Building Sensus: how an EEG device became a generative art instrument
An in-depth look at the technical and conceptual process behind Sensus — the immersive experience shown at Palacio Libertad. From the biometric pipeline to the real-time visual code environment, this is how brainwave data became a living data sculpture.
The biometric pipeline
The system uses a consumer-grade EEG device to capture electrical activity across five frequency bands: delta, theta, alpha, beta, and gamma. A custom Python pipeline running on a local server processes the incoming signal in real time, using a Random Forest classifier trained on session data to identify five emotional states: focus, calm, excitement, frustration, and neutral.
Calibrating the signal
Before each session, participants sat through a short baseline recording — thirty seconds of stillness, then thirty seconds of a simple recall exercise, used to normalize the classifier against that specific nervous system. EEG signal is noisy by nature: muscle tension, blinking, even the residue of a previous session can bleed into the reading. Most of the development time went into filtering that noise out before it ever reached the detection model.
From data to image
The classified state is sent via OSC to a visual code environment built specifically for the piece. Each emotional state triggers a different visual mode — the generative system responds by shifting color palette, particle density, flow fields, and deformation patterns. No two sessions produce the same output.
The installation
The piece ran in the immersive room on the second floor of Palacio Libertad in Buenos Aires, as part of Rituales Tecnologicos II, curated by Marianella Baladan and Natalia Uccello, with NotSociety as technical partner. Visitors sat in front of the projection, wore the EEG device, and watched the piece transform in response to their own inner state. The result was a living data sculpture — consciousness made visible as form.
In the room
What surprised me most wasn't the emotional reaction — it was the curiosity. I'd watch people's faces mid-dialogue with the system, and what came up wasn't just feeling, it was questions: how does it know, is it really reading me, what is it seeing right now. People showed up ready to take part, not just to watch — adults, elderly visitors, kids, all of them finding their way into the same encounter: a few minutes alone with their own state, watching it become something they could see.
What stayed with me
What stayed with me after the run wasn't the code — it was watching people sit still long enough to notice their own state changing the image in front of them. Sensus doesn't ask anyone to perform anything. It just reads what's already there, and gives it back as something they can watch happen. That's the whole piece, really: proof that there's still a human signal underneath all the automation, one visitor at a time.