Attention, It's Not a Channel, It's a State - What an RSVP-BCI Reveals About Distraction

EEG ERP P300 BCI RSVP

Article:
Álvaro Fernández-Rodríguez, Velasco-Álvarez, F., Francisco-Javier Vizcaíno-Martín, & Ron-Angevin, R. (2026). Evaluation of video background and stimulus transparency in a visual ERP-based BCI under RSVP. Medical & Biological Engineering & Computing. https://doi.org/10.1007/s11517-025-03498-5

What I understood from the study:

I'm trying to select a command using only my attention, in a visual ERP (P300) type BCI using RSVP — everything appears in the same place on the screen, quickly, one after the other. This is useful when the person has limited eye or muscle movement.
But here's a very "real-world" detail: there's a video playing in the background (with audio), as if I were watching TV and, at the same time, needed to "click" with my brain. They tested a white background vs. video, and also opaque vs. transparent pictograms (alpha 255, 85 and 28).

EEG ERP P300 BCI RSVP

What happens to me when the background becomes a video?
When the background is white, my brain "finds" the target more easily. When the background becomes a video, I feel like my attention has to fight for space — and the system performs worse. The article states directly: the video in the background impairs BCI performance, target detection, and subjective experience, and also increases P300 latency.

And this is shown in very concrete numbers:
With a white background (A255W), the final accuracy was higher than with video (A255V).
To reach close to ~80% accuracy (an "ok" level of control), the white background required fewer sequences than the video — and the information transfer rate (ITR) dropped significantly with the video. In my vocabulary: the video injects noise into my state, not just into my vision.
Transparency: when “freedom” becomes a loss of form.

They tested transparency with video in the background:
A028V (very transparent) worsens the situation — that's when the pictogram loses contrast and the world “swallows” the command.
A085V (intermediate transparency) becomes the point of balance: it doesn't lose performance compared to opaque in the video (A255V), but people report that it's better for “watching the video” without being so irritated by the overlay.

In my head, this seems like a simple law:
if I remove too much contrast, I remove the “body” of the signal.
So the “freedom” (transparency) needs to be such that the command still exists as a form.
What I keep as the “morality of the body”

I don't see this as just “vision.” I see it as a state:
Dynamic background + audio = my system enters a mode of attentional conflict.
I need more repetition to become stable.
The P300 arrives later.

And I liked the design of the method because they were pragmatic: EEG with few channels (including parietal/occipital), 250 Hz, and a typical ERP-BCI pipeline (filter + ASR/EEGLAB, classifier).
Direct connection to “Freedom of Expression in the Completeness of Movement”
On its axis, I would translate it like this:
The video is the moving “biome”.
The pictogram is my intentional gesture.

If the biome screams too much (dynamic background), my gesture becomes incomplete: I try to choose, but I get “stuck” in micro-corrections, repetition, effort — this smells like digital anergy (energy of adjustment without closure).

And the finding of the article becomes a very Brain Bee-like sentence:
When the environment pulls me too much, my command loses form. I don't make mistakes due to lack of “capacity,” I make mistakes due to lack of space.

If I were to design the "next experiment" (right here at BrainLatam):
Adaptive Alpha: the article itself suggests approaches like "dynamically adjusting alpha based on the background." I would test this as "visual breathing": the system opens and closes transparency to give me back space.
Combining pupil size and RMSSD as a "load thermometer": when my body leaves Zone 2 and goes into exertion, I see it before it turns into an error.
Comparing Zone 2 vs. Zone 3 states: even with the same stimulus, performance changes because my state changes.