Army finds untapped brain power with EEGs
- By Mark Pomerleau
- Nov 19, 2015
When humans need more brain power, they can’t just add more RAM or attach a hard drive. Instead, they must learn to use what they were born with more effectively.
Researchers at the Army Research Lab are trying to make such efforts easier, however -- using electroencephalograms, or EEGs, to help soldiers with mental tasks and potentially gain insight into the processing power of the brain. According to neuroscientist David Hairston, fatigued soldiers produce a unique EEG pattern, and if commanders knew when their troops are flagging, they could rotate in better-rested soldiers for missions requiring a high level of alertness.
Image analysts, for example, examine thousands of images captured by unmanned aerial vehicles for patterns such as enemy locations or other key aspects of intelligence. Given the intense cognitive processing required for this task, an analyst could miss something through an instant’s inattention. If he were wearing an EEG cap, however, scientists might be able to discover the moment when something caught his attention subconsciously, but his brain’s internal communications did not elevate that information to the conscious level.
Researchers in UC San Diego constructed a brain-computer interface that speeds up the identification of mines in sonar images beneath the ocean’s surface with an algorithm that filters pictures of mines based on analyst unconscious reactions detected by the EEG. As the numbers images the analyst scans grows -- providing data on positive identifications and false positives -- that information can be used to help computers learn to make an initial evaluation of the images, cutting down the number of images the human analysts must evaluate.
Soldiers’ gut instincts can be measured as well. Researchers can flash a series of images on a screen to a soldier wearing EEG sensors. The soldier, without pressing any buttons or even providing a verbal response, will be able to quickly tag the image as threatening or not using only his brain waves.
Besides the speed advantage, such a system can use those image assessments to refine a machine learning algorithm that can learn to distinguish between threatening or non-threatening images. That technology can help pass along some of what had been considered uniquely human analysis to computers – or even help train other, less experienced soldiers.
Mark Pomerleau is a former editorial fellow with GCN and Defense Systems.