
Adrien Doerig
@adriendoerig
Computational neuroscience, machine learning, psychophysics, consciousness. Currently Professor at Freie Universität Berlin.
ID: 986855817516404736
19-04-2018 06:35:36
1,1K Tweet
1,1K Followers
1,1K Following

"Simple distribution shifts break DNN models of the visual cortex, consistent with broader findings that the underlying DNNs are brittle to OOD shifts." Cool twist on brain-score-ology from Spandan Madan, Marge Livingstone, Gabriel Kreiman arxiv.org/abs/2406.16935





Hey ECVP, please come join me for my talk this afternoon! Using fmri & eyetracking, we (Marius Peelen & Floris de Lange) find preparatory activity in visual cortex can reflect attention-guiding anchor-objects rather than search targets (preprint: biorxiv.org/content/10.110…) #ECVP24



Excited about this new paper! Suppose you lost half a field of view, or color vision, would you notice? In a new paper in Trends in Cognitive Sciences, we show that you probably wouldn’t! You can find out more about those striking failures of visual metacognition here: authors.elsevier.com/a/1jsFQ4sIRvTB….


Functional responses in the brain to linguistic inputs are spatially organized -- but why? Neil Rathi (at ICLR) Hannes Mehrer @ICLR2025 et al. show that a simple smoothness loss added to language model training explains a range of topographic phenomena in neuroscience


In 2023, Rowan Sommers & I hosted a reading group on “the intersection of Neuroscience and Machine learning” for the UG/PG students Uni Osnabrück. We selected papers trying to maximize our coverage of the field of NeuroAI. Perhaps this list is useful to the broader community 🧵

🧵Join us for a short tour of how a large-scale MEG dataset collected during active vision made us question it all. New work with Carmen Amme Philip Sulewski Eelke Spaak 🦋 @EelkeSpaak.bsky.social Martin Hebart Peter König biorxiv.org/content/10.110…

Neural networks need a dorsal stream to learn structure in natural scenes! really proud of this work led by Jess Thompson


🚨 New paper at NeurIPS Conference w/ Michael Lepori! Most work on interpreting vision models focuses on concrete visual features (edges, objects). But how do models represent abstract visual relations between objects? We adapt NLP interpretability techniques for ViTs to find out! 🔍



Structuralist approaches are becoming popular in consciousness science. Marlo Paßler and I propose criteria for which kinds of neural structures can be reliably linked to quality spaces derived from reports.


Next week at #SCIoI: What drives human collective foraging? 🥜 📅 Tue 22/4, 14:00 – Ralf Kurvers How do LLMs align with brain representations? 🧠 📅 Thu 24/4, 10:00 – Adrien Doerig (#ThursdayMorningTalk) 🔗 More info: scienceofintelligence.de/events/


🚨 New preprint alert! Our latest review/opinion, led by Drew Linsley, examines how deep neural networks (DNNs) optimized for image categorization align with primate vision, using neural and behavioral benchmarks.