Brain implant beams images into monkey brains

A brain electrode matrix can make monkeys see letters
08 December 2020

Interview with 

Pieter Roelfsema, Netherlands Institute for Neuroscience


Human eye


Artificial vision has made a great leap forward this week: the shapes of letters were directly transmitted into the brains of monkeys, that were able to “see” and respond to them without actually looking at them for real. It means we’re a step closer to developing ways to restore vision for some of the millions of people worldwide who’ve lost their sight. The team behind the breakthrough, who are based in the Netherlands, implanted a matrix of tiny electrodes into the region of the brain that decodes inputs from the eyes. Passing small currents into those electrodes turns on the nerve cells nearby, in the same way that signals from the retina in the eye would. This creates the impression of seeing spots of light - resembling pixels - each of which corresponds to one active electrode. And by varying the pattern of electrodes that are turned on, you can change the pattern of lit spots the animals see. Katie Haylor heard from Pieter Roelfsema how they’re doing it.

Pieter - The implant is composed of one thousand electrodes, basically tiny wires. Through one of them we pass electricity, electric current, and then you activate the nerve cells that are in the direct vicinity of the electrodes you're passing current through. If you stimulate one electrode, a person is going to see a dot of light. If you then stimulate a pattern of electrodes, you can create a pattern of these dots of lights and that can then become a recognisable pattern like a letter.

Katie - And tell me a bit more about the experiments that you've done. ‘Cause this is in macaque monkeys.

Pieter - After we implanted the 1000 electrodes, in the first experiments, we stimulated one electrode at a time and we basically found that for each electrode, the monkey detected it, and we trained the animal to make an eye movement to the location where he saw something. There is a very good correspondence to where we expected these artificial perceptions to emerge and where the monkey made his eye movements. So if there was an electrode where we expected that the perception would be here, the monkey made an eye movement there. So we were able to create perceptions at many different positions. And then the trick is you can basically work with it like a matrix works along the highway. So if you turn on one bulb, you'll see a dot of light, but you can also create patterns. And that's what we did. So we created patterns by stimulating multiple electrodes at a time. For instance, in the shape of a particular letter, the letter A, and then we found that the monkey was able to recognise those letters.

Katie - Do you know what these macaques actually see? Is it sort of all one colour of light?

Pieter - Only indirectly because you can't ask the monkey, "Hey, what did you see?" So therefore we have to do some tricks. So we train the animals to recognise letters and say for the letter A, they will make an eye movement to the right. For the letter B they will make an eye movement to the above and so on and so forth. Then at some point we replaced the letters by directly activating nerve cells in the visual cortex. And we're really excited that it worked. Then we know at least that the animals recognise these letters, but we can't ask them, "what do they look like?" Now one of our collaborators on the publication is Eduardo Fernandes. And last year he also implanted a human with the same type of electrodes. And of course, then you can ask, what does it look like? She reported most often, it looked like yellow-white,small dot of light, but sometimes it even has a small shape, like a like small edge.

Katie - Do you need a sort of certain number of electrodes to make a pixel?

Pieter - Yeah. So every electrode becomes a pixel. And if you stimulate it, the person - can also be a person who has been blind for several years - will see a single dot of light, like a pixel. And, um, the number of pixels you create or can create is then equal to the number of electrodes that you can stimulate. So in this particular experiment, we had 1000 electrodes. So we could create 1000 pixels.

Katie - Do you think we can get to the stage where you can get useful discernible information for human beings?

Pieter - Basically a question of how many pixels you can produce. The more pixels the better. There have been experiments in humans with only 60 pixels, for instance, in the eye. And that gives rise to slight vision. Of course, it's not terrific. If you go to one thousand, you can do a recognition of letters, maybe a very simple reading task where you see only a few letters at any one time, you can recognise simple objects. If you then go to 10,000, you could probably also navigate in a room, maybe even outside. So the more pixels you can produce, the better it will be. Now the resolution of the human eye is 1 million. We're far from that.

Katie - If the number of electrodes relates to the number of pixels, is there a point at which it kind of stops being practical?

Pieter - Yes, definitely. So 1000 is feasible and we think 10,000 maybe more is still feasible. It of course depends on the size of the electrodes. The electrodes that we're envisioning for the future are really, really tiny, and you can pack a lot of them on a single wire. And then it becomes feasible to get thousands of electrodes into the brain. I think that should be possible in the near future.




Add a comment