Jungsun Yoo, Ph.D.

Are world models necessary to achieve human-level agents, or is there a model-free short-cut?
Our new #ICML2025 paper tackles this question from first principles, and finds a surprising answer, agents _are_ world models… 🧵
arxiv.org/abs/2506.01622

Distal tuft dendrites predict properties of new hippocampal place fields Hippocampal place fields emerge during exploration to support navigational learning and memory. The subcellular mechanisms supporting place field formation remain unclear. O’Hare et al. report that di...

More evidence that distal tuft dendrites provide some form of target/credit signal in hippocampus:

www.cell.com/neuron/abstr...

#neuroscience 🧪

Finally hooded, by my wonderful advisor @aaronbornstein.bsky.social !

Reasoning Goals and Representational Decisions in Computational Cognitive Neuroscience: Lessons From the Drift Diffusion Model The appropriate form of the drift diffusion model depends on how one wishes to reason about their target with the model. If the goal is to parsimoniously explain the speed-accuracy tradeoff, the appr...

super excited that the first paper from my PhD is now out! we develop a "philosophical toolkit" for computational cognitive modeling & use it to conceptually re-analyze a long-standing debate about evidence accumulation models of decision making 🧠📈 onlinelibrary.wiley.com/doi/full/10....

And thanks for the celebration dinner 🥺 It was the best sushi I had!

Thank you! I really tried hard to not tear up when I was mentioning you & the lab during the acknowledgements 😢 I am so fortunate to have been part of such a beautiful community!

Using machine learning to simultaneously quantify multiple cognitive components of episodic memory - Nature Communications Mirjalili and Duarte use EEG and machine learning to simultaneously investigate how perception and attention contribute to episodic memory encoding. The study provides insights into episodic memory’s ...

Why do we remember some events but forget others? In our recent work published at @naturecomms.bsky.social, @audreyduarte.bsky.social and I used machine learning to simultaneously quantify multiple cognitive components of
episodic memory.

www.nature.com/articles/s41...
1/11

Accepting “the bitter lesson” and embracing the brain’s complexity by Dyer and Richards. Great article! www.thetransmitter.org/neuroai/acce...

New preprint from Yining Ding (@liliand.bsky.social)!
"Temporal order memory in naturalistic events is scaffolded by semantic knowledge and hierarchical event structure"
osf.io/preprints/ps...

Excellent resource here, including a slide deck, for teaching students about science funding! Any similar additional resources out there? I would love to add some graphics quantifying the consequences of the current attacks on science.

Large language models can segment narrative events similarly to humans - Behavior Research Methods Humans perceive discrete events such as “restaurant visits” and “train rides” in their continuous experience. One important prerequisite for studying human event perception is the ability of researche...

So happy that our paper on event segmentation in large language models is now out in Behavior Research Methods! tinyurl.com/2j76882b With @mtoneva.bsky.social, @ptoncompmemlab.bsky.social, and Manoj Kumar, we show that LLMs can segment narrative text into meaningful events similarly to humans.

What papers do you like that demonstrate the use of 'orthogonal subspaces' for encoding information in neural populations? #neuroskyence #compneuro #neuroAI

I love the uncompromising defense of absolute stupidity. I sometimes explain it to students using Keats' phrase 'negative capability'. Can you be comfortable with not knowing?

Temporal-Difference Learning Using Distributed Error Signals A computational problem in biological reward-based learning is how credit assignment is performed in the nucleus accumbens (NAc). Much research suggests that NAc dopamine encodes temporal-difference (...

Could we get away with (1) region specific RPEs and (2) no end-to-end training? Is that how the brain works?

My guess is yes to (1), no to (2).

This paper from our "cousins" at @vectorinst.bsky.social is exploring thes issues and has some promising results!

arxiv.org/abs/2411.03604

#NeuroAI 🧪

An inductive bias for slowly changing features in human reinforcement learning Author summary Learning experiments in the laboratory are often assumed to exist in a vacuum, where participants solve a given task independently of how they learn in more natural circumstances. But h...

Our work on inductive biases in reinforcement learning is out: tinyurl.com/yfc74e3u

By the fantastic
@noahedrich.bsky.social

Ppl learn faster from slowly changing features; we argue this reflects an inductive bias. Teamwork w @ericschulz.bsky.social & S HallMcMaster
#neuroskyence #compneuro

Important paper!

www.biorxiv.org/content/bior...

Im not sure the Discussion fully delineates its radical implications.

No more...

* Place cells
* Grid cells, splitter cells, border cells
* Mirror neurons
* Reward neurons
* Conflict cells

(continued)

Space is a sensory-motor sequence in the hippocampus A guide for cognitive/neuro scientists for reinterpreting space and cognitive maps

ICYMI on X/Twitter...if you are into space, cognitive maps, place cells, planning, replay, etc., I predict that this blog+paper will change the way you think about those ideas.

Goes against the conventional Kantian thought that space is apriori ...

blog.dileeplearning.com/p/space-is-a...

BrainFacts

Do you know www.brainfacts.org?

Terrific, accessible (free!) articles there about what’s new in brain research.

Like this timely piece! The neuroscience behind conspiracy theories.

www.brainfacts.org/thinking-sen...

I’m delighted to soon step into an editorial role at this terrific effort!

Free Neuroscience Textbooks and Online Tutorials Free Neuroscience Textbooks and Online Tutorials (Compiled by Carlos Aizenman, Brown University, Dept. of Neuroscience) Maintaining a current list of digital educational resources is tricky. Softwar...

Do you teach neuroscience? Here’s a list I’ve compiled of mostly free online neuroscience textbooks, simulations and datasets you can use for your course! If I’m missing something, let me know so I can add it! docs.google.com/document/d/1...

Recurrent Inhibitory Dynamics in the Entorhinal Cortex Support Pattern Separation The entorhinal cortex (EC) provides the major input to the hippocampus (HPC). Numerous computational models on the EC propose that its grid cells serve as a spatial metric, supporting path integration...

Super-psyched to share this preprint from my student Yicong (Alan) Zheng: "Recurrent Inhibitory Dynamics in the Entorhinal Cortex Support Pattern Separation" www.biorxiv.org/content/10.1... It's a new biologicaly-based computational model of the entorhinal-hippocampal system. Thread (1/?)

@aaronbornstein.bsky.social

OK folks, I've started an episodic memory starter pack. By no means a full list so reply below if you're an episodic memory nerd! go.bsky.app/2mtNK43

Did you know #RLDM2025 is coming to Dublin next June? rldm.org Abstract submission open now ☘️ 🇮🇪

Registration for #COSYNE2025 is now open!

www.cosyne.org/registration

#neuroscience #NeuroAI

A hierarchical active inference model of spatial alternation tasks and the hippocampal-prefrontal circuit - Nature Communications How cognitive maps of physical and task space interact when executing cognitive tasks is not fully understood. This paper models how the hippocampal-prefrontal circuits solves memory-guided spatial al...

It's very nice to see this computational theory of hippocampus-neocortex interaction building on our CSCG model. New paper from Giovanni Pezzulo's lab in NatureComms.

CSCG + hierarchy + active inference! Check it out!

www.nature.com/articles/s41...

I’m presenting my poster soon! Please stop by ;)

I’ll be presenting my poster today! Please stop by :)

One more week to apply for our joint faculty position in Princeton Neuroscience Institute and Computer Science Dept, in neuroAI and intelligent systems, broadly construed. Do you fit? Yes. But feel free to contact me with q's. puwebp.princeton.edu/AcadHire/app...

Human and Machine Learning Lab logo, with three converging boxes around a gear. The gear represents learning mechanisms, and the boxes represent a process of converging on common understanding of human learning and high-performance machine learning.

The Human and Machine Learning (HaML) Lab is recruiting a PhD student to start in Fall 2024. If you’re interested in human learning, machine learning, and neuroimaging, apply to work with me at UWM! See hamllab.org/positions.html for more details about PhD training and the application process.