arXiv:2606.00129v1 Announce Type: new Abstract: Large language models (LLMs) have emerged as powerful representation learners whose internal features increasingly align with human cognition. We study whether modern LLMs can serve as a lens for understanding neural representations in the human brain, focusing on emotional valence in EEG. We first build a one-dimensional valence direction, the V-axis, from modern LLMs using only nine emotion-evocative sentences. We validate it through zero-shot transfer to sentiment benchmarks and cross-model consistency across fourteen LLMs. We then show that t
The accelerating development of advanced LLMs and increasing research into their alignment with human cognition make this a timely study.
This research suggests LLMs can serve as a novel tool for decoding complex human brain activity, specifically emotional valence, which could have significant implications for neuroscience and AI alignment.
The ability to derive a shared emotional valence axis across LLMs and human EEG could fundamentally alter how we study both AI and the human brain, offering new avenues for understanding consciousness and emotion.
- · AI researchers
- · Neuroscience researchers
- · Computational psychiatry
- · Deep learning companies
- · Traditional cognitive neuroscience methods (potentially, long-term)
- · Black-box AI models (as interpretability grows)
Researchers gain a new interpretative tool for understanding emotional processing in the human brain via LLMs.
This methodology could lead to more robust, emotionally intelligent AI systems and advanced brain-computer interfaces.
A deeper understanding of shared 'valence axes' might contribute to theories of consciousness and allow for more ethical and nuanced AI development.
This signal links to a primary source. Continuum Brief monitors and indexes it as part of the live intelligence stream — we do not republish source content.
Read at arXiv cs.LG