The question
Vision often sounds passive: light enters the eyes and the brain processes an image. But that isn’t how we actually use vision. We move our eyes, head, and body to acquire the information we need, and then use that information to guide our movements.
So what does vision actually look like when a primate is solving a difficult real-world problem?
How we did it
We went to northeastern Brazil and recorded wild marmosets hunting insects in their natural habitat. We analyzed 288 high-resolution videos and used AI-based markerless tracking to follow the animals, their hands, and flying insects frame-by-frame.
What we found
Catching a flying insect is far more complicated than simply seeing it and reaching toward it. The insect moves unpredictably, the marmoset is often perched on an unstable branch, and the animal must coordinate its eyes, head, body, and hands in a fraction of a second. We found that marmosets actively reorganize their behavior to meet these challenges. They adjust their body position to obtain a better view of their prey, visually track insects immediately before attacking, and coordinate their movements to intercept a rapidly moving target.
One particularly striking finding was that the marmosets did not simply launch their hands along a predetermined trajectory. Instead, their reaching movements contained multiple corrections as the attack unfolded. These adjustments occurred remarkably quickly, allowing the animals to respond to changes in the insect’s position even after the reach had begun. Two-handed attacks were especially successful when capturing flying prey, illustrating how the animals adapt their motor strategies to the demands of the task.
These observations reveal that natural vision is an ongoing interaction between perception and action. The marmoset doesn’t first build a complete picture of the world and then decide how to respond. Instead, its movements continuously change the visual information available, which in turn guides subsequent movements.
Why it matters
Neurological disease and injury are often measured with simplified tests of vision or movement, even though everyday behavior requires the two to work together continuously. After stroke, traumatic brain injury, or disorders that impair motor control, a person may be able to see an object and move a limb yet still have difficulty using visual information to guide actions through the world.
By revealing how a primate coordinates gaze, body position, and rapid hand movements during a demanding natural behavior, this work helps define the integrated visuomotor functions that clinical tests and rehabilitation ultimately need to restore. The goal is not simply to ask whether the eyes or limbs work, but whether the brain can use vision to guide effective behavior in the real world.
