I finally finished writing my fun RL project during sabbatical in France. I first presented on this a year ago, but I'm still puzzled by the results, and what they mean for human reinforcement learning! Feedback welcome! osf.io/preprints/psya…

📊Check out our paper at the CCN conference, where we show that focused attention is paradoxically detrimental to representation learning under uncertainty, due to a closed-loop relationship between attention and RL: 2024.ccneuro.org/pdf/128_Paper_…

And here we go the last part of my Grand Tour d'Allemagne (German Grand Tour) starts today in Münich under very good auspices 🤣 Once again hosted by the great CrowdCognition and ophelia.deroy 🇩🇪🇪🇺🇫🇷

All good things come to an end, as my grand tour of Germany and my Humboldt-Stiftung fellowship. It has been amazing to spend time in these amazing laboratories. Forever grateful to these guys for the opportunity and especially Bahador for having the idea in the first place!

I'm looking for a postdoc interested in computational cognitive modeling and deep reinforcement learning! Submit a CV and references here: apply.interfolio.com/150309 Please share with anyone who might be interested!

New #RLC2024 paper Three Dogmas of Reinforcement Learning joint w/ Mark Ho and Anna Harutyunyan! arxiv.org/pdf/2407.10583 We reflect on where our scientific paradigm needs adjustment, and suggest three departures from previous conventions. Curious to hear what folks think! 🧵

Reinforcement learning is a powerful scientific framework for studying how people and machines make decisions 🤔🤖. I use it all the time! But what does it miss out on? In this paper with David Abel and Anna Harutyunyan, we try to identify some gaps and point to some solutions!

Almost three years in the making, the preprint is out! We take another stab at the question: How do humans learn from reinforcement? (Spoiler alert: the answer is _not_ Reinforcement Learning). Huge thanks to the dream team: summerfieldlab @summerfieldlab.bsky.social Nathaniel Daw Kevin Miller

Just saw a fascinating talk by Michal Kosinski at #ecp21berlin on how LLMs may implement a similar intelligence to humans. We (Zak Hussain, Rui Mata) agree so far that it is worthwhile to seriously consider LLM cognition: osf.io/preprints/osf/…

Beyond discrete-choice options Review by AH Hadian Rasanan , NJ Evans, laura fontanesi, Catherine (Cathy) Manning, C Huang-Pollock, Dora Matzke, Andrew Heathcote, Jörg Rieskamp, Maarten Speekenbrink, MJ Frank, Stefano Palminteri, C Lucas, J Busemeyer, R Ratcliff, & JA Rad tinyurl.com/3dsbs3zm

Very happy to see the idea of reward centering (a very robust finding in human reinforcement learning experiments) to finally break through the machine learning literature! sciencedirect.com/science/articl…

the Human Reinforcement Learning Team letting me know they miss me ❤️💔🤣🤣

📖 New #PolicyBrief ! "Motivated beliefs may root in biased (reinforcement) learning" by Maël Lebreton and Stefano Palminteri (Mael Lebreton Stefano Palminteri). Within the framework of the Opening Economics Chair. calameo.com/read/007484431…

📣 Our new theory paper "Group Coordination Catalyzes Individual and Cultural Intelligence" w/ Robert Hawkins & Rick Dale is now live! We introduce 2 powerful feedback loops 🔄 that link individual cognition 🧠to cultural evolution 🏛️➡️🏙️ via group dynamics 🤝direct.mit.edu/opmi/article/d…

🍾🥂 Proud & humbled to be awarded an #ERCStG! W/out the support of my amazing lab hmc-lab.com & colleagues Universität Tübingen Max Planck Institute for Biological Cybernetics ML for Science none of this would be possible. Here’s to the next big steps 🚀 with C4🧨uniting compositionality and cumulative culture🤯

Since I have followed the statistics, they have always confirmed that Italy is the greatest exporter of #ERCStG grantees. Why such a massive brain drain? (rhetorical question) #fugadeicervelli Ministero dell'Università e della Ricerca IIT Il Post Internazionale IlSole24ORE il manifesto Corriere della Sera

Very timely piece by Dirk Wulff and colleagues about the issues related to closed LLM research. Glad to see we converged to the same conclusions based on observed instabilities of OpenAI LLM's performance at different time points in our study nature.com/articles/s4427…

Happy to share our book chapter on sequential sampling models (SSMs) in economic choice, with Kianté Fernandez and Taro Xiaozhi Yang, forthcoming in Neuroeconomics: Core Topics and Current Directions. We review work unraveling the dynamics of economic choice. osf.io/preprints/psya…

Non proprio: ITALIA quinta, ITALIANI secondi. Differenza sottile ma importante. La Repubblica It