he has turned to investigating the complexities of ant societies. What, you might ask, do neurons and ant colonies have to do with physics? The answer is: They involve non-linear, collective phenomena, similar to those studied extensively by physicists. (Moses has investigated everything from falling leaves and advancing wave fronts* to schizophrenia to deep patterns in literature.) So in this physics lab, instead of lasers or ion traps, there are high-tech ant farms where the inhabitants run around with miniature barcodes glued to their backs.

Photo: Lab of Dr. Ofer Feinerman

Of course a huge body of work has already been published on ant social order, and conclusions have even been applied to human society from some of that work. What new perspective will a physicist bring to ant studies? For one thing, Feinerman has the tools - experimental and computational - to conduct in-depth research into one of the basic underpinnings of any society: communication networks. These go beyond the simple divisions of labor: queen, worker, nursemaid, sperm donor, etc, and his studies are already revealing that the ants' social order is both more complex and more flexible than that painted by the more traditional models. Ultimately, it may be the differences between ant and human societies that are the most interesting. Because once a physicist detects a new pattern, he will almost certainly look around for other iterations of that pattern. Understanding how communications help facilitate life in the leaderless ant colonies may not help us to create the perfect anarchist utopia, but it could provide insight into other non-linear networks - for instance complex interrelations between immune cells or communication between swarms of mini-robots engaged in a joint task.

Dr. Nachum Ulanovsky, the "bat scientist" even has a "bat cave" in his lab, where the animals can fly around in the dark. Though he is in the Neurobiology Department, Ulanovsky's research would definitely not be out of place in the non-linear physics lab; and his analytical methods are often borrowed from physics. In fact, in his latest work, he investigated a phenomenon that is more often seen in the neat world of solid-state physics than the messy world of brain biology: cells that fire in an orderly, hexagonal pattern. These cells, called grid cells for obvious reasons, fire as an animal moves through an environment and crosses the vertices of a hexagonal lattice, and they help create maps in the brain.

The question was: What causes this regular firing pattern? Was it due to another non-linear phenomenon - time-dependent oscillations in the same region? Or could the pattern arise from something in the way these cells work together?

Once again, difference was crucial: Researchers who investigated the grid cell patterns in rats and mice had noted that regular oscillations always accompanied their activity. Ulanovsky repeated those experiments with bats, on a hunch that the seemingly solid connection between the spatial pattern and temporal oscillations in the rat brain might come uncoupled in a different animal. Recordings of the activity in this tiny area of the bats' brains showed that the hexagonal firing patterns were there - nearly identical to those of the rats. But, as he and his team reported in Nature several weeks ago, the oscillations they observed were very rare, and did not appear to be tied in any way to the spatial patterns. In other words, the bat's brain has periodicity but it ain't got that "rhythmicity" - which tends to discount the popular theory that the periodicity in the rat's brain is caused by neuronal rhythmicity.

Which brings the whole story back around to communications networks. In this case, the neurons appear to coordinate their activities using the most stable possible network: a honeycomb-like hexagonal grid in which every node is equidistant from its neighbors.

Postscript: Ulanvosky and Feinerman are both locals, and both wrote this science writer that they went to high school together and had the same math and physics teachers. Who says that high school doesn't make a difference?

*The Weizmann Wave banner is a microscope image from Moses' lab of a boundary-layer eruption caused by temperature variations in water.

jhalper Mon, 11/28/2011 - 22:00

Human grid cells tile the environment

neurophilosophy — Wed, 27 Jan 2010 11:40:04 +0000

Human grid cells tile the environment

HOW does the brain encode the spatial representations which enable us to successfully navigate our environment? Four decades of research has identified four cell types in the brains of mice and rats which are known to be involved in these processes: place cells, grid cells, head direction cells and, most recently, border cells. Although the functions of most of these cell types are well characterized in rodents, it remains unclear whether they are also found in humans. A new functional neuroimaging study, by researchers from University College London, published online in the journal Nature, now provides the first evidence for the existence of grid cells in the human brain.

Grid cells were discovered in 2005 by Edvard and May-Britt Moser of the Kavli Institute for Systems Neuroscience in Trondheim, Norway, using multi-electrode arrays chronically implanted into the hippocampus and surrounding regions of freely moving mice. Whereas place cells fire when the animal is in a unique, specified position in its environment, grid cells - which are located in the entorhinal cortex - increase their activity in multiple locations, firing periodically as the mouse traverses a space. When a grid cell's activity is correlated with the animal's position and trajectory, and then superimposed onto a map of the environment, it is found to define a repeating pattern of equilateral triangles which 'tiles' the space (panel b, below). Each cell is activated whenever the animal's position coincides with any vertex in this grid, but each has its own periodicity, and so 'tiles' the environment using a unique scale.

Christian Doeller of UCL's Institute of Cognitive Neuroscience and his colleagues first carried out a series of experiments in which grid cell activity was recorded from the brains of mice. This revealed several previously unknown properties of the cells. First, they found that grid cell activity is modulated by the direction in which the animals are moving, so that the increase in grid cell activity is greater when the animals move along the main axes of the grids. Further, they found that the activity of the cells is also modulated by running speed - the faster the animals moved, the more the quickly they crossed the virtual triangle boundaries encoded by the grid cells and, as a result, the shorter was the interval between each burst of activity.

The researchers reasoned that if grid cells exist in the human brain, the activity of the population as a whole should produce a signal that is large enough to be detected by functional neuroimaging. And, because the orientations of the triangular grids encoded by the cells in the population are all aligned to each other, the activation pattern should exhibit six-fold rotational symmetry. They therefore recruited 42 male participants, and scanned their brains while they explored a circular virtual reality environment consisting of a grassy plain bounded by a cliff and surrounded by mountains (panel a, below).

When they analyzed the fMRI data, the researchers found that the signal was modulated by direction and running speed, just as they had predicted. The grid orientation appeared to vary randomly between the participants, suggesting that the activity of grid cells is independent of landmarks in the surroundings. In each participant, however, the increase in entorhinal cortex activity was greatest when their movements through the virtual environment were aligned with the three main grid axes. That is, the signal exhibited 6-fold rotational symmetry; no activity with 4-, 5-, 7- or 8-fold symmetry was observed. The signal obtained during high speed movements was also stronger than the one recorded during slower movements (panels c and d, above).

While navigating the virtual environment, the participants were required to collect various objects and then return them to the same location. In their analyses, the researchers also found that the signal obtained was related to performance on this memory task - the more coherent the signal, the better was the participant's performance. The regions from which neuronal activity was recorded overlaps extensively with brain areas known to be involved in encoding and retrieval of autobiographical memory. This provides some insight into the neural basis of this type of memory; it suggests that the brain may use information about both time and space when encoding life events.

Unlike most functional neuroimaging studies, which simply correlate behaviours with patterns of brain activation, this one is driven is hypothesis. The researchers made predictions about the signal they would expect to see, based on known properties of grid cell activity. They only provide indirect evidence that grid cells exist in the human brain, however. Obtaining solid evidence would involve implanting electrodes into the brain, which is unfeasible in healthy participants. Epileptic patients undergoing pre-surgical evaluation afford a unique opportunity to investigate neuronal function directly; perhaps researchers will try to investigate the cellular basis of spatial navigation in this situation.

Nevertheless, this study suggests that the human brain, like that of mice and rats, uses a regularly repeating grid-like geometry to encode representations of space. The hippocampus and surrounding areas are known to be the first to degenerate in Alzheimer's Disease, so the findings present may also help to explain why disorientation is one of the first behavioural manifestations of the condition.

Related:

Doeller, C., et al. (2010). Evidence for grid cells in a human memory network Nature DOI: 10.1038/nature08704.

neurophilosophy Wed, 01/27/2010 - 06:40

Tags

This all sounds pretty logical- what a neat system based on virtual triangulation! What really got me about the work with humans is how neuroscientists now use responses to virtual reality as a straight substitute for responses to visual/spatial reality! When I was working on visual imagery in a peripheral sort of way some years ago there was only a vague recognition that images and objects could be dealth with by the brain in similar ways. You have to feel sorry for those unfortunate people having pre-op tests for epilepsy surgery- there are so many branches of science lining up to test them, they might start having to auction off the privilege!

Hi, I have stumbled upon this blog a couple of times and since it seems to keep high standards I would like to offer you a few small corrections & comments:

The two first findings you describe are old news. Cells with conjunctive grid and head-direction properties were first described by the same team that discovered grid cells (Sargolini et al., 2006). In that paper we also described the speed modulation. However, most entorhinal grid cells are omnidirectional (Fyhn et al., 2004; Hafting et al., 2005) so it is very impressive that Doeller et al. picks up the signal from the conjunctive cells.

There is no evidence that all the grid cells in one animal show the same orientation, because there have not been performed simultaneous recordings from the very different parts of entorhinal cortex.

Anyway, it is a great study, but I still look forward to see the first recording of a grid cell in a human brain. It would be very interesting to know if the response-patterns are the same in a VR task as if the people were walking around....

PS! Grid cells were discovered in rats, not mice.

Lekrot: these findings are not old news, as they are the first evidence of grid-like cells in non-rodents. In my mind it has always been an obvious question to ask, whether other mammals, such as primates, also have grid cells.
Rodents have very poor vision and rely mostly on their whiskers to gain information about the world. Tactile information of this sort is inherently short range and discontinuous, making a mechanism like grid cells essential for keeping track of movements through an environment.
We primates (and most other mammals) have excellent vision, so a system of grid cells is of less use to us as we can keep track of our movements, and odometry drift, from the stream of visual cues.
If I had the money (and ethical approval) I'd immediately go and check if primates have grid cells electrophysiologically. However, the paper described above (and the smart way they went about it) hints that we do indeed have grid-like cells, and that they help us in learning, mapping and navigating our environment.

I'm sorry I did not see your reply before. If you read your own post one time more, you will realise that you are wrong and I am right. I did not argue against the fact that Doeller et al. is the first to look for grid-like evidence in humans, but I just pointed out that you have a lot of errors in the paragraph describing Doellers rat experiments in the very same paper:
"Christian Doeller of UCL's Institute of Cognitive Neuroscience and his colleagues first carried out a series of experiments in which grid cell activity was recorded from the brains of mice. This revealed several previously unknown properties of the cells. First, they found that grid cell activity is modulated by the direction in which the animals are moving, so that the increase in grid cell activity is greater when the animals move along the main axes of the grids. Further, they found that the activity of the cells is also modulated by running speed - the faster the animals moved, the more the quickly they crossed the virtual triangle boundaries encoded by the grid cells and, as a result, the shorter was the interval between each burst of activity."

I love reading your blog. This story was of particular interest to me, as I'd just finished listening to WNYC's Radiolab podcast, the 1/25/2011 edition, which discusses grid cells and place cells (for a general audience). Thanks!