Reading Your Mind
Brain-computer interface (BCI) technologies enable people to control external devices using just their thoughts. There are two main kinds of BCIs: invasive BCIs like Neuralink, which involve the surgical implantation of sensors in contact with the brain, and non-invasive BCIs, which involve placing sensors on the scalp of the subject, in a ‘wear-and-go’ setup. Non-invasive BCIs obviously carry less risk, but because the sensors are further from the brain sources generating the signals, they come at the cost of trickier signals to decode.
Non-invasive BCI developed through the project BrainApp for controlling a motorised bed
Non-invasive BCIs have the potential to become revolutionary technologies in healthcare and rehabilitation. Researchers are working towards a future where people with severe mobility impairment can be able to control assistive devices such as wheelchairs, hospital beds, thought-to-speech devices, and personal care technologies using their thoughts. BCIs could also be integrated into stroke rehabilitation, enabling subjects to receive feedback on their own brain activity whilst carrying out therapeutic exercises.
BCIs can potentially impact other industries. Systems for monitoring alertness could be used in safety applications for long-haul drivers and pilots, indicating when a break is needed. In a completely different application, BCIs could enable teams to carry out collaborative product design in virtual environments in a completely novel way. Finally, substantial efforts have been made to bring BCIs into the gaming industry because they present a golden opportunity for more intuitive, faster control of gameplay.
The mainstream adoption of BCI technologies, however, presents a potential ethical minefield. One of the key concerns is around the processing and use of the brain-signal data being recorded. In the future, it could be quite straightforward to record brain signals to control an external device, but then use additional processing to extract personal information about the user's mood, mental health, and attention level. This data could be exploited, for example, to carry out direct marketing, to manipulate how the subject feels or to adapt gameplay in a videogame to keep the subject interested beyond a healthy timespan. This kind of manipulation and exploitation of data has already happened in the past and still happens with data harvested from applications we use - imagine if those applications had direct access to our brain data, not just our likes and follows on social media. Just like all technologies, BCIs can be misused, and as they enter the mainstream one day in the future, it is essential that people are educated about how they operate, so they can make informed decisions and elicit suitable pressure for their regulation. Legislators also need to keep abreast of these technologies, so that they are always one step ahead in keeping them as safe and beneficial to society as possible.
At the Department of Systems and Control Engineering and the Centre for Biomedical Cybernetics, non-invasive brain-computer interfaces are heavily researched. Although these technologies are brimming with potential, the nature of the signals recorded makes them challenging to work with. As the brain waves travel from the source in the brain, through the blood-brain barrier, the skull, and the skin of the scalp, they accumulate substantial amounts of noise. Furthermore, as one would expect, brain signals are already complex and challenging to decode, but they can also be affected by many variables including movement of the subject, caffeine consumption, prior sleep, and posture, making the development of reliable BCI systems a significant research problem. However, this hasn’t stopped the Department and the Centre from developing a variety of research-grade BCIs over the years for controlling devices including motorized beds, a music player, a web browser and a television, with a BCI for controlling a wheelchair currently being developed.
BrainApp - controlling a motorised bed application using brain signals
SAT : a Switch-And-Train framework for real-time training of SSVEP-based BCIs
Idle state detection with an autoregressive multiple model probabilistic framework in SSVEP-based brain-computer interfaces
Natasha is a post-doctoral researcher with the Centre for Biomedical Cybernetics and is currently working on the project 'User-Intuitive Continuous Brain Control of a Smart Wheelchair (BrainCon)' which focuses on the use of brain-controlled wheelchairs.