When To Flee
How your brain works out when to be afraid
At what point do you turn tail and flee? A shadow falls across you. A shadow of a large man. A close man, panting heavily. A shadow of a large, close, panting man dressed as a clown. Carrying rusty garden shears. A shadow of a large close panting man dressed as a clown carrying rusty garden shears with a cardboard box tucked under one arm. That’s dripping. Dripping a thick, red ooze.
Time to leg it, n’est pas?
Dave the Party Clown stares after you, bewildered, stows his novelty-sized balloon scissors, and sprints off once more to find the kids’ party, stressing that the strawberry ice-creams are melting fast.
Each new thing you noticed about Dave increased your sense of danger, added more evidence of an imminent threat. And when enough evidence had been gathered, when the cumulative creeping sensation got too much, your brain decided the threat was real — and escape was the best course of action. Flee!
New work by Tiago Branco and his team — including Dominic Evans and Vanessa Stempel — has now shown us exactly how the brain makes this decision. And at its heart is a brilliant move by evolution to turn a bug of biology — a failure — into a feature.
From your experience with Dave, you now know that somewhere in your brain adds up evidence of a threat, and somewhere then sets a threshold for when enough evidence means “now run like buggery”.
To find out where, we can’t sit people in a room and menace them with clowns. And even if we were allowed to, we can’t stick electrodes inside their brains to record their neurons while being clown-menaced. So Branco and team turned to that workhorse of neuroscience, the mouse.
Place a mouse in a box, with a handy little dark shelter at one end to feel safe in. Let it roam free, exploring, scouting its new home. Now make a shadow fall over it, a shadow that rapidly grows in size. A rapidly growing shadow that looks uncannily like a diving bird. Result: mouse tears back to the shelter, and cowers deep in the darkness, little heart thudding.
The clever bit is that the darkness of shadow is the evidence of an imminent threat: the darker the shadow, the greater the perceived threat. If the shadow is very dark, the mouse runs as soon as it starts to get bigger. If the shadow is very light, it can take four or five repeats of the swooping shadow for the mouse to finally decide to run for shelter.
With this swooping shadow controlling both evidence and escape, we can then look in the brain and ask: what adds up the evidence of a swooping shadow, and what drives “leg it” when that evidence hits a threshold? Branco and team already had clues about where to start: for we already know that it should somehow involve the superior colliculus.
The superior colliculus sits atop the brainstem, and is the privileged recipient of information directly from the retina — it gets to know what’s going on in the world long before you do. (Before you ask: yes, there is also an “inferior” colliculus. And as the superior colliculus does seeing, and the inferior does hearing, that gives you some idea of the pecking order in neuroscience). And the output of the superior colliculus definitely controls movement. So if we’re looking for a bit of brain that can quickly add up information from the eyes and then make you move, the superior colliculus is the superior candidate.
So Branco and team used the armoury of modern neuroscience to find out if this was more than mere guesswork.
They switched off the output neurons of the superior colliculus. Now the mouse did not react to the swooping shadow. Just carried on exploring. As though someone had completely removed its threat detector.
They recorded the activity of the output neurons. The neurons dramatically increased their activity right from the start of the swooping shadow. The activity increased far more when escape happened than when it didn’t. And the stronger the activity, the faster the mouse started to escape. As though the activity in superior colliculus was adding up evidence of a threat.
They turned on these output neurons of the superior colliculus, when there was no swooping shadow. This is the crucial test, to see if the brain can be fooled into thinking there was an imminent threat. And it could: turning on these neurons made the mouse run for shelter. Even better, the more the neurons were activated, the more likely the mouse ran to the shelter, exactly as though the neurons were signalling the imminence of danger.
All in all, pretty damn convincing that the superior colliculus is the threat detector. The obvious next question was: what is the superior colliculus talking to, saying “it’s time to run”?
Enter the Periaqueductal Gray. Good grief neuroscientists are so bad at names: it’s a bit of grey stuff around (‘peri’) an aqueduct in the brain. Let’s call it PeGy, shall we? PeGy does a lot of things. One bit of it controls taking a piss, for example. But it gets a massive input from the superior colliculus, and it controls a lot of rapid reactions.
So Branco and team dusted off their neuroscience magic tricks again, popped some new mice in the swooping-shadow box, and set to work.
They turned off PeGy’s neurons. Now the mice did see the swooping shadow, did react to the threat. But they didn’t run away. At all. They just froze in place (the other option: stop moving, as predator brains track movement). PeGy, it seems, controls running away.
They recorded PeGy’s neurons. Their activity increased right as the escape started, and not a moment before. So PeGy’s neurons don’t add up evidence, but sure seem to mean “run”.
And the clincher: they turned on PeGy’s neurons when no shadow was swooping, again pretending there was a threat. Beautifully, as they activated more and more neurons, they found an all-or-nothing response. If too few neurons were activated, the mouse never ran for shelter. But if just enough neurons were activated, the mouse always ran for shelter. Nothing in between — no ifs, buts, or maybes. When PeGy says go, you go.
That all-or-nothing response raises the crucial question: what sets that threshold between escaping or not? It’s definitely something between the colliculus and PeGy, for when Branco and team turned off just that connection it turned off escaping completely. It seems the superior colliculus activity goes straight into PeGy’s neurons to turn evidence of a threat into running away. So why doesn’t every increase in superior colliculus activity trigger PeGy’s command to run away?
Because it turns out the connections from the superior colliculus to PeGy are weak, and rubbish. The activity of one neuron is carried to its targets by little spikes of electricity. Each spike barrelling down from one colliculus neuron into PeGy — and there are a lot of these spikes — creates just a tiny blip in the PeGy neuron. And it would need tens or hundreds of these blips to make a PeGy neuron output just one spike. Each connection is weak.
If it does anything at all. On average, 20% of all spikes from the colliculus do nothing at all in PeGy. The spike gets to the end and — zip, nada, niente. No response at all in the PeGy neuron. Each connection is rubbish: spikes can just fail.
A weak, rubbish connection means that the colliculus needs to send a lot of spikes, all at the same time, to get a response out of PeGy. The weak, rubbishness is the threshold. Those light swooping shadows, each eliciting just a bit of activity from the colliculus, did not create enough activity to overcome the weak rubbishness of the connection to PeGy. But the dark, intense shadows drove a wave of intense activity out of colliculus into PeGy, overwhelmed the weak rubbishness — and escape was the only option.
This work has two particularly beautiful lessons for those who study the brain. The first is that we have a complex behaviour — escaping by running to a shelter — now drilled down to a single connection in the brain, and the properties of that specific connection. For that connection is the threshold: its weakness and unreliability set the threshold — they are the limits to overcome.
The second is that the rubbishness of the connections between neurons — the failure of spikes — is well known. Spike failure is seen all over the brain, and causes much head-scratching. Spikes cost a lot of energy to make and send. So why send them if they’re just going to fail? It seems failure may be an inevitable bug of biology, a price to pay for sending spikes.
But here this bug is a feature. Evolution has co-opted the failure of spikes to create a threshold between escaping and not escaping. Has used the failure of spikes as a way to filter out things that are not threatening, to make sure you don’t run away at every sudden noise, or every looming shadow.
Or perhaps you do. For if the failure of this connection is the threshold between staying or fleeing, then perhaps it defines how afraid you are, how much threat you can tolerate before you flee. For if in your brain this connection is working just fine, if it is not failing badly, then you’d flee at the drop of a pin. So the nervous, rejoice: your brain is perhaps just too perfect.
Want more? Follow us at The Spike