In the not so distant past being able to control one’s environment through thoughts alone was once imprisoned within the realm of science fiction. However, the advancement of technology has brought a new reality to fruition in the guise of brain-computer interfaces (BCIs).

BCIs acquire brain signals which are analyzed, enhanced, restored, and converted via artificial intelligence (AI) into readable data relayed to output devices to carry out desired functions. In effect, BCIs enable the interaction of the central nervous system with its external environment and the peripheral nervous system.

Linking brains using machines

There is much hope surrounding the potential of these cybernetic extensions whose capabilities continue to amaze whilst integrating neuroprosthesis into the human brain to restore senses, even allowing the control of other hosts using brain-to-brain interfaces.

Therefore, more applications are being explored with regards to BCIs, one of which is crowdsourcing, an online problem-solving model distributed between innovators whose results can be used to train AI. Many studies have attempted to implement crowdsourcing utilizing BCIs, to date, none have been successful.

Now, a study from researchers at the University of Helsinki engineers a BCI capable of evaluating opinions and drawing conclusions based on the brain activity of groups of people alone. The team states their technique, dubbed ‘brainsourcing’, can be used to label images or recommend preferred content by reading brainwaves, something never before achieved. The study is published in the proceedings of the ACM Conference on Human Factors in Computing Systems, 2020.

Linked brains to solve problems

Previous studies define crowdsourcing as a technique used to break up a more complex problem into smaller tasks, which are then dispersed amongst large groups of people to be solved individually. For example, people can be asked to identify whether an object is present in an image, and their responses used to train AI-based image recognition algorithms.

This is because even the most advanced image recognition systems based on AI are not yet fully automated, with the opinions of several people still required to program them. The current study investigates whether crowdsourcing can be applied to image recognition utilizing the immediate reactions of people based solely on the analysis of their brainwaves.

The current study translates participant’s EEGs using AI to determine the preferences of large groups of people from just their brain activity. A total of 30 participants were shown pictures of human faces on a computer screen, and were then instructed to label the faces in their minds based on what was portrayed in the images.

This could be whether an image portrayed a blond or dark-haired individual, or a person smiling or not smiling where participants were implicitly instructed not to voice an opinion or indicate any preference using a mouse or keyboard.

Results show the AI algorithm learned to recognize and label images based on participant’s EEGs, such as indicating when a blond person appeared onscreen. Data findings show the machine was able to interpret these mental labels directly from the EEG. The lab states they have demonstrated brainsourcing can be utilized to label simple and well-defined recognition tasks.

Future human-brain interfaces

The team surmises they have designed a BCI possessing the ability to read people’s opinions from brainwaves alone. For the future, the researchers state these BCIs would perform better when paired with lightweight wearables able to capture a larger percentage of total brain activity.

Source: University of Helsinki

Don’t miss the latest discoveries from the health innovator community: