How the Brain Works, Curing Blindness & How to Navigate a Career Path | Dr. E.J. Chichilnisky

Andrew Huberman introduces Dr. E.J. Chichilnisky as a professor of neurosurgery, ophthalmology and neuroscience at Stanford whose work joins two goals. The first is to understand how visual perception occurs. The second is to use that understanding to design neural prostheses, which Huberman calls "literally robotic eyes that can allow blind people to see once again." Huberman says the conversation will also cover devices that might let human brains perceive things typical brains cannot, and enhance memory and cognition. He presents this as the direction neuroscience is going. He frames all of these as topics the conversation will reach, not results described in this section.

The introduction also previews a second theme: how to choose a professional and personal path. Huberman says Chichilnisky did not follow a straight line. He moved through three different graduate programs and took several years off from school to dance. Huberman says this wandering helped Chichilnisky decide which problems in neuroscience and medicine to work on. Huberman adds that the podcast is separate from his teaching and research roles at Stanford.

Huberman notes that the two are longtime friends. He describes Chichilnisky's work with the image of an astronaut: someone who goes "places no one else has been willing to go before" and builds new technologies to get there. He then asks for a short account of how the brain works, as common ground for the audience.

I don't have a one-to-five minute version of how the brain works, but I can tell you how I think vision is initiated in the brain.

Chichilnisky narrows the question instead of answering it as asked. Vision begins in the retina, a sheet of neural tissue at the rear of the eye. The retina captures incoming light, turns it into electrical signals, processes and transforms those signals, and sends the result to the brain. He says the brain then "somehow" assembles these patterns of activity into visual experience. That experience includes responding to approaching objects, circadian rhythms that govern sleep and behavior, identifying prey, avoiding predators and appreciating beauty. He openly states the limit of current knowledge: he would love to understand how that assembly works, but "At the moment, I don't." His lab's aim is a complete understanding of how vision begins in the retina, and of how to restore it in people who have lost sight.

Huberman pushes back. He agrees that visual perception strongly shapes the quality of life. But he asks why anyone would study a thin layer of neurons at the back of the eye instead of brain centers such as the visual cortex or visual thalamus.

Chichilnisky grants the premise behind the challenge. Understanding the retina alone will not explain vision, and without the visual cortex and thalamus a person will not see. But without a retina a person "won't even have a chance to see." His reason for choosing the retina is that it can be understood completely enough to rebuild. He thinks it is possible within his lifetime to understand this piece of the nervous system well enough to build it, replace it and restore its function. He thinks that goal is "farther off" and "quite a bit harder" for the central regions of the brain.

I find satisfaction in really understanding something so well that I can write down in a mathematical formula what it's doing

He sets a high bar for understanding: a mathematical description of the circuit, hypotheses tested "up and down," and devices engineered to replace the circuit's function when it is lost. He does not dismiss other approaches. He says more exploratory work in the visual cortex, thalamus and other areas plays a "really fundamental role," because retinal signals lead nowhere unless those areas combine them to govern perception and behavior.

Huberman describes the retina as three layers of cells lining the back of the eye "like a pie crust." The lens focuses incoming light onto it, and contact lenses or spectacles help when the lens does not. He suggests that the retina may be the best-understood piece of the brain, partly because of Chichilnisky's work. Chichilnisky answers that this is "a solid argument."

First layer: photoreceptors. These are highly specialized cells that "essentially don't exist anywhere else in the brain." They turn light energy into electrical signals. Chichilnisky calls them "pixel detectors" because each captures light from a particular location. They need a lot of maintenance and "die relatively easily," and he says this is what causes some forms of blindness. Second layer: processing. It receives input from the photoreceptors and adjusts, changes, mixes, matches and compares signals across neurons. It consists of "dozens of distinct cell types" that extract features from the elementary pixels. He says these are complex operations "that we're still trying to understand." Third layer: retinal ganglion cells. These are the "messengers" that send the retina's signals to the brain. There are about 20 types in humans. Each type picks out different features, sends "interesting stuff" to the brain and tries to leave out the uninteresting.

You can sort of think of them as Photoshop filters, each cell type in the retina.

The key point of the analogy is coverage. Each of the roughly 20 ganglion cell types represents the full scene, the entire visual world, but each picks out different features from it.

When Huberman asks for examples, Chichilnisky lists a few. Some cells signal fine spatial detail, "tiny little points of light, almost." Some signal movement. Some carry information about different wavelengths captured by the photoreceptors, which gives rise to color sensations. He adds a qualification: there are "probably more things" encoded by the 20 types "that we don't fully understand."

It's not one picture that comes out of the retina and gets sent to the brain. No. It's 20 different pictures

In his picture, one image has edges highlighted, another has colors highlighted, and another encodes movement. These streams go to many different targets in the brain. The brain then produces a single cohesive sense of the visual world, which Chichilnisky calls "the remarkable feature that we really don't understand." The account of the retina is therefore detailed, but the step that combines these signals into experience remains unexplained.

For listeners who don't use Photoshop, Huberman offers a different image: parallel movies. One shows the outlines of objects and people, one shows moving blobs, and one shows only color. The brain combines them so we can recognize faces, cars and objects. Chichilnisky accepts this: "That's exactly how I think about it."

Chichilnisky treats the retina as an example, "we think," of how all sensory systems work. A specialized cell type first extracts physical features from the world, and neural circuits then use that information in different ways. In hearing, specialized cells transduce sound energy into neural signals, and later stages extract features. Huberman suggests frequency and direction, and Chichilnisky adds movement and loudness.

He then states a broader view of the brain, presented as a philosophical approach: there is the sensory world and there are the actions we take, and "there's almost nothing else that we really know other than those two things." On this view, the brain's job is to take sensory input and finally turn it into action.

This also explains his focus on human vision. Many researchers study rodents, which he calls interesting animals. But rodents identify objects largely by smell and navigate largely with their whiskers. Humans do neither. Huberman recognizes him by sight, not by smell, when he walks into a room. Because humans rely so heavily on vision, he calls it "a really fundamental aspect of who we are as biological creatures."