Measure1 / 8
Two mice meet.
A resident male meets an intruder in his cage. We score every second of what happens: sniffing, chasing, attack. The behaviour is the outward sign of a state we cannot see yet.
Read2 / 8
We look inside the brain.
A lens reaches down to a small nucleus in the hypothalamus, whose neurons are needed for aggression. Through the microscope each cell lights up as it fires, while the mouse behaves.
Read3 / 8
And here is what it actually looks like.
A real recording through the lens: the bright spots are single neurons, flaring as they fire.
Read4 / 8
We listen to many neurons at once.
We record the activity of many individual neurons together. Each line is one neuron, rising and falling over time, aligned to the behaviour.
Model5 / 8
Many neurons, one moving point.
Together, the activity of all these neurons at one moment can be drawn as a single point. As the mouse's state changes, the point moves.
Model6 / 8
The point falls into a valley.
Brain activity behaves like a ball on a landscape. In most places it rolls straight back to rest. Here there is a long, flat valley: the ball falls in and lingers. Where it sits along the valley is how aggressive the mouse feels, which is how the feeling outlasts the encounter. We call this valley a line attractor.
Write7 / 8
We push the state with light.
With precisely aimed light, we switch on the neurons that form the valley: the ball moves along it, and it stays, as if the feeling grew stronger. Switching on other neurons pushes it out, and it rolls straight back. The valley is real: it is how these neurons hold on to a feeling.
Map8 / 8
We find what builds the valley.
Remove the receptors for certain brain hormones from these neurons and the valley collapses into a small bowl. Activity still flickers with each attack, but the feeling no longer builds up or lasts, and the mice attack less.
