At the backs of your eyeballs, on the living projector screens called retinas, your corneas display upside-down 2-D images of the world around you. With some complex mental origami, your brain transforms those flat worlds into a beautiful 3-D model of everything you see. In a new study, researchers changed how monkeys perceived 3-D optical illusions by stimulating particular clusters of neurons in their brains. The researchers think the region they tweaked is where 3-D modeling happens.

Peter Janssen of Katholieke Universiteit Leuven in Belgium and his colleagues trained two rhesus macaques to recognize 3-D shapes created by an arrangement of dots on a computer screen—somewhat like the illusions in the popular Magic Eye book series, except that the monkeys wore goggles called stereoscopes to make the images pop. Sometimes the 3-D image seemed to bend into the computer screen, as though the monkeys were peering into the mouth of a gramophone speaker (a concave image); other times the image bulged out toward the monkeys like the protruding end a traffic cone (a convex image). Janssen made it easier or more difficult to recognize the 3-D images by changing the density of dots on the screen, sharpening or blurring the images.

First, Janssen trained the monkeys to move their eyes to the left when they saw a concave image and to the right when they saw a convex image. Then he probed their brains with microelectrodes while they completed their visual tasks, searching for groups of neurons that fired in response to either a convex or a concave image. Previously, Janssen and others had found that neurons in a brain region called the inferotemporal cortex respond to complex features of images and objects, like their shape, so Janssen probed this region specifically. Sure enough, he found clusters of neurons that fired when the monkeys saw a concave image, and others that responded to a convex shape, and Janssen stimulated those clusters with a mild electric current.

The stimulation changed how quickly and accurately the monkeys reported what they saw. If the monkeys were looking at a convex image and Janssen stimulated a cluster of neurons that had a preference for convex images, the monkeys correctly moved their eyes to the right faster than they did in trials without stimulation. But if the monkeys were looking at a convex image and Janssen stimulated a cluster of concave-responding neurons, the monkeys generally took longer to report what they saw and made more mistakes. Sometimes, when the image was fuzzy enough, the electrical "whispers" persuaded the monkeys to consistently report the opposite of what was actually on the screen. Janssen thinks the electrical stimulation changed what the monkeys saw: Stimulating a group of neurons wired to recognize convex shapes made a convex image look even more convex than usual; stimulating the same group of neurons made a concave image look convex—especially if it was fuzzy to begin with. The work appears in the January 12 issue of Neuron.

Ed Connor at John Hopkins University, who studies how the brain recognizes objects but was not involved with Janssen's experiments, was impressed with the study as clear evidence that neurons in the inferotemporal cortex actively participate in 3-D modeling. "Lots of studies are descriptive. They measure neural activity and behavior and conclude that if neural behavior X is always associated with behavior Y, then X produces Y. But it's a powerful thing to go beyond that and demonstrate the causal relationship between neural activity and behavior," Connor explains. "This is a striking first example that demonstrates causality of a particular neural code in the inferotemporal cortex." Researchers have directly altered object recognition using electrodes only a few times in the past because pinpointing neurons that respond to complex features of objects is difficult. Hossein Esteky at Shahid Beheshti University of Medical Sciences in Tehran successfully used microstimulation to change how accurately monkeys distinguished faces from other images.

Janssen's ambitions are to manipulate not only the brain regions that control what monkeys see, but also the regions that determine what the monkeys report—whether they move their eyes right or left. "There must be a decision stage in the brain where it says go left or go right—neurons that look at the activity of inferotemporal cortex and make a choice," Janssen says. "Manipulating that region would allow us to chart the whole process from the purely sensory side to the decision stage."