The brain can recognize objects faster than many people think

The human brain lags far behind computers in the ability to recognize faces and objects, with varying degrees of difficulty in size, color, direction, lighting conditions and other factors. However, it is still unknown how our brain processes this image.

Researchers at Boston Children's Hospital, using brain maps in patients who are about to undergo epilepsy surgery, have demonstrated for the first time that the human brain, right at the stage of treatment Initially, objects can be recognized under different conditions quickly. This finding was published in Neuron on April 30.

Visual information streams from the retina of the eye, through a system of image processing areas in the brain, eventually reach the temporal lobe. The temporal lobe, which is ultimately responsible for image recognition, gives the signal back to previous processing areas. This reciprocal exchange helps strengthen the image perception process.

'What is not entirely clear is the level of the contribution of each signal, transmitted signal or retransmission signal,' said Dr. Gabriel Kreiman of the Boston Children's Hospital Department of Eye, senior investigator of the study. rescue said. 'Some people think that if you don't have feedback, you won't have a picture. But we have shown that having an initial wave of activities can give a strong impression from the beginning. '

Although feedback from higher brain regions may appear later, and often this information is important, rapid image processing will have advantages in critical situations, wallet. For example, when faced with beasts, Kreiman said.

Previous studies based on non-interventional brain investigations, which do not use electrodes placed on the scalp as well as imaging techniques, so they obtain results on brain activity through each time interval of a few seconds - significantly slower than the actual processing speed of the brain. Moreover, these techniques only collect data from general brain regions. With the placement of electrodes directly on the brain, the Children's Hospital researchers will get the data at the correct time, receiving the signal in 1/10 second after the appearance of the imaging agent - and investigation of brain activity in specific areas.

When brain activity is recorded, images appear as 5 groups of objects, with 5 examples each.Each of these objects has its own size and orientation.(Photo: Gabriel Kreiman, PhD, Boston Children's Hospital)

Kreiman collaborated with neurosurgeon, ophthalmologist Josepth Madsen at Children's Hospital; He used to create a brain map in epilepsy patients - a process that helps ensure that surgery to remove destroyed brain tissue will not harm the vital functions of the brain.Their group implanted electrodes in the brains of 11 teenagers and young people with epilepsy (48 to 126 electrodes per patient) in areas where people began the seizures. While the electrode records brain activity, the patient is shown different images in size and rotation angle; These images are divided into 5 groups: animals, chairs, human faces, fruits, vehicles.

The results show that certain areas of the brain's cortex will recognize each group of images, the reaction is so strong and consistent that just checking signals from neurons, the researcher can Use algorithms to decide what the patient is seeing.

Kreiman and Madsen are currently expanding their research to show patients movies, because watching movies is similar to how we see images in real life. Each patient is allowed to choose a favorite movie, so Kreiman group must analyze the image content of each frame, then link those data to the patient's brain activity.

So why study the image processing process in this way? Kreiman thinks of using human visual algorithms to "teach" computers how to look like humans, from which they can help in practical applications like pointing out terrorists on machines. fly, help drivers avoid collisions with pedestrians who are difficult to see on the road, or analyze tumor samples to find malignant tumors. Another far-flung application is to design a computer - a brain that allows people who can't see it to somehow recognize the image.

For the past decade, Kreiman and Itzhak Fried, Ph.D., a medical doctor at the University of California in Los Angeles (UCLA) have been studying fish codes, the front part of the brain is related to memory, and has found Single brain cells respond consistently when one sees specific images such as Jenifer Aniston and Bill Clinton. Kreiman is currently interested in the next discovery of the relationship between image processing and memory, and will translate physiological knowledge into computer algorithms.

The current study is funded by the Epilepsy Research Fund, the Government Fund, the Klingenstein Foundation and the Children's Hospital Fund of Boston Children's Hospital.

References:
Liu H;et al.Timing, timing, timing: Fast decoding của đối tượng từ hệ thống tập tin intracranial potential.Neuron, (2009) DOI: 10.1016 / j.neuron.2009.02.025