Scientists discover brain cells that remember where escape routes are

These neurons track safe places so mice can escape threats in a split-second

Samuel J Walker


Harvard Medical School and Beth Israel Deaconess Medical Center

You’re at the supermarket when the fire alarm goes off. The noise grabs your attention, your senses are on high alert, and you know immediately that you have to get outside. Before you’ve had time to register where you are going, without searching or even looking around, you brain has carried you towards the nearest exit. What happens in this split second, and how does your brain know where you should go?

In a recent study, currently posted as a pre-print, researchers at the Sainsbury Wellcome Centre in London, UK investigated these questions. To do so, they studied mice, which like many animals, will respond to a threat by running toward a safe place in their environment. Previous research showed that mice memorize the locations of shelters in the environment to guide these escape runs. In fact, just 20 seconds of exploring a shelter is enough for a mouse to form a memory that it can later use when a threat appears. 

But how is the brain of a mouse able to continually keep track of its own position relative to safety while it is involved in other behaviors, and how does it use this information once a threat appears to guide escape towards shelter?

To understand how the brain keeps track of safe places in the environment, they recorded the brains of mice as they explored their environment and responded to threats. They implanted electrodes that can record hundreds of neurons in several parts of the mouse brain at the same time, targeting areas thought to be involved in spatial navigation and escape. Specifically, they recorded from the retrosplenial cortex (RSP), which is thought to play a role in navigating in space; and the superior colliculus (SC), a region that integrates information from sight, sound, and other senses to coordinate orienting behaviors.

When the researchers placed these mice in an arena containing a shelter and exposed them to threats, small groups of neurons in both the RSP and SC jumped right out at them. These neurons represented where the shelter was – specifically, the angle between the direction the animal was facing and the direction of the shelter – so they named these neurons “shelter direction cells.” Within these groups of neurons, different neurons were activated as the animal faced in different directions relative to the shelter. Together, they represented the whole range of possible shelter directions.

This diagram shows the locations of the RSP and SC in the middle of our brains, near labels 29 & 30.

Henry Vandyke Carter, Wikimedia Commons

While it appeared that the shelter direction cells represented the direction of safety, they could also have been pointing toward the shelter by sheer chance. To distinguish between these possibilities, the scientists simply moved the location of the shelter. If the neurons were specifically keeping track of the shelter, they surmised, their activity should shift to track the new location of the shelter. Sure enough, once the mouse explored the new shelter location, the activity of the shelter-direction cells shifted, so that they now represented the angle relative to this new shelter location.

Next, the researchers wondered whether the shelter-direction cells were activated to show the direction of safety specifically or if they would follow the direction of any interesting feature in the environment. To test this, the scientists introduced a closed shelter, which mice could investigate but could not enter. This simple experiment uncovered a striking difference between the RSP and the SC. In the RSP, similar numbers of neurons responded to the closed shelter as reacted to the open ones. In the SC, however, very few neurons encoded the direction of the closed shelter, while many more tracked the open shelter. This indicated that the SC is specifically tuned to safe places, while the RSP might track the direction of any interesting feature in the environment.

Based on that finding, the researchers thought that RSP and SC might interact to represent shelter direction, and that this interaction might be important in directing escape behavior. After further investigation they deduced that information likely flows from the RSP to the SC. So, to test the importance of this RSP-to-SC connection, they removed it: the scientists used a genetic approach to "silence" the activity of RSP neurons that connect to SC. 

When they did, the representation of shelter-direction in SC completely disappeared. What's more, the mice almost entirely lost their ability to orient towards safety. Mice would still run when a threat appeared, but in a completely wrong direction. In fact, the mice would often stop running too early, as if they knew they were running in the wrong direction but weren’t sure where to turn. And when the scientists played a non-threatening sound, the mice were still able to orient toward it even when the RSP-SC connection was silenced. It seems, therefore, that the RSP-SC connection helps direct orientation and escape in response to a threat.

A diagram showing the locations of the retrosplenial cortex and superior colliculus

Another view of the locations of the RSP and SC

Allen Brain Map

Although animals respond to threats instinctively, in a way that may seem simple and hardwired, escape is actually a complex behavior. Memory guides the direction of escape, and two specific brain areas cooperate to maintain this memory. But, many other parts of the brain must work together to coordinate the different parts of escape: detecting a threat; determining that it necessitates a response; deciding whether to run away or freeze in place; turning and running; and stopping and evaluating the surroundings again once safety is reached. 

In much the same way as the mice in these experiments, we have all experienced situations where we feel threatened and we responded instinctively. This is the first time these shelter direction neurons have been explored in mice, but the areas of the brain where they exist are also present in our brains. So it is likely that once you have noticed a fire escape, even once you’re no longer thinking about where it is, your brain is still keeping track of it – just in case you need this information in a split second when the alarm sounds.

Comment Peer Commentary

We ask other scientists from our Consortium to respond to articles with commentary from their expert perspective.

Meredith Schmehl


Duke University

This is a great article! Your opening example of the fire alarm was a great way to introduce the topic of the paper. The paper you described is focused on navigating toward shelters to avoid threats, but do you think the same results would hold when mice navigate toward rewards in a neutral environment? This would be interesting because it would suggest that these brain regions help mice find the most pleasurable things in their environment, which might have implications in goal-directed behavior.

Samuel J Walker responds:

Interesting question. With regards to whether these neurons are involved in navigating towards reward: the authors show convincingly that this pathway is necessary for orienting towards a learned shelter location, but not for innate orienting towards a neutral sound cue. However, it could be the case that these neurons are involved keeping track of and orienting towards learned locations in general — for example, a location where food is available. I guess this would require a more complex learned behavioral task to test — for example, where mice learn that a particular sound indicates the availability of food at a particular location. Then, you could inhibit this pathway and see if it is also required for orienting towards the learned food location in response to the sound.
To the best of my knowledge, this specific pathway hasn’t been investigated in the context of navigating/orienting towards reward locations. But the RSP definitely does contain neurons that carry different spatial signals, including neurons that represent reward location. So this would be very interesting to investigate!

Dori Grijseels


University of Sussex

This is a great article about an interesting study. Samuel explains really well how the scientists ruled out any alternative explanations for these cells, until they could conclude that they were truly shelter direction cells.
This study is part of a larger corpus of work looking at cells used in navigation, and seems to be almost a logical sequel to place cells, reward cells and object-vector cells. It makes me wonder whether navigational brain areas have almost a similar hierarchy as visual areas: the further you get into the system, the more specific the cells are.
In the visual system, this starts with cells recognizing light, then edges, movement, all the way to cells that specialize in concepts (e.g. the Jennifer Aniston cell). Could it be that in navigational areas it starts with a place cell, then an object-vector cell, and maybe these shelter cells are a more specialized version of an object-vector cell?

It is clear that there is still a lot we don’t know about the RSP and SC, and I’m excited to see where this research is going in the future!

Victoria Sayo Turner


University of California, San Francisco

This article does a fantastic job turning a set of neuroscience experiments into a narrative.

As Dori states, this study adds to our understanding of navigation and how our brains create spatial maps. Neurons which each fire in a specific part of a room can add up to a spatial map (place cells, grid cells). Incidentally, the 2014 Nobel Prize  was awarded for work discovering these unusual cells.

Reading this article makes me wonder if scientists and clinicians will start being able to combine these spatially sensitive neurons to understand how their firing gives humans a sense of direction. This is more than just an academic exercise for people with Alzheimer’s, whose sense of direction is often seriously altered. There is also a condition called Developmental Topographical Disorder which was discovered by the medical community in 2009, where patients can become disoriented even in their own homes. One patient describes her experience of drawing a map of her house not as a metaphorical bird’s-eye-view but as a spider crawling from room to room.