Neuroadaptive Generative Modelling

Sep 30, 2020

Way back in 1924, Hans Berger, a German physicist, innovated the Electroencephalogram (EEG), a device that could read human brain activity based on electrical signals. This gave scientists the ability to map the human brain and detect diseases. It served as an inspiration for many other researchers and paved the way to numerous innovations in brain-computer interface (BCI) systems, a major breakthrough in the field of neuroscience.

The University of Helsinki, Finland, made a recent development in their ongoing Cognitive Brain Research. They developed a computer that can predict human thoughts and create images by monitoring the brain signals. They termed this as Neuroadaptive Generative Modelling.

Although similar brain-computer interfaces have been developed in the past, used to type letters or move a cursor, what sets this one apart is its ability to simultaneously model brain signals and the computer-generated images using artificial intelligence. Instead of depending on pre-defined stimuli categories to read brain signals, they used a generative model that can produce artificial, yet realistic, digital information from the stimuli. This model is capable of creating new instances apart from the already existing ones to adapt to the person’s neural activity.

Thirty-one participants that volunteered to be a part of this study were shown hundreds of images of different types of people and their EEG was recorded. These images were generated by a pre-trained Generative Adversarial Network (GAN). The volunteers were asked to focus on what they perceive ‘old people’ to look like. This created a task relevance which allowed the computer to update the output according to the subjects’ focus. These EEGs were fed into a neural network that estimated what type of person the subjects were thinking of and generated an image of the same. The participants found that the images generated by the BCI were almost a perfect match to what they were thinking of.

This neuroadaptive modelling has three general principles:

  1. Generate: A generative model produces realistic and meaningful digital information that is used as a sensory input
  2. Perceive: A human operator perceives and reacts naturally to the computer-generated sensory input
  3. Adapt: The task relevance is inferred from brain responses, and the computer estimates what the human perceives of the input

After several iterations, the computer is able to give a final prediction of the human response.

“The technique combines natural human responses with the computer's ability to create new information. In the experiment, the participants were only asked to look at the computer-generated images. The computer, in turn, modeled the images displayed and the human reaction toward the images by using human brain responses. From this, the computer can create an entirely new image that matches the user's intention,” said Tuukka Ruotsalo, Academy of Finland Research Fellow at the University of Helsinki, Finland and Associate Professor at the University of Copenhagen, Denmark.


According to Ruotsalo, one of the practical benefits of this technique is to develop computers that can augment human creativity. It can also be used to understand the underlying processes of human perception and psychological perspectives.

“The technique does not recognize thoughts but rather responds to the associations we have with mental categories. Thus, while we are not able to find out the identity of a specific ‘old person’ a participant was thinking of, we may gain an understanding of what they associate with old age. We, therefore, believe it may provide a new way of gaining insight into social, cognitive and emotional processes. One person's idea of an elderly person may be very different from another's. We are currently uncovering whether our technique might expose unconscious associations, for example by looking if the computer always renders old people as, say, smiling men.” said Senior Researcher Michiel Spapé.

This technique is the first of its kind to enhance human self-actualization. Further scope of BCI includes applications in biological cure, information, and entertainment.


Author : Sravani Yenamandra

Disclaimer : The pictures in the article are for illustration purposes only. Neither the author nor IEEE Student Branch , BITS Hyderabad has any claim over them.

Great! You've successfully subscribed.
Great! Next, complete checkout for full access.
Welcome back! You've successfully signed in.
Success! Your account is fully activated, you now have access to all content.