It may not be able to catch mice yet, but software developed in the US can perceive moving images in much the same way a cat鈥檚 brain does. The researchers hope the work will one day lead to implants that make it possible for people to see without an optic nerve.
Researchers at the , San Francisco, were motivated by the fact that, until now, models of the way brains respond to visual input used simple images like dots, bars and grids. They are typically unable to cope with the complex scenes a human would usually see.
To try and develop a more sophisticated model, the team recorded the responses of 49 individual neurons in a part of a cat鈥檚 brain called the (LGN). The LGN receives and processes visual information from the retina, via the optic nerve, before sending it on to the cerebral cortex.
Advertisement
Using a mixture of simple stimuli, like dots and bars, and building up to more complex moving artificial scenes, the team tried to work out the basics of the LGN鈥檚 response to visual features.
Call in the catcam
The data made it possible to build a software model of the LGN that can approximate how the neurons would respond to real scenes. The model was tested against scenes recorded from a 鈥渃atcam鈥� camera attached to a cat鈥檚 head.
鈥淲e chose the catcam because it was the most natural stimulus we could think of, the closest to what a cat would see when walking around,鈥� Matteo Carandini told 麻豆传媒. Because the catcam footage lacked elements moving independently from the rest of the scene, the researchers also used a scene from Disney鈥檚 animated film Tarzan.
The model鈥檚 predictions proved to be 80% accurate when shown artificial scenes, but this figure fell to 60% with the natural scenes or the Tarzan movie.
鈥淭his is still impressive, but shows a way to go,鈥� says , a neuroscientist at Imperial College London, UK.
Predicting LGN activity from moving images at all is significant, he says. 鈥淸The researchers] recognise that the perceptual world is not a single frame at a time but a constant stream of data,鈥� he told 麻豆传媒.
Bionic eyes
The ultimate goal of the research, still years distant, is to develop an implant that uses visual data to directly stimulate the LGN of blind people whose optic nerve or retina has degenerated from lack of use.
鈥淔or these people, a prosthesis in the eye doesn鈥檛 help,鈥� Carandini explains. Only people who have recently become blind can benefit from such implants 鈥� currently being tested in humans 鈥� that stimulate the retina or optic nerve,.
Work on monkeys last year showed it is possible to stimulate the LGN using electrodes to alter their vision, something previously thought impossible. Software models like that developed by Carandini and colleagues would be vital for an implant to stimulate the right neurons to create a mental impression of vision.
Journal reference:
