How Bats Track Obstacles in 3d
How do our brains keep track of all the things around us? By studying bats flying around a specially adapted laboratory, avoiding obstacles as they went, Johns Hopkins scientists Melville Wohlgemuth and Ninad Kothari explained to Chris Smith how they've ventured where science doesn’t seem to have dared to tread in the past...
Mel - Well, first think about if you're driving on the highway and you are merging from lane to lane, passing cars. You're trying to figure out where you are with respect to those vehicles, and you're trying to move forward so that you don't hit those vehicles, so you have to pay attention to one vehicle move around it change your attention to another vehicle and move around that. And this is a very common thing that we do quite frequently in everyday life, but it actually has not been studied in great detail. Very few studies actually record from the brain of an animal as it moves through space and attends to different objects to avoid them as it's doing their motion. So we found the echolocating bat to be an ideal animal for this particular study because we can train them to fly and navigate around obstacles much like they would in the wild when they're hunting and foraging for insects.
Chris - So, Ninad, how did you actually do this?
Ninad - So the way echolocating bats function they produce very loud, intense high-frequency sounds, which travel into the environment from their mouth; when they hit obstacles, some of these sound energies are reflected back; the bat listens to these echoes that are returning. And this is how the bat can form a three dimensional image of the environment that it is flying in. One very surprising and very interesting thing about echolocation is that it enables bats to estimate extremely accurately - as accurate as one millimetre - the way bats do this is they record when they produce the vocalization, and they record the time when the echo comes back. The time delay gives the bats an accurate measurement of how far an object is.
Chris - But do we know where that - for want of a better phrase - that neurological radar screen is playing out in their nervous system and, when there are blips on that radar screen, how the bat is encoding those blips so it knows where the objects or obstacles are and how they are moved relative to the bat as it flies around the environment? How do you try and probe that?
Ninad - That is where we go into the methods, and so a lot of research in the past has looked at this brain region which is called the superior colliculus. Past research has shown that this midbrain region - the superior colliculus - encodes for two dimensional space; it not only makes a sensory map of the region around the animal, but it also enables the animal to direct its gaze to locations in space. And so that is why we decided, in order to understand how the brain actually represents three dimensional space, to look at this brain region.
Chris - So what did you do, put some electrodes into that region so you could eavesdrop on what the nerve cells there were saying to each other?
Ninad - Exactly. So what we did is train the bats to navigate in a room around obstacles. Now once the bats are trained to navigate in space, we put electrodes in the superior colliculus. We now have a real time view of what the neurons are telling each other and now we are faced with a problem that's how you understand what stimulus, or what echoes, the bats are experiencing as they're flying around.
Chris - Because what you're recording is a bunch of electrical impulses coming off different nerve cells and they're going to change as the bats fly. So presumably you've got this problem: you've got to marry up where the bat is relative to each of the obstacles to then look for some kind of association between the bat in a certain position and the obstacle in a certain position, and it producing a specific and characteristic change in the nerve activity, so you can tie all those things together?
Ninad - Exactly. So in the flight room we have high speed motion capture video cameras, so we put markers on the bat's head, and as the bat flies around we can now get the instantaneous three dimensional position. In addition to this we can also get exactly where in space the bats head is pointed. So once we get the head direction of the bat, we can now record the ultrasonic vocalizations of the bat. So we have an area of 32 ultrasonic microphone channels which are lined all around the room. So wherever the bat flies we can record sonar vocalisations; wherever the bat is in space we can record its position and its heading; and now it is basically going back to a mathematical model, which we call as the Echo model, and now we can start asking questions like how do the neurons in the superior colliculus create a map of three dimensional space in the bat's brain?
Chris - And Mel, do you know the answer to that? How do the neurons in the superior colliculus - based on your recordings - seem to be encoding where the bat is?
Mel - When we actually have the bat fly in three dimensions, what we find is that these neurons in the superior colliculus fire for unique locations in three dimensional space; so if the bat is at a particular orientation with respect to one of these objects, there are neurons in the superior colliculus that fire when these objects are at a particular horizontal elevation and distance location with respect to the bat?
Chris - So is it tunable? Obviously the bat's not always going to be in your room. So there's going to be a different set of stimuli, in a different set of locations. So all these relative to the bat's present location, these cells, that fire off when the bat is at, say, 2 o'clock and five metres from something there will be a bunch of cells - a population - that will go off like a machine gun when it's in that orientation, in that position?
Mel - That's exactly right. So we talk about it being responsible for keeping track of egocentric space, and like the name suggests ego-centric is centred on yourself. So this particular part of the brain the superior colliculus it's all about the relative location of objects with respect to the animal, and that's why Ninad was saying that we were reconstructing the head aim of the bat, because as you move your head around the position of one object in space is going to change with respect to you. So we always need to know where the bat is with respect to these objects as it flies around in order to reconstruct the three dimensional tuning of these neurons in the superior colliculus. So it doesn't really matter what the object is, it's just whether an object is at a particular distance horizontal and elevation location with respect to the animal.
Chris - Ninad, what does the bat do with that tuned signal from a superior colliculus? How does that in turn translate into better attention for that target on the part of the animal?
Ninad - That's a very good question. So the superior colliculus is a very important part in the attention network. It sends out projections to the frontal cortex and it also sends projections to the motor nuclei, which actually drive behaviour. So now you can consider the superior colliculus computing this three dimensional information. It can send this to the cortex for further planning behaviour; and the sort of continuous feedback between sensory input and the superior colliculus back into the cortex can now help the animal plan its next movement, or the next location where in space it needs to pay attention to...