EJ Chichilnisky — Professor of neurosurgery, ophthalmology and neuroscience at Stanford University. A leading researcher on visual perception who is engineering retinal prosthetics ('robotic eyes') to restore sight to the blind.
The gist
Andrew Huberman interviews longtime friend and colleague EJ Chichilnisky about how the brain creates vision, beginning in the retina's three layers of cells. Chichilnisky explains the roughly 20 distinct retinal ganglion cell types, each acting like a 'Photoshop filter' extracting a different feature of the visual world, and how his lab records from living human donor retinas on a 512-electrode array. The discussion turns to neuroengineering: building a smart, adaptive retinal implant that respects cell-type diversity to restore high-quality vision, and the longer-term prospect of augmenting human vision. The second half explores Chichilnisky's unusual path, including wandering through three PhD programs and years spent dancing and playing music, and his philosophy of making decisions by feeling rather than thought. They close on knowing, being, and loving oneself, the meaning of 'ease,' and the beauty of beholding a human retina.
Big reveals
- The retina contains about 20 different ganglion cell types, each like a Photoshop filter representing the whole visual scene but extracting a different feature.
- Chichilnisky's lab gets human retinas from brain-dead organ donors whose hearts still pump, and must process them in 48-hour non-stop marathons.
- A breakthrough by lab researcher Alexander Sher revealed about 15 additional poorly-understood cell types with strange spidery and blobby light-sensitivity profiles.
- The vision-restoration concept bypasses dead photoreceptors with an electronic implant that stimulates retinal ganglion cells directly.
- Nothing learned about the retina since the National Eye Institute's founding in 1968 is incorporated into existing retinal implants.
- His proposed smart implant works in three steps: record electrical activity, calibrate by stimulate-and-record, then drive cells in the correct sequence.
- Gradual, incremental sensory changes may let the adult brain develop plasticity to handle augmented vision that abrupt changes cannot.
- Chichilnisky studied math at Princeton, lived a Bohemian life of music and travel, and started three different Stanford PhD programs before finding neuroscience.
Things worth remembering
- Vision is initiated in the retina, a sheet of neural tissue at the back of the eye that converts light into electrical signals.
- The mantis shrimp can sense 60-100 variations of each color humans are essentially blind to.
- Humans have only three photoreceptor types, which is why TVs need only red, green and blue primaries to recreate any human color sensation.
- The lab's custom physics-built apparatus can record and stimulate the retina through 512 channels simultaneously at high density.
- To map cells efficiently the lab plays an unbiased flickering checkerboard 'TV snow' pattern and works backward to find each cell's preferred stimulus.
- About seven well-understood cell types make up roughly 70% of all neurons sending visual information from eye to brain.
- Even crude vision-restoration cameras must actively suppress infrared sensitivity, so augmentation is 'right around the corner.'
- The neural retina is literally two pieces of brain extruded into the eyes during development, so looking at someone's eyes is looking at their brain.
- Imaging the retina can provide a window into brain neurodegeneration such as Alzheimer's that can't be imaged through the skull.
- Chichilnisky meditates informally each morning over a meticulously made cup of coffee and maintains an Ashtanga yoga practice.