An Awake Patient Did Math While Surgeons Removed His Brain Tumor

Andrew Huberman, a neuroscientist and podcast host, stood close enough to watch his childhood friend operate, and to talk with the man on the table. The patient's skull had been removed, and a tumor was coming out of the front part of his brain piece by piece, onto a plate for testing. The patient was awake for all of it.

The surgeon was Eddie Chang, chair of neurosurgery at the University of California, San Francisco, a bioengineer and member of the National Academy who studies how the brain produces speech. He has known Huberman since they were seven, and he still calls him Andy. He had invited Huberman because the patient was devoted to jujitsu and wanted him there.

This is the central problem of removing a tumor that sits inside parts of the brain needed for speech and movement. The tumor does not look different from healthy tissue, but it feels different to the instruments of a highly trained surgeon. If the surgeon takes too little, a tumor that can kill remains. If he takes a quarter centimeter too much in the wrong place, the patient can lose the use of his hand completely.

So Chang mapped as he worked. He touched a spot with a small electrical pulse, then waited to learn what happened. The patient reported what he felt, in his toe, in his face, in his hand. At other moments he solved math problems, because he had agreed to let the team study which areas help with mathematics while the brain was exposed.

The same mapping idea drives Chang's other work, on people with locked-in syndrome, who are fully paralyzed but conscious and could communicate only by blinking. His lab found early signals linked to getting ready to speak, as well as signals that drive speaking itself, and used bioengineering and artificial intelligence to decode them.

One patient, named Pancho, could only blink before. His first words through the system were "Hello" and "Doing well." Speech is an exhale shaped by the throat and tongue, and the system reads the brain instructions that would normally go to those muscles. The artificial intelligence learns his patterns, and patients now wheel around with a tablet beside them showing a face that speaks. The system works only after training on a person's attempts to speak, and it cannot read random thoughts.

Artificial intelligence is essential, Chang told Huberman, because surgeons cannot place unlimited sensors in the brain without destroying the tissue they want to save. The software combines a few signals to work out which word the brain was trying to say.

The man on the table is alive and thriving. The jujitsu patient, age 60, is now training most days of the week. Doctors removed the life-threatening tumor, and he is doing well. He has invited Huberman to come to jujitsu class.