By careful targeting of specific neurons mice have been fooled into thinking they are seeing a pattern. The work has some exciting potential medical applications, but might be the first stirrings of a liar's paradise, a real-life version of video deepfakes, where it becomes impossible to tell whether what you see, hear or touch is real or simulated.
Optogenetics involves making brain cells produce light-sensitive proteins and stimulating them with pulses of light. Using infrared lasers Professor Karl Deisseroth of Stanford University has made it possible to target specific neurons far more precisely than with the traditional blue-green light.
Deisseroth identified the neurons activated in the visual cortex when mice were shown specific images, in this case, parallel horizontal or vertical lines onscreen. The mice were trained to lick a tube when the lines were in one orientation, and not lick for the other.
In Science, Deisseroth reports that when as few as 20 neurons were stimulated optogenetically to match the responses produced by the desired orientation, they induced activity in the brain cells around them. This, in turn, made the mice respond as if they were seeing the real thing.
Three years ago another team used the same technique to induce visions in mice, but back then the work was less targeted. They knew they were making the mice see something that wasn't there, but couldn't be sure what it looked like.
Deisseroth can't tell quite how convincing his hallucination is. Perhaps it still appears blurry or otherwise wrong to the mice. Nevertheless, the hallucinations he has induced are similar enough to the real thing that the mice recognize them and know to lick when appropriate, suggesting the match can't be too bad. "Not only is the animal doing the same thing, but the brain is, too," Deisseroth said in a statement. "So we know we're either recreating the natural perception or creating something a whole lot like it."
Such a simple pattern is obviously easier to replicate than a complex scene, but it's early days for this work.
The frightening aspect of this work is that it may mark the start of a march to where it becomes possible to stimulate neurons to make someone believe they have witnessed something important. Wars have been fought when governments have produced faked evidence of another nation's crimes. Imagine the consequences if they could make their citizens believe they witnessed these things first hand.
Before we have to grapple with those dangers, however, the work could bring major benefits, damping down psychotic hallucinations, for example, rather than inducing them. Another team conducting similar work have expressed the belief they can stimulate neurons that decay in dementia patients, possibly reversing some of the effects.