Computational neuroscience, brain modeling
Ruben Laukkonen, Seb Krier, Chloé Bakalar et al.
For AI alignment researchers, this paper introduces a complementary agenda to safety-focused alignment, aiming to broaden the scope of alignment to include proactive support for human flourishing.
Mete Ismayilzada, Simone A. Luchini, Abdulkadir Gokce et al.
For researchers in cognitive neuroscience and AI alignment, this work provides evidence that LLM representations can be tuned to match human creative thought processes, with implications for understanding creativity and improving model interpretability.
Michal Olak, Tommaso Boccato, Matteo Ferrante
This work addresses speech brain-computer interfaces for individuals with speech impairments, offering incremental improvements in decoding accuracy and robustness.
Konstantin F. Willeke, Polina Turishcheva, Alex Gilbert et al. · stanford
For computational neuroscience, this work provides the first systematic scaling analysis of brain models, revealing that current models are data-limited despite massive recordings.
Stéphane d'Ascoli, Jérémy Rapin, Yohann Benchetrit et al.
For cognitive neuroscientists, this provides a unified model to predict and explore brain function across modalities, replacing fragmented specialized models.
Mu Nan, Muquan Yu, Weijian Mai et al.
This addresses the problem of neural variability across individuals in brain decoding for neuroscience and computer vision applications, representing a critical step toward a generalizable foundation model.
Fabio Cuzzolin
For AI researchers, this formalization clarifies a vague concept and could guide future work, but it is primarily a conceptual contribution without empirical validation.
Sohan Shankar, Yi Pan, Hanqi Jiang et al.
It proposes an integrative agenda for researchers in neuroscience, AI, and hardware to address critical challenges in developing brain-inspired AGI systems, though it is incremental as a survey and position paper.
Hubert Banville, Stéphane d'Ascoli, Simon Dahan et al.
This framework addresses the need for standardized evaluation in neuroimaging AI, enabling fair comparisons and identifying bottlenecks for the research community.
Jiawen Kang, Kun Li, Dongrui Han et al.
This addresses the need for interpretable and clinically valid AD screening systems, representing a novel approach rather than an incremental improvement.
Junfeng Xia, Wenhao Ye, Xuanye Pan et al.
For fMRI analysis, this work provides a more generalizable foundation model that outperforms prior approaches by leveraging diverse brain states and a novel pretraining strategy.
Akshay K. Jagadish, Younes Strittmatter, Nori Jacoby et al.
For cognitive scientists, AutoCog automates the bottleneck of theory generation, enabling cumulative and executable theory-building.
Hua-Dong Xiong, Li Ji-An, Robert C. Wilson et al.
For researchers studying LLM capabilities, this work provides a geometric account of ICL, establishing representational geometry as a mechanistic constraint on ICL effectiveness.
Jucheng Hu, Zhangquan Chen, Yulin Chen et al.
This work addresses the challenge of decoding animal intent for computational ethology, with potential applications in veterinary diagnostics and wildlife conservation.
Haofei Yu, Yining Zhao, Lenore Blum et al.
Provides a principled blueprint for general AI inspired by consciousness, addressing the narrow scope of current AI systems.
Veith Weilnhammer, Kevin YC Hou, Lennart Luettgau et al.
This provides a scalable safety evaluation framework for mental-health chatbot interactions, addressing an urgent need for rigorous auditing of consumer AI chatbots.
Zhongxiang Sun, Haolang Lu, Qiang Ma et al.
Provides a system-level framework for mapping functional organization in LLMs and relating it to human cognition, addressing the need for interpretability and failure analysis in large language models.
Zexuan Chen, Sichao Liu, Runhao Lu et al.
This work provides a robust, semantically-grounded method for decoding visual stimuli from non-invasive EEG, advancing brain-computer interfaces and neuroscience.
Yohann Benchetrit, Marlène Careil, Simon Dahan et al.
For researchers in brain decoding, this work demonstrates a method to improve data efficiency using synthetic data, though the approach is incremental as it relies on an existing large encoding model.
Melika Honarmand, Samin Mahdipour Aghabagher, Martin Schrimpf
For AI alignment and neuroscience, this work provides a causal mechanistic link between model internals and reward valuation deficits, but the approach is incremental as it applies known perturbation methods to a new domain.