Self-Calibrating BCIs: Ranking and Recovery of Mental Targets Without Labels
Jonathan Grizou, Carlos de la Torre-Ortiz *, Tuukka Ruotsalo · 2025
Read Self-Calibrating BCIs: Ranking and Recovery of Mental Targets Without Labels by Jonathan Grizou, Carlos de la Torre-Ortiz *, Tuukka Ruotsalo (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2506-11151.txt.
Description
Jonathan Grizou, Carlos de la Torre-Ortiz, and Tuukka Ruotsalo read a face out of a person's mind without ever being told what anyone was thinking of. Brain-computer interfaces that recover mental targets usually train on labeled sessions, an expensive calibration ritual. Their framework, CURSOR, self-calibrates instead. Paired brain responses and viewed faces from interactive sessions are all it gets.
The system learns to score image similarity in a way that tracks human perceptual judgment, with no labels anywhere. Those scores rank new stimuli against the unknown target a participant holds in mind. Generation then produces fresh stimuli matching the target. A fifty-three person user study found the generated faces indistinguishable from the mental target. Self-calibration removes the labeled bottleneck that has kept such interfaces in the lab.
Catalog record
| CREATOR | Jonathan Grizou, Carlos de la Torre-Ortiz *, Tuukka Ruotsalo |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2506-11151