As we move through the world, our brains do more than plan our actions - they anticipate potential disruptions. A new Western study published in the high impact journal Nature reveals that we rely on sensory expectations to get prepared for unexpected disturbances, helping us react faster and more accurately.
This transformative discovery deepens our understanding of the body's motor system and could lead to new approaches in stroke and injury rehabilitation. It also holds promise for the development of brain-computer interfaces, like those currently pioneered by Neuralink and Synchron based on decades of neuroscience research, that could one day harness the brain's remarkable ability to expect the unexpected.
"We've discovered that motor circuits don't passively wait for sensory signals, but proactively configure themselves to meet new challenges," said Andrew Pruszynski, Canada Research Chair in Sensorimotor Neuroscience and Schulich School of Medicine & Dentistry professor.
For the study, the Western research team, including senior author Pruszynski and lead author Jonathan A. Michaels, used a robotic device to nudge participants' arms in different directions, sometimes giving them a cue about which way the push was most likely to come. Participants adjusted their movements based on these probabilities and their muscles responded more effectively when the disturbance matched the brain's expectation.
"This study, which took years of effort, highlights how much we still have to learn about how the brain works - and it underscores the importance of basic research in making such discoveries," said Michaels, former Banting and BrainsCAN postdoctoral fellow at Western and now an assistant professor in the Faculty of Health at York University.
The research collaborators also included Western Research Chair for Motor Control and Computational Neuroscience Jörn Diedrichsen and psychology professor Paul Gribble, Pruszynski's co-principal investigators in Western's Sensorimotor Superlab.
To uncover the underlying neural mechanism, the researchers recorded thousands of neurons in monkeys performing a similar task. The data revealed that motor circuits don't wait passively but instead adopt a preparatory state that actively anticipates each possible disturbance and links it to the appropriate response that would be needed.
Computer models of the arm trained under similar conditions developed the same predictive strategies, revealing how using expectations naturally improves arm control.
"We are 100 per cent committed to sharing this data as broadly and openly as possible. It will be mined, and already has been mined other researchers, because it is unique and one of the most thorough datasets that exists today in the pursuit of understanding how the brain controls movements," said Pruszynski, associate director of the Western Institute for Neuroscience. "Our hope is that sharing this data will lead to additional discoveries related to how motor activity is organized in the brain and, in the long-term, instigate practical advances like new strategies for stroke rehabilitation or better algorithms for driving brain-computer interfaces."
While Pruszynski predominantly studies upper body motor skills like reaching and grasping in his lab, all human behaviour is based on the coordination of large populations of neurons making this dataset all the more valuable to other researchers.
"It is now more evident than ever that you can't understand the brain fully by studying one neuron at a time," said Pruszynski. "Much of how the brain works depends on the coordinated patterns of how neurons change and react to other neurons."
For the study, the researchers used Neuropixels for the recording, an advanced technology - now being used in human patients - that features thousands of recording sites arranged in two rows on a thin, 4.5-cm long probe. The results represent a quantum leap in knowledge advancement even from 15 years ago when Pruszynski was completing his PhD.
"When I was in grad school, I was recording one neuron at a time and during my entire PhD, over seven years, I recorded about 1,000 neurons in total," said Pruszynski. "Now we routinely get 1,000 neurons in just a day or two of recording. This productivity, with respect to the yield, allows us to ask questions that are just completely different and that you really couldn't have asked before."
