Your new apartment is only a few blocks from the bus stop, but today you’re running late and you see a bus pass you by. You break into a full sprint. Your goal is to get to the bus as quickly as possible, then stop in front of the door (which is never in the same place along the curb) to get in before it closes. New MIT research in mice has shown that the mammalian brain is perfectly wired to implement the principles of computation to stop quickly and accurately enough.
One might think that stopping at a target after running on flat ground would be as simple as a reflex, but catching a bus or running to a visually marked landmark to earn a water reward (as mice did) is a learned, visually guided, goal-directed feat. In such tasks, which are of great interest in the lab of senior author Mriganka Sur, the Newton Professor of Neuroscience at MIT’s Pickover Institute for Learning and Memory, the crucial decision to switch from one behavior (running) to another (stopping) is made by the cerebral cortex, where the brain integrates learned rules of life with sensory information to guide plans and actions.
“The target is where the cortex comes in,” said Suhr, a professor in MIT’s Department of Brain and Cognitive Sciences. “Where do I have to stop to reach this goal – get on the bus.”
And here it also gets complicated. Mathematical models of behavior developed by postdoctoral researcher and lead study author Eli Adam predicted that the stop signal, which goes directly from the M2 area of the cerebral cortex to the brainstem regions that actually control the legs, would be processed too slowly.
“You have an M2 that sends a stop signal, but when you model it and do the math, you find that the signal itself is not going to be fast enough to make the animal stop in time,” said Adam, whose work with appears in the magazine Cell reports.
So how does the brain speed up the process? Adam, Suhr and co-author Taylor Jones found that M2 sends a signal to an intermediate region called the subthalamic nucleus (STN), which then sends two signals down two separate pathways that converge again in the brainstem. why? Because the difference between these two signals, one inhibitory and one excitatory, arriving one behind the other, turns the problem from integration, which is the relatively slow addition of input, to differentiation, which is the direct recognition of change. The shift in computation implements the stop signal much faster.
Adam’s model, which uses engineering systems and control theory, accurately predicted the speed needed to stop correctly and that differentiation would be required to achieve it, but it took a series of anatomical studies and experimental manipulations to confirm the model’s predictions.
First, Adam confirmed that M2 did indeed cause a spike in neural activity only when the mice had to achieve their trained goal of stopping at the landmark. It has also shown that it sends resultant signals to the STN. Other stops did not use this path for other reasons. Moreover, artificial activation of the M2-STN pathway caused mice to stop, and artificial inhibition caused mice to cross the landmark somewhat more often.
The STN, of course, then had to signal the brainstem – specifically the peduncle nucleus (PPN) in the locomotor region of the mesencephalon. But when the scientists looked at neural activity starting in M2 and then quickly leading to the PPN, they saw that different types of cells in the PPN responded at different times. In particular, excitatory cells were active before the stop, and their activity reflected the animal’s speed during the stops. Then, looking at the STN, they saw two kinds of bursts of activity around the stops—one slightly slower than the other—that were transmitted either directly to the PPN through excitation or indirectly through the substantia nigra reticularis (SNr) through inhibition. The end result of the interaction of these signals in the PPN was inhibition, which was exacerbated by excitation. This sudden change could be quickly detected by differentiation to implement a stop.
“A braking surge followed by excitation can create a sharp [change of] signal,” Sur said.
The study is in line with other recent papers. Working with Pickover Institute researcher Emery N. Brown, Adam recently created a new model of how deep brain stimulation in the STN rapidly corrects the motor problems that result from Parkinson’s disease. And last year, members of Sura’s lab, including Adam, published a study showing how the cerebral cortex overrides more deeply rooted brain reflexes in visually guided motor tasks. Together, such studies contribute to understanding how the cerebral cortex can consciously control instinctually related motor behaviors, and how important deep regions such as the STN are for the rapid implementation of goal-directed behaviors. A recent review of the laboratory explains this.
Adam suggested that the “hyperdirect pathway” connecting the cerebral cortex to the STN may have a broader role than rapid action termination, potentially extending beyond motor control to other brain functions such as pauses and switches in thought or mood.
The study was funded by the JPB Foundation, the National Institutes of Health, and the Simons Foundation’s Autism Research Initiative.