The 'brain reading' device can predict what people see

A computer program can pair brain activity with visual images and predict what people are seeing. The work raises the possibility that one day computers can " read " the human brain to recreate digital memories, dreams and fantasies.

Previous attempts to decode images in this way have only isolated simple information about the image, such as physical orientation, and could not identify the images that the test participants looked at. see for the first time.

According to the team member, Kendrick Kay of the University of California, Berkeley: 'Our approach has overcome this drawback, and we have proven that we can identify imaginary imagery. . '

The new computer model is described in the latest edition of Nature magazine.

The house predicted the master

The researchers used functional magnetic resonance imaging (fMRI) to measure activity in the visual cortex of testers when they looked at photos of animals, food, people and Other common objects.

The fMRI method is a relatively new way of monitoring changes in oxygen levels in the brain, which is closely related to neural activity. The data collected will be used to ' teach ' a computer program to relate certain blood samples to specific images.
Testers were then asked to look at a second series of photos, which they had never seen.

The program is built to apply what it is new or from the previous pairing and visualize what is in the second set. For a group of 120 photos, this program recognizes 90% of what they are seeing. When the set is extended to 1000, the accuracy is about 80%.

Resonance imaging from the human brain.A new study shows that a computer program can pair brain activity with visual images and even predict what people see.(Photo: National Geographic)

Brain reading works

Scientists say their work opens the door to brain-reading devices - similar to those envisioned by science fiction writers - that can display visual experiences an inner sense of a person on the screen.

Before such a device was born, scientists had to answer important questions about dreams, memories and imagination. Kay said: 'Perhaps the content of imagination does not show in the way other senses perceive reality. In this case, we will have to study how imaginations appear and design the corresponding computer programs. '

Technology also needs to be improved.

Critics of fMRI suggest that this method does not measure brain activity directly. As a result, it lacks recorded data directly from brain cells.

A small step forward

Frank Tong, a neuroscientist at Vanderbilt University in Tennessee, said he was quite surprised because the group's method was effective for the above reason.

'Most people assume that fMRI is a rather' rudimentary 'method, but the data obtained contains a surprising amount of information, enough to predict hundreds of thousands of photos without success beyond randomness. of course. '

The work is also based on a number of other groundbreaking studies, including a scientific work that appeared last year in Current Biology magazine, that decoded the intended success of testers to 70% based on reading fMRI fruit.

Robert Dougherty, a neuroscientist at Stanford University, said inventing a brain-reading device is possible, but he reminds that the software is just a small step toward that goal.'Their software does not work backwards, it cannot produce a unified image from measured brain activity. However, in relation to strong assumptions about natural image analysis, a more complex software can produce predictive images of an object. '

Brain reading devices may be useful in studying phenomena that are difficult to observe by conventional methods, the difference in human perception.

But the team added that such a device could be abused for bad purposes. The personal and moral issues related to the brain reading device will go hand in hand with the issues surrounding the human gene arrangement. In this case, it is necessary to ensure that personal interests will not be violated. According to the group's statement, 'the authors assume that no one is forced to read the brain but not fully informed.'