Decoding the Brain: Can AI help predict human behaviour?
What if it were possible to read the brain like a book? Paolo Burelli and his colleagues at the IT University’s brAIn Lab work at the cutting edge of digital technology and neuroscience. On 5 November, Paolo Burelli will present their research at Digital Tech Summit, in a talk titled “Decoding the Brain: How AI Unlocks Human Behavior.”
Paolo BurelliEventsResearchartificial intelligence
Written 27 October, 2025 09:28 by Theis Duelund Jensen
As head of the brAIn Lab, Paolo Burelli and his colleagues are exploring how artificial intelligence can help decode human cognition. Their work spans neuroscience, machine learning, and user experience, with applications ranging from neuromarketing to long-term visions of synthetic brain repair.
“We’re using machine learning as a lens,” says Paolo Burelli. “A magnifying glass, a radar – something that lets us disentangle the complexity of brain signals and understand how people respond to digital media.”
Mapping the human mind
The brAIn Lab’s research begins with the body. Using biometric sensors – heart rate monitors, sweat detectors, EEG headsets – the team tracks how people react to stimuli like music videos, advertisements, or video games. These physiological responses are then analysed to uncover patterns of stress, attention, and emotional engagement. “There’s strong evidence that these tools can give real-time feedback on how people are reacting,” Burelli explains. “We can measure impact – not just what people say, but what their bodies reveal.”
But the real frontier lies in the brain itself. While neuroscience has made great strides in understanding individual neurons, the leap from cellular mechanics to cognition remains vast. Burelli compares it to the gap between quantum physics and astrophysics: two domains of knowledge that operate on entirely different scales.
“We know how a single neuron fires,” he says. “But how that leads to intelligence, emotion, or decision-making remains a mystery.”
This is where machine learning enters the picture.
AI as a cognitive lens
Traditional neuroscience relies on highly controlled experiments: dots on screens, repetitive stimuli, and isolated variables. These setups are useful for extracting clean data, but they don’t reflect how people behave in the real world. Paolo Burelli’s team wants to change that. “Machine learning unlocks the ability to work in naturalistic conditions,” he says. “Watching a movie, playing a game – these are complex, messy environments. But with enough data and the right models, we can start to make sense of them.”
If we can understand how a region of the brain processes information, we might one day be able to build a chip that mimics it.
Paolo Burelli
Using deep learning techniques, the brAIn Lab can process vast amounts of biometric and contextual data simultaneously. These models don’t just filter noise – they identify patterns across time and space, linking what a person sees with how their brain responds.
“The brain isn’t a muscle,” says Paolo Burelli. “It’s a network of entangled signals, shaped by context, memory, and emotion. Deep learning helps us separate what’s unique to an individual from what’s universal.”
From neuromarketing to brain repair
The applications of this research range from immediate to speculative. In the short term, the team at ITU is working on neuromarketing and UX research – helping companies understand how users respond to design changes, product updates, or advertising campaigns. “We can track how people react to different features in real time,” he says. “That’s incredibly valuable for user research and product development.”
In the long term, the possibilities are more radical. One vision involves simulating damaged parts of the brain – such as the visual cortex – with AI-driven models that replicate lost functionality.
“We’re not there yet,” the researcher says. “But if we can understand how a region of the brain processes information, we might one day be able to build a chip that mimics it.”
This kind of synthetic cognition could revolutionise treatment for brain injuries, strokes, or degenerative diseases. But it also raises profound ethical questions.
Privacy in the age of biometrics
One of the most immediate ethical challenges in Paolo Burelli’s work is privacy. The data collected – biomarkers, brain signals, emotional responses – is deeply personal. Ensuring that this information is protected is a top priority. “We’re working with very intrusive data,” he says. “Our goal is to create practices that allow biological data to be shared safely – without anything traceable back to the individual.”
A new PhD project launching in December will focus specifically on privacy-preserving methods for biometric research. The aim is to avoid the pitfalls of other AI domains, where data was collected first and ethical questions asked later.
“We want to solve the problem at the source,” says Paolo Burello. “Not retroactively.”
Theis Duelund Jensen, Press Officer, phone +45 2555 0447, email thej@itu.dk