Taking Flight

Understanding how bats fly could shed light on how the human brain controls our most complex movements

Buttoning a shirt may seem like a simple task, but it requires complex, real-time coordination between our brain, eyes and fingers. And while researchers know that the brain’s motor cortex plays a central role in controlling these types of complex movements, how exactly it does this remains unclear, since a lot of what we know is based on studies of simple, more restrictive behaviors.

Now, with help from the Egyptian fruit bat, a UC Berkeley-led research team may be able to shed some light on how the brain supports these complex behaviors.

In a study published in Cell, the researchers recorded large-scale neural activity from the motor cortex of freely flying bats while tracking their wing movements in 3D. They discovered that during natural dexterous flight, the bats’ motor cortical activity is far more dynamic and varied than is apparent in simplified tasks.

“Much of what we know about motor control comes from simplified behaviors, such as an isolated, simple reaching movement,” said Michael Yartsev, professor of bioengineering and of neuroscience and the study’s principal investigator. “Our findings, however, suggest that when animals are free to perform truly natural, complex actions, the neural code may be much richer and more flexible than those paradigms reveal.”

Bats are the only mammals capable of self-powered flight, a form of locomotion that imposes exceptional demands on motor control. And unlike birds, whose wings are built around a fused and relatively rigid skeletal structure, bats fly with highly articulated, multi-joint mammalian hand-like wings. In a sense, said the researchers, they are continuously “playing piano” in midair, reshaping their wings on every wingbeat to precisely maneuver through 3D space at high speeds, sometimes exceeding 40 km/h.

During the study, researchers were surprised to discover that, despite the complexity of flight control, each wingbeat along the flight is performed with extreme accuracy. And in return, every wingbeat exhibited a unique neuronal activity pattern in the bat’s motor cortex.

“The motor cortex essentially gives a different set of instructions or corrections for each wingbeat,” explained Boaz Styr, a postdoctoral researcher and the study’s lead author. “This occurs because the bat is constantly adjusting parameters while flying, such as changing the right-left symmetry or tucking in certain fingers during the downstroke.”

To track flight at the level of the wingbeat, the researchers developed a multi-camera automated 3D markerless pose estimation system using deep learning-based neural networks. They also used tiny wireless neural recording devices to measure activity from the bats’ motor cortex. This allowed them to track bats with high temporal and spatial resolution and measure motor cortical activity from hundreds of neurons simultaneously, all while the bats flew freely inside a large flight room.

The researchers noted that these cutting-edge technologies were instrumental to the study.

“Until recently, recording from the brain of a small freely flying animal was limited to only a few neurons at once, so we could never even access the full pattern of neuronal activity,” said Styr. “With these new technologies — these very fine, very dense yet lightweight electrodes — we finally could wirelessly record many hundreds of neurons at the same time while the bat was flying.”

According to Yartsev, a better understanding of how the motor cortex functions during complex tasks could open the door to advances in both the medical and engineering fields.

“Our discovery is potentially transformative for basic neuroscience, but it also could have translational implications for humans,” he said. “This includes the design of future brain-machine interfaces that aim to restore complex natural movements in people affected by stroke and other conditions — especially movements requiring rapid feedback, many degrees of freedom and precise continuous control.”

In addition to Yartsev and Styr, co-authors of this study include Kevin K. Qi, Xing Chen and William A. Liberti III. Researcher affiliations include the departments of Bioengineering and of Neuroscience, the Helen Wills Neuroscience Institute and the Biophysics Graduate Group, all at UC Berkeley; the Howard Hughes Medical Institute; and Lawrence Berkeley National Laboratory.

This research was supported by the NIH National Institute of Neurological Disorders and Stroke and the Howard Hughes Medical Institute.