Trends in NeuroAI - CorText: Chat with your brain data
10 views
Skip to first unread message
Paul Scotti
May 18, 2026, 8:30:17 PMMay 18
to Paul Scotti's Trends in NeuroAI
Hi all (sorry it's been so long!), we are resuming our Trends in NeuroAI reading group (which is now merged with the broader MedARC journal club) by announcing that tomorrow we have a presentation from Victoria Bosch, PhD student in Dr. Tim Kietzmann's lab, presenting her CorText paper.
Victoria Bosch will present her work, CorText: Brain-Language Fusion Enables Interactive Neural Readout and In-Silico Experimentation
Date: Tuesday, May 19
Time: 7:15 AM PT | 10:15 AM ET | 2:15 PM UTC | 3:15 PM London | 7:45 PM India | 10:15 PM China
Meeting link: meet.google.com/bof-ikcz-ygh
Paper link: https://arxiv.org/abs/2509.23941
Add MedARC journal club to your calendar: GCal event | ICal | Public calendar | Email notification
Abstract:
Large language models have been extended by embedding diverse modalities such as images into a shared language space. Yet, neural decoding has remained constrained by static, non-interactive methods. We introduce CorText, a framework that integrates neural activity directly into the latent space of an LLM, enabling open-ended, natural language interaction with brain data. Trained on fMRI responses to natural scenes, CorText generates accurate image captions and can answer detailed questions about image content, while having access to neural data only. CorText achieves zero-shot generalization beyond semantic categories seen during training, and in-silico microstimulation experiments reveal a consistent and graded mapping between brain-state and language output. These results highlight a shift from passive decoding toward generative brain-language interfaces.
Bio:
Victoria Bosch is a PhD student in the Machine Learning group at the Institute of Cognitive Science, University of Osnabrück, supervised by Prof. Tim Kietzmann. She previously completed a Master’s degree in Cognitive Computing at the Donders Institute. Her research lies at the intersection of cognitive neuroscience, philosophy and machine learning, with a current focus on neural decoding and developing neural network architectures for modelling vision.
Hope to see you there!
Victoria Bosch will present her work, CorText: Brain-Language Fusion Enables Interactive Neural Readout and In-Silico Experimentation
Date: Tuesday, May 19
Time: 7:15 AM PT | 10:15 AM ET | 2:15 PM UTC | 3:15 PM London | 7:45 PM India | 10:15 PM China
Meeting link: meet.google.com/bof-ikcz-ygh
Paper link: https://arxiv.org/abs/2509.23941
Add MedARC journal club to your calendar: GCal event | ICal | Public calendar | Email notification
Abstract:
Large language models have been extended by embedding diverse modalities such as images into a shared language space. Yet, neural decoding has remained constrained by static, non-interactive methods. We introduce CorText, a framework that integrates neural activity directly into the latent space of an LLM, enabling open-ended, natural language interaction with brain data. Trained on fMRI responses to natural scenes, CorText generates accurate image captions and can answer detailed questions about image content, while having access to neural data only. CorText achieves zero-shot generalization beyond semantic categories seen during training, and in-silico microstimulation experiments reveal a consistent and graded mapping between brain-state and language output. These results highlight a shift from passive decoding toward generative brain-language interfaces.
Bio:
Victoria Bosch is a PhD student in the Machine Learning group at the Institute of Cognitive Science, University of Osnabrück, supervised by Prof. Tim Kietzmann. She previously completed a Master’s degree in Cognitive Computing at the Donders Institute. Her research lies at the intersection of cognitive neuroscience, philosophy and machine learning, with a current focus on neural decoding and developing neural network architectures for modelling vision.
Hope to see you there!
-Paul
Reply all
Reply to author
Forward
0 new messages